# Who Is This?: The State of Advertising Identity in 2026

*When a seller says it found a match, did it find the same person, a device or a home? This paper asks how buyers can tell, what the seller counted, and which choices follow the ID when it changes hands.*

Evgeny Popov · No Fluff Advisory · Research paper · November 2026 · Research cutoff 11 November 2026

## A match to what?


*Deep Sync says it linked 73.0% of the transactions in one US grocer’s monthly file to an inferred buyer [12]. It says what went in, what it found and how it made the link [12]. The case gives no separate test of who paid for each purchase [12]. A buyer needs both the rate and a way to check the links.*

> **In short.**
> Ask what the firm matched. Then ask who could use the match and who checked it. The Identity Quotient scores the public proof for a named US buying task [11, 13]. It gives a band only when the row and its set of peers meet the rules [10, 218, 11].

### Start with the thing that changes hands

Do you get a code in a bid, a file of linked rows, a set of ad slots or a key for a home? Those are the four kinds of purchase in this study [11, 13]. Start where the output leaves the firm. Who takes it? What can they do with it?

In LiveRamp’s ATS guide, a site can pass a sealed ID to a supply platform, which opens it and sends a code made for that bidder [133]. Yahoo’s US terms permit a site to send a sealed ConnectID on bids to Yahoo and limit who else may get it [160]. Before you pay for reach, trace the code to the seat you plan to use.

### What the tests found

The grocer’s case breaks the claim that no scored firm had given the base, the matched group and the method for a US match rate [14, 13, 12]. The search left the test of link accuracy open [14, 13, 40, 41, 43, 45, 62]. It could not settle whether an independent test had checked a named graph’s US links to people or homes against a known truth set [14, 13, 40, 41, 43, 45, 62]. Ask to see both the count and the truth test.

What happens when someone opts out? Among the IDs known to be sold in the US, too few showed how a choice reached the next party and changed what it could do in at least one market where the ID was offered [14, 13, 149, 150, 186, 175, 128, 182, 219]. Even if each open case in that group passed, the share would still fall short of half [14, 13, 149, 150, 186, 175, 128, 182, 219]. If all the IDs whose place in the group is unclear were added, the range would run across half [14, 13, 149, 150, 186, 175, 128, 182, 219].

![Adding uncertain products changes the consent result](figures/fig-choice-path.svg)

*Figure fig-choice-path. **Adding uncertain products changes the consent result** Each range runs from documented paths to the most that could qualify if every unknown case succeeded. The vertical line marks half. Units: products and share of frame. Scope: US-available IDs through November 11, 2026; a qualifying path in any applicable jurisdiction. Sources [13], [14]. Caveat: The intervals are logical bounds, not confidence intervals. A documented path does not establish compliance in live traffic. Data: data/figure-data.json#choice_bounds*

Only Blockgraph met the US buying test for a CTV household graph [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. Its guides tell buyers how far back they can look up a past IP address, but give no age limit for the live link [114] [115]. H5 was contradicted for that one service (0 of 1) [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. Eligibility remained open for 8 other services, so the study could not settle what most US sellers disclose [14, 13, 109, 114, 115, 227, 228, 229, 230, 231].

The test of agent buying found code that can use an ID in a check of who is in an audience [92, 93, 94] [95, 93]. That breaks the claim that the work had gone no further than the fields in a spec [14, 13, 92, 95, 108, 101, 232]. The record did not show that this check drove a live sale [14, 13, 92, 95, 108, 101, 232].

The study could not settle whether most of the web ID guides moved back to cookie sync, whether an order led to a verified change in a named identity service, or how often a new offer kept links from an older graph [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226] [14, 13, 65, 68, 70, 79] [14, 13, 160, 214, 215, 216, 217].

### How to read the score

Of 27 US products that met the buying test, 24 had enough proof to enter the IQ [10, 218, 11]. The register holds 57 records in all [10, 218, 11]. The score uses records found by 8:37 a.m. EST on 2026-11-11 [364]. Read a band within the group for the task you plan to buy [11, 13]. A buyer who needs to know how often a link is right should ask for a test of that link.

### How the paper was made

OpenAI models ran all seven stages, from the first rules to the final source check [364]. Other model runs checked the sources and scores within the same process [364]. No independent human peer review is documented [364].

### Interests

I have worked at Samba TV since May 2025 [232, 233]. Samba sells TV measurement and ACR data and has a household graph [232, 233]. I left it out of scoring [232, 233]. I have worked at Lotame, Verve Group and Hearts & Science [232, 233]. The register marks my ties to Lotame Panorama ID and Verve Moments.AI [232, 233]. I co-lead the Signals & Measurement working group at AgenticAdvertising.org [232, 233]. My views on match rates and graph reuse helped set the questions [232, 233]. The answers use the rules set before the search [11, 13].



## 1. Defining what the buyer buys


*Start with what comes back. Does a brand get a link it can add to its own list? Does a publisher get a code it can pass to a bidder? Does the buyer get the right to serve an ad to a group? Put the answer in the brief. Let it shape the test and the price a buyer can defend. It will also shape what the buyer can claim when the work is done.*

> **In short.**
> Group the services by the job they do. One job passes an ID from firm to firm. One links a brand’s own list to a graph. One sells a group as a way to replace an ID-based buy. One links TV sets or ad views to homes. Who gets the result? What does each row stand for? What can the firm that gets it do with it?

### Read the return file first

Take a brand that wants to sign up new members for its club. It wants to skip the people who have joined. It has each member’s name, email and date of joining. Start at the far end of the job. Must the brand know which rows matched so it can fix the rest? Or would it be enough for the seat to skip a group while the ads run? Write the brief both ways. The first asks for a file and a way to work on it. The second asks for an action at a named destination. A firm could meet one brief and leave the other job undone.

Draw a line from a row to the thing the team will act on. Give each end a plain name. If the firm links an email to a device, who gets the key for that device? If it links the member to a home, should the seat skip an ad for each person there? If it builds a group in a seat, can the brand take it to a new seat? Set the test for the job the brand needs. Then choose which firms to hear from.

A useful clause would say, “For each input customer key, return an accepted link at person level or a reason the row was left out.” Another would say, “Make the group available to exclude in the named seat for the agreed dates.” Ask the bidder to mark the clause it can deliver. For a file, show the fields and one filled example. For a segment, show where the trader can see its name, select its use and check that the campaign applied it. Keep those two forms of proof with their respective clauses.

### Name the thing at each end of the link

Call a row from a source a *record*. Call a code used to refer to a thing an *identifier*, or ID. That thing might be a browser, a phone, a home, a person or an account. A *link* claims that two rows refer to the same thing, or that they have some stated tie. Call the set of links and the things they join a *graph*. These terms let the buyer name the claim it needs the seller to make.

Take a shared inbox. In this example, two adults have separate store accounts and use the same email address. A test for the same email would find the same text on both rows. Who placed the order? That would take more proof. Now change the task. If the retailer wants to send just one email to the inbox, the common address may be all it needs. Ask the team to state which decision would change if the two adults were merged. The cost of an error follows from that decision.

Call the system in which a key has its meaning a *namespace*. Suppose a store and a streaming service both assign the string “A17” to an account. In that case, keep the keys apart until you know what joins them. Ask for a source label for each end of a link. How will the team know when a new key has replaced an old one? Try this on the sample. Strip the source from a person key and put it next to the store’s own club ID. Could a new member of the team tell which was which?

### A common ID that other firms can use

For cohort A, look for an ad ID that can pass from firm to firm. Who can issue it, send it or make sense of it? Follow one ID across the boundary. LiveRamp says the sell-side platform can open its ATS envelope and pass a RampID for a specific buy-side platform in a bid request [133]. Who gets the envelope? Who can open it? Which buyer will know what the code at the end means? Put a firm’s name next to each handoff.

UID2’s guide names The Trade Desk as core administrator and a public operator [142]. Public or private operators make the raw IDs and encrypt them for use in tokens [142]. Its guide sets out who can do what [143, 144]. A publisher can ask for a token, a mapper can submit an email, phone number or its hash, and a bidder can open a token [143, 144]. Which role is the buyer taking? Want keys to join the agency’s own list? Read the terms for the mapper. Want to send a token that a bidder can use? Read the terms of the route at both ends.

The Trade Desk describes a CRM route that lets a client target ads in its own seat without receiving the IDs or the map from client rows to IDs [188, 203]. UID2 says a firm under contract can use a public operator at no charge [188, 203]. A private operator has to host and run the service itself [188, 203]. What right did the brand gain? What work must it do to make use of that right? What comes back? On the budget, give the cost of that work its own line. List the ad spend and any fee for the service on their own lines too.

A code that comes back can mean more than one thing. LiveRamp says its API can issue a maintained RampID from a graph match or a derived RampID when it finds no match [234, 181]. Its ATS security guide also describes a hash producing a derived RampID when that input has no representation in the offline graph [134]. How can a buyer tell the two cases apart? A test for a filled cell might pass in both. If the brand needs a link that was in the graph when the job began, write a test for that.

### A service that resolves the buyer’s records

For cohort B, start with the rows a brand wants to link. Which graph will do the work? What kind of link will the brand get? How will it get it? Experian offers its US Digital Graph as a full dataset or as links appended to the buyer’s own records [117]. If the team buys the full set, which fields can it keep? How will it spot the latest set and replace its old links? If it buys an append, how does it join the new fields to its rows? How does a failed row look? What if the team sends the same row twice?

Experian describes clustering digital IDs into individual or household links using signals that include timing, IP addresses and browser or device IDs [36, 117]. Its Offline Graph is a separate offer for names, addresses and emails, and a joint offer pairs the Digital Graph with Marketing Attributes [120, 121]. Put the name of the product on the order. If the brand wants a street address or a new fact about a shopper, which part of the order gives it that field? The list of things a firm sells is a good place to start. Get the name of the one that gives the brand the field it wants.

LiveRamp describes an API that takes personal information and returns AbiliTec Links, and a cloud route that returns Known IDs [235]. The AbiliTec API accepts single or batch calls and returns identifiers and match details for matched records [243]. Microsoft documents a US-only preview that requires active LiveRamp and AbiliTec subscriptions and uses mapped profile fields to retrieve person-level IDs [212]. Keep the preview label on the Microsoft feature. For an API buy, ask who can use it and what link comes back. If a firm adds to a list held in a second system, where will the new field show up? Who can read it there?

Two apps can sit side by side and do quite distinct jobs. Zeta’s US-only app guide describes Customer Growth Intelligence taking first-party data through a Snowflake share and exposing enriched views [244]. The same guide describes Customer Matcher pairing records in two client tables and scoring those pairs [244]. For the first job, which new fields come back? How may the brand use them? For the next, what does a pair score mean? What if one row has two strong links? Keep the sample next to the name of the app that made it.

One more distinction belongs on the order. Define *observation* as a piece of input, such as a time-stamped use of a device. Reserve *assignment* for the decision to tie that device to a person or home. Imagine a customer first signing in at a shop and then signing in again from home. A buyer could ask for the details of both observations, a single person link, or a new address appended to the customer file. Specify which came from the buyer and which the seller added. If the address came from the buyer, counting its return as new coverage would give the append service credit for work it never had to do.

Set a rule for ambiguity as well. Suppose a phone number appears on both a business account and a personal account. The buyer might prefer two possible links, each with its own flag, over a single link chosen by the service. Or it might want the service to leave the row open for review. Put that choice next to the field definition. A person identifier, a candidate identifier and a reason for declining the match should each be recognizable in the sample without a call to the sales team.

### An audience the buyer can trade

For cohort C, ask what group the brand can buy in place of an ID-based buy. Take a brand that wants to reach new runners on sports sites. It may have no need to link each one to a store list. What would its trader choose in the seat? Who would set the terms for ad space to count? What would tell the buyer that the deal ran? A list of pages tagged for running would meet a more limited brief. For this group, look for a seller that makes the replacement offer and spells out what the brand can buy.

Permutive says its publisher groups can be used in deals [205, 206, 207, 208]. Google says TelevisaUnivision sent Permutive’s groups to Ad Manager [205, 206, 207, 208]. On its US ad space, the publisher saw a mean of 1.24 audience categories per impression [205, 206, 207, 208]. OpenX separately announced a US route for buying Permutive audiences without third-party cookies or device IDs [205, 206, 207, 208]. Keep the partner’s name with its case. Use the first to ask how the seller put a group up for sale. Use the next to ask which seat can buy the OpenX route.

Audigent describes SmartPMP as a curated media route using TransUnion audiences, while Hadron ID has its own documented identity module [236, 237, 238, 239]. What right does the deal give the brand? It might want to buy ad space. It might need the ID too. If that calls for a new set of terms, put both sets in the brief. Ask to see the item the trader will pick. Which terms can change the group while ads run? Where will the brand see when it changed?

Name the place too. Ogury described an airline campaign that targeted travelers arriving in Japan and included visitors from the United States [248]. Want to show ads in the US? Ask for a case or an offer where that is where the ads ran or will run. What does “US” mean in a brief? It could name the source, the buyer, the people, the ad space or the place the ads ran. In this case, check the place where the trip ends.

### A TV purchase with the household as its unit

For cohort D, look for a service that links TV sets or ad views to homes for a TV buy or a test of that buy. The home can be the right place to end the link. Which homes went to a site once an ad had run on TV? A join at the home could serve that brief. Who watched the ad and then bought the goods? Get proof of the extra link to make that claim. Give the two claims their own lines.

Spectrum Reach describes a working arrangement in which it provides household TV exposure, Innovid provides household website visits and Blockgraph matches them [38, 167]. Spectrum’s account describes counting visits within seven days of a TV exposure [167, 38]. Put that span next to the visit measure. Blockgraph separately describes historical IP lookups that can use a lookback of 30, 60 or 90 days when the provider maintains household signal history [114]. Give each clock a job. How far back can the search go? How old can a link be and still count? How much time can pass from the ad to a site visit?

TransUnion described TV audience activation through its marketplace and the use of its graph to build segments for campaigns on Comcast [245, 246, 247]. If the team wants to buy a TV graph, which item on the order will it get? If it picks a group, ask who owns the group and where it can run. If it gets a link from a TV event to a home key, who can check it or join it? How will a new link replace one that has gone stale? The teams may have asked for things that come back in quite different forms.

### Decide what the buyer can keep

Settle who holds the result when a campaign ends. Imagine a brand that parts ways with its agency after a TV flight. If the brand holds a dated file of accepted household links, the handover could include that file and the rules for using it. If the result lives in an agency seat, handover might instead require that a segment be shared or rebuilt. If the only output is a report, specify which cuts of the results the brand needs before access ends. Ask the parties to agree on this point before the flight starts.

Apply the same test to tools used alongside identity services. AWS describes a Clean Rooms query as access to configured tables for analysis within a collaboration [240]. OneTrust describes its consent management platform as capturing and signaling purpose-based consent [242]. Microsoft documents a Curate workflow in which a curator adds its assets to seller inventory [241]. If the brand needs a link to a person or a home, who will supply it? Put the query result, the choice signal and the media deal on their own lines.

Return to the membership campaign. Let the membership team own the rule for entrance and exit. Ask the graph firm what each link means. Have the trader name the seat that used the group. Did it use the group to reach those members or to skip them? Agree on what to do with a shared inbox, an expired member and a row the provider cannot place. If a member changes their address, identify the list or link that should change and who is able to check it. The signed brief can then say what the team bought in terms that survive the first change of data, staff or seat.



## 2. Standards and specs


*Read the source note next to the code. It can help a buyer ask who put it there and how the match was made. OpenRTB 2.6 has fields for who inserted an ID, whose technology supplied the match method, and which method was used [252, 1]. Want proof about one ad? Ask for a trace from the path it took. Keep the build, the settings and the fields the bidder saw.*

> **In short.**
> Read the spec to learn what the field means. Read the guide to learn how the code uses it. Keep a copy of the settings in use. From Prebid.js 11.14, ID modules skip retrieval if a configured TCF or GPP consent-platform lookup fails [239, 255]. If time runs out, they run with the consent data at hand [239, 255]. Write both cases into a release test.

### Read a bid from the outside in

Take a site that wants to sell ads with an ID. Which spec does the seller use? Ask that before you look at the ID. Which release does it use? In which field will it put the ID? Where does it put the source? Which fields can the bidder see? Start with a copy made for a test. If the test asks where the ID goes, take the ID itself out. Keep the parts around it in place. Have both firms say what each part means.

OpenRTB has an extended identifier object, or EID, for a group of IDs from one source [1]. Its `uids` field holds a list of UID objects from that source [1]. The `source` field names the canonical domain of the ID [1]. In UID2’s own example, the encrypted token appears in an EID entry with `uidapi.com` as the source [143, 144]. Prebid’s guide names `criteo.com` as the EID source for the Criteo ID for Exchanges module [250]. Check each name against the guide for that ID. What is the source field meant to name in that guide?

What can a count of these fields tell you? Make a test with two source groups. Give the first one two UIDs. Did the right sources show up? Count the groups. Did the adapter build the right list for its source? Count the UIDs in that group. How many people did the buy reach? First show what links the IDs to people and how the team checked those links. Store a copy of the test in the shape it had when it ran. Each count should still trace back to the right part of the test.

### Who inserted the value and who matched it?

IAB Tech Lab added provenance fields to OpenRTB 2.6 to describe who inserted or bridged an ID and the match method [252, 1]. In the EID object, `inserter` names the party that caused the entry to be added [1]. The `matcher` field names the technology that supplied the match method, and `mm` holds a code for that method [1]. Ask one thing of each field. Who owns the namespace? Who put the ID in the request? Whose tool did the work? How did it match?

An empty field deserves a close read. The OpenRTB text says to assume that a blank `matcher` equals the source, and it permits omission for certain match-method values [1]. Keep what came in from the sender. Add a field for what the spec says it means. Which source did the sender name? Read the first field. What should the buyer use when a name is blank? Read the next field. A team that puts both in one place might take a name it filled in by default as one that came from the sender.

Take an exchange that adds a key from a namespace a different firm owns. Ask the exchange why it put the key there. Ask the firm that built the match how it made the link. If the two firms differ on what the method means, ask for its terms and a test of the link it claims. The match code gives the team a place to start. To find out how many links are wrong, get a set of cases with known answers. How were those cases picked? How did each link fare?

### Find the part that issues a usable ID

Use the name of the module to find its guide. Prebid says SharedID stores an ID in the site’s first-party domain and lets adapters use it [258, 239]. Prebid says Parrable’s ID works across browsers and webviews on one device [249]. Its guide says the Parrable cookie holds the ID and the user’s opt-out state [249]. Does the claim cover one device, site or source? Ask the seller. Set up a test based on its answer. Which things should stay the same when a user comes back?

Read the guides at both ends of the route. UID2’s permission matrix says a publisher can request a token and a bidder can decrypt one [143, 144]. LiveRamp’s ATS route gives the supply-side platform an envelope to decrypt before it passes a bidder-specific RampID [133]. At each step, give the code a name. Which firm has the right to open it or change it to a key the next firm can use? If the handoff is meant to change the key, a test for the same text at both ends could flag a fault where there was none. Check that the next firm got the key meant for its role.

Say the team gets two fault reports. One says, “The publisher created the ID, but the bidder could not use it.” The other says, “The publisher created no ID for this request.” Keep a log at each end. For the first report, trace the handoff. Which kind of ID did the bidder expect? For the next report, start with the sender’s settings and inputs. Make up a test key that can join the logs. That way the ID itself can stay out of the report. Give both firms the time and the test key so each can check its step.

### Turn the default into a test case

Prebid documents a publisher option to require TCF Purpose 4 consent before passing user IDs and EIDs, and the option defaults to false [2]. Ask how the site set that switch. Run a case with it off. Turn it on and run one with the choice due to be there. Leave it on and run one where that choice is missing. Give each case a row. Keep the exact build and settings for each row so the team can go back to a case that failed.

Prebid.js distinguishes a failed consent lookup from a timeout when a TCF or GPP consent management module is configured [239, 255]. From version 11.14, its ID submodules skip retrieval and add no IDs to the bid request when the lookup fails [239, 255]. On a timeout they run with the consent data available then [239, 255]. For a test of the first case, make the test system report an error. For the next case, let the clock run out. Save what the code had when time ran out. Did it still have a choice from a prior check? Test with that choice in place and again with no old choice to read.

Keep the release note with the code it covers. ID5’s Prebid Server Java guide describes forwarding available privacy signals to the ID5 API during ID generation [256]. A browser-side Prebid.js result and a Java server module description belong to different test paths [239, 255] [256]. If the site runs both, give each path a known input. Which code read the choice? Which code asked for the ID? Which code put the ID in the bid request? If one team runs the site code and one runs the server, each can show the step it took.

For the acceptance test, decide what the system should count when it skips a lookup. A publisher could report a request with no ID as one failed match. Under the failure branch just described, the module never asked for an ID in the first place [239, 255]. Give that case a place in the report. How many checks could start? How many sent back an ID? How many IDs did the wrapper put in a request? Ask for counts at each step in the same test. If the first count drops, the team should have a clear place to look. If just the last count drops, send the case to the team that adds the ID to the bid.

### Give each choice signal its own meaning

IAB Tech Lab describes the Global Privacy Protocol, or GPP, as a way to carry privacy and choice signals from sites and apps to ad technology providers across supported jurisdictions [253]. Which signal was on the path? Where did it come from? Where did the sender put it? Who could read it? Make a test where one part is blank. Run a case where the string cannot be read and one where a user has made a clear choice. Ask both firms what they expect in each case before they see the result.

IAB Europe says newly created TC strings that lack the disclosed-vendor segment are invalid from March 1, 2026 [254]. Under the transition it describes, earlier strings can remain valid until the user renews or changes the choice [254]. To test the change, give the date the string was made. Has the user made a new choice since then? Check what this part of the string says. Want to know if the choice covers a set use? Ask for a check on that use too. Give the check on which firms were named its own name and result.

The W3C document dated September 24, 2026 is a Working Draft of Global Privacy Control, or GPC [3]. It describes a signal exposed through HTTP and the browser’s document object model that requests sites and services refrain from selling or sharing personal information [3]. Keep the word “draft” with the date. If a firm says it supports GPC, ask what its code reads. What does the site do once it sees the signal? If the firm also supports GPP, ask how it deals with both inputs for this use. Check each answer against the spec the firm named.

### Trace a change of state

A choice made after an ID was issued calls for a new test. UID2 says a user can opt out for a supplied email address or phone number, publishers learn the result on token generation or refresh, and an operator sends DSPs a separate webhook [149, 186]. Its DSP guide instructs the bidder to stop using an opted-out UID2 for bidding and allows another ID or a decision to skip the bid [186, 149]. Set the test for the ID it covers. If an ad comes back, which ID did the bidder use for that event? Did it need one at all? The ad alone would leave that open.

Look for a change the site team can see. LiveRamp says its ATS JavaScript removes the stored RampID envelope when consent is rejected under the US GPP National or fallback CCPA check [257]. Yahoo documents a domain-specific opt-out flag in local storage that stops the ConnectID JavaScript module generating new IDs on that domain [201]. In a site test, save what the site held just before the choice. What did the module do next? To find out about a copy held by a firm at the far end, ask that firm for a receipt. Use the terms that apply to its copy.

Set the clock for the test. Say the buyer refreshed its copy just before the user changed a choice. The site checked just after the change. Those are two points in time. Mark when the choice came in. When did the issuer next check it? When did it send a note? When did the firm at the far end act on it? Mark each step the teams can show and leave a space where the proof ends. The site team can show what its own code did. The buyer can then test the part that took place in its seat.

### Keep names and versions with the evidence

A change of name can leave a reader unsure which document a buyer used. IAB Tech Lab says Seller Defined Audiences was renamed Curated Audiences in December 2024 [251]. If a brief cites the spec, keep both names with its date. Who gets to set the group? Which deal can the buyer pick? If the seller cites an ID too, ask how to get it and on what terms. A buyer can look up the old name and find the new one. If a firm says its own service has stayed the same through the change, what proof does it have for that claim?

A version change can alter the data a buyer needs to keep. In July 2025 UID2 announced v3 mapping for advertisers and data providers, replacing v2 salt-bucket monitoring with per-ID refresh timestamps and making the prior raw ID available for 90 days after rotation [145, 146]. To test the move, take one map built with the old code and one built with the new. Read the guide for each. How long can the team use an old map? Which field tells it when to check next? What if the team misses a check?

UID2 also distinguishes the raw ID from a generated advertising token and says each generated token differs even when participants mapping the same email at the same time receive the same raw ID [144, 146]. Put the name of the thing next to how long it should last. To test the guide, have the team make the map again and ask for two tokens. Say up front which should match and which should differ. What would make the team take a closer look? Agree on that first. A change the guide predicts could look like a failed test if no one wrote down what to expect.

### Hand over a case that someone else can run

Build a small test pack. Start with the dated spec and the fields due to show up on the path. Add the sender’s build, the module that was on and its settings. Who was meant to get the request? Add a test with the group in place. Add one where it should be missing. If the team takes out a code to protect the person in the test, mark that spot. Leave enough in place for the next reader to tell a code that was cut from a field the sender left out.

Send the receiving team a matching copy of the test case. Ask it to return the source it read, the version of its decoder and the action it took when a field was missing. Include one deliberate mismatch between the expected source label and the example value. Ask the engineer whether the buyer would accept the entry, ignore it or set it aside for review. Agree on what this build should do. Run the same test when the build next changes. Give the bidder a place to say it could read the field yet lacked the rights or keys to use it.

Mark where each sample came from. Was it in the guide? Did a test make it? Did someone see it in the buy? Have the person who gave it to the team mark the box and date it. A sales deck could show a request straight from the guide. A trace with the settings could tie one to the site build that made it. Which question does the buyer want to ask of each?

Then state the question in a form the packet can answer. Did this build put this source group in a request for the chosen recipient? Did a timeout produce the documented branch under the settings supplied? After a choice changed, did the local site stop producing a new value? Each question has an actor, an input and an observable result. Give a second reviewer the packet and ask them to identify the step that would fail if just one setting changed. Keep the reviewer’s answer with the original result so the next release can be tested on the same terms.



## 3. Match rates and accuracy


*Deep Sync describes a US grocer’s monthly sales file [12]. Deep Sync says it linked 73.0% of the file in three steps [12]. The last step used the home and the area a store serves to infer who bought the goods [12]. You can see how the share grew [12]. The case gives no check against known buyers for that last step [12]. Start with what went in, what came back and how the seller made the link. Then ask who checked that it was right.*

> **In short.**
> TransUnion says it linked 99.5% of a US insurer’s records to a person in its offline graph [17, 18, 19]. Its report lists fields for a second, smaller figure [17, 18, 19]. Experian puts a 70% rate on a link from MiQ's IP addresses to cookieless IDs [20, 21]. Its case gives no US-only rate or full rule for accepting a link [20, 21]. A seller can show how much of a file it linked. To learn how often it got the right person, give the links a test with known answers.

### Start with the file

A brand gives you a list. What share can you reach? First choose what you want to count. Make a row for the whole file. Then count the rows that have the right fields. Count each distinct key. Count the keys the brand has asked you to send. Count the ones that linked and the ones the seat can use. Keep each count and its unit on its own row. A row could give you one device or ten. Ask which line the seller will divide by which. Settle that before the job runs.

Consider LiveRamp's standard rate for activation at a destination [15]. Start with unique records that can be identified and are in scope for the send [15]. Count those that link to at least one online device at that destination [15]. The base leaves out records the client is not sending, and Facebook has a separate rule [15]. What if your brief asks about the share of the whole file that the seat can use? Keep your own base as well. Give both rates to the planner. Use the first to see how the send went. The other would tell her what share of the list she can plan to reach.

LiveRamp gives a simple example with one person and four devices [16]. It finds one of those devices, so its rate for the devices is 25% [16]. The person's email can match at the destination, giving a rate of 100% for that one person [16]. These are made-up figures to show how the rule works [16]. Now put those units in your own plan. If you want to show one ad per person, which devices share the cap? If you want no ads to that person, what happens on the three devices that did not match? Write the answer in the same units as the goal.

### Where the grocer gained the most

The first pass in Deep Sync's US grocery case used the loyalty signal to link 22.8% of the month's till records [12]. A pass that added links from cards took the share to 30.6% [12]. In its last pass the seller used the store's catchment and a household to infer further links [12]. It put the share at 73.0% by the end of that pass [12]. All three shares use the monthly till file as their base [12]. Do not add them. A sale found on the first pass is still in the count at the end [12].

The seller's headline total is about 32 million matches [12]. The rounded counts of new matches on each pass add up to about 31 million [12]. Its summary gives a range of roughly 64% to 73% of monthly sales volume and does not name the month [12]. Keep 73.0% as the stated end of the three-pass run [12]. If you need exact counts for a bill, ask the seller for them. Get new links on each pass, plus all rows left when the last pass ends. Check that they add back to the full file. Keep rows sent twice apart from sales made twice.

Think about what the final pass may be worth to a store. It might help choose homes to send a broad offer. Would you use it to claim that a named person bought a box of tea? The case gives no test against known buyers for that last pass [12]. Ask for a flag that shows which pass first found each sale. Plan one buy with the loyalty rows alone. Add the card rows in a second plan. Add the last group in a third. Find out what each new group does to your cost per home. You could agree to buy one group and run a test on the next.

A second US case from Deep Sync deals with an outdoor retailer's customer file [27, 28]. It says its Profile Hygiene & Enrichment service resolved 83.5% of that file, with all 19 rules on [27, 28]. The case gives about 885,000 records and does not define the 19 rules one by one [27, 28]. A count of rules could tell the team that it ran the right job. How many rules gave one clear match? How did the team break ties? Which rule could turn one buyer into more than one home? Ask for the first rule that worked for each row. The grocery case cannot fill those gaps for this file [27, 28] [12].

### Read the next line in the case

TransUnion's report for a US insurer says 99.5% of its records found a person in the firm's offline graph [17, 18, 19]. It gives another figure of 97.4% for matches on first name, last name and at least one more item of personal data [17, 18, 19]. The source gives no full match rule for the 99.5%, count of rows sent or dates for the test [17, 18, 19]. Keep both lines in the brief. Ask which rule applied to each part of the 99.5%. Ask which service a client could buy to run that same match. If a client needs a home, find out whether the person match can yield a home as an agreed output.

Experian says it linked 70% of the IP addresses MiQ gave it to cookieless IDs [20, 21]. The case gives no US-only group, row counts or rule for deciding an IP had a link [20, 21]. Experian says MiQ licensed hashed emails, UID2s and partner cookies for its own Identity Spine [21, 118]. Ask when an IP counts as one input. Once in the whole file? Once on each day? Each time it shows up in a bid? Ask whether one IP can count as a success if it points to a long list of IDs. To plan a US buy, get the US cut first.

Cadent put its average rate above 90% for segments using its Aperture Viewer Graph in March 2022 [24]. Its release says little about which rows counted in each group or how it weighted the groups [24]. Try a small example. Take one list of ten rows and one of a thousand. All ten rows match on the first list. Half the rows match on the next. Give the lists equal weight and the mean rate is 75%. Pool the rows and 510 of 1,010 match. Which figure would you use to buy each list? Ask for both views and the rate on the group that matters most to the brief.

A claim about growth needs a base too. Kantar said its rates with TV partners grew more than tenfold after it took up Blockgraph's IDoS [112, 113]. It gave no rate for the start or end, or count of the rows in scope [112, 113]. A TransUnion-funded Forrester study used interviews to model a household rate rising from 60% to 80% for a composite firm [26]. That gain of 20 points is a model input [26]. Ask for a starting count, an end count and the same set of homes on each side before you put a cash value on a gain.

### Ask who did the work

Deep Sync compared a US mass retailer's file before and after adding new data [31]. In the row for UID2, it reports 73.2 million matched records before the work and 94.4 million after [31]. The case labels those two results 51% and 66% [31]. Against the stated 141.7 million input records the figures would be about 51.7% and 66.6% [31]. The source does not explain how it rounded or which exact base the seat used [31]. Credit the firm that changed the file. Ask for a new test before using the change as a rate for UID2 itself.

Draw two circles for your own file. Put the row keys the seat could use on the first run in the first circle. Put the keys it could use on the next run in the second circle. Which keys are in both? Which are new? Which have dropped out? The size of the two circles alone cannot tell you all three counts. Ask for the overlap if the seat can only share totals. Your team might pay for the new part. It might also lose a group that did well in past buys. Look at both sides before renewing the order.

Read API examples as guides to the count. ID5's Match Preview shows how many distinct hashed emails in a file match at least once for each type of output ID [32]. Its sample table gives no rate from a live US bidstream test [32]. LiveRamp says a hash can get a derived RampID even when the hash has no match in its offline graph [134]. What should the buyer do with each kind of reply? Ask the team to mark a prior link, a new code and a failed lookup in separate ways. Check what the buyer can infer when a field simply comes back filled.

The last line on a media plan may count quite a different thing. The Trade Desk says HP reached 8% of an estimated 12.5 million target homes in its US Disney+ and Hulu campaign [147]. Google reported an estimated 81% revenue gain on restricted-cookie browsers in a New York Times Advertising case that used both UID2 and RampID [59]. Keep the first figure with HP's target and the next with publisher revenue. Neither case gives the count of wrongly linked homes [147] [59]. Use the reach and revenue figures for their stated purposes. Give it a row of its own in the plan.

Set one last trap for the rate in your own test. Take the same permitted file and split it by the field that first made a link possible. Use one batch with emails alone, one with emails and phones, and one with no usable contact field. Keep the same buyer, seat and dates. Run those groups together as well as apart. Can the seat explain its total from the counts in the three groups? If it keeps only one copy of a key seen in two groups, ask which group gets the credit. That rule can change the share shown for a group even when the seat has found the same people.

### Which links did the graph get right?

Use *precision* for the share of accepted links that are right. Use *recall* for the share of true links in scope that the graph found. CIMM's April 2026 buyer guide asks buyers to set up a truth set and to check those two measures [259]. Try a test with a hundred known links. A graph claims 80 links. Of the 80 claimed links, 60 are known to be true and 20 are known to be false. Its precision is 60 out of 80. Its recall is 60 out of 100. Keep the 20 wrong links as well as the 40 true ones missed. An email that passed a format check would tell you little about who used a phone.

Pick a source of known answers that does not come from the graph itself. Say a login helped the graph learn a link. Would its own login rule be a fair way to mark that link right? To test homes, say who lived there at the date you care about. To test people, find out who used the device. Keep a group for cases with no known answer. Ask who picked the pairs and who could see the answers. Did the seller get a chance to drop hard cases? Agree on how to treat them before the test starts. Ask for the error counts as well as a range for the result.

Truthset ran a US study for CIMM and Go Addressable covering December 2024 to February 2025 [40, 41]. Its model put the mean accuracy of IP-to-postal links at 13% across six unnamed providers [40, 41]. It put IP-to-email links at 16% on a different base [40, 41]. The postal figure is a mean of the firms' scores, even though the firms supplied 938 million distinct IP-to-postal pairs in all [40, 41]. Ask for each type of link in a separate test. You cannot give any one provider the study's mean as its own score [40, 41] [40, 41].

There is one more way for a match to change a result. In a CIMM TV study the matched group skewed older, had higher incomes and watched more TV [42]. The authors had no independent data to check the truth of each link [42]. Say every link in a future file is right. You could still keep more heavy viewers than light ones. Ask for the traits of the homes that stayed and of the homes that dropped out. Use fields held for both groups. If that is out of reach, say which group you can describe when you report the result.

### What would an old test prove today?

LiveRamp credits comScore with a graph precision range of 95% to 98.9% and names a pool of about 12 million verified devices and people [43]. The pool gives no count of accepted links behind each rate [43]. A PDF with a 2017 copyright and a help page last changed in 2020 repeat the same result [44, 43]. They do not date the test [44, 43]. The papers leave the US group, the funding and the link to current US ATS in doubt [43]. Keep one entry for the test. Ask for the old version, its links and what part of that version the buyer would get today.

Tapad said its Device Graph had twice the precision of the next best graph in a 2016 Hotels.com test [45]. That case gives no absolute rate, base of known US links or version of today's Experian Digital Graph [45, 46]. A fund that held a stake in Tapad said in a separate account that Tapad hired Nielsen to test the graph [53, 54]. Keep the tests apart. A sponsor for the second test cannot settle who paid for the first. A buyer can still use the old report to find the right questions to put to a seller.

When a seller brings back a study, ask for the smallest claim the test would bear. It might read, “For these homes, during this month, this version linked these two kinds of device with this rate of error.” Try replacing “homes” with “people.” Would the same answer still hold? Try replacing the month with the date of your planned buy. What could have changed by then? Keep each of those changes in a note to the team that will sell the results to the client. That team needs to know when a sound, narrow claim would turn into a much larger one.

The grocery case overturns the claim that no admitted product gives a US rate, a clear base, the matched group and the method [14, 13, 12]. The paper could not settle whether a public, independent test verified an admitted graph’s US person or household links against known truth [14, 13, 40, 41, 43, 45, 62]. That leaves a buyer two distinct jobs. Sign off on the yield when you can trace it to the file and rule. Sign off on a claim about who matched only when a test can show who got linked in error. Keep the case, test and decision together so the next buyer can repeat the check.

**Reported match rates and reach**

| Source of the estimate | Estimate vintage | Metric as published | Value | Year | Geography | What the estimate includes | Exclusions | Can these values be combined? | Src |
|---|---|---|---|---|---|---|---|---|---|
| Activation match rate definition | Guide revised 2026-08-06 | Activation match (definition) | No observed percentage | Guide revised 2026-08-06 | Not specified | Unique identifiable records intended for distribution → At least one device at named destination | Distinct from recognition and true-match accuracy. | No. Read the named base and case. | [15] |
| Worked illustration | Guide revised 2025-07-10 | Device recognition and person activation (worked example) | 25% recognition and 100% activation | Guide revised 2025-07-10 | Hypothetical | Four devices versus one submitted person/email → One recognized device versus one person with a destination device | A worked example rather than an observed counterexample. | No. Read the named base and case. | [16] |
| US insurer assessment | Case PDF copyright 2025; assessment period not reported | Assessment file records resolved to a person | 99.5% | Case PDF copyright 2025; assessment period not reported | US | Records in the unnamed US insurer’s assessment file; absolute count, exclusions and deduplication unspecified → A person in the TransUnion offline identity graph | TransUnion reports a file-to-person result from its assessment. The diagnostic assessment and commercial resolution/enrichment service have separate catalog entries. The remaining acceptance paths and direct scored-service attribution are not shown. | No. Read the named base and case. | [17, 18, 19] |
| Experian / MiQ Digital Graph | Case materials read 2026-11-11 | Submitted IPs matched to cookieless identifiers | 70% | Case materials read 2026-11-11 | Unspecified | IP addresses provided by MiQ; counts and filters unavailable → Cookieless identifiers | Global operations do not identify the US test stratum. | No. Read the named base and case. | [20, 21] |
| IRI transaction data / CORE ID | Article 2022-09-09 | Transaction data linked to CORE IDs | 95% | Article 2022-09-09 | US publisher page; data scope unknown | IRI transaction data; record unit and eligible population unstated → CORE IDs | Missing denominator unit and sample scope. | No. Read the named base and case. | [22] |
| Intent Messaging, known email subscribers | Undated as of 2026-11-11 | Match for messaging among known email subscribers | 77% | Undated as of 2026-11-11 | US publisher page; case scope unknown | Known email subscribers; count and exclusions unstated → Unspecified match for messaging | CORE ID is described for the distinct unauthenticated-reach outcome. | No. Read the named base and case. | [23] |
| Cadent audience segments | Release 2022-03-30 | Mean reported segment match | Above 90% average | Release 2022-03-30 | US release; segment scope unspecified | Audience segments; denominator within segment and averaging basis not disclosed → Graph link on match keys | No file counts, uncertainty or segment weights. | No. Read the named base and case. | [24] |
| Viant/Acxiom direct match | Release 2019-09-11 | Count of matched unique IDs | More than 275 million unique IDs | Release 2019-09-11 | US | Eligible input base not reported → Matched identifiers | Count only; no percent and no licensed-SKU mapping. | No. Read the named base and case. | [25] |
| Forrester composite identity match gain | Study July 2026 | Modeled change in enterprise identity-match rate | 20 percentage points | Study July 2026 | Four US-headquartered customers: three North America-focused, one US-focused. Composite has no file-level US stratum. | Modeled organisation; no submitted file → Enterprise identity-match rate | Modeled change, not a product-level observed percentage. | No. Read the named base and case. | [26] |
| US RV retailer CRM file | Calendar test period unknown; case linked on 2026-09-01 | Full CRM records resolved | 83.5% resolved | Calendar test period unknown; case linked on 2026-09-01 | US retailer | Full file, around 884,000–885,000 CRM records as stated in different places → Records resolved against graph with appended identifiers | Case also displays a 700K incoming audience; no raw counts to derive exact global matched numerator. | No. Read the named base and case. | [27, 28] |
| US grocery chain monthly POS | One undisclosed month; case linked on 2026-09-01 | Cumulative monthly POS match after three passes | 73.0% cumulative, after pass 3 | One undisclosed month; case linked on 2026-09-01 | United States | Full monthly POS file in the methods section → Buyer linked using loyalty, payment and household signals | Vendor also reports about 70% of anonymous transactions and 64–73% of monthly volume. Approximate pass counts total 31.0m, against a 32m headline. The final pass infers buyers from geography and households; no labeled accuracy. | No. Read the named base and case. | [12] |
| Kantar / Blockgraph panel overlap with TV partners | Announcement 2022-11-01 | Relative increase in panel-to-partner matching | Above 10× before/after | Announcement 2022-11-01 | North America report; individual test geography unstated | Eligible panel and TV partner file counts not given → Partner-specific matching, unreleased baseline | Relative change, no absolute levels. | No. Read the named base and case. | [29] |
| Yahoo DataX sample response | Docs as read 2026-11-11 | Partner-ID match in a sample upload response | 80 of 100 (80%) | Docs as read 2026-11-11 | Example unlocated | Illustrative partner upload job → Example partner-ID match | No measured result or automatic transfer to ConnectID. | No. Read the named base and case. | [30] |
| Mass retailer, UID2 activation | Undated measurement; directly linked in 2026-09-01 article | Matched CRM records at UID2 before and after enrichment | 73.2M / 51% before; 94.4M / 66% after | Undated measurement; directly linked in 2026-09-01 article | United States | Retailer CRM, nominally 141.7M; comparison of original versus enriched file → Records matched at UID2, as labeled by Deep Sync | Deep Sync’s labeled UID2 row gives 51% and 66%. Dividing by the 141.7m nominal CRM size gives about 51.7% and 66.6%; rounding or destination eligibility is unspecified. Meta’s upper estimate exceeds 100%. | No. Read the named base and case. | [31] |
| ID5 onboarding preview | Unknown where not stated by the publishing party. | Distinct input hashed emails with at least one match, by output ID type (definition) | Definition only | Unknown where not stated by the publishing party. | Unspecified | Distinct input hashed email addresses → Input with at least one match, by output ID type | Preview gives a definition; a US measured numerical case for this particular metric is absent from the inspected evidence. | No. Read the named base and case. | [32] |
| Merkury Data Enhancement / DataSource to Data Cloud | 2024 presentation; underlying cases unstated | DataSource matches to Data Cloud profiles | At least 95% | 2024 presentation; underlying cases unstated | US footprint on other slide; cases not separately US | Prior clients; exact eligible base unspecified → DataSource consumer data matches to Data Cloud profiles | Published by Merkle; no defined US client file or rule. | No. Read the named base and case. | [33] |
| Joint onboarding and Lexicon | 2024-11-19 announcement; measurement period unstated | Average onboarding match, joint Lexicon claim | 90% average | 2024-11-19 announcement; measurement period unstated | US dateline, sample geography not given | Eligible bidstream or file base not disclosed → Onboarded first-party records linked to online identifiers | Joint production claim; no denominator or averaging weights. | No. Read the named base and case. | [34] |
| US QSR exposure/control campaign | Study period not disclosed | Household match of exposure and holdout records | 66.1% | Study period not disclosed | United States | About 1.7m exposure plus 300k holdout, hashed primary IPs → Household graph match against both groups | Reported by Kochava for a measurement engagement; original screened unit is not a separately verified licensed US graph. | No. Read the named base and case. | [35] |
| US insurer field-agreement slice | 2025 case copyright; sample period unstated | Agreement on names and at least one other personal field | 97.4% | 2025 case copyright; sample period unstated | United States | Unnamed insurer assessment file; exact denominator of subset not stated separately → First and last names and at least one other personal field agree | The 97.4% is a field-agreement slice. The 99.5% is a wider file-to-person assessment result. Exact slice denominator and attribution to a licensed TruAudience service are unverified. | No. Read the named base and case. | [17, 18, 19] |
| Unnamed DSP cookie-sync case | Within weeks; dates unspecified | IDs synced by a DSP | 84% synced | Within weeks; dates unspecified | No verified US-only sample | Unspecified eligible IDs; raw counts/valid input exclusions undisclosed → IDs the DSP synced in the first weeks | Secondary issuer report on Digital Graph with a different input and output from MiQ. | No. Read the named base and case. | [36, 37] |
| Spectrum own MVPD service | Unknown where not stated by the publishing party. | Addressable match for Spectrum’s own service | 95% addressable match rate | Unknown where not stated by the publishing party. | US Spectrum TV service; sample geography not separately stated | Unknown incident base for Spectrum service → Addressable result for the MVPD | Adjacent Spectrum service. No attribution to a Blockgraph match rate. | No. Read the named base and case. | [38] |
| 33Across publisher traffic resolved to DSP ID | Undated | Issuer-reported traffic-to-DSP-ID match | Nearly 90% | Undated | Unstated; UK hosting is not a case location | Publisher traffic; eligible counting unit unstated → Traffic linked to the participating DSP’s own ID | No identified US sample, exact eligible traffic unit or assignment rule. Distinct from the MediaWallah joint offering. | No; no comparable units or test population. | [39] |

**How a study checks an identity match**

| Study | Year | Design | Sample and denominator | Study setting | Media format or environment | Measure tested | Reported result | Uncertainty reported | Sponsor | Where the result applies | Src |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Truthset IP linkage benchmark | 2025 | Vendor-submitted IPv4 linkage pairs scored by models trained and tested on sampled ISP/vMVPD subscriber records and lease-time data | 938 million distinct vendor IP-to-postal pairs; six masked provider scores, based on reference operators. The provider mean does not use all 938m as its denominator. | US subscriber reference samples from two ISPs and one vMVPD; six link suppliers masked | IP-to-postal | Mean provider accuracy under proprietary Truthscores | 13% | Product results, weighting and pairwise uncertainty unresolved | Jointly commissioned by CIMM and Go Addressable; Truthset provided proprietary Truthscores. Participants are masked. | No named graph inference | [40, 41] |
| TV match rate and match bias | 2023 | Pre-match versus post-match TV household profiles | Five identity providers; two TV datasets | US national TV input | Household viewing sample | Composition and match yield | Matched viewers skewed older, wealthier and heavier viewers | No independent truth dataset | Commissioned by CIMM. Data contributors and identity firms participated. | Bias estimate is specific to these input files | [42] |
| comScore assessment summarized by LiveRamp | 2017 copyright; PDF created 2018-10-05; online help last modified 2020-08-19 | Seller reports external panel comparison | Panel of about 12 million devices and individuals; tested-link counts unspecified | Unreported region and version | Device/person links | Precision as described by LiveRamp | 95%–98.9% reported by vendor | Unreported in note | Product seller reports a comScore result. Who commissioned the test and how the panel was drawn are unstated here. | Original study, sponsor and per-sample denominator needed | [43, 44] |
| Hotels.com cross-screen comparison as described by Tapad | 2016 | Customer partner selection reported by seller | Three graph offerings; comparison denominator undisclosed | Sample geography unresolved | Cross-screen device links | Relative precision | Twice next provider, as stated by Tapad | No absolute counts or uncertainty disclosed in announcement | Vendor published a customer evaluation during partner selection. Study funding not disclosed. | Historical Tapad test; not transferable to Experian graph | [45, 46] |
| USENIX participant-device matching experiment | 2017 | Authors matched mobile and desktop devices using IP and history | 126 participants; 107 complete mobile/desktop pairs, split into 63 training and 44 test pairs | Participant cohort | Mobile–desktop device pair | F1 | 0.91 for authors’ matcher | Convenience recruitment and 44 held-out pairs; Table 4 supplies point estimates without confidence intervals | Comcast Research Grant for the user study; Jebara acknowledges DARPA and NSF. Sponsor commissioning terms unreported. | Research matcher only; interested grant disclosed | [47] |
| Reasons Dynamic Addresses Change | 2016 | Longitudinal observation of probes and public IP reassignment | 3,038 probes across 929 autonomous systems | 156 countries | Probe / customer equipment | Network events associated with observed address changes | Reassignments coincided with some outages | Geography and ISP policy differ | NSF grants acknowledged. Authors affiliated with research institutions and RIPE NCC. No advertising graph under test. | No person/household ground truth | [48] |
| Economic consequences of online tracking restrictions | 2023 | Author manuscript; longitudinal auction and cookie records with censoring treatment | 54,127 unique exchange cookies, sampled among active cookies on April 29, 2015 | European ad exchange | Exchange cookie | Estimated mean cookie lifetime | Censoring-adjusted mean 279 days; observed mean 216 days | Exchange-specific selection and historical observation window | NET Institute, Hi!PARIS Center and European Research Council grant 833714. Data supplied by a commercial exchange. | No transfer to current US cookies | [49] |
| comScore Cookie Deletion Analysis | 2007 | Metered home-PC data vs. cookies at a selected site and ad server | 400,000 home PCs in panel study | US homes | First-party cookie / home computer | Monthly clearing of cookies for studied site | 31% of panel home computers had at least one first-party cookie reset at the selected site; login cookies excluded | Vendor-defined panel and selected site | Market measurement vendor presenting its own panel analysis. Funding beyond publisher undetermined. | Historical home-PC context only | [50] |
| Forrester / TransUnion Total Economic Impact | 2026 | Interviews used to model a composite enterprise | 4 interviewees | Decision-makers from four US-headquartered customers, recruited from names supplied by TransUnion | Interview and financial model | Modeled identity match-rate change | 20 percentage points | No observed file-level estimate from this model | TransUnion | Composite illustrative enterprise; no standalone product-level estimate | [26] |
| Forrester / Neustar identity resolution survey | 2019 | Online survey | 175 respondents | US, UK, Germany pooled | Organizational survey | Self-reported practices and priorities | No link-error test | Selection and self-report limitations | Neustar | Professionals responding in three countries during April–May 2019 | [51] |
| Truthset IP linkage benchmark — email | 2025 | Vendor-submitted IPv4 linkage pairs scored by models trained and tested on sampled ISP/vMVPD subscriber records and lease-time data | 5.6 billion IP-email pairs; separate linkage population. Reported value is mean of six provider scores. | US subscriber reference samples from two ISPs and one vMVPD; six link suppliers masked | IP-to-email | Mean provider accuracy under proprietary Truthscores | 16% | Product results, weighting and pairwise uncertainty unresolved | Jointly commissioned by CIMM and Go Addressable; Truthset provided proprietary Truthscores. Participants are masked. | No named graph inference | [40, 41] |
| Nielsen historical Tapad verification | 2015 evaluation; 2017 program overview | Sponsor’s investor says Tapad commissioned Nielsen; independent construction of labels and per-unit denominator not published in the reviewed sources. | Accepted-link denominator and raw counts undisclosed. | Historical Tapad links; geographic test population unverified. | Cross-device identity | Reported cross-device accuracy; later program distinguished precision, recall and match rate. | An investor reported 91% accuracy for Tapad. The original test protocol and current product mapping were unavailable. | Not stated. | Tapad commissioned the evaluation. | Known interested commission; not an independent test, regardless of whether the test population was US. | [52, 53, 54] |
| Intent IQ BidPulse release (October 2026) | 2026 | Competing seller scores location consistency across identifiers and fixed-line IP geography. External implementation used Intent IQ data. | Person- or household-labelled panel size unreported. | Intent IQ benchmark and fixed-line IP location data. | Programmatic identity signals | Geographic agreement between signals. | The competitor reports geographic consistency. The release does not give an independent person- or household-link error rate. | Not stated. | Intent IQ competes with ID providers. | Geographic agreement does not supply an independently known person/household association. | [55, 56, 57, 58] |
| Google/New York Times Advertising Secure Signals | 2025 publisher results | Google Ad Manager impact estimates after integration of more than one ID at NYTA. | US-only sample and single-ID control undisclosed. | New York Times Advertising on restricted-cookie browsers. | Publisher revenue with Secure Signals | Ad Manager estimate of revenue change. | Google reports an 81% gain with RampID and UID2 both in use. The contribution of either ID alone is unknown. | Not stated. | Google publishes its customer case. | Outcome is publisher yield; distinct from person-level or household-link accuracy. | [59] |
| ORG/Cracked Labs technical review: LiveRamp | 2023 review; published 2024 | Review of vendor software documentation with a European focus (research September 2023). | Documentation review; no test panel. | European RampID processing documentation. | Identity service documentation | Technical and governance analysis. | ORG reported complaints to the UK ICO and French CNIL. The review supplies no independently labelled link accuracy or regulatory finding. | Not stated. | Open Rights Group commissioned the authors. | No US labelled panel or correctness benchmark. | [60, 61] |
| Deep Sync grocery and RV retailer studies | Cases linked September 2026 | Seller counts resolved file records and transactions, including a probabilistic final pass. | Separate bases: full monthly POS file and full retailer CRM file. | Two separate unnamed US retailers. | Retail POS and CRM identity enrichment | Cumulative transaction match and CRM record resolution. | Seller reports 73.0% of monthly grocery POS after three passes and 83.5% of a separate RV retailer’s CRM file. | Not stated. | Deep Sync markets both services. | Addressability and device expansion provide neither independent truth nor verified correctly linked buyers. | [27, 28, 12] |
| Conversant/comScore cross-device announcement, April 2016 as reposted in May | 2016 | comScore validated a subset of historic cross-device links; four of six performance tiers are reported. | Subset of cross-device links; accepted-link test denominator unavailable. | Conversant cross-device validation; US sample unestablished. | Cross-device identity | Accuracy across performance tiers. | The issuer describes a range for the top four of six tiers. The source does not connect these results to current Epsilon CORE ID. | Not stated. | Interested Conversant communication describes it as comScore’s client; payment and evaluator independence unestablished. | Original protocol, US identifiers, test denominator and transfer to the present CORE ID service are unresolved. A client study is a lead, not an independent replication. | [62] |



## 4. Consent and the bidstream


*A site can stop making new UID2s after a local opt-out [149, 150]. An exchange can pass on a choice [262]. A bidder can act on a withdrawal [186, 149]. The evidence question is whether the published record joins those steps for the same ID, use and place [14, 13, 149, 150, 186, 175, 128, 182, 219]. A buyer who knows where the chain ends can ask the right party for the missing receipt.*

> **In short.**
> The test covered 13 IDs available for use across firms [14, 13, 149, 150, 186, 175, 128, 182, 219]. The record traced a choice through to the buyer for 2 of them [14, 13, 149, 150, 186, 175, 128, 182, 219]. It fell short for 9, and 2 cases were still in doubt [14, 13, 149, 150, 186, 175, 128, 182, 219]. UID2 and ID5 met the test in different ways [14, 13, 149, 150, 186, 175, 128, 182, 219]. Each has a rule for the buyer [14, 13, 149, 150, 186, 175, 128, 182, 219]. The study did not watch live bids to test those rules [14, 13, 149, 150, 186, 175, 128, 182, 219]. A wider test that admits all the firms still in doubt could put the share above half [14, 13, 149, 150, 186, 175, 128, 182, 219].

### Name the use that should stop

Take one ad buy. If a reader withdraws a choice on a publisher’s site this morning, could a buyer still use yesterday’s ID to select an ad this afternoon? Name the product, the place and the use. Suppose the publisher can show that it has stopped making new IDs. The buyer still needs to know what becomes of a value it received yesterday. Suppose instead the buyer can show that it has dropped a value. The publisher still needs to know whether the action covered only its site or the same reader’s broader use of the service.

What did the reader ask the site to do? Stop storing data on this device? Stop using an email for ads? Take an address off a list? Delete a record? Give each of those requests its own test. A buyer could stop using a key for one campaign yet keep it for a use the reader still allows. A site could clear a key from a browser and have no way to tell who last got it. Ask the firm that makes the promise to name what it will do and which copy of the data it can reach.

On Yahoo’s server route, the site sends a hashed email and its publisher ID [165, 189, 163]. An opted-out lookup returns an empty object [165, 189, 163]. The supplied email and the user’s state therefore serve different roles in that route [165, 189, 163]. If an advertiser’s test starts with a signed-in reader, ask for the choice the reader made as well as the event that proved control of an account. A valid email could explain how the site got the ID. What right would the buyer have to use it?

Ask what value the buyer should see. Does the site leave out the ID, send a code for a refusal, or send the ID with a separate rule on its use? Ask the buyer what it should do in each case. What part of that result could the site see? For a trial, use a made-up key and a choice the teams know in advance. Save the result at the buyer. The teams can rehearse this flow without a real reader’s email. For a live test, agree who can run it and what it covers.

### Read the signal at the right boundary

The Global Privacy Platform can pass a user’s choice from a site or app to firms that serve ads [253]. A second signal asks sites and services to refrain from sale or sharing [3]. The September 2026 W3C draft calls it Global Privacy Control and sets out how to send it by HTTP and read it in the browser [3]. Ask which signal the US site sent. Who read it? Which use did it cover? Use the draft to plan a test. Ask the firm that got the signal for the result.

IAB Europe set a March 1, 2026 rule for new Transparency and Consent strings [254]. A new string that lacks the disclosed-vendor segment is invalid from that date [254]. An older string can stay valid until the user renews or changes a choice [254]. One test would save the time at which the choice was logged with the time at which the buyer evaluated it. Under that rule, the same missing field could mean two different things [254]. A log with just the time of the bid would leave the cause in doubt.

From Prebid.js 11.14, lookup failures can change which IDs its modules add to a bid request [239, 255]. If a lookup of a configured TCF or GPP consent platform fails, the user-ID modules skip retrieval and add no IDs to the bid [239, 255]. If the lookup times out, the modules run with the consent data at hand [239, 255]. A site can also require Purpose 4 consent before it sends IDs or extended IDs [2]. Name the version and the site setting. Save each test result. One line that says “consent missing” could hide the cause.

OpenRTB 2.6 has fields that can name who added or bridged an extended ID and how it was matched [252, 1]. If the values are there, the buyer can use them to trace the bid back to the firm that put the ID on it [252, 1]. Log the reader’s choice and the buyer’s reply as well. Two firms might send IDs with the same source name. Did each get the same choice from the site? Did each pass it on? Use the name on the bid to find the firm that can show the next step.

### When a separate message follows the ID

A UID2 user can ask to opt out through a portal [149, 186]. The user sends each email or phone number as its own request [149, 150]. A site learns the result when it next asks the operator to make or renew a token [149, 186]. The operator tells DSPs through a separate channel [149, 186]. Look at the timing from each side. A check of the site’s page could miss a notice the buyer had already received. It could also miss a delay before the buyer got that notice.

The UID2 notice to a DSP comes by webhook [186]. The notice carries the raw ID and the opt-out time, and UID2 says it typically sends it within a few minutes [186]. The bidder must stop using that UID2 for real-time bids [186, 149]. It may still use another ID or make no bid [186, 149]. The DSP guide tells the bidder to check opt-out status after it decrypts the token [186, 149]. Under the study rule, the choice may reach the buyer on a second path if it refers to the same ID and says what use must stop [14, 13, 149, 150, 186, 175, 128, 182, 219].

For a test, first get a token the bidder can use. Ask to opt out for the input that made it. Note when the buyer gets the notice. Now send the earlier token to that buyer. Does it use the key? If the buyer bids, save the key that backed the bid. It may have used a different ID, which the UID2 guide allows [186, 149]. If the buyer makes no bid, ask why. A test with no bid at all could hide a budget cap. The buyer’s log should tell those cases apart.

The US terms and the bidder guide set duties [150, 144, 186, 187]. The sample bid and the guides do not show who enforced TCF or GPP on a given impression [150, 144, 186, 187]. Keep the claim made here at the level the record supports. The DSP guide ties the raw UID2 to a withdrawal notice and a rule for bidding [186]. An observer would still have to test a bidder to show that it enforced withdrawal for a given impression. A site-level opt-out stays with that participant and is not sent to the wider UID2 system [149, 150].

Think of a reader who uses one email for a site and a different email for a store. For a test of an address-based opt-out, use two made-up addresses in place of those accounts. Submit the request for the first one. Leave the second one alone. Ask the issuer to say which key it expects the buyer to block and why. Now run both keys through the same bid path. If the buyer blocks both, the test team should ask how it joined them. If it blocks just one, check the scope of the promise made to the reader. The right result depends on the promise and the input the service accepted.

### When the ID field remains present

ID5’s Prebid guide gives the restricted value an exact form [175]. When consent is absent or the user opts out, the guide says the id5-sync.com EID should stay in the bid with its ID set to the string “0” [175]. For a usable ID, the SSP sends the field in OpenRTB [176]. The DSP gets its keys and merge instructions through a separate route [176]. ID5 says only firms with rights under the user’s and site’s choices can decrypt the ID [176]. Ask to see the value. A count of fields alone would miss the zero [175].

For US traffic, ID5 asks the site to include it in the flow for a choice against sale or sharing [177, 131]. Its US examples include GPP and GPC [177, 131]. For the EEA and UK, its guide gives TCF or prior opt-in instructions [177, 131]. Its Java module for Prebid Server sends the privacy signals it has to the ID5 API when it requests an ID [256]. Ask the site which route it uses. Write the country on the test sheet. A team that tests the wrong branch could get a sound result for the wrong rule.

ID5 met the public record test [14, 13, 149, 150, 186, 175, 128, 182, 219]. Its record links the choice at the site, the value sent in the bid and the rules for access to that value [14, 13, 149, 150, 186, 175, 128, 182, 219]. The study did not see a buyer deal with the zero on a live bid [14, 13, 149, 150, 186, 175, 128, 182, 219]. In a pilot, ask to see two bids. One should carry the zero. The other should carry an ID the buyer can use. Use the same keys and site settings. If the buyer lacks a key for both bids, a drop in use would tell the team little about the reader’s choice.

ID5 says it posts opt-out updates every 30 minutes and routine deletion updates once an hour [128]. It runs a separate best-effort feed for on-demand deletions [128]. A partner is meant to load the lists [128]. The records did not show when each firm did so [128]. Start two clocks at the user’s request. Stop the first when the issuer puts the change on the list. Stop the next when the firm that got the list acts on it. If the file went out on time, the second clock might still be running.

### Local controls and downstream gaps

In the US, LiveRamp’s ATS JavaScript checks the national GPP string first [182]. It can fall back to the older CCPA string [182]. If the user rejects consent, it clears the stored envelope [257]. LiveRamp tells a site that runs its own opt-out to delete the local envelope and leave a flag in place until the reader opts back in [138]. On the bid path, an SSP reads the envelope and sends a DSP-specific RampID to the buyer [133]. The site’s act is clear. The public record leaves open what the buyer does with an older copy [14, 13, 149, 150, 186, 175, 128, 182, 219].

LiveRamp makes opt-out and deletion requests available to destinations that might receive the consumer’s data [139]. For first-party advertiser audiences and third-party marketplace segments, its guide names a route to The Trade Desk [139]. LiveRamp expects destinations to respect the requests and leaves processing to their discretion [139]. Ask how the buyer first got the key. Did it come in a bid or in a file? If both sent it, ask for a receipt for each route. A file sent to the right firm could still leave a question about the copy that arrived with a bid.

A site that looks up an opted-out email can get an empty object from Yahoo’s server [160, 200, 165]. Yahoo has a form that asks for an email to confirm the opt-out [160, 200, 165]. A site can also set a flag that stops the JavaScript module from making new ConnectIDs on its own domain [201, 164, 195]. Google lists ConnectID as a web secure signal [156, 157]. Yahoo names its App & Display and Video bidders in its guide for sites [156, 157]. Those names tell the site whom to ask. The study could not trace the user’s choice to a decision on that bid [14, 13, 149, 150, 186, 175, 128, 182, 219].

Lotame says its browser opt-out stops its use of Panorama ID [261]. A stored ID can take up to seven days to expire [261]. Lotame’s case names the firms involved in a Michigan media buy [264]. It describes a publisher using its PubMatic seat to target an audience and lists Xandr and Adelphic for delivery [264]. The case does not show how a bidder dealt with an opt-out for a given Panorama ID [14, 13, 149, 150, 186, 175, 128, 182, 219]. Ask who can find a copy sent before the user’s choice. Which key would the buyer need to show that it had stopped using that copy?

LiveIntent says its module gets a US privacy string and honors opt-outs [260]. Its lookup endpoint also accepts a GPP string and the sections that apply [219]. TripleLift lists the LiveIntent provider for US bridging [263]. In a 2020 California FAQ, TripleLift said it sent consent status with the bid and left targeting to the DSP [262]. These records show where the ID can go [263] [260]. They leave open what the bidder must do with a nonID after the reader says no [219] [14, 13, 149, 150, 186, 175, 128, 182, 219]. A buyer could ask that one question without making a claim about a breach.

### Which copy of the ID is at stake?

SharedID stores a key on the site’s own domain and makes it available to bid adapters [258, 239]. TripleLift lists SharedID for US passthrough [263]. For a test, draw a line around the site that supplied the value. If a reader acts there, which copy can the site reach? Draw a second line around the firm that got the bid. If that firm kept a copy, how could it find it next week? Let each side answer for its own copy. Then ask what would link the two replies to the same user request.

For a shared device, start with the person who made the choice. Suppose one reader signed in, shut the browser and gave the device to a second reader. Which reader now asks the site to stop? Which key should the site clear? A choice tied to an account could reach beyond the browser used to send it. A choice on a device could carry over to the next user there. Agree on the expected outcome before the test. Without that step, the team might see the right change in the next session and credit it to the wrong request.

Test this with made-up keys. Ask for no one’s web history. Keep one test publisher key active on a test site and withdraw a second. Keep the same buyer, site, place and version. Have the receiver report its action for both values, then repeat after the next scheduled update. If the report ends at the site, ask the firm that next got the bid. If it shows a receiver decision without the input that produced it, ask the sender to supply the missing test reference. That brief exchange would yield a more precise public statement about whose choice affected whose action.

### How the result changes with scope

The paper set a strict majority test [14, 13, 149, 150, 186, 175, 128, 182, 219]. Would more than half of the IDs in scope show a full choice path in at least one place where the service was sold [14, 13, 149, 150, 186, 175, 128, 182, 219]? The test kept IDs that lacked enough evidence to enter the scored field [14, 13, 149, 150, 186, 175, 128, 182, 219]. Now give each case still in doubt the benefit of the doubt. At most 4 of 13 would pass [14, 13, 149, 150, 186, 175, 128, 182, 219]. That falls short of the stated bar [14, 13, 149, 150, 186, 175, 128, 182, 219]. The test counts paths in the public record [14, 13, 149, 150, 186, 175, 128, 182, 219]. It did not count live bids on which a buyer acted on the choice [14, 13, 149, 150, 186, 175, 128, 182, 219].

In the US, 2 of 13 had a full path [14, 13, 149, 150, 186, 175, 128, 182, 219]. The search fell short for 10, and 1 remained in doubt [14, 13, 149, 150, 186, 175, 128, 182, 219]. In the EU, 1 of 3 had a path, 2 fell short and 0 were open [14, 13, 149, 150, 186, 175, 128, 182, 219]. In the UK, 1 of 3 had a path, 2 fell short and 0 were open [14, 13, 149, 150, 186, 175, 128, 182, 219]. Each group leaves out a service if the named place was out of scope or its sale there could not be shown [14, 13, 149, 150, 186, 175, 128, 182, 219]. Use the US line for a test on a US site.

The wider test adds 6 names whose place in the study was still in doubt [14, 13, 149, 150, 186, 175, 128, 182, 219]. That would bring the group to 19 [14, 13, 149, 150, 186, 175, 128, 182, 219]. Give each new case a full path and allow each open case to pass. The upper end becomes 10, more than half the wider group [14, 13, 149, 150, 186, 175, 128, 182, 219]. To narrow that range, ask first whether the firm sells the named ID for use on a US bid. Then ask what the firm that gets the ID must do when the reader says no. Ask the firm to show where its policy covers that exact route.

For a buying review, give the site and the buyer the same test event. Ask each to write just what it saw. What did the reader choose? Which value left the site? What reached the buyer? Could it use that key, and which rule told it what to do? If it used a different key, note that too. Join both accounts by test ID and time. Leave a step open if neither party saw it. The gap will show the site what to fix, whom to ask for a log, or which promise to narrow when it next speaks to a reader.



## 5. After Chrome kept cookies


*On April 22, 2025, Google kept the way Chrome lets users choose what to do about third-party cookies [4]. It dropped the plan for a new prompt [4]. Did most US web ID products give cookie sync a bigger role in their guides after that choice [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]? The test of public guides leaves that open [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Ask the site a plain question. Which part of our deal would change if this browser stopped sending the cookie?*

> **In short.**
> Google later chose to retire some Privacy Sandbox tools and keep CHIPS and FedCM [220]. Notes on a release, a change in the code and a small trial mark different steps [265, 266] [268, 5] [270, 267]. ID5 kept the same cap on syncs in three guides for its SDK [179, 223, 180]. UID2 told sites to move some calls to a server on the paths hit by a change from Apple [151, 152, 145]. Neither case shows a rise in third-party-cookie sync on Chrome in the US [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226].

### A choice in settings

The April post said users could still make their choice in Chrome’s privacy and security settings [4]. It said Incognito already blocked third-party cookies by default [4]. The post gave no share of users who let those cookies through and no count of the times a site sent one [4]. What could a buyer of US Chrome ads that week learn from the post? It could date the change to the plan for the prompt [4]. To put a size on its own buy, it would need a count from the sites and the firms that took the bids.

Take a site that wants to set the price of a test. Say two visits load the same page on the same build of Chrome. On one visit the cookie the test needs gets through. On the next the browser blocks it. Keep both in the page-view count if the aim is to price all the ads on that page. For a test of the sync itself, keep a count of the visits that could take part. Now say a request did go out but had no match. Give that event its own line. One loss might call for work on the match. The other could be the choice the user made.

The date on a sales plan matters too. In April 2021, Mediavine said it had links to LiveRamp ATS, UID2 and ID5 in use [280]. A launch note from that time can show there was work in place years before the Chrome choice [280] [4]. Ask which route the site ran in the week before the April 2025 news. Which did it run in the week after? To link a shift in sales to a change at the site, ask when the site changed its code. Ask when the sales team began its test. Put each date next to the change it marks.

### When a feature leaves Chrome

On October 17, 2025, Google named the technologies it planned to retire [4, 220]. Each would go through the steps needed for its platform [4, 220]. The list included Topics and Protected Audience [282] [4, 220]. Google said it would keep supporting CHIPS and FedCM [220]. It had once proposed FLoC, run trials, then named Topics as its replacement [284]. If an old media plan refers to a trial, find out which trial it meant. Ask which feature the team built for and which version the publisher put on its site.

Chrome 144 said it planned to remove the Private Aggregation API [265, 266]. Chrome 152 said it had removed that API [265, 266]. Use the first note to date when a team could plan the switch. Use the next note when writing a test for what that build should do. Then ask when the site saw a change. Which builds did it serve? Did its own code use the feature? If a buyer’s charts went blank, get the last report with data and the first one with a gap. On which build did the last good event run? Which part of that claim can be checked from a release note alone?

A July 2026 Chromium change says the Topics interfaces that sites could call had already become stubs [267]. In September 2026 the team said a trial to take out the stubs had reached 1% on milestone 153 and later in the Stable channel [270, 267]. The 1% was the rollout on those Stable builds [270, 267]. The team gave no device count or US share [270, 267]. A US buyer can keep the figure if it keeps the trial and the builds with it. To price the effect on a buy, ask how many of its bids had used Topics. Did the same set of sites send bids after the change? How many of them sent a field the desk could use?

### A call can survive a change in its job

On July 14, 2026, Chromium changed its Protected Audience renderer calls, including `runAdAuction` [268, 5]. The change made them return rejected promises or default values [268, 5]. Chrome 153 still listed deprecation and removal as planned [283]. On October 30, 2026, Chromium removed the last web bindings for Protected Audience from its source tree [5]. When testing one build, ask what the page got when it made the call. Did the test just find a method of that name? Did it get back a result that the buying code could use? If the buyer wants to apply the test to the whole deal, get the share of live bids that came from that build in the week at issue.

A July 14, 2026 change to Attribution Reporting kept the setter used with `XMLHttpRequest` but made it do nothing [269]. A test could find the setter even if no report was made [269]. Check the piece the buyer needs. If that is a report, start with a test event that should count. Write down which report it should make and check what the party at the far end got. Ask where the team could have lost the event even if the browser had done its part.

On October 8, 2026, the Attribution Reporting team said its fix for `window.open` had reached Chrome 155 Stable [271, 269]. It planned to roll out the rest of the change in steps, with full removal on Stable set for after October 24 [271, 269]. The update did not say that last step was done [271, 269]. If a buy ends that month, get the note for the fix. For the wider change, look for a later note or ask for a test of the build the site used. What did the buyer need the build to do on the day its ads ran?

### Where the browser makes a difference

In March 2020, WebKit said Safari 13.1 on macOS and the iOS and iPadOS 13.4 updates blocked cross-site cookies by default [276]. Firefox gives each site its own store for cookies, according to Mozilla [277]. Say a page reads a stored value when a user comes back. What does that show about use of the value on a second site? Test both visits in the same browser. Where was the value saved? Who could read it on each visit? What did the bidder see? A count of users who came back to the first site would answer just one part of the brief.

In October 2026, WebKit described a new block on third-party domains used to link a user across sites [151, 152]. It covered WebKit on iOS, iPadOS, macOS and visionOS 27 [151, 152]. UID2 named a domain that the new rule blocked [151, 152, 145]. All browsers on iOS 27 or later, and Safari on macOS 27 or later, would block client calls to `uidapi.com`, its guide said [152, 151]. UID2 told publishers on those paths to use a server to make and refresh the tokens [151, 152, 145]. Where the client route still worked, three editions of its guide told the publisher to look for the token in local storage [224, 225, 226]. Ask which calls were moved. Try a first visit and a visit with an old value still in place.

In an app, start with what the user did. Apple’s rule for the 14.5 releases of iOS, iPadOS and tvOS took effect on April 26, 2021 [275]. An app had to ask for leave to track or gain access to the ad ID [275]. On Android, Google Play services says a request for a deleted ad ID gets zeros back [274]. Google dates the deprecation of the Android Privacy Sandbox to October 17, 2025 [272, 220]. Its API guide marks the Topics classes deprecated at level 37 and says calls will be rejected in the future [273]. Ask for separate app and web test results. Test for zeros in the app. On the web, check whether the request went out and what the server sent back.

### How to test a return to cookie sync

Did most of the scored US web products give third-party-cookie sync a bigger role in their guides after the April decision [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]? For each product, the test compares the same job on both sides of the date [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Start with the latest dated guide in the year before Google’s decision [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Find the latest guide for the same route after that day and through the cutoff [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Hold the role of the firm and the place of the buy fixed [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Where more than one path exists, use the one the seller calls its main US route [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. What did it tell a publisher on that path to do?

Say an old guide lets a site choose whether to sync. A new guide for the same job says the site must sync. That would be a change the test could count. Now take a guide that says the site can work when the browser blocks third-party cookies. What should the site do when the browser lets them through? The seller might still tell it to sync. Or it might skip sync for all bids. Both paths fit the words on the page. Leave the result open until the guide tells you which path the site should take. Keep the date with the text. For an old guide that still loads, record when its words last changed as well as when someone read them.

For the products known to be in scope, the test left the majority in doubt [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Both sides of the threshold were still possible when the uncertain products were included [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. There was a second test for the Sandbox changes [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. For each relevant API, could the study pair the date of its announced or effective removal with a change in sync advice for the same main US route [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]? It found no such pair [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226]. Keep the gap with the product name. One copy of a dated guide might close a gap about advice. To answer how much traffic changed, ask for a count. To say the policy caused that change, ask what the publisher would have done had the policy stayed as it was.

### ID5 left a number in the guide

ID5 has a route that syncs with other ad-tech firms where third-party cookies are available [221, 179, 180]. Its SDK guide has three dated versions, on April 18 and October 15, 2025, and November 6, 2026 [179, 223, 180]. Across all three versions, when sync is needed, `maxCascades` caps the syncs at eight by default [179, 223, 180]. Set it to minus one and the guide says sync is off [179, 223, 180]. A cap does not tell the buyer how many calls a site made [179, 223, 180]. The guide also leaves open whether this was the main US Chrome route [179, 223, 180].

Take a site with that default in place. Say its test page needs no sync on the first visit. On the next, it tries a sync but gets no value it can use. How many visits should be in a count of pages with the setting turned on? How many in a count of calls made? How many in a count of joins that worked? If the tester just writes down “eight,” the desk lacks an answer to all three. Keep the cap with the copy of the code that ran. Put the count of calls with the traffic that caused them. Ask the firm at the far end what it got for those same test visits.

ID5’s Prebid example gives 90 days for local storage and two hours for refresh [175, 131]. Its Prebid guide calls for the string zero in the ID field if consent is missing or the user opts out [175]. So check what is in the field as well as whether it is there. Count an expected zero on its own. What would the team expect for a signed-in user? What about a visit with no first-party value, or a visit from a user who opted out? Agree which group a visit goes in before the firms compare their logs. A correct result could look like a loss to a desk that just counts codes.

### LiveRamp puts the pixel in a particular job

LiveRamp says ATS can serve signed-in users without third-party cookies [133, 222, 134]. It also lists an optional ATS endpoint that uses a third-party cookie [133, 222, 134]. On one ATS route the supply partner decrypts an envelope and puts a RampID for that DSP in the bid [133]. For a file that maps cookies to RampIDs, LiveRamp tells the buyer to turn on cookie sync with a pixel [222]. If an engineer cites that rule in an ATS review, look at the order. Does the buyer expect a file with pairs of values? Or should its bidder get one value with an ad request?

Trusted Media Brands told readers about a LiveRamp code that it might keep in its own cookie [281]. It gave no share of visits on which the code was issued [281]. LiveRamp gives ATS envelopes a life of 15 days in California and 30 days elsewhere [137]. Its guide tells sites to update the stored expiry when they refresh an envelope [137]. If a bidder got no useful value, check the age on the envelope. Ask when the site last refreshed it. Ask the bidder which field it got and whether its key could read the value. An old envelope, a cookie the browser blocked and a bid that never reached the desk would need three different tests.

The public guides did not date a switch to or from cookie sync on these LiveRamp routes [133, 222, 134]. To close that gap for one site, ask for a copy of the old guide for the same job. Ask for the settings in use now. If the seller added a path the site could choose, when did the site turn it on? If the old how-to page just describes a map in a file, keep it with that file buy. The team may have changed what it bought. That alone could give it very different steps to follow.

### A working ID can have a much older history

Prebid’s April 2025 guide says Yahoo ConnectID can work without third-party cookies [202, 199]. Yahoo gives sites more than one way to use it [202, 199]. Both date checks found the same April 4 guide [202]. The later state of the main US sync path stayed open [202, 199]. An older code change is easier to date. Prebid 7.51.0 took either an email hash or a site’s own user ID [164, 189, 191, 192, 193]. Release 7.52.0 on June 1, 2023, dropped the need to supply either one [164, 189, 191, 192, 193]. Test that change by leaving both inputs out in each build. A test of sync would need to check the sync path itself.

In June 2021, Prebid.js 5.0.0 merged the PubCommon and SharedID modules [285, 286]. Prebid’s guide shows the earlier third-party sync and the first-party path that followed [285, 286]. On the header-bidding path reviewed, the site stores the ID on its own domain [278]. The software change dates to years before Chrome’s April decision [285, 286] [4]. When a site shows the new setup, keep the old release too. It tells the buyer when the code changed. Ask the publisher which week it put the new code on its own pages.

Prebid’s guide to Lotame Panorama describes an ID for a device that a publisher can use to enrich data or activate an audience [279]. Both date checks led back to the same January 2025 guide [279]. Suppose a seller also gives a buyer a pixel for a different product. That pixel would leave the gap in Panorama’s own guide open [279]. Save each guide with the name of the product, the route and the release. A new member of the trading team should be able to see which page applies to the deal without having to piece together its full history.

### Read a revenue claim at the end of the path

Google put an 81% gain in revenue on record for New York Times Advertising on browsers with cookie limits in its 2025 results [59]. The site sent both RampID and UID2 through Secure Signals [59]. The figure was Google’s estimate for that joint setup, with no test of one ID by itself or US-only group [59]. It did not check who the ID belonged to [59]. A site may want to know what happened to its sales when the signals could reach more bids. For a new buy, ask which sites, bids and prices went into each arm of the test. Keep the browser settings with the counts.

Sovrn and ID5 ran a test with US audiences on sites that took part [129, 130]. They compared a control with two paths for a linked email hash [129, 130]. In one it went into the auction, and in the other it was held back [129, 130]. The case gives CPM differences for the combined treatments [129, 130]. It gives no count of right-person links and no stated way to assign the groups at random [129, 130]. If the aim is yield, set the price or sales measure first. If the aim is to test the ID, agree on a way to tell a right link from a wrong one. If the aim is to test a shift back to cookie sync, count that step on the same path before and after the change. Get the site’s test and ask the bidder which results it could use.



## 6. Platforms and walled gardens


*Who gets the match? Google says a match on an email or phone adds the linked Google Account to a Customer Match segment [213]. Meta’s list terms say the brand does not learn which people make up its matched group [302]. A brand can buy ads for that group. What can it take back to its own store? Settle that point before it pays for the match.*

> **In short.**
> Trace the direction of travel. A brand can send a list in, a publisher can pass a signal through, and a store can send a group out to a media seat [213] [289] [295]. Google’s guide describes sending a publisher’s registered signals to approved bidders [289]. Kroger has described making its purchase-based audiences available through existing seats on The Trade Desk [295]. Write down what to test for each path.

### Begin with the account that can act

Take a store that wants its old shoppers back. Its club has a list of those who have stopped buying. The ad team wants to reach them. The sales team wants to count new trips. Give each team the name of the place where it will see its result. For the ad buy, can the seat use the group? For the sales check, which trips will count and for how long? To add to its own list, the store would need a key it could join back to that list. It would also need the right to use what came back.

Sketch the path with ordinary verbs. Who sends the list? Who links it to an account? Who sees a group or gets a count? Who picks which ads can run? Ask to see the screen where the list went in. Then ask to see the group on the screen where the trader works. Get a copy of the report that will go to the brand. A list can go in as planned and still fail to reach the seat. The sales team can use those screens to trace each count back to its source.

Say what each row means before the buy. Take a home that shares one club card. That card might be just the right unit for a store keen to win back a lost club member. To send one gift to each person, the store would need more. It might still want the ads even if it can see only a group count. What could it say about the people in that group? Agree on the answer now.

### Google Customer Match links to a Google Account

For email and phone inputs it can match, Google Ads says it adds the corresponding Google Account to the Customer Match segment [213]. Google defines its match rate as the share of an upload it can connect to logged-in Google users [287]. Put that unit in the brief. Who was meant to be on the list? Did the brand drop rows that were there twice? Which rate did the screen show when the job was done? Keep the terms used to build the list next to that rate.

Google says a match rate is displayed for an API upload with at least 100 matched unique users [300]. Try two test runs. Both get a receipt. One shows a match rate and one shows a blank. What does the blank mean in this case? Find out before the team logs a failed job or puts a zero in a chart. Keep the receipt, the terms for showing the rate and what the screen showed. When the trader tries to use the group, log its state then. The help desk can use each of those facts.

Google says most advertisers see rates between 29% and 62%, without supplying the period, geography or advertiser sample for that range [287]. Ask how the brand’s list compares with those behind that range. If the answer is unknown, use past runs of the brand’s own list as a check. Keep a note of what went in each time and which rows were in scope. If the rate moves, first find out which of those things changed. A claim that the links got better would need its own test.

Which tool sent the list, and when? Google directed most new Customer Match integrations to its Data Manager API from April 1, 2026, while eligible existing integrations could continue to use the Google Ads API [288]. If the team moves the job, name the old and new tools. Keep the source fields and the list names from both runs. Check that the same rows went in before you look at what came back. What if both the tool and the list have changed? Check them one by one. The first test of the new tool is whether the list got there.

### A signal can pass through Google to another bidder

Google says a publisher can tell Ad Manager to send an obfuscated signal to a third-party bidder it has approved [289]. The signal must be registered, and bidders select the signals they want to receive [289]. Name the publisher, the signal and the bidder due to get it. Ask the bidder to show that the signal reached its end for the test ads. A buyer that wants to match to a Google Account would ask for a different path [213] [289].

Send each question to the team that holds the switch. Suppose a signal went missing in a test. What did the publisher turn on for this bidder? What did the bidder choose to get at the time of the test? Ask the seller to check that the test took the right path. Even if the match worked at the start, the handoff might have failed. Mark the first place where what the team saw broke from the plan.

Roku announced on March 23, 2026 that it was a launch partner for Google’s Confidential Publisher Match, describing a route for DV360 advertisers to match first-party data to Roku Media inventory [299]. Roku also documents a separate route using UID2 to activate against Roku first-party data through The Trade Desk and Roku Exchange [154]. Name the seat, match path, ad space and flight dates in the brief. Ask for proof from that path. News of a launch in one seat can be hard to act on from a desk set up for a different one.

### Amazon distinguishes an audience from an analyst’s answer

In its terms dated November 20, 2024, Amazon sets out a way for a buyer to send hashed data to match to Amazon IDs [290]. Under those terms the brand also sends IDs for its segments [290]. Amazon Ads tells the brand or its agency to prepare and hash the list, then choose the group to use for ads [291]. Give three things their own names in the brief. What key does the brand use in its books? Which field goes in for the match? What name does the group have in the seat? The group name should tell the buyer what it chose. To check who a single member is, ask what proof the brand could see.

Take a brand with two groups. One has members due to renew. The other has those who just paid. The buyer wants to reach the first and skip the second. For each list, state who goes in, the date that takes effect and the name that shows up in the seat. Which list wins if a late sale puts someone on both? A trader could choose the right group and still send the wrong offer if last night’s list was stale. Log this as a fault with the source list when the check shows that is where it went wrong.

Amazon Marketing Cloud describes anonymous, aggregated results for analysts and a separate route that sends custom audiences to ad activation [301]. Which output did the brand buy? To check a count, the team needs its dates and a clear sense of what each row counts. To run ads, the trader needs a group it can choose in the seat. It also needs a way to check which ads used it. If a brief calls for “clean-room results,” ask who gets what for each use.

A brief could ask for a count by month of sale. It might also ask for ads to reach the group from just one month. How should the count treat a return? What if the same shopper placed two orders? What if a member left the group while the team was running the query? Give each count a date. Put the same date on the list if it has the same cutoff. If the two dates differ, keep both. They may help the trader explain why the two sums do not match.

### What a Meta customer list can tell its owner

Meta lists email and phone as inputs to its Marketing API for Custom Audiences [292]. Its customer list terms say it does not tell the brand which people make up the group [302]. Use that clause when the team writes the brief. Does the brand want a list of its own shoppers who have a Meta account? Ask how it could get that list and what else would have to be bought. Does the brand want ads to run for a group? Say which ads it wants to run and which facts it needs once the buy is done.

Say a club member disputes a bill. Was she in the group used on the day she called? The brand could go back to its source list and check if she met the terms for that day’s group. It could look up the receipt for the list and the name of the group used for the ad buy. Those checks could show what the brand sent and meant to use. Did the site match this one person? The team needs a source for that next step. Find out what the help desk can show before it makes a promise to her.

Now take a buy that spans two sites. Both show a count for the same brand list. All the people in the smaller group could be in the larger one. Very few could be in both. The two sums leave both answers open. Can the brand count how many are in both? Has it got the right to make that check, and is it counting people or accounts? If a planner makes a guess, keep the guess beside the plan. Show how the reach figure would move if more of the group were shared. Give a count based on a test its own field.

Consider what happens if the same agency takes over a second brand’s seat. In this example, both brands use the label “recent buyer.” One wants to exclude anyone who bought in a store. The other wants to include anyone who bought from its website. A table of audience names would make the rows look alike. Put the account owner, source list, qualifying action, date range and intended campaign use beside the names. Have the trader verify the seat before selecting either group. If a seat changes hands, hand over those definitions as well as the settings. Those terms will help the next team build the right group if its source changes.

### When a retailer sends an audience out

What if the store sends the group out? In November 2023, Kroger Precision Marketing said its customers could activate Kroger purchase-based audiences through their existing seats on The Trade Desk [295]. Walmart Connect announced an offsite path for Walmart audiences that named Yahoo DSP, VIZIO inventory and Magnite [293]. Roundel says its managed Google search ads can use Target groups, such as people who bought on Target.com [294]. Name whose purchases define the audience and where the brand can use it. Use those terms to write the order, even if the brand has no list of its own to send.

For a Kroger buy, name the type of sale that should put a shopper in the group. How far back should the store look for one? Which seat will use the group? Can the trader find it in that seat now? Ask who can change the terms while the ads run. For the Walmart buy, name the ad space and the firms in the path offered. If the plan calls for a new seat or a new source of ads, check that path too. Who can confirm that the group and the controls will work there?

For the Roundel search offer, start with the retailer’s own audience and the managed buying service [294]. If a brand also wants to send its customer list to the retailer, add that as a separate request. Who will take the list and on what terms? Which search ads will use it? Ask to see the way the team set up the buy. A generic request for access to “retail data” could be met by a sell-side audience, a buyer’s list upload or a report on sales. Write down who sends what to whom. Put that verb in the order.

Give sales their own line in the brief. If the retailer reports purchases after the buy, ask whether it counts units, orders, shopper accounts or households, and which channels it covers. If a shopper returns an item, agree whether the count will change. If a product goes on sale midflight, retain the dates and terms of the sale with the media report. Knowing who was meant to see the ads leaves a sales question open. How many sales were due to the ads? Agree on a way to test that claim.

### A clean room can direct its result back to the retailer

Instacart said in January 2026 that selected agencies and consumer goods firms had piloted its Data Hub clean-room service for joining marketer and grocery data [296]. Its setup guide illustrates an analysis that intersects hashed emails from two companies while keeping the full datasets hidden [297]. For activation queries, Instacart directs results to its own account and says partners cannot see the specific matched identifiers [298]. Read each for the step it covers. One names the pilot, one shows a join and one says where the group goes [296] [297] [298]. Check that the brand can use the route before it books the ads.

Take a food brand keen to reach past buyers of its soup who now buy soup again. Who will send the old list for the brand? Who will show the new sales for soup as a whole? How far back should each side look? Does a row mean an email, a club account or a sale? If the query gives the brand a count, say what the count is of. If it sends a group to the store, write down which account gets it. Show where the trader can use it to run the ads.

Give the join and the ad buy their own tests. A buyer could ask whether both tables contained the same hashed email in the agreed period. It could also ask whether a group built from that intersection saw an ad, then bought the product. For the join, get the inputs and the match terms. For the ad buy, the team also needs the ad logs, which sales count and what to compare them with. Ask what the brand can check and what the store can share only as a sum.

### Close the loop in the media order

Write the promised return next to the destination. Disney’s 2023 LiveRamp destination terms define a custom segment through overlap between advertiser data and Disney data, and require the submitting party to warrant rights and consents for its data [303] [303]. Choose a firm that can give that warranty to send the list. Give the trader the name of the group and where it will go. Tell the sales team which sums it will get and which dates they cover. If a third firm holds the seat, who can get the proof once the ads stop?

Go back to the store that wants its old shoppers back. Keep a dated note of who made the list and on what terms. Keep the name of the group, its state in the seat and the terms for the sales report. Ask the store to log when the list changed. Which ads first used the new list? Have the agency name the flight. Ask the sales team what a new trip means and which trips it will count. Which comparison could show how much the ads helped? If the site can share only group sums, settle the buy on those terms. Let the club team retain the questions still open about its own members.



## 7. CTV and households


*What can a TV report tell a brand about site visits from homes that saw its ads? Spectrum Reach says it supplies the TV ad exposures for each home [38, 167]. Innovid supplies the web visits, and Blockgraph matches the two [38, 167]. The join leaves open who watched and who went to the site [38, 167]. For a buy aimed at homes, ask how the set came to belong to that home. When would the firm stop using the link?*

> **In short.**
> Blockgraph asks for a household record ID and a postal address [109]. A firm can add an IP or set-top-box ID [109]. The guide sets a time span for a search of past IP links [114]. Only Blockgraph met the US buying test [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. Its guides gave no age limit for the live link [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. LtvAdx does give an age rule for a device link [227, 228, 229, 230, 231]. It sells ads aimed at US homes, but the terms do not show a right to buy the join on its own [227, 229, 230, 231]. The study could not settle the market-wide result [14, 13, 109, 114, 115, 227, 228, 229, 230, 231].

### Start with the home someone paid to match

What can a firm buy? In 2020, Blockgraph’s general manager described a fee to join its network [171]. Its 2021 notice named the US as the market for the TV match service [170]. In February 2022, TVSquared told NBCUniversal it used Blockgraph’s Identity Operating System to link data across kinds of TV [168]. They show a route to buy the service, a market and a firm that said it used the match [170] [168] [171]. TVSquared named other firms too [168]. It used hashed IP as a fallback but did not tie that step to Blockgraph [168].

Viant says its DSP uses data from the bid to find a household profile [312, 313]. The Household ID itself does not travel in the bid [312, 313]. Cadent names its IP-to-home Viewer Graph in a guide to contextual targeting and in its US CTV media offer [306, 314, 315] [306]. FreeWheel describes a built-in network with identity signals from Comcast and IDs from its publisher partners [305]. Buyers can connect their own data too [305]. Ask what the deal gives the brand. Can its team take home the links? Can the team reach a group only in the seller’s seat? Will the brand get a report once the ads run? Write a test for the job on the order.

The name on an old press release can be a poor guide to a new buy. In January 2020, Skydeo announced a deal to map mobile data to the Tru Optik Household Graph in the US [310]. In 2021, TransUnion described its identity service across channels and named the former Tru Optik market for streaming data as a separate offer [246]. TransUnion and Epsilon later named the TruAudience Data Marketplace as the way to buy Epsilon groups for connected TV [245]. One account dates a deal to map data [310]. Another names the route a buyer can use to reach homes [245]. The terms the study found still leave open how to buy the TV household links on their own [245, 246, 247]. List the rows the buyer wants and ask who has the right to take them away.

### Which observation belongs to which home?

Blockgraph says a home is tied to its postal address [166]. In version 2.29, a firm needs a record ID and a postal address for the home [109]. It can add an IP address or an ID for a set-top box [109]. Start with the firm that sent the row. Did the street and unit number come from an account it runs? Which fields came with the TV event to be joined? For a test, try two flats on the same street and at the same number. Use a new record for a customer who moves from one flat to the other. What would the firm do if both rows claimed the same set? Ask to see the result for each row and the rule that picked the home.

In August 2026, Blockgraph added a preset for SHA-256 hashed email [116, 172]. A firm could use it in an upload or for a batch match [116, 172]. The release did not give a count of new homes or a test of wrong links [116, 172]. To check the effect, give both builds the same test homes. Keep each result with the inputs the firm used to make the match. Did an address and the firm’s own record of a set both point to the same home? Could the match also be made from a shared IP alone? If more rows went through in the new build, which input made the difference for those rows?

CIMM describes return-path data from set-top boxes and automatic content recognition, or ACR, from smart TVs [6]. Its example of ACR checks that the ad was shown on a set that was on [6]. First ask what played. Get the time, the ad and the source that saw it. Then ask which home should get the ad view. Which rule puts the set in that home at that time? If the claim names a person, ask who can place that person in front of the set. Suppose two people share the account. Either one might have watched the ad. The home’s row could look the same in both cases.

### Time can qualify a link or end it

Viant says it checks the type of internet provider, the device count and the time of day when it screens a home IP [304]. It uses a location-enabled device’s coordinates to match to a home address on file [304]. Its US patent gives an example in which an IP assignment must be more than 45 days old to go to the next step [311]. That is a test for entry [311]. The 45-day example gives no expiry date for a link after its last sighting [311]. Its use in Viant’s live product remains unconfirmed [311]. Ask which rule is in use, who set it and where the buyer can see it work.

Try both clocks on the same case. Say an address must have been assigned for more than 45 days to pass a test. At 46 days, it might pass. Now take a rule that drops a device 30 days after the last time the firm saw it. If that was 46 days ago, the link would fail the test. A count in days needs a start event and an action at the end. What can set the count back to zero? Could a phone on the same IP keep a TV set’s link alive? Ask when the firm that uses the link would first see the change. To check an ad that ran last week, ask which clock is in force at the date of the ad.

LtvAdx describes a shared-IP fallback to find a home and says it drops a device link 30 days after the device was last seen [227, 228, 229, 230, 231]. Its Identity API can return a known HouseholdID or make a new one [227, 229, 230, 231]. On the VAST ad route, IP fallback is off by default [227, 229, 230, 231]. It needs exactly one known home at that IP [227, 229, 230, 231]. For a test of the ad route, try a case with two known homes at one IP. For a test of the device link, keep the home fixed and vary the last-seen date. Ask what each test should return. The record has the firm’s account and no independent test of how the firm applied the rule [227, 228, 229, 230, 231].

### Rebuild, look back, count forward

On Blockgraph, a firm can keep a history of the signals for each home [114]. It can then look up an IP from the past with a time on each row [114]. The span to search can be 30, 60 or 90 days [114]. Suppose an IP comes in today with the date of an ad that ran last month. How far back can the team search? The span in the guide answers that question [114]. It leaves the age for a link in current use open [114]. Try a test with a move in the middle. When the ad aired, the address belonged to the first home. Which home is returned if the same IP is looked up today with the date of the ad?

Blockgraph’s version 2.29 guide says a full manual refresh replaces the identity layer at that node with a new file [115]. Rows left out of the new file drop out of that layer [115]. The client can start the change when it wants, with no stated age limit for each link [115]. The guide also lets the client delete a record or opt it out [115]. Old groups and matches stay available after a delete [115]. An opt-out marks them stale, and they must be refreshed before further use [115]. For an agreed test row, ask to see the old match, its new state and the step a team must take to use the group again.

Ask when a file arrives and what gives each row its date. LiveRamp’s guide to IP-based CTV events recommends new data each day or week once the programmer is set up [307]. It can look back up to six months for past events [307]. Factori says it links encrypted mobile ad IDs and IP addresses and sends a feed each day [308, 309]. Its AWS listing offers a contract for the data and describes global scope with no US breakout [308, 309]. If a file is due each morning, a buyer can check the time it lands. To check how old a link is, ask when the device last gave the firm cause to put it in that home. A row could stay in each new file for a week, with no new sighting through the week.

### A visit in the week after a TV spot

Spectrum supplies the TV ad exposures for each home, Innovid supplies home visits to a site and Blockgraph does the match [38, 167]. Spectrum’s showcase counts a visit within seven days of a TV ad view [167, 38]. The clock starts from the TV ad exposure and sets a time for the visit to count [167, 38]. It gives no term for the life of the match to the home [167, 38]. Criteo describes a shared-home fallback when it cannot get an exact match [204]. Its TV guide gives a sale up to 30 days to count after an ad view [204]. Suppose a trader gets more visits in a revised report. First check whether the team let later visits count. The report could change without a change to a single home link.

Ask what happens if two ads ran in one home before a web visit. Will both ads get credit? Does the report count site visits, homes with a visit or all the homes the spot reached too? Try a test family with two spots, followed by two site visits from two people. Write down what you expect in each column. To see how many visits the ads caused, set up a fair test of what would have happened with no ad. Spectrum’s report puts the TV view and site visit in the same home [38, 167]. It does not say who watched or what caused a visit [38, 167]. It does not test whether the link to the home was right [38, 167].

Spectrum says its own data across screens comes from more than 30 million homes [38]. It claims a 95% addressable match rate for its own pay-TV distribution service [38]. A CIMM report from April 2025 says Blockgraph’s system works across more than 130 million homes [6]. It gives no region, time span, rule for repeat homes or base to which that count applies [6]. How many US homes could one brand reach? None of the figures gives that count for a specific buy [6] [38] [38]. Start with the homes served by the named TV firm. Which qualify for this deal? How many could the ad reach in its planned run? Which figure would change if the brand chose a different TV firm?

### More matches can change who is in the sample

Kantar and Blockgraph say Kantar’s match rates with TV partners rose to more than ten times their old level after the firms linked their systems [112, 113]. The firms gave neither rate, the counts that fed it nor a test of false links [112, 113]. Blockgraph’s CEO discussed what a match rate means but did not say what Kantar counted in the case [113, 112]. Both accounts cover the same work [113, 112]. Did each firm send the same set of rows both times? Did the base count rows, homes or members of a panel that qualified for the test? To test the added links, take a sample of those that failed the first time and passed the next. Ask for a second source that can check them.

CIMM and Go Addressable hired Truthset to study US links from December 2024 through February 2025 [40, 41] [40, 41]. For links from IP to postal address, the mean modeled accuracy across six unnamed firms was 13% [40, 41]. For IP to email, the mean for the six was 16% [40, 41]. Each type of link had its own base [40, 41] [40, 41]. For postal links, there was a large pool of pairs from the firms and a smaller set of checks from operators [40, 41]. Keep those groups apart when using the figures. The study does not give the score for a named graph on a media plan [40, 41] [40, 41]. A seller that cites a better rate for its own graph should show which links it tested and how it knew which ones were right.

In another CIMM study, five masked firms matched two sources of TV data [42]. The mix after the match skewed toward older people, higher incomes and heavier TV viewing [42]. The study did not have an independent set of labels to check each link [42]. Suppose that every link a firm kept was right. Say homes that watch less TV still make it through the match less often. A plan based on the matched homes could give too much weight to heavy viewers. List the mix of homes that went in and the mix that came out. Test the links on a separate set with known homes. Ask the planner to show how the proposed buy would change if it kept the viewing mix of the full input.

### What the disclosure test can bear

H5 asked whether most US CTV household-graph services disclose how IP helps assign a device to a home and when to refresh or expire that same link [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. Only Blockgraph met the US buying test [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. Its manual lists IP as an optional signal, and its lookup windows apply to past IPs [109] [114]. The guides reviewed gave no time rule for the live link [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. The registered majority claim was contradicted in the confirmed group (0 of 1) [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. When all 8 other candidates of uncertain scope are included, the share can range from 0 of 9 to 8 of 9 [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. That range leaves the US market result open [14, 13, 109, 114, 115, 227, 228, 229, 230, 231].

LtvAdx could change the count. It sells US ads aimed at homes and charges for the ads served [227, 229, 230, 231]. It gives both a shared-IP fallback and a time rule for the device link [227, 228, 229, 230, 231]. The offer did not show a separate right to buy the home match, so this case stayed out of the confirmed group [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. If only this case were added, the share that met both tests would be exactly half [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. The claim calls for more than half [14, 13, 109, 114, 115, 227, 228, 229, 230, 231]. A contract giving the buyer a right to the home match would close the gap on scope. Another price per ad would leave it open. A time rule from a firm already in the group could change the answer even if the set of firms stayed the same.

For a deal on the table, ask for a trace from one known TV ad view to the row the buyer gets. Blockgraph’s Universe Match guide lists partners that gave leave to match [110, 111]. It describes a file that holds one column of Blockgraph IDs [110, 111]. If the buyer needs proof of why one of those IDs was chosen, ask for that with the file. Which firm can show the home address and the device signal under the terms of the deal? How long ago did the firm get the evidence it used? Who can change the link when a customer moves? Which old reports could be fixed? Put the answers with the name of the firm that can act. A brand buying reach and a team asked to stand by a sales report might need to see very different things.



## 8. Market and ownership


*Publicis closed its purchase of LiveRamp on November 2, 2026 [141, 318]. The buyer described a new offer that uses both RampID and Epsilon identity [216]. Publicis markets both in that offer [216]. It still needs to know which service it will get, who can use the output and what the fee covers.*

> **In short.**
> ID5 offers free encrypted IDs to publishers that register [56, 351]. A platform needs a licence for commercial use of those IDs [56, 351]. A buyer can target its own CRM list in a Trade Desk account [188, 203]. That route gives it no UID2s or map to take home [188, 203]. With Experian, the buyer can license the whole US Digital Graph or have links added to its own file [117]. Count what the buyer gets. A budget for reaching people and a budget for a set of links that a buyer can reuse may pay for quite different things.

### Follow the deal to the party that signs

On May 16, 2026, LiveRamp signed a deal with MMS USA Holdings [317]. The deal closed on November 2 [141, 318]. MMS, which Publicis owns in full, bought LiveRamp Holdings [141, 318]. Trading in the shares stopped before the New York Stock Exchange opened that day [141, 318]. The filing set a cash price of $38.50 for each share that met its terms [141]. Read the filing for the shares it leaves out [141]. Treat the figure as a price per share. To price a match on a client file, a media buyer needs a quote for the work.

Check the name on the US terms. The group’s terms name LiveRamp, Inc., a Delaware firm with its main place of business in San Francisco [345, 141]. The merger filing names the parent, LiveRamp Holdings, Inc., as the acquired firm [345, 141] [141, 318]. Start with the name beside the signature on the deal you signed. Use the merger record to trace the owner. Then check who undertook to run the service. If the party to the deal has changed, who still owes the buyer support? Who must deliver files already due? Who will act on a request sent to the old firm before the deal closed?

### Three purchases and one new offer

The seller filed July 1, 2019 as the date it sold Epsilon to Publicis [319, 320]. Publicis announced the close the next day [319, 320]. In March 2025 the group announced a deal for Lotame [214, 215]. Its accounts say it bought all of Lotame Solutions in April [214, 215]. The November 2026 closing statement named RampID and Epsilon identity as inputs to a new client offer [216]. A buyer can now trace these names to one group [141, 318] [214, 215] [319, 320]. To find out what a client can buy from that group, return to the terms of the offer.

Suppose an agency gets bids from two firms in that group. Does its risk policy set a cap on how much it can pay one owner? If so, put both bids in the same group for that check. For a test of what each bid buys, keep them apart. What goes in? What comes back? What rights go with it? If both firms bid to match the same client file, ask which rows each one could link. For the rows both firms linked, ask how they did it. Did they draw on the same source? Did they share a link? Did each make its own match? The client could then weigh the extra reach from the second bid against the risk of paying twice for the same links.

### What LiveRamp booked

LiveRamp booked $812.9 million in group revenue for the year to March 31, 2026 [7]. It had one operating segment [7]. It split sales into Subscription and Marketplace and Other [7] [7]. For the quarter to June 30 it booked $214.0 million [7, 344]. Keep the year and the quarter apart. Both cover more than RampID [7] [7, 344]. On November 10, LiveRamp told the SEC that its September report was late [318]. It had to review its link to its auditor after the deal [318]. A reader looking for the missing quarter gets no set of results from that notice [318].

To price ATS, turn to the terms for the publisher. LiveRamp limits use of its envelope to the parties enabled under a statement of work [352, 353, 345, 354]. Its service terms put any fees in the order form [352, 353, 345, 354]. In 2021 its CEO said the firm did not plan to charge publishers or media-platform partners for ATS [352, 353, 345, 354]. It expected more use and sales of other services [352, 353, 345, 354]. Ask for today’s order form. Which step has a fee? Which party pays it? Can the buyer use the same envelope if it moves the buy to a new bidder? The answers would give the buyer a cost for the route it plans to use.

### An operator can be free to use

Under the UID2 guide, a party with a contract can use a Public Operator at no charge [188, 203]. Those who run a private operator supply their own hosting and engineers [188, 203]. The guide names The Trade Desk as Core Administrator and a public operator [142]. The core supplies keys and salts [142]. The rights also vary by route [188, 203]. The licence matrix lists approved API access and CRM onboarding [188, 203]. With CRM onboarding, a buyer can target its list in the platform account [188, 203]. It does not get the IDs or the map back [188, 203]. Ask each bidder to price the right in the brief. Can the client take the map and use it with a second platform?

For its 2025 financial year, The Trade Desk booked $2,896,284,000 in worldwide group revenue [331, 346]. It reported one ad tech segment [331, 346]. Gross spend includes clients’ ad inventory, data and value-added services, plus the platform fee [331]. The filing gives no UID2 sales figure [331, 346]. Suppose a buyer pays one bill for a campaign. Suppose the platform fee is part of that bill. A spend survey should ask whether the buyer gave the full bill or the fee on its own. Add both and part of the bill gets counted twice. To assign all of it to an ID, the buyer would need to show why none of it paid for the ads.

### Capital, control and licences at ID5

The UK register names ID5 Technology SAS of France as the firm that controls ID5 Technology Limited [339, 340]. It holds at least 75% of the shares and voting rights [339, 340]. In April 2024 ID5 named TransUnion and S4S Ventures as investors in a $20 million Series B [341]. Its UK firm had £702,598 in cash at bank and in hand at June 30, 2025 [340]. It left its profit and loss account out of the public unaudited filing [340]. Keep these records apart. The UK register shows who has control [339, 340]. The funding release names two investors [341]. The cash figure applies to the UK firm on one date [340]. To work out how long the group could fund its work, ask for group cash and the rate at which it spends that cash.

ID5’s guide offers publishers a way to get an encrypted ID at no charge [56, 351]. Its terms for paid use include the right to decrypt an ID, link two IDs, or use ID5 data [56, 351]. Ask which of those rights come with the quote. A site might want to send a token to a bidder. A data firm might want to read the ID and link it to a row in its own file. A team might log these as two new links. To count sales, ask who pays a fee and what they can do once they pay it. If the licence ends, can the firm still use the links it has on file?

### What moved with the TrueData name

ID5 dated its statement on buying TrueData October 30, 2025 [132, 217]. It posted the English text on November 4 and later said it had closed the deal [132, 217]. The next deal brought in a new buyer [342]. On September 30, 2026, Cloud Technologies agreed terms with TrueData Solutions Inc. for a new US firm [342]. The terms set $1.1 million for its shares and a technology licence, subject to conditions [342]. On October 1, ID5 said the deal covered TrueData’s third-party audience business and the TrueData brand [217]. Some client contracts had to move before the sale could close [343]. Cloud Technologies reported that it bought all the shares in the new firm on October 5 [343].

For a client, the first question is which contract moved. Start with the SKU and the name on the last bill. Who now ships the audience? Who runs the graph that helped build it? Who can end the right to use it? Ask to see the scope of the licence if the new firm relies on code it does not own. A deal can move a brand, rights to use code and a set of client contracts at once. Here, the firms named an audience business, the brand and a licence [217] [342]. Those terms do not assign the whole of ID5’s graph business to the buyer [217].

### A full graph or a set of appends

Experian bought Audigent on December 4, 2024 [321, 322]. Its annual report says it bought all the equity [321, 322]. The group parent is based in Dublin and listed in London under EXPN [337, 338]. For the year to March 31, 2026, it reported worldwide statutory revenue of US$8,445 million [337, 338]. The US Digital Graph sheet gives its copyright to Experian Marketing Services, Inc., in its footer [119, 336, 337]. The US notice on marketing data names Experian Marketing Solutions, LLC [119, 336, 337]. Both names are worth keeping. Neither tells the buyer who will sign its graph licence [119, 336, 337]. Get the draft deal. Does the contract the buyer has on file cover this job too?

With the US offer, a buyer can take the whole graph or links added to its own file [117]. Experian also sells an Offline Graph [120, 121]. A further offer pairs Digital Graph with Marketing Attributes [120, 121]. Experian says MiQ licensed hashed emails, UID2s and partner cookies for use in its Identity Spine [21, 118]. The case says MiQ traders and analysts use the data [21, 118]. The case has no price for the deal [21, 118]. To compare a new bid, ask how many rows the buyer can send. How often can it get a new file? Can it keep the full graph or just the links added to its list? Price the form of delivery that the draft terms allow.

### Two names with histories of their own

IPG bought Acxiom assets in October 2018 [325, 326]. The deal left out LiveRamp [325, 326]. Omnicom said it closed the Interpublic deal on November 26, 2025 [327, 328]. It later named Acxiom RealID in its media plans [327, 328]. Publicis closed the LiveRamp deal in November 2026 [141, 318]. If an old client list is headed “Acxiom,” ask what year the list covers. Get the name of the product that served the client. For an old deal, check who could pass the rights to a new firm. An invoice can show that a buyer paid the then owner. Ask for the new order or the terms that moved it. Which firm now earns the fee?

TransUnion and Neustar said their $3.1 billion deal closed on December 1, 2021 [323, 324]. It left out Neustar Security Services [323, 324]. In January 2024 TransUnion said it combined data from both firms [316]. It grouped IDs by person and household and scored the links [316]. That later account gives a buyer a claim about how the graph was built [316]. Use it when asking about input sources and why the graph might accept a link. For a TV proposal, first check which part of the business made the offer. TransUnion and Epsilon named TruAudience Data Marketplace as a way to buy Epsilon audience data for connected TV [245]. Ask for an extract if the brief calls for a set of household links the client can use elsewhere.

### The parent on the date you need

Apollo’s funds bought Yahoo on September 1, 2021 [329, 330]. The close left Verizon with a 10% stake [329, 330]. Verizon reported $4.1 billion in cash proceeds for Verizon Media, net of cash transferred [329, 347, 348]. In September 2025, Yahoo Ad Tech LLC named College Parent, L.P. as its corporate parent in court [329, 347, 348]. Its US ConnectID terms name Yahoo Ad Tech LLC as the contracting party, with its main office in New York [160]. Keep the 10% figure with the date of the close. The net cash figure covers Verizon Media [329, 347, 348]. To value ConnectID on its own, the buyer would need a breakdown the sale record does not provide [329, 347, 348].

The US terms describe a value sent back under a key specific to the firm or the publisher [160]. It is encrypted, and the right to share it is tied to the supply deal [160]. Ask which deal the quote covers. Which other supply paths would need fresh terms? A buyer who wants to use the output with two partners should put both in the brief. Ask if the licence permits this use. Run a test. Can the other firm use what it gets? If the brand might move its spend to a new bidder next year, ask whether it could take its rights with it.

### Shared ownership in television

In May 2020, Comcast, Charter and ViacomCBS said they had agreed to own equal shares of Blockgraph [332, 333]. Their joint release dates the first FreeWheel project to 2017 [332, 333]. Blockgraph later named Comcast NBCUniversal, Charter Communications and Paramount as its owners [334, 335]. Comscore’s 2025 filing called Blockgraph private [334, 335]. Keep each set of owner names with the date of its source. They leave current shares open [332, 333] [334, 335]. In February 2020, Blockgraph’s general manager said firms could join the network with a software subscription [171]. The price of a licence today is not in that record [171]. Ask for the term, fee and rights that apply to the next buy.

What does a buyer get when it joins the network? In its Universe Match guide, Blockgraph lists the partners that have granted access [110, 111]. The guide describes a file with one column of Blockgraph IDs [110, 111]. Spectrum Reach describes a different job [38, 167]. It supplies TV ad exposure by household, Innovid supplies site visits by household, and Blockgraph makes the match [38, 167]. A buyer might want to check its own homes against one partner’s list. Or it might want a report on the homes that show up in both an ad file and a site file. Request the actual output and the name of the partner that can use it. To claim access to an owner’s data, ask for terms that grant it.

### What changes hands in a substitute buy

Seedtag offers groups as a way to reach an audience without third-party tracking [355]. It scores URLs against the brand’s brief [355]. Its October 2026 US political offer gave a Michigan example [349]. That example paired Seedtag’s context segments with voter audiences from Aristotle [349]. If the seller can split the quote, ask for two prices. What does it charge to select content for the brief? What is the extra charge for the voter audience? If only the package has a price, ask what the buyer could drop and how the bill would change. Do not assign the full price to both firms. In that example, the buyer would pay the package price once.

Ogury lets brands and agencies plan and buy persona groups on its platform or use its API [210]. NIQ names the US among the markets it serves with purchase insights for Ogury’s persona engine [356]. Ask the planner for the name on the bill. Then ask for the names of firms whose data helped build the groups. Did each send the client a bill? Or did one firm charge the client for the whole buy? One list can show how a group was built. The other can show where the money went. Use the two lists to map the firms. When a source shows up on both, has a second sale taken place?

A buyer can test the scope of a quote with a small change in the brief. Keep the same input file, run date and set of IDs. In the first bid, ask for a single file back. In the next, ask for the right to use that file with two named firms. Ask for the same service term in both bids. Has the fee moved? If so, which right set off the change? If there is no change in the fee, does the draft still grant the right? Read the terms. Does the price include the right to send the file on? Keep the two quotes with their terms so the team can go back to the exact right it chose.

### Put a boundary around the budget

Set a boundary for any estimate of market size. Start with the sum the buyer paid. Was it for software, a file, a set of links, or the media on which those links were used? More than one item can fall under the same fee. Ask how the buyer splits it. If it cannot split a bundle, keep the whole sum on one line of its own. Note the time span and currency. Check whether the same sum appears in a reseller’s gross bill and in its supplier’s sales. Are we trying to measure what brands spent? Or are we trying to measure fees that a vendor booked? State which question the estimate will answer.

Ask what would happen to the budget if the buyer dropped the ID. Comscore offered US political buyers a choice on The Trade Desk [350]. One option used an ID and the other was an ID-free predictive audience [350]. Which line in the plan would the new buy replace? If it takes the place of another media buy, compare the spend and the reach that each would get. If it takes the place of a file licence, compare that fee plus the cost to put the file to use. Leave a fee of unknown size open. The public filings give group totals and leave US sales of the named identity products open [7] [331, 346] [337, 338]. A buyer can still ask what it will pay for the right it needs.



## 9. Regulatory and legal record


*In France, Google changed a notice shown to people who open an account [80, 79]. It also changed how ads looked in Gmail [80, 79]. In September 2026 the French regulator closed its order on both points [80, 79]. Could we trace such a change for the firms that sell US ad IDs? The records we reviewed did not link an order in force to a change that a separate source confirmed for the named product [14, 13, 65, 68, 70, 79].*

> **In short.**
> The French court upheld a sanction against Criteo [64]. The partner audits cited in the case had begun while the probe was still open [65]. Mobilewalla told investors it had applied a third-party location filter in its US business after an FTC order [71] [72, 73]. Its own account did not show that a named identity product had changed [72, 73]. Read the order to see who must do what. To find out what the service does now, look for a test. The product-change claim, H6, stays open [14, 13, 65, 68, 70, 79].

### Read the duty, then look for the change

Did a case cause a change in a named ID service? That is the test for H6 [14, 13, 65, 68, 70, 79]. The search starts on January 1, 2022 and runs through November 11, 2026 [14, 13, 65, 68, 70, 79]. It covers both current offers and past services that were active in the period [14, 13, 65, 68, 70, 79]. Some have a listing but no score [14, 13, 65, 68, 70, 79]. A team might know what the law requires. It may still lack proof of how two devices were linked. Keep the legal finding out of the product score. Read it for what the party named must do.

Find the part of the order that sets the duty. Who must act? Which data fall within its terms? Where do those terms apply? What starts the clock? Next look for a dated account of what changed after the order. Who saw the change? Which version did they test? Could they see what left the service? If the duty was to change a notice, a check of what the user saw could meet the test. If the claim is that a graph changed, seek a check of the link. Keep the product name next to the name of the firm. A case can tell a buyer a great deal about a screen or a device setting and still leave the graph in doubt.

**Advertising identity in the legal record**

| Jurisdiction | Official record | Requirement, if any | Effective date, where applicable | Scope of the record | Type of record | Src |
|---|---|---|---|---|---|---|
| France / EU | CNIL SAN-2023-009 and Conseil d’État 482872 | Retargeting operator sanctioned. The appeals court upheld the €40 million sanction. | 2023-06-15 decision; 2026-03-04 appeal dismissed | CNIL examined Criteo’s partner-site cookies and retargeting. It recorded partner audits during the investigation. The decision and appeal do not establish a later change to ID for Exchanges. | Administrative sanction and appeal judgment | [63, 64, 65] |
| United Kingdom | Information Commissioner v Experian, [2024] UKUT 105 (AAC) | Substituted notice requires specified transparency for open-source data used in direct marketing, subject to exceptions. | 2023-02-20 substituted notice; 2024-04-22 appeal dismissed | The UK ConsumerView direct-marketing database is named. Experian projected compliance by July 24, 2025. Separate verification of implementation and a change to the US Digital Graph remain open. | Substituted notice; appeal judgment | [66, 67, 68] |
| United States / Idaho | FTC v Kochava, 2:22-cv-00377-BLW | The stipulated order requires withdrawal controls for CDS’s use or disclosure of precise device location. | 2026-06-25 signed and filed order | The complaint describes household mapping. The order imposes duties on Collective Data Solutions for precise device location. The record does not show a separate household-graph change. | Amended complaint and stipulated injunction | [69, 70] |
| United States | FTC v Mobilewalla, docket C-4811 | Limits secondary harvesting in auctions. Sensitive-location and other controls under distinct provisions. | 2025-01-13 issued | Mobilewalla’s SEC filings describe its own US data-screening program. The filings do not independently verify compliance or establish a change to a separately identified identity product. | Final consent order; subsequent issuer representation | [71, 72, 73] |
| United States / California | People v Disney DTC LLC and ABC Enterprises Inc., 26STCV04425 | Effectuate logged-in opt-outs across associated streaming services; different instructions when not logged in. | 2026-02-11 judge-dated; the order takes effect upon entry | Disney streaming services associated with a logged-in Disney account. | Stipulated judgment | [74, 75] |
| United States / Texas | Texas v Sony, original petition | State requested relief about smart-TV data controls. Figure 5 identifies the Samba Interactive TV opt-out screen. | 2025-12-15 petition; no operative duty inferred | Texas named the Sony entities. The petition depicts a Samba Interactive TV screen. Texas reported the case ongoing on May 11, 2026. The filing establishes no change to the Samba household graph. | Complaint and requested relief | [76, 77] |
| United States / Texas | Texas v LG Electronics U.S.A., Inc., Bell County, 169th District Court, 25DCV358507 | The agreed terms call for affirmative express consent for covered viewing-data practices beginning 180 days after judicial signature and entry. | 2026-05-11 settlement announced. Judicial signature and entry date unverified. | Texas announced a settlement concerning LG viewing data. The posted agreed terms lack a verified judicial signature or entry date. Neither the trigger date nor a product rollout is established. | AG-announced settlement and published agreed judgment terms | [78, 77] |
| France | CNIL Google account and Gmail decision and closure | Article 82 limb concerns account-creation advertising-cookie notice; the regulator closed the injunction after review. | 2025-09-01 injunction; 2026-09-23 closure | Google account creation in France and Gmail. Distinct from the Chrome API entries IDN-P035–037. | Operative injunction and verified regulatory closure | [79, 80] |
| Belgium / EU | IAB Europe TCF decision and Brussels Market Court ruling | The regulator records the procedural annulment of decision 21/2022, confirmation of its fine, and a limit concerning downstream OpenRTB operations. | 2025-05-14 judicial outcome as reported | IAB Europe and the TCF; consent-framework boundary. | Appeal ruling reported by regulator | [81, 82] |
| France | Conseil d’État 494300, Yahoo EMEA Limited | The court rejected the company’s appeal concerning the cookie-placement and withdrawal practices in the decision. | 2025-10-07 judgment | Yahoo website practices; commercial IDN-P005 ConnectID was not identified as the subject in the passages screened. | Appeal judgment | [83] |
| United States | FTC X-Mode Social and Outlogic decision and order | Sensitive-location program due within 180 days of issuance. | 2024-04-11 issued | X-Mode and successor Outlogic; sensitive location-data program. | Final consent order | [84] |
| United States | FTC InMarket Media decision and order | Prohibits sale or licensing of defined Location Data for valuable consideration. | 2024-04-29 issued | InMarket Media, LLC. Separate provisions address targeting and categorisation. | Final consent order | [85] |
| United States | FTC Gravy Analytics and Venntel, C-4810 | Sensitive-location program must be established within 90 days of the effective date and maintained thereafter. | 2025-01-13 issued; time runs from effective date | Gravy Analytics, Inc., and Venntel, Inc. | Final consent order | [86] |
| United States / Federal Trade Commission | FTC, Publicis Groupe / LiveRamp, C-4841, file 261 0115 | The issued Order to Maintain Assets requires continued access to LiveRamp standalone services and safeguards for confidential information. The separate proposed five-year remedy was subject to public comment at cutoff. | 2026-10-19 interim order issued. A separate proposal was posted for comment through 2026-11-23. | Publicis and LiveRamp, competition and supply access. The interim order protects standalone services and confidential information. The record does not independently verify a completed RampID data-policy change. | Operative interim asset-maintenance order and separate proposed consent remedy | [87, 88, 89, 9] |
| United Kingdom | Open Rights Group complaint to UK Information Commissioner’s Office concerning LiveRamp | The complainant reported a submission. No obligation imposed by the authority is established by that submission. | Complaint reported 2024-02-28. No operative outcome verified. | Open Rights Group reported the submission about European RampID processing after commissioning the Cracked Labs review. No linked regulator ruling or separately verified product change was found. | Reported complaint, no verified operative disposition | [60, 61] |
| France / EU | Open Rights Group complaint to French CNIL concerning LiveRamp | The complainant reported a submission. No obligation imposed by the authority is established by that submission. | Complaint reported 2024-02-28. No operative outcome verified. | Open Rights Group reported the submission about European RampID processing after commissioning the Cracked Labs review. No linked regulator ruling or separately verified product change was found. | Reported complaint, no verified operative disposition | [60, 61] |
| United States / Southern District of New York | Baker v. Yahoo Inc., 25-cv-2797, orders 103 and 104 | The court sent five plaintiffs’ claims to arbitration and stayed those claims. It separately directed consultation about limited discovery for the sixth plaintiff. | 2026-08-26 orders issued | Private procedural case. The orders resolve no merits issue and establish no change to ConnectID. Outside the H6 public-enforcement population. | Private suit, arbitration and procedural orders | [90, 8] |

### France and the Criteo retargeting decision

In June 2023, France’s CNIL ruled on requests to erase data used for ads on partner sites [63]. It found that the firm had stopped the tailored ads but had yet to meet the requests to erase data [63]. On March 4, 2026, France’s Conseil d’État rejected Criteo’s appeal against the €40 million sanction [64]. The court dated some of the firm’s changes to the time before the sanction [64]. The case summary says its checks on partner sites began during the probe [65]. The dates of those checks put them too early to show a change the firm made under the June 2023 ruling [64] [65].

How could a publisher check the result of an erasure request? Suppose the next page load shows an ad selected without retargeting. The test would show how the site chose that ad. What happened to the data held before the page load? The operator may have dropped a link or kept it and changed how it served the ad. For this test, trace the request to the keys it covers. Inspect the record held before the request and again once the request has taken effect. For a claim about an exchange ID used in the US, find the US route that uses those keys. The French case did not show a later change to that exchange product [14, 13, 65, 68, 70, 79].

Put the two dates on a line. When did the firm start the work? When did the duty take effect? Ask for a third date if the firm went back to the same tool after the ruling. It may have changed which sites it checked, what the check found or what it did with a failed site. A test from that later point could pick up a real change. A copy of the old test could show only that the tool was still in use. If the dates are clear but the scope of the change is not, leave that part of the case open. This keeps work done before an order in view without giving the order credit for it.

### The UK notice that actually bound Experian

Which notice bound Experian in the UK? The First-tier Tribunal changed much of the Information Commissioner’s 2020 notice in 2023 [67]. Under the new notice, Experian had to tell some people that it used their data for direct marketing [66]. The duty had exceptions and applied to records from the open sources it named [66]. On April 22, 2024, the Upper Tribunal dismissed the Commissioner’s appeal [67]. In June 2025, the firm told investors it thought it would meet the new terms by July 24 that year [68]. Did the firm meet that date? The records we read did not say [68]. Nor did they show a change to its US Digital Graph [14, 13, 65, 68, 70, 79].

For a check of the UK service, ask which rows require a notice. If the firm says a row is exempt, what grounds does it keep on file? Ask for a test dated after the rule took effect. Trace one covered row and one row to which the stated exception applies. That could show how the firm got word to a person in its UK file. For a check of a US graph, start with a US input. Obtain the link that came back. What rule changed, when did it change, and what ties that change to the UK notice? The litigation identifies the place and the direct marketing use at issue [66] [67, 66, 357]. It leaves the US graph question open [14, 13, 65, 68, 70, 79].

### A household use in a complaint

The FTC filed an amended complaint against Kochava in July 2024 [69]. It said the firm had pitched its feed on the claim that the data could help map a device to a home [69]. Use that claim as a lead. The complaint did not decide how well the feed linked a device to a home [69]. A US order in June 2026 bound Collective Data Solutions [70]. The firm had to give users a way to take back consent to use or share precise data on where a device had been [70]. The order lets the firm use a qualifying device or operating-system control for that purpose [70]. If the terms it sets are met, the order also reaches Kochava [70]. To check if Kochava IdentityLink changed, ask for a test. The order does not show what it did [70].

Take the location feed first. Use a test device with consent. Change the setting, then check where the new state reached the firm. What was in the last feed before the change? What was in the first feed due after it? Keep the field name, the time and the name of the party that got the feed. Get leave to run the test. Use a row the team knows. Now ask what it would take to test a graph. Was the device linked to a home in the first place? Did the link last after the change? Could a fresh file or a new source have changed the link that day? Who holds the graph? The buyer may see the first feed and still need that firm to check the link.

### What Mobilewalla said it changed

In January 2025 the FTC limited Mobilewalla’s collection or retention of covered data from online ad auctions [71]. It could collect or keep the covered data to take part in that auction [71]. The order defines which data it covers [71]. In August 2026 Mobilewalla told investors it had applied a third party’s sensitive-location filter in its US operations [72, 73]. It said the filter checked signals on the way in and files on the way out after the order [72, 73]. Was that filter tested? The record we saw has no independent test or FTC finding on it [72, 73]. It also leaves open which named identity product changed [14, 13, 65, 68, 70, 79].

Ask what the filter sees. Which feed comes in? Which file does the client get? What happens to a row that fails? The firm might drop the row. It might drop just the location field or hold the row for review. Which of those paths fits the file the buyer got? Seek a test from the time the rule went live. Ask who could see the row and the outgoing file. If a third party ran part of the check, which part could it see and what did it report? A later filing might repeat the same account. That would show what the firm continued to tell investors. Who could see the file that reached the buyer? Get their log.

To test a filter with a buyer, choose two rows the test team has a right to use. Make one meet the stated rule for a block. Keep the other outside that rule. Hold the rest of the file fixed. Ask the firm to show what came in, when it ran the check and what it sent out. Ask the buyer what it got. If it saw no rows at all, the feed might have failed. A clean row that came through can help rule out that fault. If both rows came through, check the version and the scope of the rule before drawing a conclusion. Such a test could shed light on one feed, at one point in time. State whose feed it was.

### Three other US location orders

The FTC issued an order to X-Mode and Outlogic in April 2024 [84]. It gave them 180 days from issuance to put a sensitive-location program in place [84]. An April 2024 order bars InMarket from selling or licensing defined location data for something of value [85]. The next clause deals with products that use sensitive location data to group or target people [85]. The FTC issued its order to Gravy Analytics and Venntel in January 2025 [86]. It gave them 90 days from the date it took effect for their sensitive-location program [86]. The public file also includes the prior complaint [86]. Which page sets out the claim? Which page sets the rule? Read the part that tells the firm what to do before you put a date in your file.

Say a firm uses location data to build a graph. What would the buyer get? Is it a coordinate, a class of visits, a matched list, or more than one of these? Which firm bought or held the input and under what terms? If a rule has a due date, what starts the clock? If it says a change went live, seek the dated rule and a file shipped under it. To test a claim about a graph, trace an input into the graph and out to one of its clients. If all the buyer can see is an audience at the end, mark the point where the test stopped. A due date can tell the buyer when to look without proving what it will find.

### California followed a streaming account

California’s February 11, 2026 judgment names Disney DTC and ABC Enterprises [74, 75]. When a person is logged in, those firms must honor an opt-out across the Disney streaming services linked to that account [74, 75]. A different rule applies when the person is not logged in or has no account [74, 75]. The state attorney general gave the terms of the deal the day the judge signed the order [75, 74]. What the state said had gone wrong was its claim in the case [75, 74]. Read the signed text to see the duties the two firms took on.

Start a test of the signed-in rule with one test account on two covered services. Change the choice on one. Then check what the other does for the same account. Run a separate test for a signed-out user. At a shared TV, more than one device might be linked to the home. The viewer on the TV might match none of the users signed into the test account. A change to the link between the TV and the home needs a check of its own. The California judgment uses the linked account to set the scope of its signed-in rule [74, 75]. The case record does not show a change in an external firm’s household graph under that order [74, 75] [14, 13, 65, 68, 70, 79].

### Television settings in the Texas records

Texas filed a petition against Sony in December 2025 [76, 77]. It named Samba Interactive TV in its account of how to opt out on BRAVIA sets [76, 77]. The pleading was against Sony [76, 77]. It made no finding against Samba TV and showed no change to a Samba household graph [76, 77]. The author works at Samba TV [232, 233]. The study keeps Samba out of its scores and includes it in this legal screen [232, 233] [14, 13, 65, 68, 70, 79]. An opt-out screen can give a buyer a place to start a test. What left the TV once the user made that choice? Which firm got it?

Texas announced a settlement with LG Electronics U.S.A. on May 11, 2026 [78, 77]. Under its terms LG must get affirmative express consent for the viewing-data use it covers [78, 77]. The terms allow 180 days from judicial signature and entry [78, 77]. The posted copy leaves the judge’s signature and date blank [78, 77]. It cannot tell a buyer when the clock started or when the consent flow went live [78, 77]. Get the entered order to check the date. For a test of a TV set, get the model, the market and the software version. What happens when the viewer says yes? What happens with a no? To make a claim about a graph, trace what the TV sends and how the graph uses it.

### A case with a completed follow-up

In September 2025 France ordered Google to explain the ad cookies used when a person in France opens a Google account [79]. Google changed the disclosure and the way ads appear in Gmail [80, 79]. On September 23, 2026, the French regulator closed its injunction on both points [80, 79]. This gives a reader two records with different jobs. The first says what the firm had to do [79]. The next one says what changed and what the CNIL had seen [80, 79]. It closed just the two points it named [80, 79].

The two French records make a good test of scope [79] [80, 79]. The account-creation order required no change in Chrome’s third-party-cookie default, Topics or Protected Audience [79]. To test the notice in France, compare what a user saw before and after the change. Check the new wording against what the regulator said it accepted [80, 79]. For a claim about an API, check the records for that code. What field used to be in the message? Was it removed or changed? What did the party that got the message do with it? The CNIL confirmed the change to the Gmail ads it had named [80, 79].

### A consent string and the parties that use it

In March 2024 the Court of Justice ruled on a question about the Transparency and Consent Framework [82]. A TC String can be personal data when a person can be identified by linking the string to an ID by reasonable means [82]. In May 2025 the Belgian data protection authority gave its account of the Market Court’s ruling [81]. The court set aside the authority’s decision on procedural grounds and upheld the fine [81]. The authority said the court rejected joint control for processing that takes place wholly within OpenRTB [81]. Name the source and keep that limit in view.

For a test of a string, ask what else the party that gets it holds. Can that party link the string to a person? At which step can it do so? A string sent with a request can show what the sender supplied. To find out what the bidder did with it, read the bidder’s rule. Or observe its output in a test where one use is allowed and one is blocked. Check the case where the request has no string at all. Those cases did not test what a US bidder did with a named ad ID under each choice [81] [82].

### A court can resolve a question of procedure

In October 2025 France’s Conseil d’État rejected an appeal by Yahoo EMEA Limited against a CNIL ruling on cookie placement and withdrawal [83]. The ruling dealt with the use of cookies on the site and in mail [83]. It did not name a change to Yahoo ConnectID [83]. The US case did name ConnectID [90, 8]. On August 26, 2026, a federal court in New York sent five plaintiffs’ claims in a private suit to arbitration [90, 8]. It stayed the court case on those claims [90, 8]. A separate August 26 order asked the parties whether limited discovery was needed for the sixth plaintiff [90, 8] [8]. The ruling tells us where those claims would go [90, 8]. It did not rule on the merits [90, 8].

In February 2024, Open Rights Group said it had filed complaints in the UK and France [60, 61]. The group had commissioned a review of how European RampID documents said the system worked [60, 61]. The group that filed the claims is the source for that account [60, 61]. A later US case dealt with access to LiveRamp services [88, 89, 87, 9]. The FTC’s October 19, 2026 interim order required continued access to LiveRamp’s standalone services [88, 89, 87, 9]. The proposed five-year order was open for public comment through November 23, 2026 [88, 89, 87, 9] [9]. Did a live RampID service change how it acted on a consent flag or a request to erase data? These records leave that point open [88, 89, 87, 9] [60, 61].

A buyer can keep track of the gap with a short note on each item it sees. For a court file, mark the party, the claim and what the judge has ruled on so far. For a seller’s account, mark who said it, which facts that person could see and the date of the test. For a file sent to a client, mark the key the client got and where that key came from. If one of the links is missing, say which one. That gives the next reader a place to look. What can the buyer check on its own? What does a second firm need to check?

### What the open verdict means

The review found duties that had taken effect and some changes in practice [14, 13, 65, 68, 70, 79]. It could not tie an order in force to a change confirmed by a second source for a named identity service in scope [14, 13, 65, 68, 70, 79]. Some US state and EU national case searches by entity were incomplete [14, 13, 65, 68, 70, 79]. Some records left open which product a duty covered [14, 13, 65, 68, 70, 79]. The finding stays open [14, 13, 65, 68, 70, 79]. We cannot use that gap to judge that a seller met or breached its duties. It also leaves open the effect of past cases on how firms link rows.

For a purchase, keep two items on file. On the order, mark the duty and the date or event that starts it. On the test, mark the product, version, market, date and result. Who saw the change? How far downstream could they see? If the seller relies on a second firm to act on a choice, did anyone check what that firm sent on? A new check could settle the claim. Did the same product change? A new allegation could give the buyer a lead. A second account from the firm that made the first claim would still leave the result to be checked.



## 10. What an agent can do with an ID


*A publisher can use the AdCP Go client to send a typed ID token to the Trusted Match router [92, 93, 94]. The reference code can check whether a token belongs to an audience and return IDs for packages that qualify [95, 93] [108, 93]. A buyer who wants to know whether that path influenced a purchase must look beyond the code [14, 13, 92, 95, 108, 101, 232].*

> **In short.**
> The paper asked if work on identity in agent protocols had stopped at a schema, a plan for the shape of a message [14, 13, 92, 95, 108, 101, 232]. AdCP has code that sends the token and code that can act on it [92, 93, 94] [95, 93]. That contradicts the claim [14, 13, 92, 95, 108, 101, 232]. The records still leave open whether a buyer and a seller with distinct systems have used that path to make a buy [14, 13, 92, 95, 108, 101, 232]. Evgeny Popov co-leads the Signals & Measurement working group at AgenticAdvertising.org [232, 233]. The study applies the same rule to AdCP and the IAB protocols it screened [14, 13, 92, 95, 108, 101, 232].

### Start with the decision an ID could change

Suppose a buyer wants to reach people who have already bought from its brand. It asks an agent for one deal at a site they visit. The buyer adds a term. The deal must run on a set type of page. The site could send its whole list of deals. It could send a smaller set that met the terms. It could send no offer at all. If a short list comes back, ask what put each deal on it. Was it the choice of site, the type of page, or a check that someone belonged to the brand’s group?

In that brief, a site name tells the buyer where an ad could run. A deal ID tells it which deal to book. For this test, an identity signal is an input used to recognize a unit or check a link that matters to an ad buy [14, 13, 92, 95, 108, 101, 232]. That unit could be the device that can show an ad. The link could say that someone belongs to a group the seller holds. Keep the claim at the level the facts support. The code for a deal could be the result of a match. It need not tell the buyer who matched.

Give each party a job. Ask the publisher to record which type of token its client sent. Ask the party running the matching service what question it asked of the audience data. Ask the buyer what it did with the answer. Those requests call for different pieces of evidence. A publisher could prove dispatch even if the buyer never received the message. A seller could show an eligible package even if the buyer declined the price. A buyer could show an invoice without showing why the package qualified.

If the claim is about time saved, ask for the plan and the time the team spent on it. If it is about access to a group, ask for the request that picked the group. Look at the reply. If the claim is about extra sales, first link the ad to the buy. Then ask how the seller found out what changed sales. Each claim calls for a different test. A bill might break out each line of the price paid and still leave out how the seller chose the people to reach.

### A request with a deliberate limit

AdCP 3.2.8 defines a Trusted Match request as part of an experiment [91]. A publisher can send it through a router to buyer agents [91]. The request can carry one or more opaque ID tokens [91]. It must leave out page context [91]. The issuer also called clean-room activation experimental in its October account of the 3.2 release [358]. Keep the status and the version on the brief. A team can test the shape of the request before it decides whether a partner can use it on a live buy.

Think of two people in the same target group on two very different web pages. The buyer wants both the right group and a safe page. The Trusted Match request must leave out page context [91]. If the buyer learns that someone belongs to the group, what has it learned about the page? To check that part of the brief, the team needs a further source and the right to use it. It could also limit the stock in advance, drop the extra term or pass on the deal if it cannot check the page.

An opaque token leaves work for the service that gets it. Before agreeing a pilot, ask which part can read the token, who controls its credentials, and whether that use is in scope. Ask how the service identifies the token type when the request includes more than one. Ask what happens when the receiver recognizes the type but cannot read this value. A blank response might mean there was no audience match. It could equally mean the test never ran. Make those outcomes visible as different test results, even if the buyer receives the same empty offer in each case.

Test the router on its own. It could accept a request that went no further. Get a receipt from the buyer’s service too. Give the test a name that both firms can look up. Keep the person’s token out of that name and the shared report. If a request goes astray, the two teams can look for the same event on their own side. They can tell whether both saw it without copying a live user’s key into a ticket. Ask them to show that much before a test calls for access to live data.

### Follow the reference implementation

In the September 11 Go code, the client starts with a user token and its namespace type [92, 93, 94]. It builds the request and posts it to the Trusted Match route [92, 93, 94]. A handler at the other end can read the request and pass it to the service for a check [108, 93]. It can then return IDs for packages that qualify [108, 93]. The two sides of that code give a team a place to test a handoff [92, 93, 94] [108, 93]. A full test would still have to show which parts ran and record what came back.

The Go engine needs an audience service, IDs in the request and a list of target audiences before it runs the membership check [95, 93]. With those in place, it can send the tokens and target group IDs for a batch lookup [95, 93]. For a test, list what each part needs. State where the audience data came from. A short table of made-up members could show which branch the code takes. For a paid deal, the buyer would still need rights to the group it plans to use. A test table cannot grant those rights.

The same Go code has a UID2 decoder and an EUID decoder [96, 93]. Each calls a client set up to decrypt an ad token [96, 93]. Ask which one ran. Who issued the input token? Which keys worked? Keep the answer with the name of that service. If the buyer has rights to one system, give it a test on that system. A sound result under a different name would leave its own buy in doubt. The code path shows what could run. A receipt would show which path did run.

The reply can carry IDs for the packages that qualify [108, 93]. In a test, save the list and then save what the buyer chose. Suppose two packages fit the budget but only one fits a cap on how often to show the ad. The audience check could explain why both came back. The cap could explain which one sold. With just the final choice, the team that signed off the spend would have to guess. It needs both a record of the set and a record of the choice made from it.

For a negative test, change just who is in the audience. Keep the token type, set of deals, price and seller the same. Take the test member out of the group and try again. Did the set of deals change? If it stayed the same, ask first whether the buyer or seller called the membership service at all. If the set changed as planned, the teams can say how the code treated that known input. They still need a different test to tell how often a real brand’s list links to the right people.

### What the seller expects to match

IAB Tech Lab calls the AAMP 3.0 OpenProposal model a public-comment draft [98]. The draft lets the buyer supply segments and a suppression list [97]. The seller can give an expected match rate [97]. Who makes the forecast matters. Ask the seller which records it hopes to match, where it plans to find them and how long the offer runs. Ask what count sits below the rate. The field in the draft holds a forecast [97]. To check what happened, the buyer needs the results of the lookup.

Suppose the buyer sends a list of customers. It asks the seller to leave out people who made a sale last week. Two sellers could quote the same match rate on that brief. One could start with all the records on the list. The other could start with the smaller group it accepted for lookup. Ask each seller to put the base in words. Which records count? What must happen for one to match? Once each side can say what goes above and below the line, it can judge the number it plans to put in the bid.

Ask when the seller checks the list of people to leave out. Which key does it look up? Who can change the list? A buyer might take one test member off the target list after the deal was approved but before ads ran. Ask the seller to show the first result, the new result and the time of the change. The OpenProposal draft lets the buyer supply the list [97]. The test would answer a further question. At what point would the seller’s system stop treating the test member as a match?

IAB Tech Lab describes a demo in which a buyer and seller use AAMP tools to agree on terms [99]. The confirmed deal then goes to Google Ad Manager [99]. OpenProposal’s expected match rate is a field in a draft [97]. To test the role of identity in the demo, ask for the field the seller read and the term it changed. A deal might meet the price, format and supply terms without a check on a person or device. What was it about the input, if anything, that changed what the buyer could buy?

### A vector can enter an auction

The OpenRTB Agentic Audiences extension uses type 1 for an identity embedding [100]. One segment entry can hold more than one signal type [100]. Prebid has a module that takes the type and vector from a stored entry and copies them into the user’s segment data on an OpenRTB auction [103, 105, 104]. The project merged that work on April 14, 2026 [104]. Ask the seller for a test bid. Look for that type and vector at the spot the code would fill. Then ask which party can show that it read them.

To test this route, ask what kind of vector the seller sent. Do not infer its type from the presence of a segment. Record the declared type at the source and the type in the message. Next ask which party can explain how the vector was made and what an intended receiver can do with it. A copy step could preserve every value exactly while leaving the buyer unable to read the values. For a test, give the receiver an agreed input and one clear question to solve with it.

IAB Tech Lab gives Agentic Audiences a use in a normal programmatic auction as well as one in agentic advertising [101]. The August 2026 Agentic Audiences release is v1.1 [102]. The AAMP umbrella release in September 2026 is v3.0 [102]. Write down which version was used and which app made the choice. A bid log could show that an identity embedding crossed the wire. To show that an agent helped make the buy, get its request or decision and join it to that bid. Keep the two records together for the next team to check.

**What agent buying protocols do with identity**

| Protocol, platform or operator | Identity signal and operation | Status | Date | Independently observable? | Scope and limitations | Src |
|---|---|---|---|---|---|---|
| AdCP Trusted Match | IdentityMatchRequest | unconfirmed | October 29, 2026 (version record) | Experimental 3.2.8 schema: publisher via router to buyer agents; opaque tokens, page context excluded. | Schema evidence. The request definition supplies no live identity exchange. | [91] |
| AdCP Go reference client | Typed user-token dispatch | unconfirmed | 2026-09-11 | Source code constructs a typed token request and posts it to the router’s identity route. | Reference connector and publisher-client tests at the September 11 commit. External buyer receipt, production identity traffic and an identity-bearing trade remain unverified. | [92, 93, 94] |
| AdCP Go reference identity engine | Membership lookup | unconfirmed | 2026-09-11 | Code submits request tokens and target IDs to a batch membership lookup. | Reference connector. The lookup needs a configured audience service, identities and target audiences. No independent two-party trace. | [95, 93] |
| AdCP Go UID2/EUID decoder | Advertising-token decryption | unconfirmed | 2026-09-11 | UID2 and EUID decoder code calls the configured client to decrypt an advertising token. | Reference code for a provider-specific operation. Key access, counterpart receipt and paid use are unverified. | [96, 93] |
| AAMP OpenProposal | Expected match and buyer suppression fields | unconfirmed | 2026-09-22 | Draft seller field for expected match rate and buyer fields for segments and suppression. | Draft object model. The expectation is supplied by the seller. No measured match or executed suppression is reported. | [97, 98] |
| AAMP buyer and seller SDKs | Media plan and GAM handoff | unconfirmed | October 27, 2026 (overview revision) | IAB describes buyer–seller negotiation and a handoff to Google Ad Manager. | Issuer demonstration. The exact demo date and identity payload are unverified. | [99] |
| IAB Agentic Audiences | Embedding identity type | unconfirmed | August 2026 (v1.1 release listing) | The OpenRTB community extension defines integer type 1 for an identity embedding. | Ordinary programmatic auctions can also use the extension. A type code does not identify a buyer agent or a transaction. | [100, 101, 102] |
| Prebid Agentic Audiences RTD | Pass through stored embedding fields | unconfirmed | 2026-04-14 | Merged code copies a stored type and vector into user.data[].segment[].ext.aa. | Prebid handoff inside the auction system. A stored entry can represent identity; the code supplies no receipt by a buyer agent. | [103, 104, 105, 100, 101] |
| PubMatic Activate | Clubtails campaign | unconfirmed | 2025-12-15 | PubMatic reports an AdCP-enabled CTV campaign for Clubtails. | Seller account of a campaign. No token, identity-bearing field or transaction trace links its delivery to the protocol’s identity route. | [106] |
| Optable and PubMatic | First-party signal planning | unconfirmed | 2026-03-12 | Optable reports publisher first-party audience planning via PubMatic. | Partner account describes access without transfer of underlying data. No identity field or recipient trace establishes an identity-bearing trade. | [107] |
| AdCP Go reference HTTP identity handler | HTTP identity request and eligible package response | unconfirmed | September 11, 2026 (commit) | The handler decodes the request, calls its service and returns eligible package IDs. | Reference connector. This code supplies no two-party exchange record, named production use or identity-bearing trade. | [108, 93] |

### Read the campaign account at its stated scope

In December 2025, PubMatic said a Clubtails CTV campaign was running on Activate with AdCP [106]. Its account names a client and a buy [106]. It does not name an identity field that passed between the parties [106]. The Go code comes from a separate record dated September 11, 2026 [92, 93, 94]. A buyer needs a link between the field and the older buy before it could make that claim. Ask for the version in use on the campaign dates as well as the version an engineer can run now.

In March 2026, Optable said its PubMatic link let buyers plan with a publisher’s first-party audiences while the source data stayed in place [107]. The release did not show a message with an ID or a receipt that tied that ID to a sale [107]. A planner need not ask to see people’s names to test the claim. Write down the group the buyer asked for. Record what the site agreed to share about that group. Which part of the plan changed as a result? For a claim about a buy, ask for the accepted terms and proof of delivery too.

Do the same for a campaign report that includes an invoice. Imagine a publisher bills a buyer for a completed deal that had been proposed through an agent. Ask the parties to name the deal from both sides. Then ask them to show how the matched audience narrowed the package list. Finally, ask for the acceptance event that joins the selected package to the billed deal. If they can substantiate the second step but cannot link it to the invoice, they have evidence of audience selection and evidence of a purchase. The causal link between those records remains a live question.

Two firms could have a clean pair of logs and still have no spend. The logs might show that each side read the same ID-bearing message in the same way. A product lead could use that result to decide whether to fund a live test. A media owner asked to move a client’s budget would need the sale record as well. Give each team the record it needs for its own choice.

### What would move the evidence forward

To test whether two systems work with each other, ask two teams to run a small test. Use made-up inputs. Name who runs each side, the version, the token type and the form the reply should take. Have the seller record the expected result first. Send a valid request and one with no match. If the service could not read a token, keep that error apart from a lookup that ran and found no match. Have each team save its result, the time and a shared test reference. The reference need not show the test member’s data.

Ask one more thing about the price paid. If a deal uses a flat rate for a group of sites, a change in the package list might leave the rate the same. If the price is set per package, the same change might alter the cost. In a test, keep the offer open for the same length of time and save the terms for each package. Give the person who sets the budget both lists. They can then see which stock the ID made available, which stock they chose and how much of the cost arose from that choice. None of that calls for the buyer to see raw names or emails.

Let both sides set a time limit for the lookup. Suppose the buyer waits for an answer until its offer closes. A match that arrives a little later might still be correct for the test member. The buyer could have had no chance to act on it. Capture the time at which the publisher sent the token, the time at which the seller returned the package and the time at which the buyer stopped waiting. The sales team can then check why the buy fell through. Was it the audience, the late offer or the price?

For a claim about live use, request the same facts from a live flow the parties have the right to share. Mask the parts that could reveal a person. Record the date, buyer, seller and stock. What did the ID do? A live log could show that the ID reached the buyer. On its own it could say little about the price. A bill could show that funds changed hands while leaving open how the audience was picked. Tell the parties which claim they need to test. That will help them find the smallest shareable record that answers it.

A team can now inspect a Trusted Match request, the code that takes it in, the membership check and the code that reads a provider’s token [92, 93, 94] [95, 93] [108, 93] [96, 93]. What the study did not find was a public record that ties an identity-bearing message between two distinct systems to a completed buy [14, 13, 92, 95, 108, 101, 232]. A team that needs such a record can make a small request. Join one eligible package to one accepted deal. Then show what came back when the same request used a test ID that could not match.



## 11. The Identity Quotient


*How do you get an ID? Who can use it? How would you know if it links the right two records? Each row of the Identity Quotient follows the public record for one US buy [11, 13]. Only the peer group for IDs that pass from firm to firm has enough rows with enough proof to show bands [10, 218, 11].*

> **In short.**
> The matrix has 24 scored rows and 142 cells that count toward the IQ [10, 218, 11]. In the ID group, 7 rows have bands [10, 218, 11]. For 7 of them, the band can change when the test makes a high or low guess for cells with no score [10, 218, 11]. Find the row for the thing you plan to buy. Read the source for the cell that counts most in that deal.

### Read a row from left to right

Say a news site has email links to its readers. It wants a bidder to know when one of them turns up on the site. The site has to get an ID, put it in an ad request and know how to stop sending it when the reader opts out. The team that buys the ad may have a more basic need. Can its DSP use that ID? On which ads? How will the desk check the match? A guide to the way IDs pass between firms may help with the first step. For the last step, ask for a test of the links.

The matrix puts a service in a peer group based on what a buyer can get from it [11, 13]. Start with a name on an order. Is the firm selling a new ID, a graph of links or the right to buy a group in its DSP? A brand may use all three. Which one is at issue here? Read the row for that task. Check how many cells have scores. Open the cell that could change the deal. The source and the note tell you what the score rests on [11, 13]. Use them when two firms use the same term for a different step.

The IQ groups the cells under Depth, Reach and Standing [11, 13]. For Depth, ask how the service makes its output and when the output changes [11, 13]. Which signals go in? Who has the right to use them? Where a match rate applies, Depth also checks the base for that rate [11, 13]. Reach looks at the handoff, the systems that use it, the size of the US base and the proof of US access [11, 13]. Standing covers tests of the output, the right to stop use and the trail of changes to the service [11, 13]. Start with the group that bears on the risk in your own buy.

### Work back from the field score

Give each scored cell the same weight as its peers in that group [11, 13]. Depth gets 45 percent of the weight, Reach gets 25 percent and Standing gets 30 percent [11, 13]. A group with no scored cells drops out, and the other weights scale up to make the sum one [11, 13]. That gives the raw score [11, 13]. A gap has no value in that sum [11, 13]. Where a row has a score for a few strong cells and gaps in the rest, those strong cells can still give it a high raw score [11, 13]. Keep the count of scored cells in view.

Take the mean of the raw scores of all rows in the group, even the thin ones [11, 13]. This is the field mean [11, 13]. To get the IQ, give the raw score one vote per counted cell and the field mean 4 votes [11, 13]. Take the mean of those votes and multiply by twenty [11, 13]. For a check by hand, keep the full value as you go. The printed IQ rounds at the end [11, 13]. The band uses the full score and its distance from the field mean [11, 13]. Two rows can thus share a printed whole number and sit on either side of a line [11, 13].

The strength of that pull toward the field is an editorial choice [11, 13]. Say two peers have the same raw score above the field. One has more scored cells, with the same means in its groups. Before rounding, its IQ will lie closer to its raw score. Give both a raw score below the field and the same rule will pull the thin row up more. What could you check to find out if a link was right? Find the cell for tests of the output. Seek the names of the units linked, a test with known answers and a way to count errors. None of those facts comes from the math alone.

### Where a band is warranted

A row needs enough scored cells to enter the matrix [11, 13]. A band needs at least 6 scored cells, with at least one in each group that applies [11, 13]. Even then, its cohort needs 6 or more rows that pass that test [11, 13]. A row can meet its own test and have too few peers for a band [11, 13]. A thin row can have a score near one with a band [11, 13]. Look for the group with no counted cells. If the gap concerns how to stop use or how to test the output, the buy may turn on precisely that gap.

The ID group has 11 rows and a rounded field IQ of 53 [10, 218, 11]. Its bands put 4 rows in Well documented, 2 in Partly documented and 1 in Least documented [10, 218, 11]. No row reaches Most documented at the base weights [10, 218, 11]. Yahoo ConnectID and SharedID both sit in Well documented, yet their gaps differ [10, 218, 11]. Think about a deal that has to work on the brand's chosen DSP. Is that DSP on a list? Can it use the ID? Has it done so? Start with the cell for systems that take and use the ID. Then see what the source can prove.

The graph and resolution group has 8 rows, of which 2 pass the row-level band test [10, 218, 11]. Its field IQ is 41, with no bands [10, 218, 11]. Group C has 4 ways to buy groups in place of ID-based ads, all with thin records [10, 218, 11]. Group D has 1 TV household service, so the field mean is Blockgraph's own raw score [10, 218, 11]. Groups C and D have no bands [10, 218, 11]. If a plan calls for both a graph and a TV household link, write a brief for each buy. A score from a different peer group gives no basis for choosing between the two [11, 13].

Of 148 cells with a number, 142 count toward the IQ [10]. The other 6 carry the label “unverified” and count as n/e [10].

**Public evidence for advertising identity products** (0–5 evidence-bounded scores; n/e = insufficient evidence; n/a = outside the cohort’s mask; unverified cells are marked and counted as n/e; availability, evidence grade and verification status per cell are in data/capability-matrix-long.csv; one section per cohort; banded products ordered by IQ, the Identity Quotient, within their cohort, and unbanded rows listed A to Z)

*A · Interoperable IDs*

| Vendor | Product | Group | IQ (band) | Assignment | Inputs | Freshness | Use controls | Match rate | Interface | Destinations | US coverage | US use | Validation | Withdrawal | Changes | Counted |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| The Trade Desk | Unified ID 2.0 | A | 64 (Well documented, provisional: Well documented with n/e = 1, Most documented with n/e = 4) | 4 | 4 | 3 | 3 | n/e | 4 | 1 | n/e | 3 | n/e | 4 | 4 | 9/12 |
| Prebid | SharedID | A | 61 (Well documented, provisional: Partly documented with n/e = 1, Well documented with n/e = 4) | 3 | 3 | 4 | 3 | n/e | 3 | n/e | n/e | 3 | n/e | 3 | 4 | 8/12 |
| Yahoo | Yahoo ConnectID | A | 61 (Well documented, provisional: Partly documented with n/e = 1, Well documented with n/e = 4) | 4 | 4 | 3 | 3 | n/e | 3 | 1 | 2 | 3 | n/e | 3 | 4 | 10/12 |
| LiveRamp | LiveRamp RampID | A | 56 (Well documented, provisional: Partly documented with n/e = 1, Well documented with n/e = 4) | 3 | 3 | 4 | 3 | n/e | 2 | 1 | n/e | 3 | n/e | 3 | n/e | 8/12 |
| ID5 | ID5 ID | A | 54 (Partly documented, provisional: Partly documented with n/e = 1, Well documented with n/e = 4) | 2 | 3 | 3 | 3 | n/e | 3 | 1 | n/e | 3 | n/e | 3 | n/e | 8/12 |
| Lotame | Lotame Panorama ID | A | 53 (Partly documented, provisional: Least documented with n/e = 1, Well documented with n/e = 4) | 2 | 3 | 3 | 3 | n/e | 2 | 1 | n/e | 3 | n/e | 3 | n/e | 8/12 |
| Epsilon | Epsilon PubLink | A | 42 (Least documented, provisional: Least documented with n/e = 1, Well documented with n/e = 4) | 2 | 3 | n/e | 1 | n/e | 2 | n/e | n/e | 2 | n/e | 1 | n/e | 6/12 |
| 33Across | 33Across Lexicon | A | 55 (thin record · no standing) | 2 | 3 | 3 | n/e | n/e | 3 | n/e | n/e | 3 | n/e | n/e | n/e | 5/12 |
| Flashtalking by Mediaocean | Flashtalking FTrack | A | 45 (thin record · no standing) | 1 | n/e | 3 | n/e | n/e | 1 | n/e | n/e | 2 | n/e | n/e | n/e | 4/12 |
| LiveIntent | LiveIntent nonID | A | 53 (thin record · no standing) | 2 | 3 | n/e | 3 | n/e | 3 | n/e | n/e | 2 | n/e | n/e | n/e | 5/12 |
| Neustar | Neustar Fabrick ID | A | 48 (thin record · no standing) | 1 | 3 | n/e | 2 | n/e | 3 | n/e | n/e | 2 | n/e | n/e | n/e | 5/12 |

*B · Identity graphs and resolution*

| Vendor | Product | Group | IQ (band) | Assignment | Inputs | Freshness | Use controls | Match rate | Interface | Destinations | US coverage | US use | Validation | Withdrawal | Changes | Counted |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Deep Sync | Deep Sync Profile Hygiene & Enrichment | B | 45 (thin record · no standing) | 2 | 3 | n/e | n/e | 2 | 2 | n/e | n/e | 3 | n/e | n/e | n/e | 5/12 |
| Deep Sync | Deep Sync Transaction Matching | B | 45 (thin record · no standing) | 3 | 3 | n/e | n/e | 2 | 1 | n/e | n/e | 3 | n/e | n/e | n/e | 5/12 |
| Epsilon | Epsilon CORE ID | B | 42 (thin record · no standing) | 2 | 2 | n/e | 1 (unverified) | n/e | 2 | n/e | n/e | 3 | 1 (unverified) | n/e | n/e | 4/12 |
| Experian | Experian Digital Graph | B | 40 (cohort too small) | 2 | 2 | 3 | n/e | n/e | 2 | n/e | n/e | 3 | 1 (unverified) | n/e | 1 | 6/12 |
| LiveRamp | LiveRamp Known Identity Resolution (AbiliTec API) | B | 45 (thin record · no standing) | 3 | 2 | 2 | n/e | n/e | 2 | n/e | n/e | 3 | n/e | n/e | n/e | 5/12 |
| Merkle | Merkle Merkury | B | 34 (thin record · no standing) | 1 (beta) | 2 (beta) | n/e | n/e | n/e | 1 (beta) | n/e | n/e | 1 (beta) | n/e | n/e | 1 (unverified) (beta) | 4/12 |
| TransUnion | TransUnion TruAudience Identity Resolution | B | 38 (cohort too small) | 2 | 2 | 2 | n/e | n/e | 2 | n/e | 3 | 2 | 1 (unverified) | n/e | 1 | 7/12 |
| Zeta Global | Zeta Customer Growth Intelligence | B | 41 (thin record · no standing) | 2 | 2 | n/e | n/e | n/e | 2 | n/e | 2 | 2 | n/e | n/e | n/e | 5/12 |

*C · Cookieless targeting alternatives*

| Vendor | Product | Group | IQ (band) | Assignment | Inputs | Freshness | Use controls | Match rate | Interface | Destinations | US coverage | US use | Validation | Withdrawal | Changes | Counted |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ogury | Ogury Personified Advertising | C | 45 (thin record · no standing) | 2 | 3 | n/e | n/e | n/a | 1 | n/e | n/e | 3 | n/e | n/e | n/e | 4/11 |
| Permutive | Permutive publisher cohorts | C | 45 (thin record · no standing) | 2 (beta) | 3 (beta) | n/e | n/e | n/a | 2 (beta) | n/e | n/e | 2 (beta) | n/e | n/e | n/e | 4/11 |
| Proximic by Comscore | Proximic by Comscore Predictive Audiences | C | 41 (thin record) | 2 | 2 | n/e | n/e | n/a | 1 | n/e | n/e | 2 | n/e | n/e | 2 | 5/11 |
| Seedtag | Seedtag contextual audiences | C | 46 (thin record · no standing) | 2 | 3 | 1 | n/e | n/a | n/e | n/e | n/e | 3 | 1 (unverified) | n/e | n/e | 4/11 |

*D · CTV and household identity*

| Vendor | Product | Group | IQ (band) | Assignment | Inputs | Freshness | Use controls | Match rate | Interface | Destinations | US coverage | US use | Validation | Withdrawal | Changes | Counted |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Blockgraph | Blockgraph Identity Operating System | D | 53 (cohort too small) | 3 | 2 | 3 | 2 | n/e | 2 | n/e | n/e | 3 | n/e | 3 | 3 | 8/12 |

### What a strong interface earns

The UID2 row has scores for how it makes the ID, the sources of its inputs and its handoff to a buyer [10, 218, 11]. The Trade Desk's guide names the core, the parties that make raw IDs and tokens, and the sites and bidders that can use the tokens [142]. Its rights table shows who can ask for a token, send an email or phone number for a match or read a token to bid [143, 144]. This gives a team jobs to assign. Who calls the service? Who puts the result in an ad request? Who has the keys to read it? A deal can need all those steps before anyone sees a first bid.

The UID2 row has gaps for its US reach measure and for tests of its output [10, 218, 11]. The guide names US production endpoints where the operator accepts calls [153, 154]. How many of this brand's customers could the bidder reach on that path? Ask for the input base, the time frame and the part of that base the bidder can act on. For a claim that the links find the right person, ask who labeled the pairs and how the test counts false links. A single test plan can ask both questions. It needs a clear base and a way to judge the result for each.

Yahoo's ConnectID check looks for an ID in an ad request, while its DataX report gives the result of matching a partner's file [30, 164]. ConnectID has a scored handoff cell and a gap for a match-rate rule on the route scored here [10, 218, 11]. An ad team could use the first check to see if an ID leaves its site. For a deal priced on the share of customer records matched, set a test for that job. Name the list that went in, the records that met the rules and the firm that did the count. State the point in the job at which the rate is due.

### Why the clock and the receiving party affect the score

ID5 and Lotame Panorama ID sit in Partly documented within the same ID group [10, 218, 11]. ID5's Prebid example sets a life for the stored ID and a time to check it again [175, 131]. ID5 gives a further rule for how long it keeps data used to run its live service [175, 131]. Say an old ID still turns up in an ad request. Which clock would explain that? The stored value, the response from the firm and the link used to assign it may each need their own checks. A score for a stated refresh can stand even if a reader still needs the age of the link.

In Lotame's Michigan case, Advance Local used its PubMatic seat to set up a Panorama ID deal, and the ads ran through Xandr and Adelphic [264]. The SSP did the targeting [264]. To check the destination count, ask who used the ID itself. A site could set up a deal for selected ad space. A DSP could bid on that deal without seeing the ID used to select the space. Keep the path of the ad in the notes. To count a system as a user of the ID, look for the point where it gets the ID and acts on it.

SharedID's guide puts the ID on the site's own domain [286]. A US exchange list calls it a passthrough ID, with the site holding the link to the ID firm [263]. The scored row has a handoff cell and a gap for systems that use the ID [10, 218, 11]. Say the point of a deal is to buy through a single DSP. Ask for a test with that DSP. The exchange list can give the site a place to start. A dated log from the DSP could meet a different need if it shows both receipt of the right ID and what the bidder did with it.

### Read an unbanded row on its own terms

A graph buyer can read each row in group B on its own. Experian sells the US Digital Graph as a whole set of data or as links added to the buyer's own rows [117]. It says it groups digital signals into links to a home or a person [36, 117]. It says it builds the graph afresh each week [173]. The cells for how it makes a link and how it refreshes the output have scores [10, 218, 11]. The cells for rights at the handoff and for withdrawal have gaps [10, 218, 11]. Ask what to do with last week's list if a link is gone from this week's graph. Who would change a list that the client built from that link?

Deep Sync’s grocery case passed H1’s disclosure test because it gave the input base, the group it linked those inputs to and the three steps it used [14, 13, 12]. The same case earns 2 out of 5 for how it defines the rate because the seller names a one-month run but gives no calendar month or as-of date for that run [10, 11, 12]. The cumulative rate after the last step is 73.0%, the top of the page says about 70%, and the results give a range of about 64–73% of monthly volume [12]. The rounded step counts do not add up to the reported total [12].

Deep Sync's retail case runs through more than one pass to link a sale to a buyer [12]. Later passes use a card signal or infer a buyer from the store and the home [12]. In its profile case the input is a retailer's CRM [27]. Both rows are thin in group B [10, 218, 11]. If the job is to link a receipt to a shopper, open the row for sales. If the job is to clean a customer list, open the row for profiles. In each case, start with the record the firm had to work with. A result for one type of record needs its own check before it can help with the other.

In cohort C the match-rate definition is outside the mask [11, 13]. For example, Proximic offers an ID-free predictive audience, and a US political buying route has offered that approach alongside an ID-based option [209] [350]. The scored row is the ID-free buy [10, 218, 11]. Ask which audience a buyer can select, how the supplier tests whether members or placements meet the brief, and which systems honor the selection. For a campaign that buys a group, those questions are more relevant than a request for the share of submitted email addresses matched to a person.

In the TV household cohort, Blockgraph's v2.29 guide requires a household record ID and postal address and allows additional signals such as IP [109]. A permitted partner match can return a file of Blockgraph IDs [110, 111]. Assignment and interface count in the row, while the evidence falls short of a scored validation of the sold output [10, 218, 11]. A TV buyer might accept that the household match ran and still ask who was in the room. Set the unit of the contract and then ask for evidence appropriate to that unit. Giving a valid home a single key does not label which member used a screen.

### Test which assumptions matter

The stress tests change the group weights, change the pull to the field, drop scored cells one at a time and try high and low values in the gaps [11, 13]. Which test suits the deal? Say a brand puts a great deal of weight on its right to stop use. It can check what happens when Standing gets more say. Say a reader finds a flaw in a source for one cell. Check what happens when that cell drops out. A team may also have a private test the public study could not see. Check the gap in the row before using that test to make a new score.

Changing the group weights leaves 4 of 7 observed bands unchanged in cohort A [10, 218, 11]. Under the Standing-led weighting UID2 moves from Well documented to Most documented, and ID5 moves from Partly documented to Well documented [10, 218, 11]. RampID moves down one band under the Reach-led weighting [10, 218, 11]. Nothing about the source records changes in these calculations [11, 13]. The choice changes what an analyst wants the summary to emphasize. Read the cells first if the deal rests mainly on one group.

Removing the pull toward the field changes 3 cohort A bands [10, 218, 11]. Removing a counted cell can even raise a row's number if the cell being removed was pulling down its group means [11, 13]. In the stock calculation, deleting the low destination-coverage cell moves UID2 from 64 to 66 [10, 218, 11]. The study has learned less in that scenario. A reader should flag the lost evidence rather than treat the higher display value as an improved product. It is the quickest way to see why a score with no coverage count can mislead.

To test the gaps, hold the field and the other rows fixed [11, 13]. In the row under review, fill all the gaps that apply with the same low value, then with the same high value [11, 13]. Keep the count of real scored cells and the rules for bands as they were [11, 13]. A thin row stays thin in both tests [11, 13]. The observed band can change for all 7 banded ID rows [10, 218, 11]. If a seller has a new test, check which cell it bears on. Does the test meet the same rule that was used on the peers? If so, score the actual proof and run the sums again.

### What to open when the figures look alike

A buyer reading the table may see peers with the same display score and assume that the rows answer the same questions. Test that assumption on the gaps. SharedID and Yahoo ConnectID have the same rounded IQ in cohort A [10, 218, 11]. The rows both have counted evidence on changes to the product and neither has a counted output-validation cell [10, 218, 11]. Yahoo has a counted US coverage measure, while the SharedID record leaves that cell open [10, 218, 11]. Even the presence of a score needs a closer look. Yahoo's own US user counts do not supply a common time window, a rule for removing repeats or the base of people a buyer could reach [197, 198]. If reach is the point of the buy, ask for the count of eligible people in the proposed campaign and the count reached in the same period.

A near tie calls for the same care. ID5 and Lotame carry the same band in the baseline, but their bands react differently when the test gives more weight to Standing [10, 218, 11]. Does the planned use depend on which party can revoke an ID? Is a named destination more important than the range of destinations advertised? Might a new release change the input a publisher has to send? The weight in a research index cannot settle those choices for a buyer. Keep the firm's answer and the document used to check it next to the condition in the order. If two teams use the same index to plan different deals, each should be able to explain which cell led it to ask a different question.

### Blockgraph (active)

*Blockgraph, LLC*

**SWOT (2026-11-11).** US TV partners can obtain a household match with a defined file structure, while the public record leaves its error rate and active US footprint unresolved.

- **Strengths.** The enterprise guide identifies the household key, required postal address and additional signals a distributor can supply. (vendor documentation) [109] A permitted universe match returns a downloadable, headerless column of household BG IDs. (vendor documentation) [110, 111]
- **Weaknesses.** Kantar’s reported gain cannot be reproduced from the published case because the two base rates, counts and eligibility rules are absent. (inference) [112, 113] Published historic IP lookup periods leave the maximum age of a currently accepted household link open. (inference) [114, 115]
- **Opportunities.** A measurement partner could test how timestamped IP and set-top-box signals change matched household counts within one distributor footprint. (inference) [109, 114] The August 2026 hashed-email input gives a distributor and measurement partner another documented signal to test where emails are available. (inference) [116]
- **Threats.** A buyer that treats Spectrum’s MVPD match rate as a service-wide accuracy estimate may purchase against the wrong denominator. (inference) [38] A partner using old exported results after an opt-out could retain links that the originating node now marks stale. (inference) [115, 111]

- **What is sold.** US TV identity software lets a participant establish a postal-address household record, match its permitted partner’s records, retrieve overlap counts or IDs, and resolve impression or outcome signals. GraphPort maps external graphs into IDoS. The enterprise identity node is the delivery route covered here.
- **Who buys it.** TV distributors, publishers, measurement providers and their participating partners use the platform. Spectrum Reach offers attribution reports to advertisers. TVSquared and Kochava describe consuming identity within measurement. An advertiser buying a report need not receive household identifiers directly.
- **How it earns.** In 2020, then-general manager Jason Manningham described a software subscription with fees that varied by company data type and size. Enterprise rates and service-level revenue as of the cutoff were not disclosed in the records reviewed.
- **Supply and data dependencies.** Native node identity requires the participant’s household records and valid postal addresses. Optional IP, email, mobile advertising IDs and set-top-box IDs can be mapped to these records. Batch resolution needs Client Services and a partner to enable permissions and can restrict results by a delivery scope. GraphPort can connect external graph IDs. Actual coverage depends on the chosen participant and the records and signals available to that integration.
- **Differentiation.** The purchaser can inspect both a subscriber record structure and distinct output schemas for universe and segment matches. Published controls distinguish deleting a source record from invalidating earlier segments and matches after an opt-out.
- **Risks.** Signal age and household moves can affect a timestamped IP lookup. The source collections contain no per-provider error rate for this service. Dated partner descriptions establish activity at different times. Counting all of them as active at the cutoff could overstate the number of current destinations.
- **Best fit.** A US TV distributor matching its own subscriber households against approved partners. A measurement provider linking TV impressions to website or sales records at household level inside a permissioned workflow.
- **Poor fit.** A buyer requiring a verified individual viewer for a shared television. A procurement team that requires a public, current price, a service-wide US eligible reach denominator or a validated household precision and recall estimate.
- **Limits.** Public specifications define a household association. They do not identify individual viewers in a home or whether a given viewer took a later action. The reviewed record does not give a current US active-household denominator, rate of false links or fixed TTL for the live IP-to-household edge. Partner authorization and record removal are documented at the participant node. The public sources reviewed do not show a consumer refusal delivered to a downstream destination or erasure of files previously exported.

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | C | In v2.29 a node maps IP, email and device signals to a household BG ID generated using a valid postal address and participant record ID. The upload supports an apartment/unit field and multiple signals. The guides do not specify how identity resolution breaks ties when different households share a signal or an address is incomplete. [109, 115, 114, 166] |
| Where input signals come from | live | C | For the native node upload, the subscriber record ID and postal address are mandatory. Device and IP signals are optional. In the Spectrum workflow, the distributor supplies exposure from set-top boxes and streaming devices, Innovid collects website visits, and Blockgraph matches them at household level. The precise IP or device selected for each campaign is not disclosed. [109, 38, 167] |
| Refresh and expiry of the sold output | live | C | A full manual refresh replaces the node’s identity data, removing records omitted from the new file. The v2.29 ID Resolution guide allows a row timestamp for historical IP lookup at providers that retain prior values for 30, 60 or 90 days. The Spectrum workflow uses seven days to attribute a visit after exposure. The materials do not specify a time-based expiry rule for the current IP-to-household or device-to-household link. [115, 114, 167] |
| How an allowed use reaches the recipient | live | C | Universe Match lists authorized partners. Batch resolution requires permissions enabled by the partner and Client Services, with an optional delivery-scope filter. In the v2.29 manual upload template the Data Rights column is left empty. The reviewed input and output contracts do not demonstrate transfer of a household’s consent or refusal to a destination. [115, 110, 111, 114] |
| How a reported match rate is defined | live | D | Kantar and Blockgraph reported a more-than-tenfold change in partner match rates for the panel integration. The CEO’s general definition does not identify the panel base, deduplication rule, counts, exclusions or time period for either rate. Spectrum separately claims a 95% addressable match rate for its own MVPD service. Its FAQ does not attribute that number to an IDoS-wide test. [112, 113, 38] |
| Recipient's delivery interface | live | C | An authorized node can see an overlap count and download a Universe Match CSV containing BG IDs in a single headerless column. Segment Match results can contain BG IDs, record IDs or both, depending on permissions. The batch measurement route accepts signal and record ID, optionally a timestamp for IP, and supports Blockgraph or Measurement IDs as output. [110, 111, 114] |
| Receiving systems that use the output | live | C | TVSquared named its use of IDoS in 2022. Kochava documented an implemented measurement path in 2023. Spectrum Reach documents an Innovid website-visit handoff for a US campaign route. Kantar also describes panel matching. These dated records establish distinct roles and periods. They cannot support a contemporaneous count of active, independent US operators. [168, 169, 38, 112] |
| US coverage with a stated base | live | D | CIMM’s April 2025 report describes 130M+ households without giving the figure’s geography, eligible base or counting window. Spectrum’s FAQ separately describes collecting exposure inputs from more than 30 million households. Neither count establishes unique US households currently reachable through IDoS for one buyer. [6, 38] |
| US availability and use | live | C | Blockgraph’s 2021 notice identifies the US as the service territory and refers to selling access to commercial clients. Named US Spectrum Reach TV measurement activity and v2.29 enterprise workflows for participant nodes provide an obtainable route. The historical subscription statement supplies no current price. UK and European language in the notice concerns website users. [170, 171, 38, 109, 110] |
| Evidence testing the sold output | unconfirmed | D | Kantar reports a relative match-yield change. Spectrum describes an MVPD-level match rate. Neither source reports the correctness of BG household assignments against reference labels. The commissioned IP study masks suppliers. A product-specific, externally referenced validation of IDoS household links was not established in the sources screened by the cutoff. [112, 38, 40] |
| Stopping or removing an output | live | C | Section 6 of the 2021 notice directs a customer who wants contact to stop to the client. In v2.29, Delete removes a subscriber record but leaves existing segments and matches usable. Opt Out removes the record and requires refresh of affected segments and matches. No processing deadline or mechanism to recall earlier exports was established in these sources. [170, 115, 111] |
| Version and correction history | live | C | Release 2.29 introduced pre-hashed email across several identity workflows in August 2026. Version 3.0.0 in October placed node-specific automations on the Identity Data page and displayed more information on signals in the most recent upload. Neither note reports changes in household accuracy or reachable scale. [116, 172] |

Performance evidence: Kantar reported greater than ten times its prior TV-partner match rates after using IDoS. (vendor assertion, sponsor Kantar and Blockgraph, commercial partners.) [112, 113]; Spectrum states a 95% addressable match rate for its MVPD offering and describes a Blockgraph household exposure-to-visit handoff. (vendor assertion, sponsor Spectrum Reach, a media seller affiliated with a Blockgraph owner.) [38]; CIMM and Go Addressable commissioned a US IP study covering six unnamed suppliers. (commissioned measurement, sponsor CIMM and Go Addressable commissioned Truthset.) [40, 41]

### Experian (Live US offer, as read on November 11, 2026.)

*Experian Marketing Services, Inc. is the Digital Graph sheet’s copyright holder. The US marketing business privacy notice names Experian Marketing Solutions, LLC. A US graph license identifying its contracting entity was not located in this review.*

**SWOT (2026-11-11).** Experian gives buyers a choice of graph delivery and a named customer example. The public cases do not establish the accuracy or withdrawal timing of current US links.

- **Strengths.** Buyers can choose person or household resolution and receive the whole graph or links appended to their own data. (vendor claim) [117, 36] MiQ describes how its traders use the licensed IDs. The anonymous DSP case traces buyer cookies to a weekly file. (vendor claim) [37, 21, 118]
- **Weaknesses.** The two seller cases report match or sync rates without the counts and eligibility rules needed to reproduce either percentage. (inference) [20, 36, 21] The US individual count lacks a public measurement window or deduplication rule on the materials reviewed. (inference) [117, 119]
- **Opportunities.** A buyer could label a sample of file links and test person or household accuracy alongside the reported match rate. (inference) [37, 20, 36] Separate graph and attribute offers let buyers specify which identity links and enrichment fields they need in a procurement test. (inference) [117, 120, 121]
- **Threats.** Snowflake blocked new legacy rooms in its web app on October 1. Buyers planning a new Experian setup there need a different route. (inference) [122, 123, 124] A buyer relying on a browser opt-out could misstate its coverage if it assumes the same suppression reaches every linked device and recipient. (inference) [125, 126]

- **What is sold.** Experian Digital Graph connects digital identifiers at the buyer’s chosen household or individual level. A buyer can license the full dataset or have links appended to its own records. Offline Graph connects names, postal addresses and emails. Marketing Attributes can be bought with Digital Graph as a joint offering.
- **Who buys it.** The named user is MiQ, a programmatic media partner. Its traders and analysts use graph links in MiQ’s Identity Spine for audience work and measurement. A separate, unnamed DSP used the graph to relate ad exposures to website conversions.
- **How it earns.** Experian sells graph licenses and offers to match a customer’s records. The seller describes MiQ licensing hashed emails, UID2s and partner cookies. Product rates, contract minimums, billing units and Digital Graph revenue were not located in the product, case and financial records reviewed.
- **Supply and data dependencies.** The clustering model uses device and browser signals. Its inputs can include timestamps, IP addresses, user agents and IDs. Some buyers contribute their own cookies or IPs. Experian describes adding IPv6 and phone-based UID2s. Marks such as UID2 identify supported inputs or outputs, not a destination or a completed bidstream transfer.
- **Differentiation.** A buyer can choose a household or a person link and a full dataset or a match against its own records. [IDN-EXP-F01, IDN-EXP-F03] The named MiQ case shows where graph output sits in a buyer’s own targeting and measurement workflow. [IDN-EXP-F06]
- **Risks.** Snowflake removed the Experian connector from its legacy web interface in September 2026. Its web app no longer creates new legacy clean rooms. [IDN-EXP-F12] A buyer whose suppression obligations extend across recipients would need terms and evidence that tie each link removal to its own purchased delivery. [IDN-EXP-F10, IDN-EXP-F11]
- **Best fit.** An ad platform with its own identity system can license selected linkage data. MiQ is the documented example. [IDN-EXP-F06] A buyer that needs a link file to compare CTV exposures with web conversions can inspect the unnamed DSP workflow. [IDN-EXP-F04]
- **Poor fit.** An analyst requiring independently measured error rates for person or household links would still need a labeled test of the current graph. [IDN-EXP-F07, IDN-EXP-F08, IDN-EXP-F13, IDN-EXP-F14] Teams requiring a guaranteed daily graph file or an edge-age limit would need delivery terms beyond the reported weekly route. [IDN-EXP-F04, IDN-EXP-F05]
- **Limits.** Graph counts omit the eligible population and counting method. [IDN-EXP-F09] The MiQ and unnamed DSP rates use different inputs. Neither gives a US-specific eligible base or a test of correct links. [IDN-EXP-F07, IDN-EXP-F08] Group revenue, UK proceedings and a separate Tapad privacy rule have narrower uses than a US Digital Graph claim. [IDN-EXP-F11, IDN-EXP-F17, IDN-EXP-F18]

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | D | Experian describes clustering timestamps, IPs, user agents, cookie IDs and device IDs into links at a chosen household or individual level. The case materials do not publish thresholds or rules for conflicting links. [36, 117] |
| Where input signals come from | live | D | An unnamed DSP supplied cookies for sync. In another case MiQ supplied IP addresses. Product material names timestamp, IP, user-agent and device signals, but the reviewed records do not trace each signal to a collecting party and event. [36, 117, 37, 20] |
| Refresh and expiry of the sold output | live | D | Experian describes a weekly graph rebuild. It separately reports weekly file delivery to one DSP. Neither passage defines an edge’s maximum age, decay function or removal rule when signals lapse. [37, 173] |
| How an allowed use reaches the recipient | unconfirmed | C | The US notice describes how GPC opt-outs attach to a browser identifier in certain states. The product page, case files and policy do not trace permit, refuse or missing states from a Digital Graph input through its delivered file to a recipient. End-to-end handling remains unconfirmed. [125, 117, 37] |
| How a reported match rate is defined | live | D | The MiQ report identifies the input class, output and direction of its 70% rate. The anonymous DSP has a separate 84% sync figure. Neither discloses counts, eligibility exclusions and acceptance logic for a US-only test or for the whole dataset. [20, 36, 21] |
| Recipient's delivery interface | live | D | An unnamed DSP synced its cookies and received related Experian cookies, mobile ad IDs, CTV IDs and IP addresses in a weekly file. The cases do not specify browser cookie context, file schema, duplicate handling or unmatched rows. [37, 36] |
| Receiving systems that use the output | unconfirmed | D | MiQ confirms it used Experian IDs in its Identity Spine. Its case gives no US-specific sample. The DSP is unnamed. Snowflake removed the connector from its legacy UI in September 2026 and blocked new legacy rooms there in October. A US-specific recipient count remains unconfirmed. [37, 21, 122, 123, 118, 124] |
| US coverage with a stated base | live | D | Experian advertises a graph with 250 million individuals in the US. Its page does not date the count or describe deduplication, active status or the eligible US population. [117, 119] |
| US availability and use | live | D | The seller labels a US edition, describes delivery as the whole dataset or as links appended to customer records, and directs buyers to its identity sales team. Public prices, license duration and a sample agreement were not located in the sources screened. [117] |
| Evidence testing the sold output | unconfirmed | D | The 70% MiQ match rate and 84% DSP sync figure measure yield. The cited Truthset comparison covers hashed-email data and attributes. The 2016 Tapad case concerns a historical graph. None of these records publishes independent labeled-link error rates for the current US Digital Graph. [20, 36, 174, 45] |
| Stopping or removing an output | unconfirmed | C | The US notice starts GPC opt-outs at the browser ID unless already linked. Tapad describes permanent opt-out for in-graph identifiers, a 60-day conditional rule for related devices, a 12-month IPv4 rule and a recipient discard obligation on refresh. No reviewed record shows a US Digital Graph licensee applying a deletion or preventing reappearance. [37, 125, 126] |
| Version and correction history | live | D | The US marketing product lead documented the addition of IPv6 and phone-based UID2s by the October 2026 page revision. It gives no release version, effective day or buyer migration instruction. Snowflake’s dated legacy UI change is a separate interface event. [173, 123, 122] |

Performance evidence: Experian reports a 70% match rate from MiQ-supplied IP addresses to cookieless identifiers in the MiQ engagement. (vendor assertion, sponsor Experian, seller of the graph. MiQ is the customer. Study funding unreported.) [20, 21]; Experian says the unnamed DSP synced 84% of IDs within weeks of using its Digital Graph. (vendor assertion, sponsor Experian, seller of the graph. Customer unnamed. Study funding unreported.) [36, 37]; Experian’s account of Truthset’s Q1 2026 results describes its hashed-email data and marketing attributes. (vendor assertion, sponsor Experian summarizes Truthset. Payment and design for a Digital Graph test were not established.) [174]; Tapad reported that its Device Graph had twice the precision of the next competitor in a 2016 Hotels.com evaluation. (vendor assertion, sponsor Tapad reports Hotels.com’s comparison. Financial sponsorship is undisclosed.) [45]

### ID5 (Active in the UK register as of 11 November 2026.)

*ID5 Technology Limited (UK company 10735438).*

**SWOT (2026-11-11).** ID5 and two buying platforms describe how the ID is used, while public evidence on national reach, link accuracy and completed deletion remains limited.

- **Strengths.** Adobe documents placement targeting with ID5 IDs and lets a buyer select a universal ID type for a placement. (vendor documentation) [127] ID5 publishes separate opt-out and deletion feeds, their update schedules and the partner’s ingestion responsibilities. (vendor documentation) [128]
- **Weaknesses.** The published Sovrn test reports auction pricing for a combined treatment. It supplies no labeled US person-link errors. (vendor claim) [129, 130] The onboarding guide illustrates the share of uploaded emails matched. It gives no observed US bidstream match rate. (vendor documentation) [32]
- **Opportunities.** Publishers could test the value of sending hashed email separately from linking it by adapting the published three-group auction design. (inference) [129, 130] A DSP could compare its deletion completion timestamps with ID5 feed delivery to quantify the delay buyers experience. (inference) [128]
- **Threats.** A partner that processes a withdrawal file late could continue using an ID after ID5 publishes the request. (inference) [128] A buyer assuming the acquisition added Twine’s licensed records could overstate addressability. ID5’s current policy keeps those datasets separate. (inference) [131, 132]

- **What is sold.** A publisher can obtain an encrypted ID5 ID and forward it in an ad request. The agreement sells decryption, connections between IDs and use of ID5 data as additional services. ID5’s DSP guide describes out-of-auction keys and merge instructions for linking publisher-specific IDs. [IDN-ID5-F05, IDN-ID5-F08, IDN-ID5-F10]
- **Who buys it.** Publishers register to distribute the encrypted ID. DSPs may purchase decryption and linkage information. Adobe documents a targeting setting for advertisers. Adform reports a retargeting route, and Sovrn and ID5 describe a historical test with participating publishers’ US audiences. [IDN-ID5-F05, IDN-ID5-F11, IDN-ID5-F12, IDN-ID5-F14]
- **How it earns.** ID5 offers encrypted ID delivery without charge under its agreement. It offers decryption, information linking IDs and use of its data for purchase. Adobe says ID5 IDs carry no fee in its own DSP placement feature. Enterprise licence rates, revenue per ID and US ID5 ID sales remain undisclosed in these records. [IDN-ID5-F04, IDN-ID5-F05, IDN-ID5-F11]
- **Supply and data dependencies.** The issuer uses signals from the browser or device and optional publisher inputs such as hashed email. A Prebid route depends on the publisher’s partner configuration, privacy choices, an SSP forwarding the EID, and an authorized DSP integrating the off-auction keys and merge instructions. The 2026 policy puts Twine Data’s US licensing data outside this service. [IDN-ID5-F06, IDN-ID5-F07, IDN-ID5-F08, IDN-ID5-F09, IDN-ID5-F10]
- **Differentiation.** Keys and merge instructions are distributed outside the auction. The participating DSP can recognize the encrypted ID carried in an SSP request. [IDN-ID5-F10] A buyer can distinguish documented placement targeting in Adobe from Adform’s single-tag site-retargeting path. [IDN-ID5-F11, IDN-ID5-F12]
- **Risks.** Inference: an operator may publish a withdrawal file before a partner ingests it. The feed cadence alone understates uncertainty about completed removal. [IDN-ID5-F13] Inference: interpreting an affiliate acquisition as a graph migration would assign data to ID5 ID that its later platform policy places in a separate business. [IDN-ID5-F07, IDN-ID5-F16, IDN-ID5-F19]
- **Best fit.** Inference: a publisher and a DSP that can exchange encrypted IDs and integrate keys can test addressable inventory on a documented route. [IDN-ID5-F05, IDN-ID5-F10, IDN-ID5-F11] Inference: an advertiser using Adform’s tag can test an ID5-based retargeting workflow in that system. [IDN-ID5-F12]
- **Poor fit.** Inference: the published partner test does not support procurement that requires independently measured US person-level false-match rates. [IDN-ID5-F06, IDN-ID5-F14, IDN-ID5-F15] Inference: the public feed alone does not meet a buyer’s need for documented recipient deletion completion times. [IDN-ID5-F13]
- **Limits.** The two-hour refresh, local storage duration and service-side data retention do not set a uniform age limit on graph links. [IDN-ID5-F20] The partner study measured auction pricing. Intent IQ’s separate comparison uses location history as a proxy for the IDs bridged into bid requests. The reviewed reports establish neither independently labeled US identity-error rates nor national bidstream coverage. [IDN-ID5-F06, IDN-ID5-F14, IDN-ID5-F15] The public feed describes a handoff. It leaves the last mile of deletion with each partner. [IDN-ID5-F13]

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | C | ID5 describes matching deterministic and passive signals, issuing an encrypted identifier and separately giving the recipient merge instructions for publisher-specific IDs. The public method leaves the pairwise acceptance threshold undisclosed. [55, 56, 175, 176] |
| Where input signals come from | live | C | The publisher passes optional first-party identifiers or hashed email in the JavaScript partner-data parameter. ID5 names deterministic matching and use of other signals in its model. The affiliate data-use disclosure prevents attribution of Twine’s US licensing data to this product. [55, 56, 131] |
| Refresh and expiry of the sold output | live | C | Prebid illustrates 90-day local storage and a two-hour refresh. ID5’s policy separately sets a 90-day duration from the last update for its third-party cookie and a 100-day maximum for live-service data before deletion or aggregation. The records do not give a general graph-link expiry. [175, 131] |
| How an allowed use reaches the recipient | live | C | ID5 requires a US sale/sharing opt-out signal path, including GPP or GPC, and separate purpose-specific TCF or prior opt-in handling in EEA/UK traffic. Prebid says the id5-sync.com EID should remain present with the ID string “0” on no consent or opt-out. No named recipient’s enforcement was verified. [175, 177, 176, 131] |
| How a reported match rate is defined | unconfirmed | C | The adjacent Data Onboarding guide defines distinct hashed emails matched over distinct inputs for each output ID type. Its sample is illustrative. The Sovrn case compares CPMs on participating publishers’ US audiences. Neither source supplies an observed US bidstream match rate for the ID5 ID. [129, 130, 32] |
| Recipient's delivery interface | live | C | The Prebid EID uses id5-sync.com. ID5 describes OpenRTB forwarding and separate keys and merge instructions for the DSP. Adobe documents using ID5 for placement targeting, subject to geo-target selection. These records describe separate interfaces, not one traced auction. [175, 176, 127] |
| Receiving systems that use the output | unconfirmed | C | Adobe documents an ID5 placement option; Adform reported a single-tag ID5 retargeting integration on 28 May 2025 across products it operates. The records do not establish how many named recipients used this exact ID on US-supported production routes at cutoff. Sovrn’s 2022 test is historical. [127, 178, 129, 130] |
| US coverage with a stated base | unconfirmed | D | The Sovrn case identifies participating US audiences and treatment groups. No current national count or share of eligible US devices, people, households or bids with both a defined base and counting period appears in the case, buyer settings or onboarding example reviewed. [127, 129, 130, 32] |
| US availability and use | live | C | ID5 publishes requirements for US-originating publisher traffic and identified the US as an existing client market in 2025. Sovrn and ID5 reported a 2022 test on participating publishers’ US audiences. Adobe’s geography-dependent option does not enumerate US-enabled seats or transactions. [177, 127, 129, 130, 132] |
| Evidence testing the sold output | unconfirmed | D | Sovrn and ID5 compared CPM. Competitor Intent IQ separately tested how closely the locations of IDs bridged by ID5 matched bid-request locations. Its method uses fixed-line IP history and distance, without separately labeled person or household membership. Independent accuracy of the core ID5 ID remains unconfirmed in this search. [55, 56, 129, 130, 32, 57, 58] |
| Stopping or removing an output | live | C | ID5 documents S3 opt-out files every 30 minutes, routine deletion files hourly and on-demand deletions on a best-effort schedule. Partners configure delivery and are expected to process the files. The public records do not show completion latency or removal from derived segments at a named DSP. [176, 128, 131] |
| Version and correction history | live | C | The Prebid module documents a version boundary for externally configured module loading. Dated ID5 SDK snapshots preserve the same conditional cookie-sync default and disable option. Neither establishes a change in the underlying US graph. [179, 180, 175] |

Performance evidence: Sovrn and ID5 report 72% CPM uplift with hashed-email linkage and 98% with the hashed email passed. The joint case compared three publisher traffic groups and supplies no raw CPMs or counts to reproduce the uplifts. (vendor assertion, sponsor Sovrn and ID5 participated and publicized the work. No separate funder identified.) [129, 130]; Match Preview reports the share of distinct input hashed emails matched to each output ID type. Its example is illustrative. (vendor assertion, sponsor ID5, provider of Data Onboarding) [32]; In Intent IQ’s dated table, 34.3% of assessed ID5-enriched bid requests passed its strict 100-mile rule between 12 August and 10 October 2026. Its mean within-request proportion was 35.1%. (vendor assertion, sponsor Intent IQ, competing provider; the issuer says an external implementer replicated the method on its data.) [57, 58]

### LiveRamp (Operating subsidiary under Publicis Groupe; US ATS/RampID documented.)

*LiveRamp, Inc. (US service provider named in general terms); LiveRamp Holdings, Inc. (SEC registrant acquired in November 2026).*

**SWOT (2026-11-11).** RampID has a documented route from publisher to buyer; the available yield, time-spent and historical precision figures measure different things.

- **Strengths.** ATS documentation traces the publisher envelope through SSP decryption to a buyer-specific RampID. (vendor documentation) [133, 134] AWS publishes the hashed-email and placement requirements for using ATS with its MediaTailor streaming service. (vendor documentation) [135]
- **Weaknesses.** A derived RampID can be issued for an input absent from the offline graph, so issuance alone cannot demonstrate a person match. (inference) [134] The cited US reach figure uses consumer time online. Buyers cannot convert it into a count of authenticated people from the disclosed information. (inference) [136] The publisher case combines UID2 and RampID. It does not isolate how much additional revenue RampID generated. (inference) [59]
- **Opportunities.** A publisher can use the documented envelope expiry to measure how often refresh restores bidder access in its own traffic. (inference) [137, 133] A streaming publisher with an ATS agreement can test the documented MediaTailor route using its own eligible viewer base. (inference) [135]
- **Threats.** If a recipient misses a withdrawal, local envelope removal alone gives the publisher no proof that the recipient deleted an earlier copy. (inference) [138, 139, 140] The FTC’s discrimination and confidentiality provisions give rival agencies a basis to scrutinize standalone service access after Publicis’s acquisition. (inference) [88, 89, 141]

- **What is sold.** US publishers can license ATS to request an envelope from LiveRamp, store it and pass it to enabled partners. At an enabled SSP, Sidecar can translate it to a DSP-specific RampID. The separate Known Identity Resolution service (IDN-P012) returns AbiliTec Links by API or Known IDs in the cloud.
- **Who buys it.** Publishers can implement ATS, SSPs can contract for Sidecar, and separate terms cover DSPs and DMPs. Advertisers can buy enabled inventory through those platforms. Google reports that New York Times Advertising used ATS. AWS provides a MediaTailor example requiring an active ATS agreement and a domain placement ID.
- **How it earns.** In 2021, LiveRamp’s CEO said the company did not intend to charge publishers or media-platform partners for ATS and expected it to drive platform use and cross-sales. The publisher form licenses limited envelope use. General terms put any applicable service fees in the signed order. The reviewed records give no current universal per-envelope price or buyer CPM. They disclose group subscription, revenue-sharing, usage and services income without a standalone RampID line.
- **Supply and data dependencies.** A publisher must obtain an identifier and handle applicable user choices. The underlying graph uses partner-observed device linkages and offline reference data. ATS may produce a derived ID from an unrecognized input. SSPs need the applicable keys and recipient rights. Token expiry, local opt-out and destination processing control different stages.
- **Differentiation.** The documentation distinguishes the DSP-specific output from the publisher’s opaque envelope. It also explains how an unrecognized input can still yield a deterministic derived ID. US SSP and DSP country terms identify separate responsibilities. First-party documentation from Google and AWS identifies named handoff routes.
- **Risks.** Opt-out controls work at different layers. A cleared browser envelope does not establish erasure of copies already routed to buyers. Post-acquisition access and data separation fall under an FTC proceeding. The interim order is a legal duty and does not measure either discrimination or compliance.
- **Best fit.** Publishers with authenticated inventory and approved demand routes needing a contractually scoped identifier handoff. Buyers able to test a named bidder and delivery route against their own eligible first-party base.
- **Poor fit.** A buyer requiring an independently measured US person-level error rate for the exact bidstream output. A purchaser needing a public per-envelope price or audited nationwide count of unique authenticated persons available to one DSP.
- **Limits.** Public US file match rates, time spent online and publisher yield have distinct units. The screened material does not give an observed US ATS translation rate with its input base or test the correct end user. The envelope TTL applies to transport. It cannot be carried over to the underlying graph, a household or an individual’s retention period.

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | C | ATS resolves a publisher-supplied identifier against the graph. A derived ID can represent a consistent hash even when the offline lookup misses. An enabled SSP decodes the envelope and sends the appropriately encoded RampID to the DSP. [134, 133] |
| Where input signals come from | live | C | The publisher obtains a login or other offline identifier; ATS handles plaintext or hashed inputs. LiveRamp describes additional observed device-to-PII links from its match partners and multisourced offline reference data. The public guide does not enumerate each contributor or give record-level lineage. [134, 133, 181] |
| Refresh and expiry of the sold output | live | C | The storage guide sets the envelope lifetime at 15 days in California and 30 days in other regions. It recommends a timestamp and updating it on refresh. Storage and graph-edge lifetimes are separate. [137] |
| How an allowed use reaches the recipient | live | C | ATS.js checks US national GPP first and falls back to a legacy CCPA string, whose presence is optional. LiveRamp groups the UK and EU in its consent diagram and specifies v2.3 for the EU/EEA, with vendor 97 and purpose 1 for retrieval. Its US recipient terms assign rights-handling duties. The screened sources show no DSP refusal tied to a particular incoming bid field. [182, 140, 183] |
| How a reported match rate is defined | unconfirmed | E | The Connect Activation guide defines a file-to-destination rate. The US ATS handoff screen of that guide, seller method, US time-spent claim and Google publisher case found no observed numeric envelope-to-recipient match rate on a defined US population. [15, 134, 59, 136] |
| Recipient's delivery interface | live | C | The documented publisher-to-SSP handoff uses encrypted envelopes. Keyed Sidecar decrypts and translates them to a buyer-specific RampID in a bid request. A legacy Prebid example uses identityLink and idl_env. [134, 133, 184] |
| Receiving systems that use the output | live | C | Google reports an actual New York Times Advertising Secure Signals implementation with ATS. AWS separately publishes a MediaTailor request example, requiring a placement ID and an active ATS agreement. The docs do not state that every bidder sees or can decode the envelope. [133, 59, 135] |
| US coverage with a stated base | unconfirmed | E | The 2023 issuer claim uses a US time-spent base for publisher integrations. No qualifying count or share of unique US ATS persons, households, tokens or devices with a matching eligible base was identified in the issuer exhibit, methodology and Google case screened to the cutoff. [136, 59, 181] |
| US availability and use | live | C | The SSP Sidecar/RampID and DSP/DMP terms include the United States subject to the applicable agreement. Google describes NYT implementation without reporting US-only impressions. AWS supplies a streaming integration example with no named deployed customer. [140, 183, 59, 135] |
| Evidence testing the sold output | unconfirmed | E | LiveRamp’s historical graph precision claim, Google’s mixed-ID yield result and a European document review cover different outcomes. The public screen found no US labelled truth set testing people correctly recognized in the present ATS bid route. [43, 59, 61] |
| Stopping or removing an output | live | C | For a publisher-managed opt-out, LiveRamp instructs publishers to delete the local envelope and keep a suppression flag. Connect separately makes requests available to destinations and describes direct delivery to The Trade Desk for specified first- and third-party segments. Destinations control processing. No screened source documents completed erasure at every ATS recipient. [182, 138, 139, 140] |
| Version and correction history | live | C | The May 2021 rename notice identifies the earlier IDL name and the Prebid v1.0 configuration still shows the legacy key. The current API storage guide uses _lr_env. These public artifacts distinguish namespaces and integration versions, but do not give revision-level identity-edge lineage. [185, 184, 137] |

Performance evidence: More than 90% of US consumer time spent online connected through integrated publishers, as claimed in 2023. (vendor assertion, sponsor LiveRamp) [136]; 81% increase in publisher revenue on restricted-cookie browsers in 2025 publisher results. (vendor assertion, sponsor Google Ad Manager publishes the case with New York Times Advertising; any separate commissioning arrangements are undisclosed.) [59]; 95%–98.9% historical graph precision attributed by LiveRamp to comScore. (vendor assertion, sponsor LiveRamp hosts the assertion. Funding and independence of the underlying test undisclosed.) [43]; European privacy review describes the service’s architecture and raises privacy questions. (synthesis, sponsor Open Rights Group commissioned Cracked Labs and Alan Toner.) [61, 60]

### The Trade Desk (Active UID2 public operator and Core Administrator as read on 2026-11-11.)

*The Trade Desk, Inc.*

**SWOT (2026-11-11).** UID2 documents how a credential becomes a token and how a bidder can stop using it. The public cases measure narrower outcomes than national reach or person-level accuracy.

- **Strengths.** The permission matrix distinguishes token generation, raw-ID mapping and bidder decryption. Each participant has a defined role. (vendor documentation) [142, 143, 144] The v3 migration guide supplies old and new parsing examples and exposes the prior raw ID for 90 days after rotation. (vendor documentation) [145, 146]
- **Weaknesses.** Deep Sync reported matched CRM records without the exact eligible denominator. The Trade Desk’s HP case measured reach against one campaign’s estimated target households. (inference) [31, 147, 148] A site opt-out stays with that participant. A consumer must submit a separate portal request for each email address or phone number they want suppressed across UID2. (vendor documentation) [149, 150]
- **Opportunities.** Publishers on affected Apple browsers can evaluate the server-side route that the UID2 guide documents. (inference) [151, 152, 145] A buyer with customer credentials can test a specific activation through Roku Exchange and compare its own results against a defined baseline. (inference) [143, 144, 153, 154]
- **Threats.** A token in a bid request can be unusable for targeting. One private operator reported finding its encryption error after three months. (inference) [155] To keep UID2 on affected iOS 27+ browsers and macOS 27+ Safari, publishers using browser-side token generation or refresh must move those steps server-side. (inference) [151, 152, 145]

- **What is sold.** Publishers can obtain an encrypted advertising token from an email address or phone number. Approved mappers can receive raw IDs. CRM onboarding instead puts IDs into a Trade Desk account for targeting without returning them to the client. The separately branded Identity Alliance is a cross-device graph. [IDN-UID-F03, IDN-UID-F05, IDN-UID-F06, IDN-UID-F09]
- **Who buys it.** Publishers and their service providers put the token into supply. Advertisers, agencies and data providers onboard records. DSPs receive the encrypted token from the supply path. HP is a named advertiser, while Roku describes the publisher-side route through Roku Exchange. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F09, IDN-UID-F10, IDN-UID-F17]
- **How it earns.** UID2 says a participant with a contract and credentials can use a Public Operator at no charge. Private operators supply their own infrastructure and engineering. The Trade Desk earns platform fees linked to client spend and earns fees for services and data. Its filings do not disclose UID2 revenue or profit. [IDN-UID-F02, IDN-UID-F03, IDN-UID-F09]
- **Supply and data dependencies.** The operator needs an email address or phone number, or the permitted hash, and current keys and salts. The chosen publisher and recipient must support the route. A site opt-out binds that participant. The global portal requires a separate request for each email or phone. On the affected Apple browser paths, UID2 directs publishers to generate and refresh tokens server-side. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F07, IDN-UID-F12, IDN-UID-F13, IDN-UID-F15]
- **Differentiation.** Publishers can request an encrypted auction token and approved bidders can decrypt it. Mappers can use a separate raw-ID output for customer records. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F09] A dated v3 migration shows how a data holder can update its mapping after rotation. [IDN-UID-F08]
- **Risks.** A private operator can send unusable tokens. One publisher’s error escaped notice for months according to reporting. The incident does not establish a market error rate. [IDN-UID-F19] Publisher integrations depending on a browser-side UID2 request lose that path on affected Apple browsers. [IDN-UID-F15] The FTC’s general warning on hashing and the pending private case call for clear descriptions of use. The private allegations and procedural ruling sit outside the IQ. [IDN-UID-F20, IDN-UID-F21]
- **Best fit.** Publishers with usable customer credentials and a verified supply path, or buyers mapping their own identifiers for a named participating DSP or publisher route. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F09, IDN-UID-F10, IDN-UID-F17]
- **Poor fit.** A buyer requiring independently validated person-level error rates or a complete national reach estimate cannot establish either from the sources reviewed here. [IDN-UID-F16, IDN-UID-F17, IDN-UID-F18, IDN-UID-F19, IDN-UID-F22] A buyer whose platform-only CRM onboarding must return a portable customer-to-UID2 lookup would need a different approved access route. [IDN-UID-F09]
- **Limits.** The raw ID can match the same input across operators at the same time. Individual advertising tokens vary. The reviewed academic and commercial sources do not compare inferred people against independent labels. [IDN-UID-F07, IDN-UID-F18, IDN-UID-F22] HP reports reach for one campaign in a US-edition report. Deep Sync reports matches in one US retailer’s CRM. Neither publishes a national UID2 adoption share. Retailer eligibility, deduplication and rounding remain unspecified. [IDN-UID-F16, IDN-UID-F17, IDN-UID-F22] The administrator’s policy is role-specific. Its example and opt-out instructions give no observed recipient-level TCF/GPP enforcement result. [IDN-UID-F12, IDN-UID-F13, IDN-UID-F23]

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | C | An operator salts and hashes directly identifying information to create a raw ID and encrypts it for a publisher. Two participants mapping the same email at the same time should receive the same raw ID. A fresh token can differ for that same ID. [142, 143, 144] |
| Where input signals come from | live | C | A publisher or SSO supplies first-party credentials. A mapper can supply an email address, phone number or corresponding hash. HP’s campaign used device-registration data from its CDP. The identifier supplied may establish an account link without independently identifying a person. [142, 143, 144, 147] |
| Refresh and expiry of the sold output | live | C | For manual publisher token refresh, UID2 recommends hourly checks of the indicated time. It recommends daily raw-ID refresh timestamp checks for audience updates. Raw IDs change roughly yearly, and may remain unchanged across refresh checks. V3 can return the prior ID for 90 days after rotation. [144, 145, 146] |
| How an allowed use reaches the recipient | live | C | The UID2 client addendum assigns processing duties for covered customer data. The opt-out guide requires the participant to stop local issuance after a site opt-out. DSPs must screen opted-out raw IDs during bidding. The sample request and DSP guide do not show actual per-impression TCF or GPP enforcement. [149, 150, 186, 144, 187] |
| How a reported match rate is defined | live | D | Deep Sync’s one US retailer case gives a nominal 141.7 million-record CRM population and before/after UID2 match counts and percentages. It omits exact destination eligibility and rounding. The result describes an enrichment case and does not define a universal UID2 match rate. [31] |
| Recipient's delivery interface | live | C | The FAQ shows an encrypted token in user.ext.eids[].uids[].id with source uidapi.com. An approved DSP may decrypt it and must honor opt-outs before bidding on it. The separate CRM onboarding route allows in-account targeting and expressly withholds the IDs and mapping. [143, 144, 188, 186] |
| Receiving systems that use the output | live | D | Roku’s August 2024 announcement describes a TTD–Roku Exchange activation route. The 2024 platform report says HP used TTD to reach audiences on Disney+ and Hulu. The reports do not identify those properties as distinct downstream UID2-decrypting operators or quantify cutoff-period use. [153, 154, 147] |
| US coverage with a stated base | unconfirmed | D | In its US-edition report the platform says HP reached 8% of an estimated 12.5 million campaign target households. The case gives no campaign geography or dates, national UID2 eligible base or deduplication rule. The reviewed sources do not establish coverage of the US product population. [147, 31, 148] |
| US availability and use | live | C | UID2 lists production endpoints in Ohio and Oregon. The license matrix describes approved access. Deep Sync reports UID2 matches for one US retailer, and Roku describes an activation route. None of these establishes the size of the national UID2 population. [188, 153, 154, 147, 31] |
| Evidence testing the sold output | unconfirmed | D | The reviewed cases measure matches or campaign reach. The academic study examines the architecture and the reported private-operator incident tests no labelled population. The reviewed record gives no independent UID2 person-level agreement or error estimate. [31, 147, 148, 155] |
| Stopping or removing an output | live | C | The portal takes a separate request for each email or phone. The guide describes a publisher opt-out response and a DSP webhook, typically within minutes. The DSP must stop bidding on that raw UID2 and may use a different ID. Recipient deletion and compliance times were not verified. [149, 150, 186, 187] |
| Version and correction history | live | C | The July 2025 v3 mapping release changed the response used by advertisers and data providers. The migration guide shows both v2 and v3 parsing, the transition away from bucket monitoring, and the 90-day prior-ID lookup. [145, 146] |

Performance evidence: For one US retailer, Deep Sync reports 73.2 million matched UID2 records (51%) before enrichment and 94.4 million (66%) afterward. Its nominal CRM file held 141.7 million records. The case does not explain the whole-percent rounding or exact destination denominator. (vendor assertion, sponsor Deep Sync published the case and sells the enrichment service. The retailer is unnamed. Separate study funding was not disclosed.) [31]; In its 2024 US report, The Trade Desk says HP used device-registration data for a UID2 campaign on Disney+ and Hulu through its DSP and reached 8% of an estimated 12.5 million target households. (vendor assertion, sponsor The Trade Desk Intelligence produced the report with Hypothesis Group. HP supplied the campaign account. Separate campaign-study funding was not disclosed.) [147]; An unfunded 2025 academic study examined technical documentation and code for identity architectures including UID2. It described how a credential-derived token permits linkage across participants. The study did not test UID2 assignments against labelled people or households. (synthesis, sponsor The three academic authors declared no research funding or competing interests. Architectural study.) [148]; In May 2026 AdExchanger reported that an unnamed CTV publisher operating privately found an encryption error after three months. A Trade Desk engineering executive confirmed the faulty tokens could not be reconciled for targeting. (synthesis, sponsor AdExchanger reporting. Article funding unspecified. Interviewees include the publisher and The Trade Desk.) [155]

### Yahoo (Live US publisher and Yahoo DSP route documented as of 2026-11-11.)

*Yahoo Ad Tech LLC is the US signatory for this service. The contract allows Yahoo affiliates to deliver it. [IDN-YAH-F01]*

**SWOT (2026-11-11).** Yahoo documents how publishers send the ID into its DSP; the evidence for additional buyer destinations and for person-link accuracy is narrower.

- **Strengths.** On web inventory, Ad Manager publishers can configure the Google-listed ConnectID signal for Yahoo’s named Authorized Buyers bidders. (vendor documentation) [156, 157] Allen Media Group described its ATS integration, and Mediavine reported connecting its Grow publisher toolkit to ConnectID. (vendor claim) [158, 159]
- **Weaknesses.** The US agreement directs publishers to include ConnectID in bids to Yahoo and restricts sharing beyond the company and its approved publishers. (record) [160] Mediavine reports publisher yield. Truthset describes ratings of its own audiences. Neither account tests whether Yahoo linked the right person. (inference) [158, 161]
- **Opportunities.** An advertiser already buying through Yahoo DSP could use Truthset’s rated segments or Merkle’s custom audiences within that platform. (inference) [162, 161] A publisher running ATS or Magnite Demand Manager could add Yahoo demand through an existing integration. (inference) [159, 163]
- **Threats.** If required consent is missing or a user opts out, fewer eligible ConnectIDs may be available for a planned buy. (inference) [160, 164, 165] Buyers requiring interchangeable demand destinations may find the named Yahoo-only bidder configuration and contractual sharing limits too narrow. (inference) [160, 157]

- **What is sold.** Yahoo returns an identifier to a participating publisher for a bid request. The approved publisher can send it to Yahoo, including via an SSP or Ad Manager. Advertisers can use the identity within Yahoo DSP. Next-Gen is the associated offer for non-addressable supply. [IDN-YAH-F04, IDN-YAH-F05, IDN-YAH-F09, IDN-YAH-F11, IDN-YAH-F21]
- **Who buys it.** Participating publishers and approved supply partners integrate the ID. Advertisers use supported audiences in Yahoo DSP. Yahoo’s Merkle announcement and Truthset’s own release identify upstream audiences for that DSP. [IDN-YAH-F04, IDN-YAH-F09, IDN-YAH-F12, IDN-YAH-F13, IDN-YAH-F21]
- **How it earns.** Yahoo links the publisher agreement to demand in its DSP and markets ConnectID within DSP campaigns. The reviewed service agreement, integration guide and issuer announcements disclose no separate per-ID fee, take rate, product revenue or margin. [IDN-YAH-F01, IDN-YAH-F04, IDN-YAH-F21]
- **Supply and data dependencies.** The server route needs the publisher’s hashed email, Yahoo publisher ID and credentials. The web module can accept a domain-specific input and the Magnite wrapper can create a first-party cookie. Yahoo describes a graph based on consumer relationships. The integration does not expose the truth of each underlying person link. [IDN-YAH-F03, IDN-YAH-F04, IDN-YAH-F05, IDN-YAH-F13, IDN-YAH-F17]
- **Differentiation.** The publisher can select a server lookup, a Prebid integration or a brokered Ad Manager route to Yahoo DSP. The contract governs who may receive the ID. [IDN-YAH-F04, IDN-YAH-F05, IDN-YAH-F09] The buyer can select upstream Merkle or Truthset audiences in the named DSP route while the publisher can use its existing ATS integration. [IDN-YAH-F10, IDN-YAH-F12, IDN-YAH-F13]
- **Risks.** Counting SSPs, two Yahoo bidders and audience partners as separate demand destinations would overstate buyer choice. [IDN-YAH-F09, IDN-YAH-F10, IDN-YAH-F11, IDN-YAH-F12, IDN-YAH-F13] In the private US case, five plaintiffs were directed to arbitration. The procedural decision cannot establish a privacy violation or an operational remedy. [IDN-YAH-F20] A cached Prebid value can accompany a refresh request. The local opt-out can block decoding in the examined module. Neither that control nor an empty new server lookup demonstrates removal of copies already delivered. [IDN-YAH-F06, IDN-YAH-F07, IDN-YAH-F08]
- **Best fit.** Publishers with approved supply and Yahoo demand that want a documented bidstream handoff. [IDN-YAH-F04, IDN-YAH-F05, IDN-YAH-F09] US advertisers buying in Yahoo DSP that want to activate supported partner audiences. [IDN-YAH-F12, IDN-YAH-F13, IDN-YAH-F21]
- **Poor fit.** Buyers requiring independently documented reuse of the identical token by several unrelated DSPs. [IDN-YAH-F04, IDN-YAH-F09, IDN-YAH-F11] Procurement requiring verified person-link precision, a stated graph edge expiry or downstream proof of erasure. [IDN-YAH-F06, IDN-YAH-F07, IDN-YAH-F13, IDN-YAH-F16]
- **Limits.** Published interfaces document the handoff into Yahoo demand. Yahoo’s reach estimates do not establish how many unique people a particular campaign can reach. [IDN-YAH-F09, IDN-YAH-F14] A receiver can see whether a token was included. None of the screened product guides and partner case records supplied labeled person-link error rates. [IDN-YAH-F13, IDN-YAH-F15, IDN-YAH-F16] The US contract’s geographic language conflicts within the same paragraph. An API field for GDPR cannot resolve contractual rights to deploy. [IDN-YAH-F19]

| Dimension | Availability | Evidence | Observed |
|---|---|---|---|
| How the output is assigned | live | C | The service contract specifies a keyed representation of a hashed identifier. Yahoo’s server API resolves the supplied email to a ConnectID; the Prebid route can start from a local publisher ID. These are rules for issuance and encoding. Neither the agreement nor endpoint identifies how every person-level graph edge is created or reconciled. [160, 189, 165, 190] |
| Where input signals come from | live | C | The server route requires the seller’s email hash and Yahoo’s publisher ID. On the Prebid route, 7.51.0 accepted an email hash or a publisher ID. Version 7.52.0 removed the need for either. A publisher can pass a local ID, and the Magnite wrapper documents a first-party-cookie fallback. [164, 163, 189, 165, 191, 192, 193] |
| Refresh and expiry of the sold output | live | C | Yahoo’s standalone JavaScript may return a cached ID and fetch asynchronously. Yahoo recommends a 24-hour S2S cache. The March 2026 Prebid module checks an external publisher’s stored ConnectID for staleness using the response TTL, up to 14 days, or a 14-day default. A stale ID may accompany the refresh callback. No hard recipient or graph-edge expiry is established. [194, 165, 195] |
| How an allowed use reaches the recipient | live | C | The agreement requires suppliers to forward applicable preferences. Yahoo documents optional GPP inputs and an empty HTTP 200 response on an opted-out S2S lookup. In the March 2026 Prebid module the local flag gates getId and decode. Disabled local storage and a storage exception take different branches. No single rule for all missing signals or recipients is established. [160, 164, 165, 195] |
| How a reported match rate is defined | unconfirmed | E | Bounded evidence gap. The DataX guide describes partner-file matching and return of matched partner IDs. The ConnectID guides check issuance and bid inclusion. Neither the guides, Truthset nor Mediavine supplies an observed rate for US ConnectID publisher inputs with a defined eligible denominator and period. [164, 158, 161, 165, 30] |
| Recipient's delivery interface | live | C | Yahoo specifies the yahoo.com EID at user.ext.eids for OpenRTB 2.5 and user.eids for 2.6. Prebid supplies a payload check. Ad Manager’s Authorized Buyers setup names two Yahoo bidders; Open Bidding uses a sharing toggle. The guides provide no transaction log. [164, 156, 157, 165] |
| Receiving systems that use the output | live | C | Yahoo DSP is the documented buyer destination. Merkle, LiveRamp and Truthset offer upstream routes. Google Ad Manager and publisher SSPs distribute the signal. Neither the partner announcements nor the selected bidder settings document a second DSP consuming the exact Yahoo-issued value. [156, 157, 159, 162, 161, 196] |
| US coverage with a stated base | live | D | The latest issuer article describes 213 million US users; its opening text describes logged-in relationships. The April 2025 CAPI release used 232 million. Neither includes a comparable counting window, eligible-universe base or deduplication method. No trend is inferred. [197, 198] |
| US availability and use | live | C | The US signatory publishes a publisher agreement, whose terms link its supply to Yahoo demand. The integration guide offers onboarding and the DSP article addresses US reach. US-based Mediavine described a deployment. The agreement restricts sharing; it is not a universal decoding license. [160, 199, 157, 158, 198] |
| Evidence testing the sold output | unconfirmed | E | The issuer describes a browser payload check. Mediavine reports publisher yield; Truthset grades its upstream audiences. The screened issuer, publisher and Truthset records do not provide a labeled holdout or false-link result for the correctness of ConnectID’s person assignment. [164, 158, 161, 198] |
| Stopping or removing an output | live | C | Yahoo offers an email opt-out form and specifies an empty opted-out S2S result on a new lookup. A domain-local flag stops new JS generation. In the examined Prebid code it also blocks decoding a stored ID. Yahoo recommends a daily server cache refresh. The reviewed interfaces provide no receipt or deadline for deletion of earlier recipient copies. [164, 200, 201, 165, 195] |
| Version and correction history | live | C | Prebid.js 7.51.0 accepted an email hash or a publisher user ID. Version 7.52.0, released June 1, 2023, removed the requirement for either. Yahoo’s current help table overstates the older email requirement. These versions document a publisher configuration change. They do not date a change to Yahoo’s graph edges or the primary US Chrome sync route. [164, 189, 202, 191, 192, 193] |

Performance evidence: Mediavine reported an 80% eCPM increase on Safari and Firefox in November 2021. (vendor assertion, sponsor Mediavine, the integrating supply partner.) [158]; Yahoo reported different US logged-in user counts at its April 2025 and October 2026 publication dates. (vendor assertion, sponsor Yahoo, identity operator and DSP seller.) [197, 198]; Truthset describes ratings of its upstream audiences and an activation integration with Yahoo. (vendor assertion, sponsor Truthset, the participating data seller.) [161]

### Take the record into diligence

Use the profiles to trace a decision back to an actor. If access is the issue, identify who grants the key, where the receiving system gets it and which agreement covers that route. If a partner reports higher yield, ask for the input population, rejected cases, inventory and test period. If a buyer needs to remove an audience, identify who can stop future issuance, who can remove a prior export and what confirmation each can return. Bring the open cell into the deal brief as a request for a particular record. A buyer can then decide whether to run a test, narrow the use or accept the remaining uncertainty for that purchase.

![Read the band, not the rank: no product clears the most documented line, and 17 rows are not banded.](figures/fig-quotient.svg)

*Figure fig-quotient. **Read the band, not the rank: no product clears the most documented line, and 17 rows are not banded.** The Identity Quotient for the products with enough evidence to band, one section per cohort. The IQ uses the kit’s Depth, Reach and Standing weights and its documentation bands. Its field mean is the average raw score of admitted products within a buying task. Earlier Contextual and Native Quotients used a pooled group mean. Depth, Reach, Standing group means are combined 45/25/30 and shrunk toward the cohort’s field mean (the mean Raw score of its admitted rows) with a prior strength of 4; bands come from the unrounded distance to that field score, and only verified or qualified cells count. A band needs at least 6 counted cells with one in every applicable group, in a cohort with at least 6 such rows. Across the reweightings, positions move by up to 3 places within a cohort, and 4 of the 7 banded products keep their band under every weighting. IQ measures publicly documented capability within a cohort, not product performance. Units: IQ points on a 0–100 scale, pulled toward the cohort field mean; shown rounded, plotted unrounded. Scope: 7 banded products of 24 in the capability matrix, scores from evidence read in November 2026; recheck on 11 December 2026. Caveat: Single scorer. The index inherits every limit of the dimension scores and is sensitive to the weights; the bands, not the positions, are the finding. Data: data/iq.csv*



## 12. Buyer’s decision guide


*What will the buyer get? Disney's terms define a group made from the overlap of a brand's data and Disney's data [303]. Experian offers its US Digital Graph as a full dataset or as links it adds to a buyer’s records [117]. Either could fit a brief. The team that takes the work needs a different test for each. Name the seat, file, deal or report where you expect to see the result. Agree on a check you can run there before you sign.*

> **In short.**
> CIMM asks buyers to check how much of a matched list their main ad platforms can reach [259]. It also asks them to set up a truth set to check how often the links are right [259]. Put both checks in the brief when the job needs them. Give each a named owner and a due date. Tie the price to the part that passed. Set a time to check it again and agree on what would pause the buy.

### Write down what would make the job a success

Ask the team that will take the work to fill in one line. “We can use this file or group here, for this job, until this date.” Pick a verb that has a test. Find old customers in a new file. Keep people on the loyalty list out of a buy. Choose a group for a deal. Link an ad on TV to a visit in a report on homes. If two of these matter to you, write two lines. Get a price and a sign-off for each. Put the name of the person who owns the seat or file next to the line.

Say a brand wants to keep club members out of its offer for new shoppers. What should happen when someone joins the club at lunch? How soon must the seat stop showing them the offer? What about a member whose record has no key the seat can use? Which person on the team will check that the list on the seat is up to date? For this brief it might be enough to act on a list in one seat. If a new brief asks you to merge two files the brand holds at home, write a new line. Ask for rights to take the link back.

Disney's terms put the duty to warrant rights and consents on the party that sends the brand's data [303]. Meta's customer-list terms say a brand does not learn which people make up the matched group [302]. Amazon Marketing Cloud describes a data report that gives back sums for groups and a separate path to use a custom group for ads [301]. Decide what sort of receipt you need. For a seat, ask for a group ID and a way to check it in the account. For a file, get a sample row, a key you have the right to keep and the uses your terms allow. Ask the team to show where it would store that row.

### Put names on the route

Who sends the keys? Who does the match? Who passes the result? Who can use it to buy an ad? Write down the names as the traders know them. Include the seat and who owns it. Put a contact next to each handoff. Ask which party can still reject an audience after the seller has accepted it. Run the test with the rights the team would have on the first day of the buy. If the team borrows an admin account for the test, record the step it will need to run again from its own seat.

Roku described a path from The Trade Desk via Roku Exchange for ads that use UID2 and Roku's first-party data [154]. In its guide for Google Ad Manager, Yahoo names the Yahoo App & Display and Yahoo Video bidders for Authorized Buyers [156, 157]. Its Open Bidding route uses a sharing switch [156, 157]. Yahoo also named Amazon Publisher Services, InMobi, Magnite and PubMatic as ways for publishers to take part [196]. Use each name for the role it plays. A name on the supply side could help you find stock. Before you promise an audience to a buying team, ask that seat to read an allowed test value. Save its answer.

The Trade Desk says a client can onboard a CRM list for use in its own account without getting the customer map back [188, 203]. It sets out separate rights for UID2 API access [188, 203]. Which option does your quote cover? Can the buyer keep what comes back? If all it sees is a group on a screen, what will tell it the send has worked? Ask for the reply on a good send and on a failed one. If one agency buys and a second one reports, test that both can see the part they need.

A team can bring good evidence and still be early in the sale. In January 2025 dentsu reported a test of its Merkury and Trade Desk route with Hy-Vee's RedMedia [359]. An October 2025 agency announcement called for Merkury to link Hy-Vee’s enterprise media, RedMedia and loyalty programs [360]. RedMedia's own site describes using shopper data with outside data for off-site ads without naming UID2 [361]. Put a test, a plan and a live run on separate lines. To sign off on the run, ask for the date, the account and a receipt from the team at the far end.

### Pick the unit you really need

Does the buyer need a person at all? Google says TelevisaUnivision sent Permutive’s standard groups in ad requests [205, 206, 207, 208]. Its US figure counts groups per impression [205, 206, 207, 208]. OpenX also announced a US way to buy Permutive audiences without third-party cookies or device IDs [205, 206, 207, 208]. Seedtag says its custom audiences can score URLs from the brief the brand gives it [355]. Ogury describes groups that brands and agencies can use in Ogury One [210]. Ask what gets put in the group. Is it a page, a chance to show an ad, a device or a login? Does the unit stay the same when you send it to the seat?

For a page-based buy, ask to see one page the rule should accept and one it should reject. Compare both with the brief. If a page is a poor fit for the brand, tell the seller why. You may want all sports stories, or just those about a team in your market. Ask what the buyer sees in the deal and what it gets back from a win. Is there a cost to update the rule as the news changes? If you want to cap the number of ads one person sees across sites, ask for a way to test the cap. Pay for a page match on terms that fit a page.

For TV, say which claim about a home the team needs to make. Spectrum Reach describes supplying household TV exposure while Innovid supplies household website visits and Blockgraph matches them [38, 167]. The visits can count within seven days of an ad in that workflow [167, 38]. What if one person in the home watches while someone else makes the visit? Could the buyer still count a home with a later visit? Would it be able to name who watched? To test if the ad drove a visit, randomly hold it back from some eligible homes where feasible, then compare visits across the groups. Have the team say how it would choose that group before it buys the spots.

### Try the service on a file you know

Pick a small, permitted part of the real file. Ask the data owner to record the columns, count and date. Put in some known good cases, some that should fail and some with no known answer. For a job that removes customers from a buy, add a test account that should stay in on the first run and come out on the next. For a job that adds links to rows, add a known duplicate and a changed email. Keep the answers with the person running the check. Give the seller only what it needs to do the job.

Keep a sheet that both ends of the route can read. How many distinct keys went in? How many could the seller use? How many came back? Did one input count once, once per device, or each time the seat saw it? Give both ends the same date and file code. Get the failures by cause as well as the good rows. If a screen hides small groups, find out how large the test needs to be. Google displays a match rate on an API upload only once at least 100 unique users match [300]. Keep a blank with its reason. If the result is below the limit for display, a zero would give the trader the wrong answer.

Choose a test that fits the cost of a mistake. You may be able to plan broad reach with a range. If your job is to keep ads away from a group, you may want a strict cap on how many of its members could still see one. Set the bar as a buyer. Look at how the test file was picked. Does it have the same mix of frequent and rare users as the full buy? Keep wrong links and missed links on separate lines. Leave space for cases where no one can check the answer. Agree on how to count that group before you see the rate.

For a media test, hold the buying rules as still as you can. Use the same dates, bid rules and time for the buyer to act. Which effect do you want to buy? An extra chance to bid? A better chance of winning the same stock? A gain in sales? If the new route lets you bid on a different set of sites, split those sites out. If the bid price changes, keep the price and the win count. Decide how much gain would cover the fee. A test of media value can tell you where to spend. Keep a test of links to known people on its own line.

### Follow a change all the way to the seat

Set up a test account you have the right to use. Start with a permitted use. Get the value and ask the seat what it may do with it. Change the choice at the agreed source. Log when the source accepts it. At what time does the seat find out? What does it do when it next sees the old value? Ask for the output from the first run after the change. If a saved group also needs to be cleared, name who can clear it. Agree on what proof of the change the buyer can keep once the data itself is gone.

UID2 says a site must stop making the ID for a user who opts out at that site [149, 150]. For an ecosystem request, the user supplies each email or phone number to the service [149, 150]. The UID2 guide calls for a separate message to the DSP and tells it to stop using that UID2 to bid, while allowing another ID or no bid [186]. Which request did the team use in the test? Which ID did the seat stop using? If the brief calls for all targeted ads to stop, add a test for any other path the seat may still use.

ID5 says partners should pick up opt-out updates it issues every 30 minutes and routine deletion updates it issues once an hour [128]. It also describes a separate, best-effort feed for deletion on demand [128]. LiveRamp tells a publisher handling its own opt-out to clear its stored envelope and keep a flag to stop it being made again until the user opts back in [138]. Ask to see the time of the last good import at the seat. What would happen if that import failed? What can the seat still do with an old copy? If that step needs a separate service, get a price and name the team that will check it.

Give the age of a household link a test of its own. Blockgraph offers historical IP lookbacks of 30, 60 or 90 days where the node keeps household signal history [114]. In version 2.29, a full manual refresh replaces the node’s identity layer and leaves out rows missing from the new file [115]. Experian says it builds its Digital Graph anew each week [173]. Ask for a permitted test of a move. Which version first shows the new home? When does the old home stop being used? What happens to a file that has already reached a client? Let the team that uses the file show its own rule.

### Pay for the step that passed

Break the quote into lines you can check. Does the brand pay per row sent, per home the seat can use, per thousand ads shown, or per month of access? What does a second run cost? What fee does the seat take? Put a price on work to set up the file, read the ID and keep the job going. Ask finance to price the trial and a full term. If you get a credit for the trial, say which step earns it. Set a date when the buyer can cut its loss if the job will not work.

LiveRamp's publisher form limits use of ATS envelopes to parties enabled under a statement of work [352, 353, 345, 354]. Its general terms place any fees in the order form [352, 353, 345, 354]. ID5 offers encrypted IDs to registered publishers at no charge and has a commercial licence for platforms and other covered uses [56, 351]. UID2 says a firm under contract can use a public operator at no charge, while a private operator provides its own hosting and engineering [188, 203]. Put each price next to the party and step it covers. Get the cost for the whole route from the team that has to pay it.

Write an exit clause that a trader could use. Which files can the team keep after the term ends? Which old keys must it clear? Who will close the account? Can it still read reports from the months it paid for? Set a rule for a change of source data, owner, terms or route. Which test would you run again? Who would pay for the work? How much time would you have to challenge the result? Treat the test file and the signed receipt as part of a handover when a new agency takes the seat.

Set one last check on the bill. Pick a date on the invoice. Have the media lead get the count for that date from the seat. Have the file owner get the count from its send. Ask finance to price both on the units in the order. If the bill uses a third figure, who can explain it? A repeat send, for instance, might incur a fee even if it put no new homes in the group. Settle how to treat that case while both sides still have the run sheet.

### Keep the test with the decision

| Buyer question | Test that fits | What the inference needs | Where it fails | What it cannot tell you |
|---|---|---|---|---|
| What does the quoted rate count? | Ask the seller to take one dated file through the same route you plan to buy. Request raw rows, distinct inputs, eligible inputs, matched inputs and unmatched inputs. | The same population, market and counting rule apply at each step. Each excluded record has a reason. | The denominator shifts from people to devices, or only the matched subset is in the report. | A high yield alone cannot tell you whether the output names the right person. [15, 17, 18, 19, 12] |
| How many matches can my destination use? | Trace one approved audience from upload to the receiving system. Request separate counts for linked, accepted and usable records at the same time. | The buyer and destination agree on namespace, deduplication, time window and what successful delivery means. | A delivered file or a field in a bid request is counted as a usable audience without a recipient check. | The receiving system’s count does not establish correct links or impressions delivered. [15, 188, 203, 156, 157] |
| How often does a link point to the wrong person or home? | Use a blinded sample with independent labels. Ask who supplied the truth, how the cases were selected and which pairs were left out. | Labels must be separate from the keys used to create the links. Record the tested product, market and unit. | A device-ID match, a common IP, campaign lift or model agreement is treated as independent truth. | A household test cannot tell you which household member watched or bought. [40, 41, 14, 13, 43, 45, 62] |
| Which party stops use when someone opts out? | Follow a permitted request and a refused one through a named route. Check issuer response, recipient interpretation, the treatment of cached values and logs from each party. | Use approved test accounts, a defined market and a documented response. Avoid real personal records in a demonstration. | A consent field is populated but nobody can trace what the far end did. | A successful local opt-out cannot prove deletion of every exported copy. [149, 150, 186, 128, 160, 200, 165, 138, 139] |
| When should the household link stop being used? | Ask what happens after a move, new router, record removal and full rebuild. Record the actor, affected link, start time and reuse rule. | Keep the same household definition and note whether the buyer uses a live lookup, a snapshot or a downloaded file. | A lookback window, attribution window or delivery schedule is offered in place of an edge-level rule. | A documented clock does not measure the number of stale or incorrect links in your campaign. [109, 114, 115, 14, 13, 204] |
| Can this group help when a person has no usable ID? | Request the assigned unit, the input evidence, a sample allowed placement and one rejected placement. Test the same audience under the buying route you plan to use. | A group delivered to a buyer must mean the same thing as the group the publisher assigned. Keep the original inventory and model version. | An ordinary contextual placement or forecast is counted as a verified user match. | A valid group assignment cannot identify which person read the page. [205, 206, 207, 208, 209, 210] |
| Will the offer work in our market and on our platform? | Trace the US service from the seller’s offer through a named partner or platform. Ask for production state, checks on receipt, and a date. | The records refer to the same brand, service, namespace and geography. | A corporate office, a global partner list or a beta feature in another country stands in for availability. | A working route does not establish the size of its usable US audience. [211, 212, 213] |
| Can an agent act on this identity signal? | Ask for the versioned field and the code path that reads it. Then ask for a trace between two implementations and a record of a buy where identity influenced the outcome. | Keep the schema, connector, environment and receiving actor identifiable. Use blinded, permitted test inputs. | A media-buy identifier or a buyer’s reported match-rate field is mistaken for a resolved audience or completed identity-linked transaction. | A connector can prove implementation while production volume and transaction effects remain unknown. [92, 93, 94, 95, 96, 97] |
| Does the new offer use the same underlying graph? | Compare dated inputs, linking rules, treatment of stored edges and output continuity across the predecessor and successor. Ask the provider to name the exact links retained or rebuilt. | Compare the same commercial unit, market, route and assignment level on both sides. | A shared owner, input, namespace or new brand is substituted for evidence about the linkage itself. | A new token or integration cannot by itself establish a new graph. [14, 13, 160, 214, 215, 216, 217, 190] |

Give each question in the tool a name and a date. Ask the owner to attach the result it saw at its end of the route. Write down what passed and for which use. If only half the brief passed, get the buyer to say which half it will buy. If no one could see the answer, record that and a price for the next test. A team coming back at renewal should be able to tell what it bought and where the proof lies. That is a test you can run on the paperwork itself before you sign.

For a buy run by an agent, follow the ID to the point where the agent makes its choice. The experimental AdCP Trusted Match spec defines a request with opaque identity tokens and no page context [91]. Its Go client builds a typed request, and the reference engine can check group membership if it has the required inputs and service [92, 93, 94] [95, 93]. Ask for a trace from the version you will run to the system that should act on it. What did that system decide? Request a separate receipt if the seller also claims a paid buy. Keep both with the terms so the next trader can see what the team agreed to pay for.



## 13. Method


*This paper starts with what a buyer can get [11, 13]. It follows the named US service from the first input to the last output that a public source can show [11, 13]. The plan called for a search through the end of November 11, 2026, in New York [364]. The final source sweep closed early, at 8:37 a.m. EST that day [364].*

> **In short.**
> The register keeps both the services that passed the screen and the leads that fell short [11, 13]. A source can show what a firm said it would sell, what its code does or what a buyer used. The test must fit the claim. A score applies to the named US buy, and each research test has its own rules for which cases count [11, 13]. The author discloses his work in TV, past roles in the field and a role in the group behind AdCP [232, 233].

### Who did the work

OpenAI models set down the rules from the author’s brief, read the sources, wrote the profiles, scored the products and drafted the paper [364]. In the refute pass, other model runs reopened the records and challenged the findings and scores [364]. The last pass checked the cited sentences against the ledger and reopened the sources for names, dates, figures and verdicts [364]. These were internal AI checks [364]. No independent human peer review or second human scoring exercise is documented [364].

After the handoff, a separate Claude Code session checked a sample of 36 claims [364]. It found support as written for 28, slight overstatement in two and partial support in one [364]. It could not open the sources for the other five [364].

### Start with the thing a buyer gets

For each lead, write the line you might see on an order. Does a site get a token it can put in an ad request? Does an agency get a set of links to add to client records? Does a TV team get a list of homes that match a partner's list? Does a desk get the right to bid on ads for a group, with no right to see the names in it? Use that line to trace who gets what. It tells you which end of the trade to start with and what proof to look for on the way back.

The register starts with named leads and adds firms and services found in the source sweep [11, 13]. For each one it keeps the bought service, who runs it, who owns it, its names, its markets and the reason it is in or out [11, 13]. Treat a new name as a lead. Did the buyer get a new output? Can it still buy the old one? A new field in a bid request could give the buyer a new way into the same service. To count a new row, look for proof of a new buy.

Take Epsilon's publisher link. A site can sign up for PubLink, join its own data to CORE ID and return a PubLink ID to a bidder [362]. That gives the study a way to trace a specific handoff while it looks at the graph as a buy in its own right. Merkle gives us a second case. Prebid says its Merkle ID module is a route from a site to the Merkury private graph [363]. The short name a DSP uses for a source may help us find the right guide. It takes more to make a case for a second buy.

### Set the market and record the dates

The study looks first at the US and uses EU and UK records when the named service has a link to that market [11, 13]. Read the terms, the country list, the guide or the case. What does the source place in the country? A US office could run a service in many markets. A US traveler could see an ad abroad. A data center could take records from clients in several countries. Each fact could help in a study of how the firm works. Only some of them show where a client could buy the named service.

Ogury describes ads shown to people on trips to Japan, including US visitors [248]. NIQ names the US as a market for purchase insights it supplies to Ogury for its personas [356]. Keep three columns. Where did the traveler come from? Where did the ad run? For which market did the data firm say it could supply those inputs? If a source shows a US buying route, add it to the case for that buy. You can preserve each fact about the trip without turning a US visitor into proof of an ad shown on US soil.

The study keeps the day a source went public, the day of the event and the day of the check apart [11, 13]. A guide may be right about how to use the service now but give no clue to how it worked last spring. A release may set out a plan for work still to come. Look for a dated copy of the guide, the note on the planned change and the record that says when it went live. When a firm edits a page, try to find the text that was live on the day in question. If that fails, leave the old state open.

### Trace the number from the submitted file

For a match rate, ask who sent which records, who checked them and who did the count. Seek the rows sent in, the inputs left once repeats are gone, the ones that met the rules and the ones that matched. Two email addresses might belong to the same client. If they are merged before the match, the base can change. A bad address could be dropped for good cause. State that rule so the reader can use the same base. Give the top of the fraction the same care. One person, one phone and two devices linked to that phone could each yield a different count.

Say a buyer sends a list with some of the same addresses on it twice. The graph firm drops the repeats, checks the rest and returns a key for some of the rows that pass. The DSP can use just some of those keys. To check the first step, count what is left after the clean-up. To check the match, take the rows that matched as a share of those that met the rules. To check the handoff, follow the keys to the named DSP. To check for error, take pairs for which a source apart from the graph can say what link ought to be there.

For its standard activation rate, LiveRamp counts unique identifiable records due for distribution and leaves out those which are not being sent [15]. Yahoo's file-match report and its check for an ID in an ad request concern two different steps [30, 164]. If a rate came from a file upload, name the job. Which firm got the file? In what time frame? What was the base? Add a line for any test of whether the returned links were right. A clear rate for a file can stand on its own terms. A gap on the line for link tests gives the buyer a clear next question.

### Read the original and try to break the claim

For each finding the ledger ties the claim to a source, a passage and the limits of that source [11, 13]. A fresh review tests the words against the source, logs the result and carries the revised claim forward [11, 13]. Two news sites may carry the same press release. Go back to the firm that made the claim. If a partner who ran a part of the test adds a new fact, use that partner's record for its own part. What is the most a reader could learn from each? Be clear about the interest of a firm that sells the service and of a buyer named in a joint release.

The Yahoo module shows why a line of code can matter. Yahoo's guide says it needs an email hash in versions below Prebid 7.52 [164, 189, 191, 192, 193]. In the released 7.51.0 code, an email hash or a publisher user ID will do [164, 189, 191, 192, 193]. Release 7.52.0 removed the need to send either one [164, 189, 191, 192, 193]. The code answers a narrow question. What will this version accept? To find out which inputs a live site sent, ask for a test from the site. To check who was linked, ask for a test with known answers. The source that settled the first point has already done its job.

A source can stay on the list when a claim based on it has to change. Keep the words at issue, what the source says and the row or cell where the change was made. For a sales figure, check which firm filed it. For a code change, check the release itself. For a claim about IP, does the guide refer to a home's current link, a past link or a network signal seen once? If the source can only show what was true in the past, keep that time in the sentence that uses it. A reader can then see what still needs a new check.

### Preserve the difference between a disclosure and a test

The study grades the proof for the score it gives to each cell [11, 13]. A firm may give precise guidance on the fields its API takes. A partner may know which part it ran. Neither role alone makes it a third party with no stake in the test. If a score needs several facts, check them one at a time. The grade goes with the weakest source needed to meet the rule for that score [11, 13]. A strong source for the date of a launch may settle the date. You still need the right source for a claim that a link worked.

Say a team wants to check links from two records to the same person. Start with a pair that should match. Add a pair that should stay apart. Who set the true label for each pair? Did the team that built the graph see those labels while it trained the model? Include true pairs the service missed if the aim is to estimate recall. Add pairs the service turned down if the rate you seek needs them in its base. A test of just the links that the graph accepted can answer a question about that set. It leaves the number of true links missed open.

Experian describes a DSP that got a weekly graph file with links from cookies to other digital IDs [37, 36]. Experian says the DSP synced 84 percent of IDs within weeks, with no precise base for that share [36, 37]. The case gives a reader a reported route to the DSP and a claim about how many IDs synced [37, 36] [36, 37]. Ask for the inputs that went into the rate, the ones that failed and the pairs for which someone checked the true link. Each would fill a different gap. The arrival of a new file each week would settle none of those questions on its own.

### Choose the test population before its outcome

The tests set different rules for what can settle the claim [11, 13]. In some, one case can overturn the claim [11, 13]. In others the result turns on whether more than half of the products meet a test [11, 13]. Read the group first. Is it made up of scored services, US TV household buys, dated graph launches or cases where a public order names an in-scope service? The same case may belong in one test and fall outside the next. State which rule put it in the group before looking at how the case might change the result.

For a test of more than half, ask two things. Do we know whether this product is in scope? If so, do we know its outcome? A case that is in scope but has an unknown result stays in the main count [11, 13]. A case whose scope is in doubt goes into a wider check [11, 13]. For the low end of the range, assume that none of the unknowns qualify. For the high end, assume they all do. If the high end is exactly half, the claim of more than half fails [11, 13]. If the low end is half and the high end above it, leave the result open [11, 13].

LtvAdx shows why both steps count. Its guide lets a shared IP help place a device in a home [227, 228, 229, 230, 231]. It says the device link drops out thirty days after the device was last seen [227, 228, 229, 230, 231]. The US ad plan bills for ads shown and lets a buyer target a HouseholdID [227, 229, 230, 231]. The reviewed terms did not establish a separate US purchase of household resolution [227, 229, 230, 231]. Keep the rule for the link in the record. It is a clear example of what a TV buyer might ask for, even as the right to buy that service on its own stays open [227, 228, 229, 230, 231] [227, 229, 230, 231].

### Follow the clock, the field and the duty

For a time claim, name the thing, the firm that acts, the event that starts the clock and the time it has to run. Does the clock start when a device is seen, when a file comes in, when an ID is made or when a user opts out? What should happen at the end? Should the firm stop using the link? Should the row drop out of a file? Should it get a new key or just be checked again? If the job then falls to a new firm, trace that handoff. The trade may depend on what the next firm does.

Blockgraph's resolver can look up past IP records in a set time frame [114]. Read the manual-refresh job separately. A manual refresh replaces the node's identity layer with the supplied file [115]. Spectrum Reach gives a length of time after a TV ad during which site visits can count for that ad [167, 38]. For each job, write the name of the clock next to its start date. What is the age limit of the current home link itself? To settle that, look for a rule that ages or drops that link. A buyer can use this wording to get the request to the right team.

For a rights claim, follow the choice from the site to the firm that gets the ID. LiveRamp tells a site handling an ATS opt-out to drop its local envelope and set a flag to stop the envelope from coming back till the user opts in [138]. Its guides for a different workflow describe requests to ad partners and leave the work to the firm that gets the request [139]. Ask for the ID, the job and the system at each end. A record of the user's choice can show where the process starts. A record of a change to a past export could show where it ends.

### Freeze the scoring rule and retain the exceptions

The study froze the buying tasks and the scoring rules before the team opened the cells [11, 13]. To enter the matrix a service had to be live or in beta in the US, fit the task in its peer group and have enough scored cells [11, 13]. The mask sets which cells apply to a group [11, 13]. If a cell applies but lacks a score, keep the reason in the note. Use zero only when the source meets the rule for zero. A price kept in a private agreement or a test that a reader cannot open does not show how well or how poorly the service works.

The second pass opened the sources for the proposed scores and logged which scores held up, needed a change or lost their support [11, 13]. The paper has no independent second scorer [11, 13]. There are 324 cell checks, 15 changes and 12 proposals that lost their support [10, 218, 11]. To challenge a score, name the product, the cell and the version you used. Which premise in the note does your source change? If it puts a new fact on the record, state the date. That helps distinguish a fact that was wrong at the cutoff from a change that took place after the study closed.

### Distinguish a power from its use

For a court or regulator record, name the party, the act at issue, the date, the body that heard it and the stage of the case. Is it a complaint, an order, an appeal or a record that shows who did what under the order? If the record names a product, keep its name with the source. If it describes a type of work, can you tie that work to the service on the list? A firm may have a graph, a unit that measures ads and a business that sells other data. Which one did the order cover?

In the private Yahoo case, a US court sent named claims to arbitration and stayed the court case for those claims [90, 8]. That tells us what the court did. To code an actual change, ask what changed in the way the product collects data, keeps it, shares it, links it or acts on a request to stop use. A duty with a due date may tell us what to look for in a future report. A reader still needs proof of the work and when it was done. If the search turns up just the duty, keep the date of the change open.

Give code the same care. The AdCP Go client takes a token and its namespace type, builds a request and sends it to the Trusted Match route [92, 93, 94]. With an audience service set up, the checked code can test for group membership and the handler can send back the IDs of eligible packages [95, 93] [108, 93]. To show a live trade, look for the request and the response and ask who was at each end. To show a paid use, ask where the record shows that step. Working code can tell a reader what to test even when the source trail stops at the connector.

### Declare the author's interests

Evgeny Popov has worked at Samba TV since May 2025 [232, 233]. Samba sells TV measurement and ACR data and maintains a household graph [232, 233]. The paper excludes Samba from scoring, profile selection, cohort field means and every hypothesis denominator restricted to scored products [232, 233]. It retains Samba for the separate legal screen under that test's stated attribution rule [232, 233]. This preserves the scope of the legal review while keeping the author's employer out of scored comparisons [232, 233].

Popov co-leads the Signals & Measurement working group of AgenticAdvertising.org, the organization associated with AdCP [232, 233]. The protocol analysis carries that disclosure and applies the same evidence threshold to a schema, a connector, an observed execution and a production use [11, 13]. Popov's prior roles include EVP at Verve Group from September 2022 to August 2024 and roles at Hearts & Science and Lotame [232, 233]. The study flags Lotame Panorama ID and the separately screened former-employer product from Verve [232, 233]. The author's published doubts about match rates and relabeled graphs supplied rival hypotheses to test [232, 233]. His views supply no corroborating evidence for either result [11, 13]. The study used public records and conducted no vendor outreach [232, 233].

![What made it into the IQ](figures/fig-evidence-review.svg)

*Figure fig-evidence-review. **What made it into the IQ** The cell review includes all proposals, unknowns and masks. The last panel counts only usable evidence for products admitted to the IQ. Units: streams, profiles and cell decisions. Scope: US buying tasks through the research cutoff. Sources [11], [10], [218]. Caveat: Internal verification cannot establish whether a vendor’s links are accurate in a buyer’s own file. Data: data/stats.json*



## Sources, scores and changes


*Use this page to check a line in the paper. Start with the link next to the claim. Read the part of the source that backs it. Look at what the firm sold, who took it and what they could do with it.*

> **In short.**
> For a score, find the product and its bought task. For a test, read the rule and the result. For a claim that changed, open the old and new text side by side. Keep the scope with the claim when you cite it.

### Check a line

Follow its mark to the claim ledger. Read the final words, the short quote and the place it came from. Then open the source. Who speaks? Whose data did they use? What dates does the source give for the work? Which date belongs to the news release? Check whether a case gives you the date of a test, the day of its release or both.

For a company filing, find the line and check the name on the account. If you need sales for one product, look for that product in the line. For a case study, check who paid for the work, who ran the test and who put out the news. Ask for the full test if all you have is the firm’s account of it.

The table shows the claims used in the paper or marked as headlines. The [full claim ledger](data/claim-ledger.json) also gives the source passage and locator for each claim.

**Claim ledger**

| ID | Claim (final wording) | Ch. | Evidence class | Review | Confidence | Sources |
|---|---|---|---|---|---|---|
| IDN-HYP-H8 (headline) | H8 remains unresolved. Five confirmed change cases lacked decisive evidence that the named service continued, or replaced, a specific predecessor’s graph links. Ownership, continuity of a key, common household data and migration of software did not settle reuse. The primary and expanded ranges cross the registered threshold. | abstract | synthesis | supported as stated | medium | [14, 13, 160, 214, 215, 216, 217] |
| IDN-HYP-H2 (headline) | H2 remains unresolved. No retrieved public test established independently labelled US person or household accuracy for an admitted named graph with the required attribution. Missing protocols, sponsor and unit information for credible historical cases prevent the negative claim from passing. The expanded eligible graph frame yielded the same verdict. | abstract, 3 | synthesis | supported as stated | medium | [14, 13, 40, 41, 43, 45, 62] |
| IDN-HYP-H1 (headline) | One admitted US case contradicted H1. Deep Sync reported matching 73.0% of one grocery chain’s monthly POS transactions to an inferred buyer through sequential loyalty, payment and household/location matching. The disclosed denominator, output population and method count as yield. Correct-person accuracy was not tested by that rate. | abstract, 3, 11 | synthesis | supported as stated | medium | [14, 13, 12] |
| IDN-HYP-H3 (headline) | H3 was contradicted in the confirmed US-available product frame. Two of thirteen products documented all links in at least one applicable jurisdiction, nine did not meet the test and two remained unknown. Even counting those two as successes gives four of thirteen. Six more products had unresolved eligibility, and the expanded bounds of two to ten out of nineteen cross half. Jurisdiction-specific scopes are reported separately. | abstract, 4 | synthesis | supported as stated | medium | [14, 13, 149, 150, 186, 175, 128, 182, 219] |
| IDN-HYP-H4 (headline) | H4 remains unresolved. In the confirmed cookie-capable scored web frame, the public record did not establish a complete before-and-after comparison of the same primary guide and route. Browser announcements and code defaults alone could not establish that a majority of product guides shifted to more cookie-based syncing. The expanded bounds also span the decision threshold. | abstract, 5 | synthesis | supported as stated | medium | [14, 13, 4, 220, 221, 222, 179, 223, 180, 224, 225, 226] |
| IDN-HYP-H5 (headline) | Only Blockgraph met the US buying test for a CTV household-resolution service. Its public guides document IP as an input and give no elapsed-time rule for refreshing or expiring the live link. The registered majority claim was contradicted in that one-product frame (0 of 1). Eight other candidates had uncertain eligibility. Counting all eight as in scope gives a range of 0 of 9 to 8 of 9, which leaves the broader market result unresolved. LtvAdx documents a 30-day device-link rule and paid US household targeting. The records reviewed did not establish a separate US purchase of household resolution. | abstract, 7 | synthesis | supported as stated | medium | [14, 13, 109, 114, 115, 227, 228, 229, 230, 231] |
| IDN-HYP-H6 (headline) | The reviewed enforcement records establish duties and some operational changes. For the identity products in scope, the search could not tie an operative order to a separately corroborated change in the named service. H6 remains unresolved. | abstract, 9 | synthesis | supported as stated | medium | [14, 13, 65, 68, 70, 79] |
| IDN-METHOD-F01 (headline) | Evgeny Popov has worked at Samba TV since May 2025. Samba operates a household graph and sells TV measurement and ACR data. His former employers include Lotame, Verve Group and Hearts &amp; Science. He co-leads the Signals &amp; Measurement working group at AgenticAdvertising.org. Samba is excluded from IQ scoring. Lotame Panorama ID and Verve Moments.AI are flagged. His published positions motivated tests and did not supply product proof. | abstract, 9, 10, 13 | synthesis | supported as stated | medium | [232, 233] |
| IDN-HYP-H7 (headline) | The AdCP Go publisher client sends a typed ID token to the Trusted Match router. With an audience service, a token and a target segment, the reference engine can check audience membership. This implementation contradicts the registered schema-only claim. The reviewed records did not establish a paid purchase that used this ID path. The author co-leads the Signals &amp; Measurement working group at AgenticAdvertising.org. | abstract, 10 | synthesis | supported as stated | medium | [14, 13, 92, 95, 108, 101, 232] |
| IDN-METHOD-F02 (headline) | The frozen study evaluates public evidence for separate US buying tasks. The preregistration supplies populations and decision rules. Depth, Reach and Standing use cohort-specific scores, masks and caps. Legal records and author opinions are outside the IQ. Research, refutation and citation checks retain traceable source passages and dispositions, as recorded in the retained study files. | abstract, 11, 13, appendix | synthesis | supported as stated | medium | [11, 13] |
| IDN-IQ-F01 (headline) | The frozen register contains 57 records and 27 eligible products. Of these, 24 attained at least four counted cells. The IQ uses 142 counted cells across the admitted products and reports seven documentation bands, all in cohort A. Fourteen admitted products have thin records and three cannot receive a band because their cohort is too small. Each group, sensitivity result and cell is recorded in the official matrix. | abstract, 11, 13, appendix | synthesis | supported as stated | medium | [10, 218, 11] |
| IDN-U-F01 | LiveRamp describes RampIDs as pseudonymous identifiers. Its API can return a maintained RampID for an identity graph match or a derived RampID when it finds no match. | 1 | vendor assertion | supported with qualification | medium | [234, 181] |
| IDN-U-F12 | LiveRamp documents two known-identity resolution routes. Its API takes consumer PII and returns AbiliTec Links. Its cloud service takes PII or email addresses and returns Known IDs. | 1 | vendor assertion | supported with qualification | medium | [235] |
| IDN-U-F20 | Audigent describes a cookieless media-curation route using TransUnion audiences. | 1 | vendor assertion | supported as stated | medium | [236, 237, 238, 239] |
| IDN-U-F31 | AWS describes a Clean Rooms query as accessing and analyzing configured tables in a collaboration. | 1 | vendor assertion | supported as stated | high | [240] |
| IDN-U-F32 | Microsoft documents Curate as enabling curators to enhance a seller’s inventory with their own assets. | 1 | vendor assertion | supported as stated | high | [241] |
| IDN-U-F33 | OneTrust describes its CMP as capturing and signaling purpose-based consent. | 1 | vendor assertion | supported as stated | medium | [242] |
| IDN-SB-F01 | Microsoft documents a US-only AbiliTec preview that requires active LiveRamp and AbiliTec subscriptions and credentials. A buyer maps profile fields to retrieve person-level AbiliTec IDs. | 1 | direct record | supported as stated | medium | [212] |
| IDN-SB-F02 | LiveRamp’s OAuth2 REST API accepts personal identifiers and returns AbiliTec identifiers and match information for matched records, using single or batch calls. | 1 | direct record | supported as stated | medium | [243] |
| IDN-SB-F06 | Zeta documents US-only Streamlit applications. Its Customer Growth Intelligence workflow takes first-party data through a Snowflake share and exposes enriched views. The same guide separately describes Customer Matcher scoring and pairing two client tables. | 1 | direct record | supported as stated | medium | [244] |
| IDN-UID-F06 | UID2’s permission matrix lets a publisher request tokens, a mapper submit email addresses, phone numbers or their hashes for raw IDs, and a bidder decrypt tokens. The FAQ illustrates an encrypted token in user.ext.eids with uidapi.com as its source. | 1, 2, 11 | vendor assertion | supported as stated | high | [143, 144] |
| IDN-LR-F02 | LiveRamp’s ATS security guide says an input hash can yield a derived RampID even when its representation is absent from the offline graph. | 1, 3 | vendor assertion | supported as stated | high | [134] |
| IDN-U-F24 | TransUnion said Epsilon audiences were available through the TruAudience Data Marketplace using its US connected-home graph. A separate Comcast agreement described using TransUnion’s graph to build segments for TV campaigns on Comcast. Both records describe audience activation. | 1, 7 | vendor assertion | supported with qualification | medium | [245, 246, 247] |
| IDN-CTV-F12 | Spectrum Reach describes a production workflow in which it supplies household TV exposure, Innovid supplies household website visits and Blockgraph performs the match. | 1, 7, 8, 12 | vendor assertion | supported as stated | medium | [38, 167] |
| IDN-CTV-F13 | Spectrum Reach’s showcase describes visits counted within seven days of a TV exposure in its multiscreen attribution workflow. | 1, 7, 12, 13 | vendor assertion | supported as stated | medium | [167, 38] |
| IDN-EXP-F02 | Experian markets a separate Offline Graph for names, addresses and emails, and a joint offering pairs Digital Graph with Marketing Attributes. | 1, 8 | vendor assertion | supported as stated | medium | [120, 121] |
| IDN-UID-F05 | The UID2 guide names The Trade Desk as Core Administrator and a public operator. The core supplies keys and salts. Public or private operators create raw IDs and encrypt them for token issuance. Publishers send tokens via SSPs, and DSPs can use them for bidding. | 1, 8, 11 | vendor assertion | supported as stated | high | [142] |
| IDN-EXP-F01 | Experian offers its US Digital Graph as a whole dataset or as links appended to a buyer’s own records. | 1, 8, 11, 12 | vendor assertion | supported as stated | medium | [117] |
| IDN-UID-F09 | The Trade Desk’s license matrix distinguishes approved API access from CRM onboarding. CRM onboarding permits targeting in the platform account without returning the IDs or customer mapping. UID2 also says a contracted participant can use a Public Operator at no charge. Private operators provide their own hosting and engineering resources. | 1, 8, 12 | vendor assertion | supported with qualification | high | [188, 203] |
| IDN-EXP-F03 | Experian says it clusters digital identifiers into household or individual links using signals that include timing, IP addresses and browser or device identifiers. | 1, 11 | vendor assertion | supported as stated | medium | [36, 117] |
| IDN-U-F19 | Permutive documents activating publisher cohorts through deals. Google says TelevisaUnivision sent Permutive’s standard cohorts in Ad Manager requests. It reports a mean of 1.24 audience categories per impression on the publisher’s US inventory. OpenX separately announced a US buying route for Permutive audiences without third-party cookies or device IDs. | 1, 12 | vendor assertion | supported with qualification | medium | [205, 206, 207, 208] |
| IDN-U-F21 | Ogury says it used Personified Advertising to target travelers arriving in Japan without relying on their advertising IDs during an All Nippon Airways campaign from January 17 to February 15, 2025. The travelers included visitors from the United States. | 1, 13 | vendor assertion | supported with qualification | medium | [248] |
| IDN-A-F12 | Yahoo’s ConnectID toggle stores a domain-specific opt-out in local storage and stops its JavaScript module from generating identifiers when the opt-out flag is set. | 2 | vendor assertion | supported as stated | high | [201] |
| IDN-A-F15 | Prebid’s Parrable integration guide says the module’s cookie holds the identifier and the user’s opt-out state. | 2 | vendor assertion | supported as stated | high | [249] |
| IDN-A-F16 | Prebid names criteo.com as the EID source for the Criteo ID for Exchanges module. | 2 | vendor assertion | supported as stated | high | [250] |
| IDN-A-F18 | IAB Tech Lab reports that Seller Defined Audiences was renamed Curated Audiences in December 2024. | 2 | direct record | supported as stated | high | [251] |
| IDN-U-F11 | Prebid describes Parrable’s identifier as working across browsers and webviews on one device. | 2 | vendor assertion | supported as stated | medium | [249] |
| IDN-UID-F07 | UID2 says participants mapping the same email at the same time get the same raw ID. Each generated advertising token differs. The FAQ recommends hourly refresh for publishers who refresh tokens manually, and daily raw-ID timestamp checks for audience updates. A raw ID changes roughly once a year. | 2 | vendor assertion | supported with qualification | high | [144, 146] |
| IDN-UID-F08 | On July 11, 2025 UID2 announced v3 identity mapping for advertisers and data providers. The migration guide replaces v2 salt-bucket monitoring with per-ID refresh timestamps and makes the prior raw ID available for 90 days after rotation. | 2 | vendor assertion | supported as stated | high | [145, 146] |
| IDN-A-F01 | IAB Tech Lab added EID provenance attributes for who inserted or bridged an ID and the match method in OpenRTB 2.6. | 2, 4 | direct record | supported as stated | high | [252, 1] |
| IDN-A-F02 | GPP is designed to carry privacy and choice signals from sites and apps to ad-tech providers across supported jurisdictions. | 2, 4 | direct record | supported as stated | high | [253] |
| IDN-A-F03 | IAB Europe says newly created TC strings lacking the disclosed-vendor segment are invalid from March 1, 2026, while earlier strings can remain valid until users renew or change their choices. | 2, 4 | direct record | supported as stated | high | [254] |
| IDN-A-F04 | The September 2026 W3C Global Privacy Control Working Draft defines a signal sent through HTTP and the DOM to request that sites and services refrain from selling or sharing personal information. | 2, 4 | direct record | supported as stated | high | [3] |
| IDN-A-F05 | From Prebid.js 11.14, user-ID submodules skip retrieval and add no IDs to bid requests when a configured TCF or GPP consent-platform lookup fails. On a timeout, the submodules run with the consent data available. | 2, 4 | direct record | supported with qualification | high | [239, 255] |
| IDN-A-F07 | UID2’s portal accepts an ecosystem opt-out for a supplied email address or phone number. Publishers learn the result when they generate or refresh a token, and the operator notifies DSPs through a separate webhook. Propagation can take time. | 2, 4 | vendor assertion | supported with qualification | high | [149, 186] |
| IDN-A-F08 | The UID2 DSP guide forbids using an opted-out UID2 in real-time bidding and allows the DSP to use another identifier or decline to bid. | 2, 4 | vendor assertion | supported as stated | high | [186, 149] |
| IDN-A-F09 | ID5 says its Prebid Server Java module forwards available privacy signals to the ID5 API during identifier generation. | 2, 4 | vendor assertion | supported as stated | high | [256] |
| IDN-A-F11 | LiveRamp says its ATS JavaScript checks a US GPP National or fallback CCPA string and removes the stored RampID envelope when consent is rejected. | 2, 4 | vendor assertion | supported as stated | high | [257] |
| IDN-A-F17 | Prebid describes SharedID as an ID stored in the publisher’s first-party domain and made available to adapters. | 2, 4 | vendor assertion | supported as stated | high | [258, 239] |
| IDN-B-F05 | LiveRamp attributes a 95%–98.9% graph precision range to comScore in a note on deterministic matching. | 3 | vendor assertion | supported as stated | high | [43] |
| IDN-B-F06 | Tapad reported that its Device Graph achieved twice the precision of the runner-up in a 2016 Hotels.com evaluation. | 3 | vendor assertion | supported as stated | high | [45] |
| IDN-B-F12 | A LiveRamp PDF bearing a 2017 copyright and an online help page last modified in August 2020 repeat the same comScore precision range. | 3 | synthesis | supported with qualification | high | [44, 43] |
| IDN-B-F13 | In its Winter 2015 portfolio newsletter, Firsthand Technology Value Fund said Tapad had commissioned Nielsen to test Tapad’s Device Graph. | 3 | vendor assertion | supported with qualification | high | [53, 54] |
| IDN-C-F02 | LiveRamp illustrates how one recognized device among four can yield 25% device recognition while the same person’s email yields a 100% activation rate at a destination. | 3 | illustrative scenario | supported as stated | medium | [16] |
| IDN-C-F04 | In a Data Health Assessment for one unnamed US insurer, TransUnion says it matched 99.5% of the insurer’s records to a person in its offline identity graph. The report also gives a 97.4% match on first name, last name and at least one other PII field. It does not disclose the full acceptance rule for the 99.5% rate. | 3 | vendor assertion | supported with qualification | high | [17, 18, 19] |
| IDN-C-F05 | Experian reports a 70% match between IP addresses supplied by MiQ and cookieless IDs in the Experian/MiQ engagement. | 3 | vendor assertion | supported as stated | high | [20, 21] |
| IDN-C-F08 | In March 2022 Cadent advertised an average match rate above 90% for audience segments using Aperture Viewer Graph. | 3 | vendor assertion | supported as stated | high | [24] |
| IDN-C-F10 | In July 2026, Forrester used four customer interviews in a TransUnion-funded study to model a rise in household match rate from 60% to 80% for a composite enterprise. The 20-percentage-point gain is a model assumption. | 3 | commissioned measurement | supported with qualification | high | [26] |
| IDN-C-F15 | Deep Sync reports resolving 83.5% of a US outdoor retailer’s full CRM file with its Profile Hygiene &amp; Enrichment service using 19 enabled matching rules against its graph. | 3 | vendor assertion | supported as stated | high | [27, 28] |
| IDN-C-F26 | For one US mass retailer, Deep Sync’s UID2 row reports 73.2 million matched records (51%) before CRM enrichment and 94.4 million (66%) after. The case gives a nominal file of 141.7 million records but does not explain the whole-percent rounding or exact destination denominator. | 3 | vendor assertion | supported with qualification | high | [31] |
| IDN-EXP-F07 | Experian reports a 70% match rate from MiQ-supplied IP addresses to cookieless identifiers in the MiQ engagement. | 3 | vendor assertion | supported as stated | medium | [20, 21] |
| IDN-EXP-F14 | Tapad reported that its Device Graph had twice the precision of the next competitor in a 2016 Hotels.com evaluation. | 3 | vendor assertion | supported as stated | medium | [45, 46] |
| IDN-ID5-F15 | ID5’s Match Preview defines a rate for each output ID type by dividing distinct submitted hashed emails having at least one match by all distinct input identifiers; the table illustrates the calculation. | 3 | vendor assertion | supported as stated | medium | [32] |
| IDN-LR-F12 | LiveRamp attributes 95%–98.9% precision for its historical graph to a comScore test; the reviewed note does not establish current US ATS person accuracy. | 3 | vendor assertion | supported as stated | medium | [43] |
| IDN-UID-F17 | In its 2024 US report, The Trade Desk says HP used device-registration data for a UID2 campaign on Disney+ and Hulu through its DSP and reached 8% of an estimated 12.5 million target households. | 3 | vendor assertion | supported as stated | high | [147] |
| IDN-LR-F10 | Google reports that New York Times Advertising used Secure Signals with both RampID and UID2 and reports an 81% revenue gain on restricted-cookie browsers in its 2025 publisher results. | 3, 5 | vendor assertion | supported as stated | high | [59] |
| IDN-B-F01 | In a US benchmark commissioned by CIMM and Go Addressable, Truthset reported average modeled IP-to-postal linkage accuracy of 13% across six unnamed providers for December 2024 through February 2025. | 3, 7 | commissioned measurement | supported with qualification | high | [40, 41] |
| IDN-B-F02 | In the same December 2024 through February 2025 US benchmark, Truthset reported average modeled IP-to-email linkage accuracy of 16% across six unnamed providers. | 3, 7 | commissioned measurement | supported with qualification | high | [40, 41] |
| IDN-B-F03 | CIMM’s comparative TV study reported that the post-match viewing profile skewed toward older, higher-income and heavier viewers. | 3, 7 | commissioned measurement | supported as stated | high | [42] |
| IDN-B-F04 | The CIMM TV match study expressly lacked the independent validation data needed to test whether individual links were correct. | 3, 7 | commissioned measurement | supported as stated | high | [42] |
| IDN-CTV-F11 | Kantar reported that its match rates with TV partners rose to more than ten times their prior level after integrating Blockgraph’s IDoS. The partners disclosed neither absolute rate nor the eligible record counts. | 3, 7 | vendor assertion | supported with qualification | medium | [112, 113] |
| IDN-EXP-F06 | Experian says MiQ licensed hashed emails, UID2s and partner cookies for its Identity Spine, which its traders and analysts use for targeting and measurement. | 3, 8 | vendor assertion | supported as stated | medium | [21, 118] |
| IDN-I-F02 | CIMM’s April 2026 buyer guide recommends establishing a truth set to evaluate match accuracy and distinguishes precision from recall. | 3, 12 | synthesis | supported with qualification | high | [259] |
| IDN-LR-F11 | For its standard activation rate, LiveRamp divides unique identifiable records matched to one or more online devices at a destination by eligible unique identifiable records, excluding records not being distributed. Facebook uses a separate calculation. | 3, 13 | vendor assertion | supported with qualification | high | [15] |
| IDN-A-F06 | Prebid documents a publisher option that can require TCF Purpose 4 consent before passing user IDs and EIDs. | 4 | direct record | supported as stated | high | [2] |
| IDN-A-F10 | ID5 describes its DSP decryption model as restricting access to platforms authorized under the consumer’s and publisher’s preferences. | 4 | vendor assertion | supported as stated | high | [176] |
| IDN-ID5-F09 | ID5 instructs publishers with US-originating traffic to include it in their do-not-sell/share signal flow, such as GPP or GPC, and gives separate TCF or opt-in instructions for EEA and UK traffic. | 4 | vendor assertion | supported as stated | medium | [177, 131] |
| IDN-ID5-F10 | ID5 describes an SSP sending the ID in an OpenRTB request and a DSP separately obtaining decryption keys and merge instructions that link publisher-specific IDs. | 4 | vendor assertion | supported as stated | medium | [176] |
| IDN-LR-F05 | LiveRamp says US ATS.js checks national GPP first and falls back to its legacy CCPA string, whose presence is optional. Its guide groups the UK and EU in one consent branch and specifies TCF v2.3 for the EU/EEA, with vendor 97 and purpose 1 for envelope retrieval. | 4 | vendor assertion | supported with qualification | high | [182] |
| IDN-UID-F23 | The UID2 administrator says it records opt-outs and shares them with participants. Its US-relevant client addendum separately describes the company and The Trade Desk as controller and processor for covered customer data. These terms and the bidder guide describe obligations. The FAQ’s sample bid request and DSP guide do not demonstrate per-impression TCF or GPP enforcement. | 4 | synthesis | supported with qualification | high | [150, 144, 186, 187] |
| IDN-YAH-F05 | Yahoo documents an authenticated server lookup using a hashed email and publisher ID. An eligible lookup can return a ConnectID for a bid, while an opted-out lookup returns an empty object. Its Prebid route can instead start with a domain-specific publisher ID. | 4 | direct record | supported with qualification | high | [165, 189, 163] |
| IDN-YAH-F07 | Yahoo instructs sites to honor user choices, provides an email confirmation form for opting out of ConnectID-based targeting, and documents an empty response for an opted-out server lookup. | 4 | direct record | supported as stated | high | [160, 200, 165] |
| IDN-YAH-F08 | A publisher can also set a local opt-out flag that stops the ConnectID JavaScript module from issuing further IDs on that domain. | 4 | direct record | supported as stated | high | [201, 164, 195] |
| IDN-A5-F07 | LiveIntent’s module uses a publisher-side identifier and an API to resolve nonID, linked to an active encrypted email. Adapters can receive the result for the bid request. LiveConnect receives a US privacy string and the guide says it honors opt-outs. | 4 | direct record | supported as stated | medium | [260] |
| IDN-HX-F03 | Lotame says an opt-out applies to the requesting browser and a stored Panorama ID can take up to seven days to expire. If third-party cookies are blocked, Lotame says its probabilistic recognition may fail and it may assign a new ID. The notice does not demonstrate a bidder acting on the request. | 4 | direct record | supported with qualification | medium | [261] |
| IDN-HX-F04 | LiveIntent’s guide lists a GDPR flag and consent string, US Privacy string, GPP string and applicable sections on eligible identity-resolution requests; in privacy mode, the guide says the hub blocks identity-resolution requests. It gives no downstream bidder rule for a particular nonID. | 4 | vendor assertion | supported with qualification | medium | [219] |
| IDN-HX-F05 | In a 2020 California publisher FAQ, TripleLift said it forwarded the bid request and the consent status to the DSP and left targeting to the DSP. This historical exchange policy does not identify how a particular bidder handles a particular advertising ID. | 4 | vendor assertion | supported with qualification | medium | [262] |
| IDN-ID5-F08 | Prebid documents that the id5-sync.com EID should remain present, with its ID set to the string “0” when consent is absent or the user opts out. | 4, 5 | vendor assertion | supported with qualification | medium | [175] |
| IDN-A5-F01 | TripleLift’s April 2026 exchange table lists US passthrough for Lotame Panorama ID, SharedID, Audigent Hadron ID, Criteo ID, Neustar Fabrik ID (exchange spelling) and UTIQ ID. It lists US bridging for 33Across Lexicon ID. The same US table lists Epsilon PubLink and Flashtalking FTrack for passthrough and LiveIntent for bridging. The exchange labels that row by provider. | 4, 11 | direct record | supported as stated | medium | [263] |
| IDN-A5-F02 | Lotame’s Michigan case with Advance Local and Rush Street Interactive reports that the publisher used its PubMatic SSP seat for a Panorama ID-enabled private marketplace and that the campaign delivered through Xandr and Adelphic. | 4, 11 | vendor assertion | supported as stated | medium | [264] |
| IDN-ID5-F13 | ID5 specifies 30-minute opt-out updates, hourly routine deletion updates and a separate best-effort feed for on-demand deletions. Partners are expected to ingest and process these S3 lists; their completion time is unverified. | 4, 12 | vendor assertion | supported with qualification | medium | [128] |
| IDN-UID-F12 | UID2 requires a participant to stop generating the identifier after a site-level opt-out. That request is not relayed to UID2. For ecosystem-wide suppression, a consumer submits each email or phone number separately. The guide describes opt-out responses to publisher generation or refresh requests and a webhook to DSPs. | 4, 12 | vendor assertion | supported with qualification | high | [149, 150] |
| IDN-UID-F13 | UID2’s DSP guide describes a webhook sent typically within a few minutes, with the raw ID and opt-out time. It instructs a bidder to suppress UID2 for that user but permits another identifier or a no-bid. | 4, 12 | vendor assertion | supported as stated | high | [186] |
| IDN-YAH-F09 | Google lists ConnectID as a web secure signal. For Authorized Buyers, Yahoo tells Ad Manager publishers to select the Yahoo App &amp; Display and Yahoo Video bidders. Its Open Bidding instructions use a sharing toggle. | 4, 12 | direct record | supported with qualification | high | [156, 157] |
| IDN-LR-F06 | For a publisher-managed opt-out, LiveRamp instructs the publisher to delete the local envelope and keep a flag to prevent its recreation until the user opts back in. | 4, 12, 13 | vendor assertion | supported as stated | high | [138] |
| IDN-LR-F07 | LiveRamp publishes opt-out and deletion requests for activation destinations and describes direct delivery to The Trade Desk for advertiser first-party audiences and third-party marketplace segments. Its guide leaves processing to the destination. | 4, 13 | vendor assertion | supported with qualification | high | [139] |
| IDN-E-F01 | On April 22, 2025 Google retained Chrome’s existing third-party-cookie choice and declined to introduce a separate prompt. | 5 | direct record | supported as stated | high | [4] |
| IDN-E-F02 | Google’s October decision selected named Privacy Sandbox features for retirement through Chrome and Android processes while continuing support for CHIPS and FedCM. | 5 | direct record | supported as stated | high | [220] |
| IDN-E-F03 | Chrome 144 described removal of Private Aggregation as planned; Chrome 152 later recorded its removal. | 5 | direct record | supported as stated | high | [265, 266] |
| IDN-E-F04 | A July Chromium commit reported that Topics’ web-exposed interfaces had already become stubs. | 5 | direct record | supported as stated | high | [267] |
| IDN-E-F05 | Chromium’s July 14 Protected Audience change replaced renderer behavior, including runAdAuction, with rejected promises or default values. | 5 | direct record | supported as stated | high | [268, 5] |
| IDN-E-F06 | Chromium’s July 14 Attribution Reporting change retained the XMLHttpRequest setter as a no-op. | 5 | direct record | supported as stated | high | [269] |
| IDN-E-F07 | In a September project update, the Topics engineers reported that the stable stub-removal trial had reached 1% on milestone 153 and later. | 5 | direct record | supported as stated | high | [270, 267] |
| IDN-E-F08 | On October 8, 2026, the Attribution Reporting team reported that its window.open compatibility fix had reached Chrome 155 Stable. It proposed a staged rollout of the remaining API removal to 100% of Stable after October 24. | 5 | direct record | supported with qualification | high | [271, 269] |
| IDN-E-F09 | Google’s mobile-ads SDK guidance dates the Android Privacy Sandbox deprecation to October 17, 2025. | 5 | direct record | supported as stated | high | [272, 220] |
| IDN-E-F10 | The Android API reference marks Topics classes deprecated at API level 37 and describes future rejection of calls. | 5 | direct record | supported as stated | high | [273] |
| IDN-E-F12 | After a user deletes the Android advertising ID, Google Play services describes returning zeros to attempts to access it. | 5 | direct record | supported as stated | high | [274] |
| IDN-E-F13 | For the 14.5 releases of iOS, iPadOS and tvOS, Apple required apps to request permission for tracking or access to the advertising identifier, effective April 26, 2021. | 5 | direct record | supported as stated | high | [275] |
| IDN-E-F14 | In March 2020 WebKit announced default blocking of cross-site cookies in Safari 13.1 on macOS and the iOS/iPadOS 13.4 updates. | 5 | direct record | supported as stated | high | [276] |
| IDN-E-F15 | On October 19, 2026, WebKit reported that Safari and WebKit on iOS, iPadOS, macOS and visionOS 27 block network requests at the domain level to third-party identity-resolution infrastructure used for cross-site tracking. | 5 | direct record | supported with qualification | high | [151, 152] |
| IDN-E-F16 | UID2’s Prebid.js guide says that client-side requests to uidapi.com are blocked in all browsers on iOS 27 or later and in Safari on macOS 27 or later. It directs affected publishers to a server-side integration. | 5 | vendor assertion | supported with qualification | medium | [152, 151] |
| IDN-E-F17 | Mozilla describes Firefox Total Cookie Protection as maintaining a separate cookie store for each website the user visits. | 5 | direct record | supported as stated | high | [277] |
| IDN-E-F18 | ID5 documents an integration that synchronizes with ad-tech vendors while third-party cookies are available. | 5 | vendor assertion | supported as stated | medium | [221, 179, 180] |
| IDN-E-F19 | LiveRamp’s cookie-to-RampID file guide instructs recipients to enable a cookie sync through a pixel. | 5 | vendor assertion | supported as stated | medium | [222] |
| IDN-E-F20 | In ID5 publisher SDK guides dated April 18, 2025, October 15, 2025 and November 6, 2026, maxCascades caps cookie syncs at eight by default when user syncing is needed. A value of −1 disables syncing. | 5 | synthesis | supported with qualification | medium | [179, 223, 180] |
| IDN-E-F21 | All three UID2 JavaScript guide revisions tell publishers using a supported client-side integration to look for the generated token in local storage. The October 2026 revision directs publishers to generate and refresh tokens server-side in affected OS 27 browsers. | 5 | synthesis | supported with qualification | medium | [224, 225, 226] |
| IDN-E-F22 | Prebid describes SharedID storage in a first-party domain on this header-bidding route. | 5 | vendor assertion | supported as stated | medium | [278] |
| IDN-E-F23 | Prebid’s Yahoo module guide says ConnectID can operate without third-party cookies. | 5 | vendor assertion | supported as stated | medium | [202] |
| IDN-E-F25 | Prebid’s Panorama guide describes a device-representing ID for enrichment and activation. | 5 | vendor assertion | supported as stated | medium | [279] |
| IDN-I-F11 | In April 2021 Mediavine said Grow.me was integrated with ID5 Universal ID. It also reported that Mediavine had active integrations with LiveRamp ATS, UID2 and ID5. | 5 | vendor assertion | supported with qualification | high | [280] |
| IDN-I-F14 | Trusted Media Brands’ privacy notice describes a possible LiveRamp handoff that returns a code the publisher may store in its own cookie. | 5 | direct record | supported as stated | high | [281] |
| IDN-U-F35 | Google included Topics on Chrome and Android in its announced technology retirements. | 5 | direct record | supported as stated | high | [282] |
| IDN-U-F36 | Chrome 153 release notes describe Protected Audience as planned for deprecation and removal. | 5 | direct record | supported as stated | high | [283] |
| IDN-U-F37 | Google said Topics replaced its FLoC proposal following FLoC trials. | 5 | direct record | supported as stated | high | [284] |
| IDN-ID5-F14 | Sovrn and ID5 describe a participating-publisher test with US audiences that compared a control with conditions in which hashed email was linked and either sent into the auction or withheld. | 5 | vendor assertion | supported as stated | medium | [129, 130] |
| IDN-LR-F04 | The ATS API storage guide gives its envelopes a 15-day lifetime in California and 30 days elsewhere, and advises updating the stored expiry when refreshing. | 5 | vendor assertion | supported as stated | high | [137] |
| IDN-LR-F14 | LiveRamp describes ATS for authenticated inventory without third-party cookies and also documents an optional third-party-cookie endpoint for ATS. Its separate cookie-to-RampID mapping-file workflow uses a cookie-sync pixel. These sources do not date a change in either workflow. | 5 | synthesis | supported with qualification | high | [133, 222, 134] |
| IDN-UID-F14 | Google announced on April 22, 2025 that it would retain Chrome’s existing third-party cookie choice. On October 17 it announced the retirement of named technologies including Topics and Protected Audience, subject to platform phase-out processes. | 5 | direct record | supported as stated | high | [4, 220] |
| IDN-UID-F15 | On October 9, 2026, UID2 documented that browser requests to uidapi.com were blocked in all iOS 27+ browsers and in Safari on macOS 27+. Its Prebid guide directs publishers on those paths to generate and refresh tokens server-side. On October 19, WebKit described its broader domain-level protection. | 5 | synthesis | contradicted | high | [151, 152, 145] |
| IDN-YAH-F18 | Prebid’s April 2025 module guide describes ConnectID as able to work without third-party cookies, while Yahoo’s publisher guide offers multiple implementation paths. | 5 | direct record | supported as stated | high | [202, 199] |
| IDN-VA5-F01 | Prebid.js 5.0.0, released June 10, 2021, merged the PubCommon and SharedID submodules. Prebid’s identity guide illustrates the earlier sharedid.org EID and third-party sync, and a later first-party path with a pubcid.org EID. | 5 | direct record | supported as stated | medium | [285, 286] |
| IDN-ID5-F20 | The Prebid example sets local storage to 90 days and refresh to two hours; ID5 separately reports a maximum 100-day retention period for data used to provide its live ID service before deletion or aggregation. | 5, 11 | synthesis | supported as stated | medium | [175, 131] |
| IDN-YAH-F17 | Yahoo’s guide says an email hash is required below Prebid 7.52. The released 7.51.0 module accepts either an email hash or a publisher user ID. Version 7.52.0, released June 1, 2023, removed the requirement to supply either value. | 5, 13 | direct record | contradicted | high | [164, 189, 191, 192, 193] |
| IDN-F-F01 | For matched email or phone inputs, Google Ads says it adds the corresponding Google Account to the Customer Match segment. | 6 | vendor assertion | supported as stated | high | [213] |
| IDN-F-F03 | Google defines Customer Match rate as the share of an upload it can link to logged-in Google users. Google says most advertisers fall between 29% and 62%. It gives no period, geography or advertiser sample for that range. | 6 | vendor assertion | supported with qualification | medium | [287] |
| IDN-F-F04 | From April 1, 2026 Google directed most new Customer Match integrations to the Data Manager API while retaining a Google Ads API route for eligible existing integrations. | 6 | vendor assertion | supported as stated | high | [288] |
| IDN-F-F05 | Google’s Authorized Buyers guide says publishers can instruct Ad Manager to send obfuscated signals in bid requests to third-party bidders they have approved. The signal must be registered, and bidders select which signals to receive. | 6 | vendor assertion | supported with qualification | high | [289] |
| IDN-F-F06 | Under the hashed-data route in Amazon’s November 20, 2024 advertiser-audience agreement, advertisers supply hashed data for matching to Amazon IDs and provide one or more associated segment identifiers. | 6 | direct record | supported with qualification | high | [290] |
| IDN-F-F07 | Amazon Ads instructs advertisers or their agencies to prepare and hash the source audience, then select the hashed audience for activation. | 6 | vendor assertion | supported as stated | medium | [291] |
| IDN-F-F10 | Meta’s Marketing API lists email addresses and phone numbers among the inputs used to build target Custom Audiences from customer information. | 6 | vendor assertion | supported as stated | high | [292] |
| IDN-F-F11 | Walmart Connect’s May announcement named Yahoo DSP, VIZIO inventory and Magnite for activation of Walmart audiences in an offsite buying workflow. | 6 | vendor assertion | supported as stated | medium | [293] |
| IDN-F-F12 | For its managed Google search-ad offering, Roundel describes use of Target’s own audiences, including past Target.com purchasers. | 6 | vendor assertion | supported as stated | medium | [294] |
| IDN-F-F13 | In November 2023 Kroger Precision Marketing said its customers could activate Kroger purchase-based audiences through their existing seats on The Trade Desk. | 6 | vendor assertion | supported as stated | medium | [295] |
| IDN-F-F14 | When it announced Data Hub in January 2026, Instacart said selected agencies and CPGs had piloted the clean-room service for joining marketer and grocery data. | 6 | vendor assertion | supported as stated | medium | [296] |
| IDN-F-F15 | Instacart’s clean-room setup guide uses a hashed-email intersection between two parties as an example of an analysis where complete datasets stay hidden. | 6 | vendor assertion | supported as stated | high | [297] |
| IDN-F-F16 | For Data Hub activation queries, Instacart directs results to its own account and says partners cannot see the specific matched identifiers. | 6 | vendor assertion | supported as stated | high | [298] |
| IDN-I-F16 | On 23 March 2026 Roku announced it was a launch partner for Google’s Confidential Publisher Match and described a route for DV360 advertisers to match first-party data to Roku Media inventory. | 6 | vendor assertion | supported with qualification | high | [299] |
| IDN-F-F02 | Google says API uploads need at least 100 matched unique users for the match rate to display. | 6, 12 | vendor assertion | supported as stated | high | [300] |
| IDN-F-F08 | Amazon Marketing Cloud limits the analyst-facing outputs it describes to aggregated anonymous results while directing custom audiences to ad activation. | 6, 12 | vendor assertion | supported as stated | high | [301] |
| IDN-F-F09 | Meta’s Customer List Custom Audiences terms say advertisers do not receive the individual identities that constitute their matched audience. | 6, 12 | direct record | supported as stated | high | [302] |
| IDN-I-F03 | Disney’s 2023 LiveRamp destination terms define a custom segment through overlap between advertiser and Disney data. | 6, 12 | direct record | supported as stated | high | [303] |
| IDN-I-F04 | Disney’s destination terms require the submitting party to warrant rights and consents for the advertiser data. | 6, 12 | direct record | supported as stated | high | [303] |
| IDN-I-F10 | Roku documented a buyer path from The Trade Desk through Roku Exchange to activate against Roku first-party data using UID2. | 6, 12 | vendor assertion | supported as stated | high | [154] |
| IDN-G-F01 | TransUnion’s 2021 TruAudience offering described its identity product as omnichannel and separately named the streaming-data marketplace formerly known as Tru Optik. | 7 | vendor assertion | supported as stated | high | [246] |
| IDN-G-F04 | Viant says it screens IP addresses by residential ISP, device count and time of day, then uses coordinates from location-enabled devices to match them to a registered home address. | 7 | vendor assertion | supported with qualification | high | [304] |
| IDN-G-F05 | Blockgraph says its household identity framework anchors a household to its postal address. | 7 | vendor assertion | supported as stated | high | [166] |
| IDN-G-F08 | FreeWheel says its embedded Identity Network provides access to Comcast’s deterministic signals and publisher partner IDs. Buyers can also connect their own first-party data. | 7 | vendor assertion | supported with qualification | high | [305] |
| IDN-G-F10 | Cadent says its Aperture contextual targeting offer uses a proprietary IP-to-household Viewer Graph. | 7 | vendor assertion | supported as stated | high | [306] |
| IDN-G-F11 | LiveRamp’s programmer guide accepts IP-based CTV events for matching and recommends daily or weekly delivery of new data after initial onboarding. | 7 | vendor assertion | supported as stated | high | [307] |
| IDN-G-F12 | Factori describes its cross-device offering as linking encrypted mobile ad IDs and IP addresses with daily server-to-server feed delivery. | 7 | vendor assertion | supported as stated | high | [308, 309] |
| IDN-G-F14 | Criteo’s CTV guide describes a shared household connection as a fallback when an exact match is unavailable. | 7 | vendor assertion | supported as stated | high | [204] |
| IDN-G-F15 | In January 2020, Skydeo announced an agreement to map its mobile data to the then-named Tru Optik Household Graph in the United States. | 7 | vendor assertion | supported with qualification | high | [310] |
| IDN-G-F17 | Viant’s US patent gives 45 days as an example threshold that an IP assignment’s age must exceed to qualify for further processing. | 7 | direct record | supported with qualification | high | [311] |
| IDN-U-F26 | Viant describes a Household ID service that uses IP addresses and other identifiers to represent homes. | 7 | vendor assertion | supported as stated | medium | [312, 313] |
| IDN-U-F29 | Cadent identified its IP-to-household Viewer Graph in a 2024 Aperture product sheet. A 2021 release announced its planned use with Premion. As of November 11, 2026, Cadent also names Viewer Graph in its US CTV media offer. | 7 | vendor assertion | supported with qualification | medium | [306, 314, 315] |
| IDN-U-F47 | Factori’s cross-device and CTV graph guide, marked last updated in September 2026, describes identifying household users across devices and delivering a daily data feed. Its AWS Marketplace listing offers a contract for the dataset and describes global coverage without a US breakout. | 7 | vendor assertion | supported with qualification | medium | [308, 309] |
| IDN-CTV-F01 | Blockgraph identified the United States as the territory for its commercial peer-to-peer TV resolution service in its 2021 notice. | 7 | direct record | supported as stated | high | [170] |
| IDN-CTV-F02 | TVSquared reported using Blockgraph’s IDoS for converged-TV identity resolution in its February 2022 response to NBCUniversal. | 7 | vendor assertion | supported as stated | medium | [168] |
| IDN-CTV-F08 | Blockgraph’s v2.29 manual distinguishes deleting a subscriber record from opting it out: opt-out makes existing segments and matches stale until refreshed. | 7 | vendor assertion | supported as stated | high | [115] |
| IDN-CTV-F15 | The August 2026 release added SHA256-hashed email as a preset signal for identity upload, batch resolution and other workflows. | 7 | vendor assertion | supported as stated | high | [116, 172] |
| IDN-CTV-F16 | An April 2025 CIMM report describes IDoS as operating across 130 million-plus households. It gives no geography, active period, deduplication rule or eligible household base for that figure. | 7 | synthesis | supported with qualification | medium | [6] |
| IDN-CTV-F18 | Blockgraph’s CEO gave a broad match-rate definition in the commentary about Kantar, without specifying how that case counted eligible or matched records. | 7 | synthesis | supported as stated | high | [113, 112] |
| IDN-CTV-F19 | Spectrum Reach claims a 95% addressable match rate for its own MVPD service in its multiscreen-attribution FAQ. | 7 | vendor assertion | supported as stated | medium | [38] |
| IDN-CTV-F20 | Spectrum Reach describes its first-party multiscreen data as coming from over 30 million households. | 7 | vendor assertion | supported as stated | medium | [38] |
| IDN-G-F16 | TransUnion and Epsilon named TruAudience Data Marketplace as the route for accessing Epsilon audience data for connected TV. | 7, 8 | vendor assertion | supported as stated | high | [245] |
| IDN-CTV-F03 | Blockgraph’s general manager described a software subscription to join its network in February 2020. | 7, 8 | vendor assertion | supported as stated | medium | [171] |
| IDN-CTV-F10 | Blockgraph’s v2.29 Universe Match workflow lists partners that granted permission and specifies a downloadable file containing one column of Blockgraph IDs. | 7, 8, 11 | vendor assertion | supported as stated | high | [110, 111] |
| IDN-CTV-F07 | The v2.29 enterprise guide requires a household record ID and postal address and permits additional signals for that household, including IP and set-top-box IDs. | 7, 11 | vendor assertion | supported as stated | high | [109] |
| IDN-G-F13 | LtvAdx documents a shared-IP fallback for assigning a device to a household and removes a device link 30 days after the device was last seen. It offers HouseholdID targeting in its US advertising workflow. | 7, 13 | vendor assertion | supported with qualification | high | [227, 228, 229, 230, 231] |
| IDN-U-F45 | LtvAdx documents an Identity API that returns an existing HouseholdID or creates a new one. It markets household targeting in US TV campaigns and includes HouseholdID targeting in a plan billed on delivered impressions. The plan does not establish a separately licensed graph. | 7, 13 | vendor assertion | supported with qualification | medium | [227, 229, 230, 231] |
| IDN-C-F11 | TransUnion described its January 2024 identity graph enhancement as combining TransUnion and Neustar data, clustering identifiers into people and households and scoring those identities. | 8 | vendor assertion | supported as stated | high | [316] |
| IDN-D-F01 | LiveRamp entered its merger agreement with MMS USA Holdings on May 16, 2026. | 8 | direct record | supported as stated | high | [317] |
| IDN-D-F02 | Publicis Groupe’s wholly owned subsidiary MMS USA Holdings completed its acquisition of LiveRamp Holdings on November 2, 2026. | 8 | direct record | supported as stated | high | [141, 318] |
| IDN-D-F03 | At the closing, each qualifying outstanding LiveRamp common share converted to a right to $38.50 in cash. | 8 | direct record | supported as stated | high | [141] |
| IDN-D-F04 | LiveRamp reported consolidated revenue of $812.9 million for the year ended March 31, 2026. | 8 | direct record | supported as stated | high | [7] |
| IDN-D-F05 | LiveRamp reported one operating segment in its fiscal 2026 annual financial statements. | 8 | direct record | supported as stated | high | [7] |
| IDN-D-F06 | Publicis announced the Lotame agreement in March 2025 and reported acquiring 100% of Lotame Solutions in April 2025. | 8 | direct record | supported as stated | high | [214, 215] |
| IDN-D-F07 | Alliance Data’s filing dates the sale of Epsilon to Publicis to July 1, 2019. Publicis announced completion on July 2. | 8 | direct record | contradicted | medium | [319, 320] |
| IDN-D-F08 | In its November 2, 2026 closing announcement, Publicis described RampID and Epsilon identity as inputs to a combined client data offering. | 8 | vendor assertion | supported as stated | medium | [216] |
| IDN-D-F09 | ID5 announced its acquisition of TrueData in an October 30, 2025 datelined statement and subsequently described the deal as completed. | 8 | vendor assertion | supported as stated | medium | [132, 217] |
| IDN-D-F10 | On October 1, 2026, ID5 said Cloud Technologies had bought TrueData’s third-party audience business and the TrueData brand. | 8 | vendor assertion | supported as stated | medium | [217] |
| IDN-D-F11 | Experian reported acquiring Audigent on December 4, 2024, and its annual report identifies the purchase as 100% of the equity. | 8 | direct record | supported as stated | high | [321, 322] |
| IDN-D-F12 | TransUnion and Neustar reported a $3.1 billion closing on December 1, 2021, and expressly excluded Neustar Security Services. | 8 | vendor assertion | supported as stated | medium | [323, 324] |
| IDN-D-F13 | IPG’s October 1, 2018 Acxiom transaction excluded LiveRamp. | 8 | direct record | supported as stated | high | [325, 326] |
| IDN-D-F14 | Omnicom reported closing the Interpublic acquisition on November 26, 2025 and subsequently named Acxiom RealID in its media strategy. | 8 | vendor assertion | supported as stated | medium | [327, 328] |
| IDN-D-F15 | At the September 1, 2021 closing, Apollo said Yahoo would operate under Apollo Funds and Verizon retained 10%. | 8 | vendor assertion | supported as stated | medium | [329, 330] |
| IDN-D-F16 | The Trade Desk defines gross spend in its FY2025 filing as clients’ spending on advertising inventory, value-added services and data on its platform, plus its platform fee. | 8 | direct record | supported with qualification | high | [331] |
| IDN-CTV-F05 | In a May 29, 2020 joint release, Comcast, Charter and ViacomCBS announced an agreement for equal ownership of Blockgraph and dated the original FreeWheel initiative to 2017. | 8 | vendor assertion | supported with qualification | medium | [332, 333] |
| IDN-CTV-F06 | Blockgraph names Comcast NBCUniversal, Charter Communications and Paramount as owners, while Comscore’s FY2025 filing identifies Blockgraph as private. | 8 | synthesis | supported as stated | high | [334, 335] |
| IDN-EXP-F15 | The Digital Graph sheet credits Experian Marketing Services, Inc., while the US marketing data-practices notice names Experian Marketing Solutions, LLC. | 8 | synthesis | supported as stated | medium | [119, 336, 337] |
| IDN-EXP-F16 | Experian plc, the group parent headquartered in Dublin, reports a London Stock Exchange listing under EXPN. | 8 | direct record | supported as stated | high | [337, 338] |
| IDN-EXP-F17 | Experian plc reported worldwide statutory revenue of US$8,445 million for the year ended March 31, 2026. | 8 | direct record | supported as stated | high | [337, 338] |
| IDN-ID5-F02 | The UK register names French ID5 Technology SAS as the active controlling entity for ID5 Technology Limited and places its shareholding and voting rights at 75% or more. | 8 | direct record | supported as stated | high | [339, 340] |
| IDN-ID5-F03 | On 9 April 2024 ID5 announced a $20 million Series B and named TransUnion and S4S Ventures among its investors. | 8 | vendor assertion | supported as stated | medium | [341] |
| IDN-ID5-F04 | ID5 Technology Limited reported £702,598 in cash at bank and in hand as of 30 June 2025. The UK company elected to exclude its profit and loss account from the unaudited public filing. | 8 | direct record | supported with qualification | high | [340] |
| IDN-ID5-F18 | Cloud Technologies reported a conditional 30 September 2026 agreement with TrueData Solutions Inc. covering a new US vehicle, with $1.1 million of agreed consideration for the shares and technology license. | 8 | direct record | supported as stated | high | [342] |
| IDN-ID5-F19 | Cloud Technologies reported that it completed the acquisition of all shares in the new US vehicle on 5 October 2026 after conditions including transfer of specified contracts were met. | 8 | direct record | supported as stated | high | [343] |
| IDN-LR-F17 | Publicis acquired LiveRamp Holdings through MMS USA Holdings on November 2, 2026, and trading in LiveRamp common stock stopped before the NYSE opened that day. | 8 | direct record | supported as stated | high | [141, 318] |
| IDN-LR-F18 | LiveRamp reported $812.9 million in consolidated revenue for the year ended March 31, 2026 and $214.0 million for the quarter ended June 30, 2026. | 8 | direct record | supported as stated | high | [7, 344] |
| IDN-LR-F19 | On November 10, 2026, LiveRamp notified the SEC that it had been unable to file its September-quarter 10-Q on time after the acquisition required reassessing its auditor relationship. | 8 | direct record | supported as stated | high | [318] |
| IDN-LR-F22 | The group’s US services terms identify LiveRamp, Inc. as the Delaware contracting entity with a San Francisco principal place of business; LiveRamp Holdings, Inc. is the merger registrant. | 8 | synthesis | supported as stated | high | [345, 141] |
| IDN-UID-F02 | For FY2025 The Trade Desk reported $2,896,284,000 in worldwide consolidated revenue in one advertising technology segment. Its platform fees generally depend on client spend and it also earns from services and data. The filings give no separate UID2 revenue. | 8 | direct record | supported as stated | high | [331, 346] |
| IDN-YAH-F01 | The US ConnectID agreement names Yahoo Ad Tech LLC as signatory and lists its principal place of business in New York, New York. | 8 | direct record | supported as stated | high | [160] |
| IDN-YAH-F02 | Apollo’s funds acquired Yahoo in September 2021, when Verizon retained 10%. Verizon reported $4.1 billion in cash proceeds net of cash transferred for Verizon Media. In September 2025 Yahoo Ad Tech LLC identified College Parent, L.P. as its corporate parent in a court disclosure. | 8 | direct record | supported with qualification | high | [329, 347, 348] |
| IDN-SCD-F07 | Seedtag’s October 2026 announcement describes an available managed US political targeting route and gives a Michigan example combining separate Aristotle voter audiences and Seedtag contextual segments. | 8 | vendor assertion | supported as stated | medium | [349] |
| IDN-SCD-F03 | Comscore offered US political buyers on The Trade Desk an ID-free Predictive Audience based on TV ad exposure alongside its ID-based option. | 8, 11 | vendor assertion | supported as stated | medium | [350] |
| IDN-ID5-F05 | ID5 offers registered publishers encrypted ID delivery without charge. Its guide requires a commercial licence for platforms, intermediaries or commercial uses beyond standard publisher integration. The agreement offers decryption, connections between IDs and use of ID5 data for purchase. | 8, 12 | vendor assertion | supported with qualification | medium | [56, 351] |
| IDN-LR-F16 | LiveRamp’s publisher ATS form limits envelope use to enabled parties under a statement of work. Its separate general service terms put any associated fees in the order form. In 2021 its CEO said LiveRamp did not intend to charge publishers or media-platform partners for ATS and expected more platform use and cross-sales. | 8, 12 | synthesis | supported with qualification | high | [352, 353, 345, 354] |
| IDN-SCD-F04 | Ogury offers its persona groups as a replacement for tracking individuals. Brands and agencies can plan and activate on Ogury One, its self-serve and API delivery route. | 8, 12 | vendor assertion | supported as stated | medium | [210] |
| IDN-SCD-F08 | Seedtag sells selectable audience groups as an alternative to third-party tracking. Custom Audiences derive URL scores from the advertiser’s brief. The 2026 Michigan targeting option corroborates the US buying path. | 8, 12 | vendor assertion | supported as stated | medium | [355] |
| IDN-SCD-F05 | NIQ names the US as a market for its purchase insights for Ogury’s persona engine and describes brands activating the resulting audiences in digital campaigns. | 8, 13 | vendor assertion | supported as stated | medium | [356] |
| IDN-H-F01 | In its June 2023 retargeting decision, the French regulator found that stopping personalised ads was insufficient to meet the erasure requests at issue. | 9 | direct record | supported as stated | high | [63] |
| IDN-H-F02 | On March 4, 2026, France’s Conseil d’État rejected Criteo’s challenge to the €40 million CNIL sanction. | 9 | direct record | supported as stated | high | [64] |
| IDN-H-F03 | The EDPB-hosted decision summary records that the retargeter introduced partner audits during the investigation. | 9 | direct record | supported as stated | high | [65] |
| IDN-H-F04 | The substituted UK notice required Experian to inform certain people whose records came from open sources that it processed their data for direct marketing. | 9 | direct record | supported as stated | high | [66] |
| IDN-H-F05 | The UK Upper Tribunal dismissed the Information Commissioner’s appeal in the Experian matter on April 22, 2024. | 9 | direct record | supported as stated | high | [67] |
| IDN-H-F06 | In June 2025 Experian told investors that it expected to comply with its substituted UK notice by July 24, 2025. | 9 | vendor assertion | supported as stated | high | [68] |
| IDN-H-F07 | The FTC’s July 2024 amended complaint cited household mapping as a use that Kochava advertised for its location feed. | 9 | direct record | supported as stated | high | [69] |
| IDN-H-F08 | The June 2026 US Kochava order required Collective Data Solutions to provide a means for consumers to withdraw consent to use or disclosure of precise device-location data. | 9 | direct record | supported as stated | high | [70] |
| IDN-H-F09 | The FTC’s January 2025 Mobilewalla order restricted collection or retention of covered information accessed in online advertising auctions to participation in those auctions. | 9 | direct record | supported as stated | high | [71] |
| IDN-H-F10 | Mobilewalla reported to investors in August 2026 that its US operations applied a third-party sensitive-location filter to incoming signals and outgoing deliveries after its FTC order. | 9 | vendor assertion | supported as stated | high | [72, 73] |
| IDN-H-F11 | California’s February 11, 2026 judgment requires Disney DTC and ABC Enterprises to honor a logged-in consumer’s opt-out across the Disney streaming services associated with that account. | 9 | direct record | contradicted | high | [74, 75] |
| IDN-H-F12 | The California attorney general announced the Disney opt-out settlement on February 11, 2026. | 9 | direct record | supported as stated | high | [75, 74] |
| IDN-H-F13 | Texas’s December 2025 petition against Sony cited Samba Interactive TV when describing the opt-out path on BRAVIA televisions. | 9 | direct record | supported as stated | high | [76, 77] |
| IDN-H-F14 | On May 11, 2026, Texas announced a settlement with LG Electronics U.S.A. The agreed terms call for affirmative express consent for covered viewing-data practices 180 days after judicial signature and entry. The posted copy leaves the judge’s signature and date blank. | 9 | direct record | supported with qualification | high | [78, 77] |
| IDN-H-F15 | France’s September 2025 order required Google to explain the advertising cookies involved in creating a Google account to users in France. | 9 | direct record | supported as stated | high | [79] |
| IDN-H-F16 | On September 23, 2026, the CNIL closed its injunction after Google changed its French account-creation disclosure and the format of Gmail ads. The closure applied to those two requirements. | 9 | direct record | supported with qualification | high | [80, 79] |
| IDN-H-F17 | In May 2025 the Belgian authority reported that the Market Court annulled its IAB Europe decision on procedural grounds while confirming the fine. | 9 | direct record | supported as stated | high | [81] |
| IDN-H-F18 | France’s Conseil d’État rejected Yahoo EMEA Limited’s challenge to the CNIL decision on cookie placement and withdrawal in October 2025. | 9 | direct record | supported as stated | high | [83] |
| IDN-H-F19 | The FTC’s April 2024 X-Mode/Outlogic order required a sensitive-location program within 180 days of issuance. | 9 | direct record | supported as stated | high | [84] |
| IDN-H-F20 | The FTC’s April 2024 InMarket order prohibits selling or licensing defined location data for valuable consideration. | 9 | direct record | supported as stated | high | [85] |
| IDN-H-F21 | The January 2025 Gravy Analytics/Venntel order required a sensitive-location program within 90 days of its effective date. | 9 | direct record | supported as stated | high | [86] |
| IDN-H-F22 | The Court of Justice held in March 2024 that a TC String constitutes personal data when reasonable linkage to an identifier allows the person to be identified. | 9 | direct record | supported as stated | high | [82] |
| IDN-EXP-F18 | The UK Upper Tribunal dismissed the Commissioner’s appeal in April 2024, leaving a substituted notice covering certain open-source records used by Experian Limited for direct marketing. | 9 | direct record | supported as stated | high | [67, 66, 357] |
| IDN-LR-F20 | The FTC issued an interim Order to Maintain Assets on October 19 requiring continued access to LiveRamp standalone services, while its separate proposed five-year order remained under public comment at the cutoff. | 9 | direct record | supported as stated | high | [88, 89, 87, 9] |
| IDN-LR-F21 | Open Rights Group reported complaints to the UK ICO and French CNIL in February 2024 following its commissioned review of European RampID processing documentation. | 9 | synthesis | supported as stated | high | [60, 61] |
| IDN-YAH-F20 | On August 26, 2026, the Southern District of New York ordered five plaintiffs’ claims in the private ConnectID suit into arbitration and stayed litigation of those claims. | 9, 13 | direct record | supported as stated | high | [90, 8] |
| IDN-J-F02 | The AdCP issuer described clean-room activation as experimental in its October 22 account of the 3.2 release. | 10 | direct record | supported as stated | high | [358] |
| IDN-J-F06 | The September 11 adcp-go code includes UID2 and EUID decoders that call the configured client to decrypt an incoming advertising token. | 10 | direct record | supported as stated | high | [96, 93] |
| IDN-J-F07 | The IAB Tech Lab repository labels the AAMP 3.0 OpenProposal public-comment object model a draft. | 10 | direct record | supported as stated | high | [98] |
| IDN-J-F08 | In the OpenProposal draft, the seller supplies the expected match rate, while the buyer can populate segments and a suppression list. | 10 | direct record | supported as stated | high | [97] |
| IDN-J-F09 | The IAB OpenRTB Agentic Audiences community extension assigns integer type 1 to identity embeddings and allows a segment entry to carry multiple signal types. | 10 | direct record | supported as stated | high | [100] |
| IDN-J-F10 | Prebid’s merged Agentic Audiences module copies a stored entry’s type and vector into the global OpenRTB auction fragment under user.data[].segment[].ext.aa. The code and module guide establish this handoff within Prebid. | 10 | direct record | supported with qualification | high | [103, 105, 104] |
| IDN-J-F11 | Prebid merged its Agentic Audiences RTD module into the project’s main development branch on April 14, 2026. | 10 | direct record | supported as stated | high | [104] |
| IDN-J-F12 | IAB Tech Lab explicitly describes an ordinary programmatic use for Agentic Audiences that is separate from its agentic advertising use. | 10 | direct record | supported as stated | high | [101] |
| IDN-J-F13 | IAB Tech Lab lists the August 2026 Agentic Audiences release as v1.1 and the AAMP September 2026 umbrella release as v3.0. | 10 | direct record | supported as stated | high | [102] |
| IDN-J-F14 | PubMatic reported an AdCP-enabled CTV campaign for Clubtails on its Activate platform in December 2025. | 10 | vendor assertion | supported as stated | high | [106] |
| IDN-J-F15 | Optable said its March 2026 PubMatic integration gave buyers access to publisher first-party audience planning without transferring the underlying data. | 10 | vendor assertion | supported as stated | high | [107] |
| IDN-J-F16 | IAB Tech Lab describes an AAMP reference-SDK demonstration in which a buyer and seller negotiate and push the confirmed transaction to Google Ad Manager. | 10 | direct record | supported as stated | high | [99] |
| IDN-J-F01 | The experimental Trusted Match specification in AdCP 3.2.8 defines an identity request from a publisher via a router to buyer agents. The request carries one or more opaque identity tokens and must omit page context. | 10, 12 | direct record | supported with qualification | high | [91] |
| IDN-J-F05 | The AdCP reference HTTP identity handler can pass a decoded request to its service for evaluation and constructs a response carrying eligible package IDs. | 10, 13 | direct record | supported as stated | high | [108, 93] |
| IDN-U-F09 | Prebid says SharedID sets a user ID cookie on the publisher domain. | 11 | vendor assertion | supported as stated | medium | [286] |
| IDN-U-F42 | Deep Sync describes using its identity graph to resolve the CRM of a US retailer in its Profile Hygiene and Enrichment case. | 11 | vendor assertion | supported as stated | medium | [27] |
| IDN-U-F43 | Deep Sync describes three matching passes on a US grocer’s transaction file. The second matched payment-card signals against its identity graph. The third used store geography and household data to infer a buyer. | 11 | vendor assertion | supported with qualification | medium | [12] |
| IDN-UID-F10 | UID2 lists production operator endpoints in Ohio and Oregon. In August 2024, Roku announced UID2 adoption across its premium streaming inventory and described how The Trade Desk advertisers could activate against Roku first-party data through Roku Exchange. | 11 | synthesis | supported with qualification | high | [153, 154] |
| IDN-YAH-F14 | Yahoo reported 232 million logged-in US users when it announced its Conversion API in April 2025 and described a US base of 213 million users in its October 2026 ConnectID article. | 11 | vendor assertion | supported as stated | medium | [197, 198] |
| IDN-SCD-F02 | Proximic currently markets custom ID-free predictive audiences via its activation platform. | 11 | vendor assertion | supported as stated | medium | [209] |
| IDN-EXP-F05 | Experian says it rebuilds Digital Graph weekly and has added IPv6 and phone-based UID2 identifiers. | 11, 12 | vendor assertion | supported as stated | medium | [173] |
| IDN-YAH-F16 | Yahoo’s DataX file-match reporting and its ConnectID integration check measure different outcomes: one returns partner onboarding results and the other checks whether an ID appears on a bid request. | 11, 13 | synthesis | supported as stated | medium | [30, 164] |
| IDN-I-F01 | CIMM’s April 2026 buyer guide recommends checking how many matched records can be addressed on the buyer’s primary activation platforms. | 12 | synthesis | supported with qualification | high | [259] |
| IDN-I-F06 | Dentsu reported in January 2025 that its New Stream Media team was testing its Merkury/Trade Desk integration with Hy-Vee’s RedMedia. | 12 | vendor assertion | supported as stated | high | [359] |
| IDN-I-F07 | iProspect’s October 2025 agency appointment announcement calls for Merkury to connect Hy-Vee enterprise media, RedMedia and loyalty programs. | 12 | vendor assertion | supported as stated | high | [360] |
| IDN-I-F08 | Hy-Vee’s RedMedia describes combining its first-party shopper data with outside advertising-firm data for off-site targeting. | 12 | vendor assertion | supported as stated | high | [361] |
| IDN-YAH-F11 | Yahoo’s October 2024 announcement identified Amazon Publisher Services, InMobi, Magnite and PubMatic as publisher integration routes, including bidding through Amazon’s Transparent Ad Marketplace. | 12 | vendor assertion | supported as stated | medium | [196] |
| IDN-EXP-F04 | Experian says an unnamed DSP synced its cookies and received a weekly graph file linking them to cookies, mobile ad IDs, CTV IDs and IP addresses. | 13 | vendor assertion | supported as stated | medium | [37, 36] |
| IDN-EXP-F08 | Experian says the unnamed DSP synced 84% of IDs within weeks of using its Digital Graph. | 13 | vendor assertion | supported as stated | medium | [36, 37] |
| IDN-A5-F05 | Epsilon’s PubLink registration lets a publisher connect its authenticated first-party data and an encrypted ID to Epsilon CORE ID, then supply a PubLink identifier in the bidder field. | 13 | direct record | supported as stated | medium | [362] |
| IDN-A5-F08 | Prebid identifies the Merkle ID module and its merkleinc.com source as a publisher route for Merkury’s private identity graph. | 13 | direct record | supported as stated | medium | [363] |
| IDN-LR-F01 | LiveRamp describes ATS as a publisher route in which an SSP decrypts an envelope and passes a DSP-specific RampID in a bid request. | abstract, 1, 2, 4, 5 | vendor assertion | supported as stated | high | [133] |
| IDN-CTV-F09 | For historical IP lookup, Blockgraph documents optional row timestamps and configurable lookback periods of 30, 60 or 90 days when the provider maintains household signal history. | abstract, 1, 7, 12, 13 | vendor assertion | supported as stated | high | [114] |
| IDN-C-F16 | Deep Sync reports a 73.0% cumulative match rate for one US grocery chain’s monthly POS file after loyalty, card and probabilistic household/geographic matching. Its summary gives a range of roughly 64–73% of monthly volume. Its rounded pass counts and headline total do not reconcile exactly. | abstract, 3, 11 | vendor assertion | supported with qualification | high | [12] |
| IDN-CTV-F21 | A complete manual refresh in version 2.29 replaces the node’s identity layer and omits records absent from the replacement file. | abstract, 7, 12, 13 | vendor assertion | supported as stated | high | [115] |
| IDN-YAH-F04 | Yahoo’s US agreement defines the returned value as an encrypted identifier under a company- or publisher-specific key and restricts sharing to the agreed supply relationship. | abstract, 8 | direct record | supported as stated | high | [160] |
| IDN-J-F03 | The September 11 AdCP Go publisher client builds a typed request from a user token and its namespace type and posts the request to its Trusted Match identity route. | abstract, 10, 12, 13 | direct record | supported as stated | high | [92, 93, 94] |
| IDN-J-F04 | At the September 11, 2026 adcp-go commit, the reference identity engine submits request tokens and target audience IDs for a batch membership check when an audience service, identities and target segments are present. | abstract, 10, 12, 13 | direct record | supported with qualification | high | [95, 93] |
| IDN-PROC-F01 | OpenAI models carried out registration, research, verification, profiling, scoring, drafting and the final source check. The second pass used other model runs within the same process. No independent human peer review or second human scoring exercise is documented. The plan ran through the end of the New York calendar day. The original sweep closed at 8:37 a.m. EST on November 11, 2026. A later Claude Code session sampled 36 claims: 28 supported as written, two slightly overstated, one partly supported and five whose sources that reviewer could not open. | abstract, 13 | synthesis | supported as stated | high | [364] |

### Check a product

Find the name a buyer would see. Check the route used to buy it and the market served. Where there is an old name, see what the note says about it. Use the product register to trace the service that met the buying test. Use the cell record to see which facts counted towards its score [10, 218, 11]. If the row was left out, read the reason before you draw a conclusion.

**Advertising identity product register**

| Product | Owner | Headquarters | Peer group | Media formats | Access and delivery routes | Markets covered | Status at cutoff | Src |
|---|---|---|---|---|---|---|---|---|
| LiveRamp RampID | LiveRamp Holdings, Inc., wholly owned by MMS USA Holdings, Inc., a wholly owned subsidiary of Publicis Groupe S.A. | Not extracted | A | As specified in product record | An enabled publisher supplies an ATS envelope, an SSP decrypts it and a DSP receives its specific RampID in the bid. | yes — ATS is an integration path; Known Identity Resolution is a different deliverable. US terms cover SSP Sidecar and DSP/DMP roles. | admitted | [234, 181, 133, 140, 183, 345, 185, 141, 318] |
| Unified ID 2.0 | The Trade Desk, Inc., Nasdaq-listed company, for the US public operator covered here. IAB Tech Lab stewards the separate tokenization framework. | Not extracted | A | As specified in product record | A participant can obtain a UID2 token from the public operator and a credentialed bidder can decrypt it. | yes — Private operator instances, the EUID namespace, The Trade Desk’s DSP and Identity Alliance have separate roles. Ohio and Oregon production endpoints and the Roku Exchange buying route are documented. | admitted | [203, 153, 142, 143, 365, 366, 154, 331] |
| European Unified ID | Ultimate owner unconfirmed in the cited product records | Not extracted | A | As specified in product record | Test whether an in-scope US buyer can buy or license cross-party use of EUID. | unconfirmed — EU/EEA and named additional territories; United Kingdom explicitly listed. | listed | [211, 367, 368] |
| ID5 ID | ID5 Technology SAS controls the UK company. The ultimate owner of the French company is unconfirmed. | Not extracted | A | Web / app | A registered publisher distributes an encrypted identifier; authorized demand partners obtain commercial rights to decrypt and recognize it. | yes — The SDK and Prebid modules implement the same service. Twine Data’s US licensing and the sold TrueData audience business do not constitute ID5 ID. The partner case documents participating US audiences. | admitted | [369, 132, 177, 370, 339, 56, 131, 177, 176, 129, 351, 280] |
| Yahoo ConnectID | Yahoo Ad Tech LLC named College Parent, L.P. as its corporate parent in a September 2025 court disclosure. Apollo Funds acquired Yahoo in 2021, when Verizon retained 10%. A current ownership split was not verified. | Not extracted | A | Web / app | An approved publisher receives an encrypted ConnectID and passes it through authorized supply to Yahoo’s bidder for use in Yahoo DSP. | yes — Google documents the web secure-signal path and Yahoo names its App & Display and Video bidders. Publisher software integration does not confer rights to share an ID with any DSP. | admitted | [165, 371, 372, 160, 347, 199, 156, 157, 196, 329, 348] |
| Lotame Panorama ID | Publicis Groupe; acquired April 2025 | Not extracted | A | Web / app | Lotame documents its publisher and SSP EID. TripleLift names a USA route and a Michigan case names a publisher and SSP. Lotame is an author-declared former employer. | yes — Lotame documents its publisher and SSP EID. TripleLift names a USA route and a Michigan case names a publisher and SSP. Lotame is an author-declared former employer. | admitted | [373, 374, 375, 279, 215, 263, 264] |
| Criteo ID for Exchanges | Criteo | Not extracted | A | Web / app | Prebid identifies the criteo.com EID and exchange adapter path to Criteo demand. Equativ maps that namespace separately from Criteo OB. TripleLift lists a US Criteo ID passthrough. The OneTag provisional name remains an unverified alias. | yes — Prebid identifies the criteo.com EID and exchange adapter path to Criteo demand. Equativ maps that namespace separately from Criteo OB. TripleLift lists a US Criteo ID passthrough. The OneTag provisional name remains an unverified alias. | below minimum | [250, 250, 376, 263, 377] |
| Amazon Publisher Services publisher ID (unconfirmed) | Ultimate owner unconfirmed in the cited product records | Not extracted | A | Web / app | Test whether publishers can supply a distinct Amazon publisher identifier to independent bidstream recipients. | unconfirmed — Amazon cites US publisher data for its measurement tool; proprietary interoperable ID unconfirmed. | listed | [378] |
| SharedID | Prebid project / publisher-operated implementation; no single corporate graph owner established. | Not extracted | A | Web / app | Prebid provides the publisher-scoped identifier, available to its adapters. TripleLift names US passthrough. The recipient can receive a publisher’s identifier without establishing any persistent identity across unrelated publishers. | yes — Prebid provides the publisher-scoped identifier, available to its adapters. TripleLift names US passthrough. The recipient can receive a publisher’s identifier without establishing any persistent identity across unrelated publishers. | admitted | [286, 258, 278, 263] |
| Utiq identity service | Ultimate owner unconfirmed in the cited product records | Not extracted | A | As specified in product record | Test whether a publisher or buyer can use a Utiq identifier across parties on US inventory. | unconfirmed — Named European markets; US eligibility unconfirmed. | listed | [379, 380, 263] |
| Parrable ID | Ultimate owner unconfirmed in the cited product records | Not extracted | A | Web / app | Test whether an independent seller and buyer can use the same identifier. | unconfirmed — US eligibility unconfirmed. | listed | [249, 249, 381, 382, 377] |
| LiveRamp Known Identity Resolution (AbiliTec API) | Publicis Groupe via MMS USA Holdings, Inc. | Not extracted | B | As specified in product record | Direct AbiliTec API for known US records. Microsoft Dynamics 365 is a separately named US preview integration and does not establish a completed production transaction. | yes — LiveRamp documents maintained US AbiliTec links and an available REST API for known PII. Microsoft documents US-only preview enrichment requiring active LiveRamp and AbiliTec subscriptions; this preview does not mean the standalone API is beta. | admitted | [235, 141, 318, 212, 243, 383] |
| Experian Digital Graph | Experian plc | Not extracted | B | As specified in product record | Direct Digital Graph licensing, full dataset or buyer-file append. The MiQ IP and unnamed DSP case rates remain separate geographic-unknown cases. | yes — Sales page labels US Digital Graph, advertises license or first-party file append, and provides sales access. The customer cases do not identify a US sample. | admitted | [117, 119, 336, 337, 338] |
| TransUnion TruAudience Identity Resolution | TransUnion | Not extracted | B | As specified in product record | TruAudience Identity Resolution & Enrichment for Snowflake (customer records to persistent identities); excludes Data Health Assessment, Graph Extracts, streaming media activation. | yes — TransUnion offers TruAudience Identity Resolution & Enrichment as a Snowflake native marketplace workflow that links customer records against historical and current attributes for over 250 million US consumers. The Data Health Assessment insurer case is a different service. | admitted | [384, 246, 17, 18, 19, 316] |
| Epsilon CORE ID | Publicis Groupe | Not extracted | B | As specified in product record | Core ID Services customer-file onboarding (not Intent Messaging or a combined platform). | yes — US solution page advertises Core ID recognition for US consumers; processing description identifies the USA for identity resolution and describes client file onboarding. US processing location alone would be insufficient. | admitted | [385, 386, 22, 319, 320, 216] |
| Acxiom Real ID in the cloud | Omnicom Group via Interpublic | Not extracted | B | As specified in product record | Brand-cloud Real ID app; no qualifying US route frozen. Distinct Digital Recognition noted. | unconfirmed — A US retail deployment exists for Digital Recognition powered by Real ID. The public records do not establish standalone US access to the specifically registered brand-cloud Real ID application. Do not conflate the two offers. | listed | [387, 388, 327, 328] |
| Merkle Merkury | dentsu | Not extracted | B | As specified in product record | Merkury private identity platform; US Merkury-to-The Trade Desk test with Hy-Vee’s RedMedia. | yes — US private-graph offer at launch. Dentsu reports a later US Hy-Vee/RedMedia test of the Merkury integration, and the ensuing agency appointment identifies a continuing planned use. These do not measure delivery in every commercial configuration. | admitted | [389, 359, 360] |
| Zeta Customer Growth Intelligence | Zeta Global Holdings Corp. (group; distinct graph signer unverified) | Not extracted | B | As specified in product record | Customer Growth Intelligence Snowflake native app, buyer first-party records against Zeta Data Cloud. Do not transfer Customer Matcher rules for two client tables. | yes — Zeta announced US-graph enrichment inside the Snowflake native app; the current regional guide specifies US-only Streamlit availability, a named CGI first-party data-share request, and the account-based views containing enriched data. | admitted | [390, 244, 391] |
| Permutive publisher cohorts | Ultimate owner not established by the verified records for this service. | Not extracted | C | As specified in product record | TelevisaUnivision sent Permutive’s standard cohorts in Google Ad Manager requests. | yes — Google reports a measure for TelevisaUnivision’s US inventory. OpenX separately announced a buying route for US marketers in 2022. | admitted | [205, 206, 207, 208] |
| Audigent SmartPMP | Experian reported acquiring 100% of Audigent on 4 December 2024. | Not extracted | C | As specified in product record | Test whether the named service offers audience targeting as an explicit substitute for a persistent identifier. | unconfirmed — US service terms unresolved. | listed | [236, 237, 238, 239, 321, 322] |
| Ogury Personified Advertising | Ultimate owner not established by the verified records for this service. | Not extracted | C | As specified in product record | Ogury One self-serve or API persona activation | yes — Ogury’s identifiable group-targeting offer and US market activation inputs are documented. The preceding Japan campaign remains Japan-only. | admitted | [248, 210, 356] |
| Verve Moments.AI | Ultimate owner not established by the verified records for this service. | Not extracted | C | As specified in product record | Verve managed service for Moments.AI contextual audience segments | yes — US general offering is explicit in original 2022 issuer release; Q1 2026 report again describes this product for ID-deprecation environments. Provider markets high-quality contextual segments for brands to replace personalised targeting. | below minimum | [392, 393] |
| Seedtag contextual audiences | Ultimate owner not established by the verified records for this service. | Not extracted | C | As specified in product record | Seedtag managed activation of custom contextual audience segments from a campaign brief | yes — Documented Michigan targeting option. Distinguish availability from actual US deployment. Eurail’s separate campaign documents US activity but does not establish which formal segment type it used. | admitted | [394, 395, 349, 355] |
| TransUnion TruAudience streaming household identity | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Test whether a US TV buyer can purchase household linkage across devices or streaming endpoints. | unconfirmed — US household graph and CTV audience route explicitly described. | listed | [245, 246, 247] |
| Blockgraph Identity Operating System | Blockgraph names Comcast NBCUniversal, Charter Communications Inc. and Paramount. In May 2020 Comcast, Charter and ViacomCBS announced an equal-ownership agreement. Current percentages are undisclosed in the reviewed records. | Not extracted | D | TV / CTV | Blockgraph enterprise identity node for permissioned TV household matches | yes — US commercial scope in 2021 notice; Spectrum Reach and TVSquared describe US TV matching use. Active dated v2.29/2026 product documents. | admitted | [396, 170, 332, 334, 335, 116, 333, 172, 168] |
| Viant Household ID | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Test whether a buyer can purchase TV household resolution or a household identifier. | unconfirmed — US adult scope claimed in a Viant/TransUnion partner release. | listed | [312, 313] |
| OpenAP OpenID | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Test whether TV buyers or sellers can buy cross-publisher household linkage using this service. | unconfirmed — US publisher adoption claimed in 2021. | listed | [397, 398] |
| FreeWheel Identity Network | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Test whether a TV buyer or seller can obtain a household linkage output. | unconfirmed — US service geography unresolved for the specific implementation. | listed | [305, 399] |
| Cadent Viewer Graph | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Test whether household-level resolution for TV can be purchased from a named service. | unconfirmed — Cadent markets US CTV buying using its Viewer Graph within Total Video. Separate graph access or licensing is unconfirmed. | listed | [306, 314, 315] |
| Samba TV household graph | Ultimate owner not established by the verified records for this service. | Not extracted | D | As specified in product record | Author-declared TV household graph and ACR-related interest. | unconfirmed — US eligibility immaterial to exclusion. | listed | [400] |
| AWS Clean Rooms | Amazon | Not extracted | Outside scope | As specified in product record | Boundary lead for controlled collaboration between parties. | unconfirmed — Product territory outside IQ scope. | listed | [240] |
| Microsoft Curate | Microsoft | Not extracted | Outside scope | As specified in product record | Boundary lead for curated deals and inventory packaging. | unconfirmed — Product territory outside IQ scope. | listed | [241] |
| OneTrust Consent Management Platform | Ultimate owner unconfirmed in the cited product records | Not extracted | Outside scope | As specified in product record | Boundary lead for collecting and sending preference signals. | unconfirmed — Product territory outside IQ scope. | listed | [242] |
| IAB Tech Lab Curated Audiences (formerly Seller Defined Audiences) | IAB Tech Lab | Not extracted | Outside scope | As specified in product record | Standard describing an audience signal for the seller-side implementation search. | unconfirmed — Standards document; no US product assumption. | listed | [251] |
| Chrome Topics API | Ultimate owner unconfirmed in the cited product records | Not extracted | Outside scope | Web / app | History lead for the browser’s audience-topics interface. | historical — Chrome and Android named separately in the announcement. | historical | [282] |
| Chrome Protected Audience API | Ultimate owner unconfirmed in the cited product records | Not extracted | Outside scope | Web / app | History lead for the browser’s interest-group auction interface. | historical — Chrome 153 release note; distinct from Android. | historical | [283] |
| Chrome Federated Learning of Cohorts | Ultimate owner unconfirmed in the cited product records | Not extracted | Outside scope | Web / app | History lead for a browser cohort proposal or trial. | historical — Chrome proposal; no vendor admission. | historical | [284] |
| 33Across Lexicon | Ultimate owner unconfirmed in the cited product records | Not extracted | A | Web / app | Prebid states US support and a partner-ID authorization path. TripleLift identifies US bridging. The 33Across User ID module implements the Lexicon service. | yes — Prebid states US support and a partner-ID authorization path. TripleLift identifies US bridging. The 33Across User ID module implements the Lexicon service. | admitted | [401, 263] |
| Zeotap ID+ | Ultimate owner unconfirmed in the cited product records | Not extracted | A | As specified in product record | The issuer’s North America statement refers to the underlying graph. PubMatic names five European countries for the ID+ implementation; Equativ lists its namespace without a country. US availability of the distinct ID+ offer remains unconfirmed. | unconfirmed — Historical issuer document; current US offer unknown. | listed | [402, 377, 403] |
| Proximic by Comscore Predictive Audiences | Comscore identifies Proximic as its business; ultimate beneficial ownership not determined. | Not extracted | C | As specified in product record | Proximic Activation Platform, custom ID-free Predictive Audiences | yes — US political TV exposure segments explicitly offered with the ID-free route, independently of the sold ID-based variants. The current product page offers ID-free audience creation; 2023 Trade Desk evidence is historical access, not a current seat count. | admitted | [404, 209, 350] |
| The Trade Desk Galileo | The Trade Desk | Not extracted | Outside scope | As specified in product record | Do not double-count a UID2 integration as a second interoperable ID. Reopen if a distinct service meets a registered task. | unconfirmed — US-specific eligibility not tested for distinct identifier. | listed | [405] |
| Deep Sync Profile Hygiene & Enrichment | Ultimate parent unconfirmed in verified case records | Not extracted | B | As specified in product record | Profile Hygiene & Enrichment, customer CRM file against Deep Sync graph. | yes — Explicit US outdoor/RV retailer submitted its CRM for matching and appending links under this service. A separate call to purchase offers a match test. | admitted | [27, 27, 28] |
| Deep Sync Transaction Matching | Ultimate parent unconfirmed in verified case records | Not extracted | B | As specified in product record | Transaction Matching via the monthly POS matching waterfall. | yes — Dedicated Transaction Matching case, buyer call to send distinct POS data and a monthly production pipeline establish a separate US identity deliverable: POS transactions resolved to individuals/households, with attribution an optional use. Different from CRM record resolution. | admitted | [12, 12] |
| Kochava advanced measurement household graph (QSR case) | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Trace the relation of that deployment to Kochava Collective and to independently purchasable household mapping; a CTV measurement case cannot establish a current graph licence. Track legal scope separately in H. | unconfirmed — Kochava describes a US QSR CTV campaign using household mapping. The source does not establish a Kochava Collective product or standalone graph licence. | listed | [35] |
| LtvAdx Identity API / HouseholdID | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | Identity API documents household creation and lookup. The US page and pricing evidence establish paid media access to HouseholdID targeting, without a separately priced graph licence. Distinguish device-link and household retention from VAST serving fallback. | unconfirmed — LtvAdx markets a US TV campaign route with HouseholdID targeting in a paid-impression plan. Separate licensing of household resolution is unconfirmed. | listed | [227, 229, 230, 231] |
| LiveRamp Advanced TV programmer CTV Audience Recognition | Publicis Groupe via MMS USA Holdings, Inc.; 2 November 2026 close. | Not extracted | D | TV / CTV | Check separate TV purchase and deliverable to avoid duplicating P001 or P012; the guide establishes a named workflow. | unconfirmed — US purchase or deployment unconfirmed for this programmer workflow. | listed | [307, 141, 318] |
| Factori Cross Device / CTV Graph | Ultimate owner not established by the verified records for this service. | Not extracted | D | TV / CTV | A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule. | unconfirmed — Factori markets the CTV/cross-device dataset globally through AWS Marketplace. The listing does not identify US dataset coverage or a US deployment. | listed | [308, 309] |
| Datonics Cookieless Audience Targeting | Ultimate owner not established by the verified records for this service. | Not extracted | C | As specified in product record | Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task. | unconfirmed — US eligibility unconfirmed. | listed | [406] |
| Project Rearc Accountability Platform draft | IAB Tech Lab | Not extracted | Outside scope | As specified in product record | No dated withdrawal was located in the registered IAB searches. Keep publication of the draft and possible retirement as distinct questions. | unconfirmed — Standards proposal; no US availability assigned. | listed | [407] |
| Audigent Hadron ID | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | Dedicated Prebid module and namespace. TripleLift lists the product by name for US passthrough. Google identifies a beta curation path. The ID is distinct from Audigent SmartPMP. | yes — Dedicated Prebid module and namespace. TripleLift lists the product by name for US passthrough. Google identifies a beta curation path. The ID is distinct from Audigent SmartPMP. | below minimum | [237, 238, 263] |
| Neustar Fabrick ID | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | The documented Prebid module requires a registration key. Equativ maps neustar.biz to Fabrick. TripleLift describes US passthrough with the same issuer and spells the identifier “Fabrik.” No equivalence to TransUnion’s general person graph is inferred. | yes — The documented Prebid module requires a registration key. Equativ maps neustar.biz to Fabrick. TripleLift describes US passthrough with the same issuer and spells the identifier “Fabrik.” No equivalence to TransUnion’s general person graph is inferred. | admitted | [408, 263, 377] |
| Epsilon PubLink | Publicis Groupe | Not established in cited service documents | A | See buying task and route | Publisher Link requires its own enrollment for publisher first-party authenticated inputs. It issues a distinct publisher ID linked to Epsilon CORE ID, taken by TripleLift in US passthrough. The US exchange path is selectable independently of a buyer-side CORE ID resolution engagement. | yes — Publisher Link requires its own enrollment for publisher first-party authenticated inputs. It issues a distinct publisher ID linked to Epsilon CORE ID, taken by TripleLift in US passthrough. The US exchange path is selectable independently of a buyer-side CORE ID resolution engagement. | admitted | [263, 362, 385] |
| Flashtalking FTrack | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | The provider module offers a persistent probabilistic device identifier for publishers’ bidding routes. TripleLift names FTrack US passthrough. Optional household output is outside this A row. | yes — The provider module offers a persistent probabilistic device identifier for publishers’ bidding routes. TripleLift names FTrack US passthrough. Optional household output is outside this A row. | admitted | [263, 409] |
| LiveIntent nonID | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | A publisher uses the LiveIntent module and product-registration route to obtain the email-linked nonID. TripleLift lists LiveIntent for US bridging. The ID route is distinct from general graph or media services. | yes — A publisher uses the LiveIntent module and product-registration route to obtain the email-linked nonID. TripleLift lists LiveIntent for US bridging. The ID route is distinct from general graph or media services. | admitted | [263, 377, 260] |
| Merkle ID | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | Prebid attributes the identifier to Merkury’s private identity graph. TripleLift uses Merkle ID for its own audiences, and Merkle describes US Merkury services. Distinct direct A purchase in addition to P017 B remains unconfirmed. | unconfirmed — Prebid attributes the identifier to Merkury’s private identity graph. TripleLift uses Merkle ID for its own audiences, and Merkle describes US Merkury services. Distinct direct A purchase in addition to P017 B remains unconfirmed. | listed | [263, 377, 363, 410, 389] |
| Viant ID (exchange label) | Ultimate owner unconfirmed in the cited product records | Not established in cited service documents | A | See buying task and route | TripleLift lists a US passthrough identifier but provides no unit or relationship to the separately documented Viant Household ID route, on which Viant says the household ID is not passed in the bidstream. A separate selectable A offer is unconfirmed. | unconfirmed — TripleLift lists a US passthrough identifier but provides no unit or relationship to the separately documented Viant Household ID route, on which Viant says the household ID is not passed in the bidstream. A separate selectable A offer is unconfirmed. | listed | [263, 312, 313] |
| Zeta Customer Matcher | Zeta Global Holdings Corp. (group; distinct graph signer unverified) | Not established in cited service documents | B | See buying task and route | Two input tables, threshold-based matched pairs, separate from CGI. | unconfirmed — US-only Streamlit application; matches records between two customer-supplied tables. The documented processing does not establish a sold first-party-to-Zeta-graph link for this operation, or equivalence to CGI. | listed | [244] |

### Check a result

Look up the test in the register. Read its rule for when the claim stands, when it falls and when the answer stays open [11, 13]. Start with the cases known to be in scope. Then read the result for the wider set if the rule calls for one. If your own buying job falls outside that set, use the source for the specific route you need.

The legal record sits outside the IQ [11, 13]. Look for the firm, the court or agency and the date. Read the signed text for the duty it sets. Look for separate proof that the firm made the change. A class of data in an order may be much wider than the service you plan to buy.

### Check a change

The log keeps the first claim, the challenge and the change [11, 13]. Read what the reviewer found at the source. A score can change when the date of the proof is wrong or when the source describes a different service [11, 13]. Ask what changed in the evidence before you conclude that the product itself changed.

**Corrections ledger**

| ID | Original claim or score | Challenge | Evidence checked | Verdict | Change made | Reason | Affected |
|---|---|---|---|---|---|---|---|
| X01 | Discovery records were keyed only by product names and aliases. H1 called for the same product to be followed from the universe screen to IQ admission. | A provisional commercial name could change during verification, breaking a denominator join. | Compared the draft register, H1 and the kit’s product-name join. | structural | Assigned each seed a permanent product_id. Added rules for renames, merges and splits while retaining the kit’s name/alias join. | Track the same product through the full eligible frame and admitted subset. | data/product-register.json, RESEARCH-BRIEF.md, H1–H2 sensitivity analyses |
| X02 | Some proposed instrument labels used register terminology for different kit columns, including an operating-entity heading for hq and a measurement-period heading for vintage. | A reader could mistake a headquarters for a selling entity or an estimate edition for the period measured. | Compared instrument definitions in kit/SCHEMA.md section 14 with the configuration. | structural | Restored field-specific display labels and notes for the studies, estimates, agentic, legal and universe instruments. | Preserve the meaning of existing build columns. | paper.config.json |
| X03 | H4’s first drafting pass did not fully specify tie handling for archived guidance or how to treat an API-specific decision without an effective removal. | Source choice or a pooled timeline could change the apparent direction of cookie-sync guidance. | Compared H4’s registered index event, data frame and paired-document outcome with the proposed secondary contrasts. | structural | Named the primary Chrome-choice contrast, separated announcements from effective removals, set event-day and version tie rules, and required unavailable or conflicting secondary results to be visible. | A product-document change must be traceable to the dated pair used in the calculation. | data/hypotheses.json, H4 |
| X04 | Initial wording in the public-disclosure tests did not explicitly exclude a material inaccessible lead from every bounded-support branch. | A completed search with an unread record could be coded as both support and unresolved. | Read the three decision branches for H1 and H7 using a case with a credible blocked source. | structural | Excluded unresolved material candidates from bounded support and gave a verified counterexample priority over an incomplete screen. | Each input state must have one verdict. | data/hypotheses.json, H1 and H7 |
| X05 | All majority tests bounded unknown outcomes among eligible services, but only H4 explicitly requested a sensitivity for uncertain service eligibility. | A product with unknown US or graph status could disappear from the denominator even though adding it might reverse the majority. | Applied hypothetical excluded/uncertain cases to the H3, H4, H5 and H8 denominator definitions. | structural | Introduced a distinct E count and an expanded possible-inclusion interval in each strict-majority test. The primary verdict on the confirmed eligible frame is retained. | Separate uncertainty about membership in the sample from missing evidence about eligible products. | data/hypotheses.json, RESEARCH-BRIEF.md |
| X06 | The research files initially described the cross-paper study registry as missing from the pack and awaiting a join. | The pack lacks a registry, but its source repository carries one. | Read research/_briefs/shared-registry.json at blob 914568ada71a9a6044fefef279a1952b1396ed5c. Compared its five study titles and original URLs to the extracted origins. | structural | Updated study notes to record the completed check. No corresponding REG IDs found; kept the identifiers empty. | Distinguish access to the registry from finding an applicable origin. | evidence/research/B.json and evidence/research/C.json |
| X07 | Four issuer URLs appeared in multiple workstreams and originally exceeded 25 quoted words each when their excerpts were combined. | The same source must respect its quotation limit across the complete handoff. | Checked each finding’s passage at its own source URL, normalized equivalent URLs, and summed word counts across streams. [165, 371, 305, 399, 27, 307] | structural | Shortened the universe excerpts on Yahoo ConnectID, Deep Sync, FreeWheel and LiveRamp Advanced TV; kept their exact text and locating details. | Retain evidence for each finding within the corpus-wide per-source quotation budget. | evidence/research/U.json, cross-stream passage audit |
| X08 | The Deep Sync rate table used provisional product names while the universe sweep was still running. | Unlinked names would make it harder to follow candidate cases into the admitted cohort. | Joined the actual service names to the proposals in U. Confirmed Profile Hygiene & Enrichment, Transaction Matching, and the separate role of the UID2 destination. [27, 28, 12, 31] | structural | Replaced temporary row labels with proposed IDs IDN-P042 and IDN-P043; retained IDN-P002 as the destination where applicable. | Keep publication, service tested and destination distinct when a counterexample depends on product admission. | evidence/research/C.json, match_rate_claim_screen |
| X09 | From Prebid.js 11.14, a configured TCF or GPP CMP lookup failure prevents user-ID retrieval and inclusion in bids; a timeout follows a different path using the consent data then available. | Independent re-read tested the original scope and classification. | The 11.14.0 release record confirms that userId does not run when an expected CMP cannot be found. The User ID module guide also covers CMP error responses and distinguishes a timeout: the submodules then run using the consent data available. The guarantee concerns IDs retrieved by these submodules, not every source of IDs in a bid request. [239, 255] | supported with qualification | From Prebid.js 11.14, user-ID submodules skip retrieval and add no IDs to bid requests when a configured TCF or GPP consent-platform lookup fails. On a timeout, the submodules run with the consent data available. [Source IDs added: IDN-AV-S01] | A bid may carry an ID supplied by another route. Both the release record and current guide were checked. | data/claims-overlay.json, IDN-A-F05; evidence/verify/A.json |
| X10 | The UID2 operator describes a global opt-out route in which the portal sends the choice to the operator, publishers receive an opt-out response and DSPs receive a separate webhook. | Independent re-read tested the original scope and classification. | The UID2 guide distinguishes a single participant’s opt-out, which is not relayed to the operator, from portal opt-out for an email address or phone number. For the latter, token generation or refresh calls return the opt-out status and DSPs get webhook notifications. The guide acknowledges a propagation delay. [149, 186] | supported with qualification | UID2’s portal accepts an ecosystem opt-out for a supplied email address or phone number. Publishers learn the result when they generate or refresh a token, and the operator notifies DSPs through a separate webhook. Propagation can take time. [Source IDs added: IDN-A-S09] | A site-specific opt-out is not relayed to the operator; each email address or phone number needs its own portal opt-out. | data/claims-overlay.json, IDN-A-F07; evidence/verify/A.json |
| X11 | The UID2 DSP guide forbids using an opted-out UID2 in real-time bidding and allows the DSP to use another identifier or decline to bid. | Independent re-read tested the original scope and classification. | The bidding section instructs a DSP that it may bid with an alternate identifier, or abstain, once a UID2 is opted out. The prohibition attaches to that UID2. The webhook and status endpoint describe how to learn the choice, with no published compliance rate. [186, 149] | supported as stated | The UID2 DSP guide forbids using an opted-out UID2 in real-time bidding and allows the DSP to use another identifier or decline to bid. [Source IDs added: IDN-A-S08] | The reopened DSP guide’s last-modified timestamp is November 11, 2026. This is the document revision date; the origin date of its bidding rule is unestablished. | data/claims-overlay.json, IDN-A-F08; evidence/verify/A.json |
| X12 | Prebid describes SharedID as an ID stored in the publisher’s first-party domain and made available to adapters. | Independent re-read tested the original scope and classification. | Prebid describes SharedID as a locally stored UUID in the first-party domain and says supporting adapters can use it. Cookie and HTML5 options are documented. The general user-ID module allows publishers to select recipients, so adapter access here is capacity, not universal delivery. [258, 239] | supported as stated | Prebid describes SharedID as an ID stored in the publisher’s first-party domain and made available to adapters. [Source IDs added: IDN-A-S06] | Prebid describes SharedID as a locally stored UUID in the first-party domain and says supporting adapters can use it. Cookie and HTML5 options are documented. The general user-ID module allows publishers to select recipients, so adapter access here is capacity, not universal delivery. | data/claims-overlay.json, IDN-A-F17; evidence/verify/A.json |
| X13 | {"documented_action": "Conveys matcher, inserter or method metadata"} | The table row was checked against the reopened records. | The issuer added representational fields, without evidence that an exchange populated them. [252] | structural | {"documented_action": "The EID schema can carry the inserter, matcher and match method."} | Keep the table consistent with the independently verified disclosure. | evidence/research/A.json, documented_signal_handoffs, data row 1 |
| X14 | {"documented_action": "Transmits do-not-sell-or-share request"} | The table row was checked against the reopened records. | The W3C draft specifies a signal; the table does not document a live transmission. [3] | structural | {"documented_action": "The draft defines a signal to request limits on sale or sharing."} | Keep the table consistent with the independently verified disclosure. | evidence/research/A.json, documented_signal_handoffs, data row 4 |
| X15 | 2026-08-12 | A downstream field repeated, or could imply, superseded wording. | The date identifies the GPP overview revision. The page gives separate dates for the specification releases. [253] | structural | [Field cleared; see reason] | The date identifies the GPP overview revision. The page gives separate dates for the specification releases. | evidence/research/A.json, sources.2.pub_date |
| X16 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The date identifies the GPP overview revision. The page gives separate dates for the specification releases. [253] | structural | Issuer record for documented scope. No independent implementation or performance test. Last updated August 12, 2026. | The date identifies the GPP overview revision. The page gives separate dates for the specification releases. | evidence/research/A.json, sources.2.reliability_notes |
| X17 | 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | The publisher labels dateModified in its HTML. [149] | structural | [Field cleared; see reason] | The publisher labels dateModified in its HTML. | evidence/research/A.json, sources.7.pub_date |
| X18 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The publisher labels dateModified in its HTML. [149] | structural | Issuer record for documented scope. No independent implementation or performance test. Last updated November 11, 2026. | The publisher labels dateModified in its HTML. | evidence/research/A.json, sources.7.reliability_notes |
| X19 | 2026-11-05 | A downstream field repeated, or could imply, superseded wording. | The reopened UID2 DSP guide carries a November 11, 2026 dateModified stamp. [186] | structural | [Field cleared; see reason] | The reopened UID2 DSP guide carries a November 11, 2026 dateModified stamp. | evidence/research/A.json, sources.8.pub_date |
| X20 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The reopened UID2 DSP guide carries a November 11, 2026 dateModified stamp. [186] | structural | Issuer record for documented scope. No independent implementation or performance test. Last updated November 11, 2026. | The reopened UID2 DSP guide carries a November 11, 2026 dateModified stamp. | evidence/research/A.json, sources.8.reliability_notes |
| X21 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The article provides a timestamp but does not identify it as original publication. [256] | structural | Issuer record for documented scope. No independent implementation or performance test. Article header displays October 15, 2026 at 6:44 a.m. EDT, without specifying the date type. | The article provides a timestamp but does not identify it as original publication. | evidence/research/A.json, sources.9.reliability_notes |
| X22 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The article provides a timestamp but does not identify it as original publication. [176] | structural | Issuer record for documented scope. No independent implementation or performance test. Article header displays November 25, 2025 at 8:24 a.m. EST, without specifying the date type. | The article provides a timestamp but does not identify it as original publication. | evidence/research/A.json, sources.10.reliability_notes |
| X23 | 2026-02-06 | A downstream field repeated, or could imply, superseded wording. | The topic metadata and footer describe the modification date; the original publication field is blank. [257] | structural | [Field cleared; see reason] | The topic metadata and footer describe the modification date; the original publication field is blank. | evidence/research/A.json, sources.11.pub_date |
| X24 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The topic metadata and footer describe the modification date; the original publication field is blank. [257] | structural | Issuer record for documented scope. No independent implementation or performance test. Last modified February 6, 2026. | The topic metadata and footer describe the modification date; the original publication field is blank. | evidence/research/A.json, sources.11.reliability_notes |
| X25 | 2025-08-25 | A downstream field repeated, or could imply, superseded wording. | The IAB Tech Lab overview gives the page revision and the rename separately. [251] | structural | [Field cleared; see reason] | The IAB Tech Lab overview gives the page revision and the rename separately. | evidence/research/A.json, sources.18.pub_date |
| X26 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The IAB Tech Lab overview gives the page revision and the rename separately. [251] | structural | Issuer record for documented scope. No independent implementation or performance test. Last updated August 25, 2025. The rename took place in December 2024. | The IAB Tech Lab overview gives the page revision and the rename separately. | evidence/research/A.json, sources.18.reliability_notes |
| X27 | ID5 ID Decryption; DSP route | A downstream field repeated, or could imply, superseded wording. | The linked decryption section sends readers to a login; the cited public wording occurs in the separate privacy section. [176] | structural | Privacy & Regulations / Privacy-by-Design, paragraph beginning “When ID5 returns an ID to the page” | The linked decryption section sends readers to a login; the cited public wording occurs in the separate privacy section. | evidence/research/A.json, findings.9.support.0.locator |
| X28 | EID Source  \| criteo.com | A downstream field repeated, or could imply, superseded wording. | Cite the exact value of the EID Source cell. The appended pipe had been a serialization of the HTML table. [250] | structural | criteo.com | Cite the exact value of the EID Source cell. The appended pipe had been a serialization of the HTML table. | evidence/research/A.json, findings.15.support.0.passage |
| X29 | H3 eligibility and outcome denominators are open. This stream does not resolve recipient-side actions for every registered ID or territory. The US availability of Utiq and EUID must be established separately by U. | A downstream field repeated, or could imply, superseded wording. | EUID supplies a named geographic coverage list. [211] | structural | H3 eligibility and outcome denominators remain open. Recipient-side actions are unresolved for some products and territories. EUID’s published service-area list covers Europe and other specified areas but does not list the United States. Treat its evidence separately from UID2 and retain the Utiq US availability question for the universe screen. | EUID supplies a named geographic coverage list. | evidence/research/A.json, open_questions.0 |
| X30 | EUID DSP guide at the attempted euid.eu path returned 404; do not infer that the service lacks a guide. The APS website notice is about the APS site itself and did not establish a publisher-ID consent path. Criteo OneTag and Criteo ID for Exchanges require a service-level bridge. | A downstream field repeated, or could imply, superseded wording. | The live EUID-specific guide partially closes the access gap without importing UID2 requirements. [411, 250] | structural | EUID’s DSP-guide URL returned 404 on recheck. The current EUID opt-out page documents portal-to-operator requests, token suppression for publishers and notifications or lookups for advertisers and data providers. Its participant table does not document a DSP webhook. The APS site notice did not establish a publisher-ID consent path. The service relationship between Criteo OneTag and Criteo ID for Exchanges remains unresolved. | The live EUID-specific guide partially closes the access gap without importing UID2 requirements. | evidence/research/A.json, open_questions.1 |
| X31 | Truthset reported a 13% mean across six data providers for IP-to-postal accuracy in its US operator-reference benchmark. | Independent re-read tested the original scope and classification. | Truthset scores provider-supplied IPv4/postal pairs with models calibrated against operator subscriber assignments. The study reports a 13% mean across six masked providers, with displayed provider results from 4% to 18%. [40, 41] | supported with qualification | In a US benchmark commissioned by CIMM and Go Addressable, Truthset reported average modeled IP-to-postal linkage accuracy of 13% across six unnamed providers for December 2024 through February 2025. | The reference inputs came from two ISPs and one virtual pay-TV provider. The 938 million distinct pairs are scored candidate links. The report does not give a numerator and denominator of operator-confirmed hits over those pairs or the weights used in the reported mean. | data/claims-overlay.json, IDN-B-F01; evidence/verify/B.json |
| X32 | Truthset reported a 16% mean IP-to-email accuracy across providers for the other linkage type in the same study. | Independent re-read tested the original scope and classification. | Truthset also scores 5.6 billion distinct IP-to-email pairs from participating providers. Its mean is 16%, with provider results from 7% to 30%. [40, 41] | supported with qualification | In the same December 2024 through February 2025 US benchmark, Truthset reported average modeled IP-to-email linkage accuracy of 16% across six unnamed providers. [Source IDs added: IDN-B-S02] | Data Sources and Methodology Overview distinguish the scored inference set from sampled subscriber links. The unit is a pair consisting of an IP and email identifier. No admitted product is identified. | data/claims-overlay.json, IDN-B-F02; evidence/verify/B.json |
| X33 | Miller and Skiera estimated a mean third-party cookie lifetime of 279 days in their historical European exchange sample. | Independent re-read tested the original scope and classification. | One European exchange supplied auctions observed March 3, 2014 to July 16, 2016. Section 5.2 and the lifetime analysis distinguish the observed mean, 216 days, from the censoring-adjusted mean, 279 days; both medians are 68. [49] | supported with qualification | Miller and Skiera estimated a censoring-adjusted mean lifetime of 279 days for 54,127 third-party cookies sampled on one day at a European ad exchange. The observed mean was 216 days. | The sample is 1% of active unique cookies on April 29, 2015, not a panel of 54,127 verified people or the full population of newly issued cookies. This is historical exchange evidence, not a US platform-policy limit. | data/claims-overlay.json, IDN-B-F10; evidence/verify/B.json |
| X34 | In its December 2006 US home-PC analysis, comScore reported that 31% of users cleared first-party cookies for the site studied. | Independent re-read tested the original scope and classification. | Comscore's narrative calls them users. In its first-party table, 31% is in the column for percent of computers with at least one reset. The study covers 400,000 US home PCs, one site and separately one third-party server. [50] | supported with qualification | In Comscore’s December 2006 US home-PC sample, 31% of computers had at least one first-party cookie reset at the site studied. The first-party analysis excluded login cookies. | The table supplies a computer denominator, not unique people. The third-party column is a separate 27% for one ad server. | data/claims-overlay.json, IDN-B-F11; evidence/verify/B.json |
| X35 | A LiveRamp manual bearing 2017 copyright already published the same comScore precision range that appears in its later online help. | Independent re-read tested the original scope and classification. | The PDF has a 2017 copyright and repeats the same precision range as the Connect note. This PDF's creation and modification metadata both give October 5, 2018. The online note explicitly gives August 19, 2020 as its last modification. [44, 43] | supported with qualification | A LiveRamp PDF bearing a 2017 copyright and an online help page last modified in August 2020 repeat the same comScore precision range. | Neither the copyright nor the file's creation date establishes first public release or the period of comScore’s test. These are two seller documents repeating one claim. | data/claims-overlay.json, IDN-B-F12; evidence/verify/B.json |
| X36 | Firsthand Technology Value Fund reported that its portfolio company Tapad commissioned Nielsen’s device-graph validation. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | The Winter 2015 portfolio newsletter says Tapad commissioned Nielsen, and describes 91.2% matching of related devices. The fund held an interest in Tapad. Later trade coverage says Nielsen used its panel as a reference, but provides no original test design or confusion matrix. [53, 54] | supported with qualification | In its Winter 2015 portfolio newsletter, Firsthand Technology Value Fund said Tapad had commissioned Nielsen to test Tapad’s Device Graph. [Evidence class: vendor_assertion] [Source IDs added: IDN-BV-S01] | The evidence is an interested investor’s newsletter about a portfolio company, not a regulatory or other official record. The historical study was explicitly commissioned; the claim here concerns who reported its sponsor. The newsletter supports attribution to an interested investor. It is neither a regulatory filing nor a tester-issued measurement report. Do not apply the reported percentage to today's product or substitute the later Hotels.com comparison. | data/claims-overlay.json, IDN-B-F13; evidence/verify/B.json |
| X37 | {"design": "Commercial pairs compared with time-sensitive operator records", "setting": "US providers, masked", "period": "Study includes late 2024 and early 2025 (daily analysis and December 2024 example); full window varies by comparison", "source_ids": "IDN-B-S01", "n": "Six providers; two ISPs and one TV operator supplied reference records"} | The table row was checked against the reopened records. | Operator truth data inform modeling; not a direct correctness check against every candidate pair.; The reference market is US; headquarter location and business scope of unnamed submitters are not disclosed.; Both input windows are explicitly dated. Other examples in the slides need not share these boundaries.; Go Addressable explicitly identifies both commissioners.; Distinguish candidates scored from the operator-reference sample. [40, 41] | structural | {"design": "Vendor-submitted IPv4 linkage pairs scored by models trained and tested on sampled ISP/vMVPD subscriber records and lease-time data", "setting": "US subscriber reference samples from two ISPs and one vMVPD; six link suppliers masked", "period": "2024-12-01 to 2025-02-28 for inference and truth inputs (Data Sources slide)", "source_ids": "IDN-B-S01; IDN-B-S02", "n": "Six providers; 938 million distinct IP/postal pairs and 5.6 billion IP/email pairs scored; validation from two ISPs and one vMVPD"} | Keep the table consistent with the independently verified disclosure. | evidence/research/B.json, study_design_and_scope, data row 1 |
| X38 | {"year": "2017 copyright manual; undated web note"} | The table row was checked against the reopened records. | Web note carries a modification date; PDF metadata and copyright do not identify first publication. [43, 44] | structural | {"year": "2017 copyright; PDF created 2018-10-05; online help last modified 2020-08-19"} | Keep the table consistent with the independently verified disclosure. | evidence/research/B.json, study_design_and_scope, data row 3 |
| X39 | {"n": "126 participants", "uncertainty": "Limited recruitment, sponsor disclosure"} | The table row was checked against the reopened records. | Table 4’s 0.91 is a held-out test result; 126 is the recruitment total, not its denominator.; Funding provenance is not statistical uncertainty. [47] | structural | {"n": "126 participants; 107 complete mobile/desktop pairs, split into 63 training and 44 test pairs", "uncertainty": "Convenience recruitment and 44 held-out pairs; Table 4 supplies point estimates without confidence intervals"} | Keep the table consistent with the independently verified disclosure. | evidence/research/B.json, study_design_and_scope, data row 5 |
| X40 | {"n": "54,127 in sampled user/cookie cohort", "effect": "279 days"} | The table row was checked against the reopened records. | Cookies lack participant-level identity labels.; Different estimands in the original text. [49] | structural | {"n": "54,127 unique exchange cookies, sampled among active cookies on April 29, 2015", "effect": "Censoring-adjusted mean 279 days; observed mean 216 days"} | Keep the table consistent with the independently verified disclosure. | evidence/research/B.json, study_design_and_scope, data row 7 |
| X41 | {"effect": "31% in first-party analysis"} | The table row was checked against the reopened records. | Use the denominator and exclusion shown in the first-party table. [50] | structural | {"effect": "31% of panel home computers had at least one first-party cookie reset at the selected site; login cookies excluded"} | Keep the table consistent with the independently verified disclosure. | evidence/research/B.json, study_design_and_scope, data row 8 |
| X42 | The household/IP benchmark tested commercial linkages against operator records but masked providers. A TV match study measured profile bias without an independent truth set. A named LiveRamp claim and a historical Tapad vendor comparison need original test documentation before product-level inference. Separate studies show how device links, network assignment and cookie persistence depend on the observed population, time window and unit. | A downstream field repeated, or could imply, superseded wording. | Distinguish modeled Truthscores from direct counts of correct submitted links. [40, 41, 42, 43, 45, 47, 48, 49, 53] | structural | Truthset used subscriber reference data to score commercial IPv4 links from unnamed suppliers. A separate TV study measured changes in audience composition and lacked ground truth for individual links. The named LiveRamp and historical Tapad results require close attention to funding, study design and product version. Studies of research-device matching, IP assignment and cookie survival measure different outcomes. | Distinguish modeled Truthscores from direct counts of correct submitted links. | evidence/research/B.json, summary |
| X43 | 2025-11 | A downstream field repeated, or could imply, superseded wording. | Date on report cover. [40] | structural | 2025-11-05 | Date on report cover. | evidence/research/B.json, sources.0.pub_date |
| X44 | Study includes late 2024 and early 2025 (daily analysis and December 2024 example); full window varies by comparison | A downstream field repeated, or could imply, superseded wording. | Data Sources specifies both windows. [40] | structural | 2024-12-01 to 2025-02-28 for inference and validation inputs; illustrations can show dates outside those windows. | Data Sources specifies both windows. | evidence/research/B.json, sources.0.measurement_period |
| X45 | [field absent] | A downstream field repeated, or could imply, superseded wording. | The page footer supplies a modification date, not original publication. [43] | structural | Last modified August 19, 2020; test date unreported. | The page footer supplies a modification date, not original publication. | evidence/research/B.json, sources.3.version |
| X46 | [field absent] | A downstream field repeated, or could imply, superseded wording. | PDF cover and embedded creation/modification metadata. [44] | structural | 2017 copyright; distributed PDF created and modified 2018-10-05. First publication date unconfirmed. | PDF cover and embedded creation/modification metadata. | evidence/research/B.json, sources.10.version |
| X47 | 2017 | A downstream field repeated, or could imply, superseded wording. | A copyright year is not a verified date of publication. [44] | structural | [Field cleared; see reason] | A copyright year is not a verified date of publication. | evidence/research/B.json, sources.10.pub_date |
| X48 | Study includes late 2024 and early 2025 (daily analysis and December 2024 example); full window varies by comparison | A downstream field repeated, or could imply, superseded wording. | The Data Sources slide gives both dates. [40] | structural | 2024-12-01 to 2025-02-28 for the operator reference and vendor inference sets. | The Data Sources slide gives both dates. | evidence/research/B.json, findings.0.scope_time |
| X49 | Study includes late 2024 and early 2025 (daily analysis and December 2024 example); full window varies by comparison | A downstream field repeated, or could imply, superseded wording. | The Data Sources slide gives both dates. [40] | structural | 2024-12-01 to 2025-02-28 for the operator reference and vendor inference sets. | The Data Sources slide gives both dates. | evidence/research/B.json, findings.1.scope_time |
| X50 | 126 Internet users; mobile and desktop devices linked using observed IP and browsing history; the device pair is the unit. F1 combines precision and recall. | A downstream field repeated, or could imply, superseded wording. | The included 126 people are not the held-out denominator. [47] | structural | 126 recruited internet users; 107 paired devices split into 63 training and 44 test pairs, with unmatched devices also evaluated. Authors tested their own three-stage model. | The included 126 people are not the held-out denominator. | evidence/research/B.json, findings.7.study_design |
| X51 | Copyright 2017; test dates undisclosed | A downstream field repeated, or could imply, superseded wording. | Distinguish document and study dates. [43, 44] | structural | PDF copyright 2017, file created 2018-10-05; web help last modified 2020-08-19. Neither gives the test period. | Distinguish document and study dates. | evidence/research/B.json, findings.11.scope_time |
| X52 | These documents repeat one statement by the seller. They establish the age of the claim, not a second study or an accuracy test of the 2026 product. | A downstream field repeated, or could imply, superseded wording. | Clarifies what the document timestamps establish. [43, 44] | structural | The two seller records repeat one result. Their dates bound the documents; they do not date the comScore fieldwork or establish its relevance to the 2026 product. | Clarifies what the document timestamps establish. | evidence/research/B.json, findings.11.limitations |
| X53 | These documents repeat one statement by the seller. They establish the age of the claim, not a second study or an accuracy test of the 2026 product. | A downstream field repeated, or could imply, superseded wording. | Clarifies what the document timestamps establish. [43, 44] | structural | The two seller records repeat one result. Their dates bound the documents; they do not date the comScore fieldwork or establish its relevance to the 2026 product. | Clarifies what the document timestamps establish. | evidence/research/B.json, findings.11.counterevidence |
| X54 | Abstract; §5.2 Description of Data and §5.3 sampling | A downstream field repeated, or could imply, superseded wording. | The mean is censoring-adjusted; the sample-management sections alone do not show the distinction. [49] | structural | Abstract; §5.2 Description of Data, sampling and “mean cookie lifetime adjusted for censoring” in the results. | The mean is censoring-adjusted; the sample-management sections alone do not show the distinction. | evidence/research/B.json, findings.9.support.0.locator |
| X55 | Panel size and precision range are a seller-reported lead. Same original study must be identified before classifying independence. | A downstream field repeated, or could imply, superseded wording. | The guide footer establishes the last modification. This register field preserves the distinction after version is dropped. [43] | structural | The help page was last modified on August 19, 2020. Test dates and the original publication date are not reported. LiveRamp gives a panel size and precision range; financial independence and denominator remain unconfirmed. | The guide footer establishes the last modification. This register field preserves the distinction after version is dropped. | evidence/research/B.json, sources.3.reliability_notes |
| X56 | Historic seller manual repeats the panel description in LiveRamp’s later web note. There is one evidence origin for the test. | A downstream field repeated, or could imply, superseded wording. | The printed cover and reacquired PDF metadata distinguish copyright year from file creation; these facts must survive publication without the version field. [44] | structural | This seller manual bears a 2017 copyright. The PDF creation and modification timestamps are October 5, 2018. Neither establishes when the document was first published or when comScore tested the graph. The later help page repeats the same study. | The printed cover and reacquired PDF metadata distinguish copyright year from file creation; these facts must survive publication without the version field. | evidence/research/B.json, sources.10.reliability_notes |
| X57 | In its public summary of a US insurer’s Data Health Assessment, TransUnion reports a 99.5% match to identities for that insurer’s customers. | Independent re-read tested the original scope and classification. | The teaser uses customers and potential language. In the two-page case report, TransUnion says it actually matched 99.5% of this insurer’s records to a person in its offline identity graph. It reports 97.4% matching on first name, last name and another PII field. The absolute input count, field combination or threshold for the remaining links, deduplication rule and labeled truth are absent. TransUnion separately lists its assessment and its commercial resolution/enrichment service. [17, 18, 19] | supported with qualification | In a Data Health Assessment for one unnamed US insurer, TransUnion says it matched 99.5% of the insurer’s records to a person in its offline identity graph. The report also gives a 97.4% match on first name, last name and at least one other PII field. It does not disclose the full acceptance rule for the 99.5% rate. [Source IDs added: IDN-CV-S01, IDN-CV-S02] | The teaser and full PDF are one corporate origin. Field agreement for 97.4% does not specify the acceptance rule for all 99.5%. The separately listed provision of a diagnostic does not prove the commercial resolution SKU handled the case. | data/claims-overlay.json, IDN-C-F04; evidence/verify/C.json |
| X58 | Forrester’s July 2026 TransUnion-commissioned model puts a 20-percentage-point gain in identity match rates on a composite enterprise derived from four customer interviews. | Independent re-read tested the original scope and classification. | TransUnion selected the customer names; Forrester interviewed four decision-makers at US-headquartered companies. Its July 2026 table assumes 15 million targeted households and 60% vs 80% matching in each of three model years. Twenty percentage points and 3 million incremental households follow from those inputs. These are model assumptions informed by interviews; none is a measured file-level outcome. [26] | supported with qualification | In July 2026, Forrester used four customer interviews in a TransUnion-funded study to model a rise in household match rate from 60% to 80% for a composite enterprise. The 20-percentage-point gain is a model assumption. | Keep commissioned_measurement for the study provenance; change claim_status from actual to forecast for modeled outcomes in the three-year TEI composite, consistent with the tables. | data/claims-overlay.json, IDN-C-F10; evidence/verify/C.json |
| X59 | Deep Sync reports a 73.0% cumulative match rate after three passes over a US grocery chain’s monthly POS file, using loyalty, payment-card and modeled geographic/household links. | Independent re-read tested the original scope and classification. | The grocery case identifies a US retailer and presents three cumulative percentages, 22.8%, 30.6% and 73.0%, over the monthly POS file after loyalty, card and household/geographic inference. The roughly 9.6m, 3.3m and 18.1m pass counts sum to 31.0m, versus the approximately 32m headline and 43.8m+ monthly volume. Elsewhere it gives 64–73% of monthly volume and discusses anonymous transactions. No exact common count or labeled person accuracy is recoverable. [12] | supported with qualification | Deep Sync reports a 73.0% cumulative match rate for one US grocery chain’s monthly POS file after loyalty, card and probabilistic household/geographic matching. Its summary gives a range of roughly 64–73% of monthly volume. Its rounded pass counts and headline total do not reconcile exactly. | Use vendor-published waterfall endpoint and attach its rounding and file caveats in H1. No ground truth is shown for the third pass. | data/claims-overlay.json, IDN-C-F16; evidence/verify/C.json |
| X60 | ID5 says Pubfinity tested ID5 ID on 5% of its global Windows desktop campaign traffic over 14 days in Q1 2024. | Independent re-read tested the original scope and classification. | ID5 says the Pubfinity A/B test used 5% of global traffic for 14 days in Q1 2024, with display and video inventory from over twenty publishers and over one hundred Windows desktop games and apps. The unit of traffic sampled, total eligible impressions/users and allocation procedure are unspecified. [412] | supported with qualification | ID5 says Pubfinity ran a 14-day A/B test in Q1 2024 on 5% of global traffic in a test of Windows desktop games and apps. It does not specify the traffic unit or total eligible volume. | ID5 says the Pubfinity A/B test used 5% of global traffic for 14 days in Q1 2024, with display and video inventory from over twenty publishers and over one hundred Windows desktop games and apps. The unit of traffic sampled, total eligible impressions/users and allocation procedure are unspecified. | data/claims-overlay.json, IDN-C-F24; evidence/verify/C.json |
| X61 | Zeta’s DSP pixel guide says the tag calls exchange partners to match IDs when it loads. | Independent re-read tested the original scope and classification. | The JavaScript FAQ says exchange-partner requests on tag loads help synchronize cookie IDs. The later piggybacking FAQ permits turning those requests off and states that which pixels fire depends on prior matching and the user’s cookies. The page does not date a change in behavior or describe every Zeta service. [413] | supported with qualification | Zeta’s DSP guide says its JavaScript tag normally requests cookie-ID matches from exchange partners when it loads. The guide says the partner pixels can be disabled. | The JavaScript FAQ says exchange-partner requests on tag loads help synchronize cookie IDs. The later piggybacking FAQ permits turning those requests off and states that which pixels fire depends on prior matching and the user’s cookies. The page does not date a change in behavior or describe every Zeta service. | data/claims-overlay.json, IDN-C-F25; evidence/verify/C.json |
| X62 | For one US mass retailer, Deep Sync reports that UID2 matched 73.2 million records (51%) before CRM enrichment and 94.4 million (66%) after enrichment. | Independent re-read tested the original scope and classification. | The social/identity comparison table labels the UID2 rows as matched records: 73.2m/51% before and 94.4m/66% after. With the stated 141.7m CRM total, those counts yield around 51.7% and 66.6%; no rounding, exclusion or destination eligibility rule is given. The case is Deep Sync’s account, not a Trade Desk measurement. [31] | supported with qualification | For one US mass retailer, Deep Sync’s UID2 row reports 73.2 million matched records (51%) before CRM enrichment and 94.4 million (66%) after. The case gives a nominal file of 141.7 million records but does not explain the whole-percent rounding or exact destination denominator. | The social/identity comparison table labels the UID2 rows as matched records: 73.2m/51% before and 94.4m/66% after. With the stated 141.7m CRM total, those counts yield around 51.7% and 66.6%; no rounding, exclusion or destination eligibility rule is given. The case is Deep Sync’s account, not a Trade Desk measurement. | data/claims-overlay.json, IDN-C-F26; evidence/verify/C.json |
| X63 | {"eligible_base": "Insurer customers; exact input file not disclosed", "matched_population": "Vendor-labeled verified identities", "method": "Absent from accessible summary", "period": "Not stated; summary read 2026-11-11", "qualification": "Assessment versus buyable resolution-service attribution pending.", "source_id": "IDN-C-S04"} | The table row was checked against the reopened records. | Full PDF page 2 names records; the teaser refers to customers.; The recovered issuer report identifies the graph and output unit.; Full issuer report p. 2 provides this level of field agreement, but not a uniform acceptance threshold.; The issuer PDF predates the 2026 access date.; New case PDF and issuer catalog change the basis for H1 screening.; The full two-page case supplies the input unit, output unit and partial matching method. [18] | structural | {"eligible_base": "Records in the unnamed US insurer’s assessment file; absolute count, exclusions and deduplication unspecified", "matched_population": "A person in the TransUnion offline identity graph", "method": "97.4% match on first name, last name and another PII field; complete rule for 99.5% not given", "period": "Case PDF copyright 2025; assessment period not reported", "qualification": "TransUnion reports a file-to-person result from its assessment. The diagnostic assessment and commercial resolution/enrichment service have separate catalog entries. The remaining acceptance paths and direct scored-service attribution are not shown.", "source_id": "IDN-CV-S01"} | Keep the table consistent with the independently verified disclosure. | evidence/research/C.json, match_rate_claim_screen, data row 3 |
| X64 | {"geo": "Composite, no US stratum"} | The table row was checked against the reopened records. | The TEI interview table distinguishes headquarters, market focus and modeled population. [26] | structural | {"geo": "Four US-headquartered customers: three North America-focused, one US-focused. Composite has no file-level US stratum."} | Keep the table consistent with the independently verified disclosure. | evidence/research/C.json, match_rate_claim_screen, data row 9 |
| X65 | {"qualification": "Elsewhere describes roughly 70% of anonymous transactions and 64–73% of monthly volume. Accuracy not measured."} | The table row was checked against the reopened records. | The claim and numerical rows describe different bases; a single exact numerator cannot be recovered. [12] | structural | {"qualification": "Vendor also reports about 70% of anonymous transactions and 64–73% of monthly volume. Approximate pass counts total 31.0m, against a 32m headline. The final pass infers buyers from geography and households; no labeled accuracy."} | Keep the table consistent with the independently verified disclosure. | evidence/research/C.json, match_rate_claim_screen, data row 11 |
| X66 | {"qualification": "Do not attribute the experiment to The Trade Desk or compare destination-specific populations. Meta’s upper reach value in this case exceeds 100%."} | The table row was checked against the reopened records. | The issuer’s rates should remain attributed; the eligible UID2-specific denominator is unclear. [31] | structural | {"qualification": "Deep Sync’s labeled UID2 row gives 51% and 66%. Dividing by the 141.7m nominal CRM size gives about 51.7% and 66.6%; rounding or destination eligibility is unspecified. Meta’s upper estimate exceeds 100%."} | Keep the table consistent with the independently verified disclosure. | evidence/research/C.json, match_rate_claim_screen, data row 14 |
| X67 | {"setting": "TransUnion customer decision-makers"} | The table row was checked against the reopened records. | The original discloses sponsor input to recruitment; three organizations have North American geographic focus and one is US-focused. [26] | structural | {"setting": "Decision-makers from four US-headquartered customers, recruited from names supplied by TransUnion"} | Keep the table consistent with the independently verified disclosure. | evidence/research/C.json, sponsored_identity_studies, data row 1 |
| X68 | summary only | A downstream field repeated, or could imply, superseded wording. | The teaser contains a direct issuer download; the accessible evidence now exceeds the teaser. [17, 18] | structural | Public summary opened; full two-page issuer PDF recovered separately as IDN-CV-S01. | The teaser contains a direct issuer download; the accessible evidence now exceeds the teaser. | evidence/research/C.json, sources.3.access_limits |
| X69 | Landing page opened. Form gates full case; no personal information submitted. Publication day not stated in the opened record; page observed on the cutoff date. | A downstream field repeated, or could imply, superseded wording. | Omitting the downloadable report leaves input and output units ambiguous. [17, 18] | structural | Corporate case summary; the linked full report describes records matched to a person in the offline graph, not a deduplicated count of customers. See IDN-CV-S01. | Omitting the downloadable report leaves input and output units ambiguous. | evidence/research/C.json, sources.3.reliability_notes |
| X70 | Summary read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Full text gives a document year but no observation date. [18] | structural | 2025 copyright on case PDF; date of insurer assessment undisclosed; opened 2026-11-11. | Full text gives a document year but no observation date. | evidence/research/C.json, findings.3.scope_time |
| X71 | Public summary lacks file size, matching rules, fielded period and clear mapping from assessment to a scored resolution service. | A downstream field repeated, or could imply, superseded wording. | The newly recovered PDF supplies partial method and unit. [18, 19] | structural | The full case provides a record denominator category, person-level output and a field-agreement slice. It gives no absolute file count, measurement period, acceptance threshold covering 99.5%, independently verified identity labels or link to a specific commercial resolution SKU. | The newly recovered PDF supplies partial method and unit. | evidence/research/C.json, findings.3.limitations |
| X72 | Public summary lacks file size, matching rules, fielded period and clear mapping from assessment to a scored resolution service. | A downstream field repeated, or could imply, superseded wording. | Do not repeat superseded teaser-only limitations. [18, 19] | structural | The full case provides a record denominator category, person-level output and a field-agreement slice. It gives no absolute file count, measurement period, acceptance threshold covering 99.5%, independently verified identity labels or link to a specific commercial resolution SKU. | Do not repeat superseded teaser-only limitations. | evidence/research/C.json, findings.3.counterevidence |
| X73 | Recover source PDF or archive for the TransUnion US insurer’s gated Data Health Assessment. A product-specific method and a link to the registered service remain missing. | A downstream field repeated, or could imply, superseded wording. | The full PDF was recovered without submitting a form. [18, 19] | structural | For the insurer assessment, determine the acceptance rules covering all reported links and whether the commercial resolution product performed that test. The full case and a distinct diagnostic catalog listing were verified. | The full PDF was recovered without submitting a form. | evidence/research/C.json, open_questions.2 |
| X74 | Commissioned by TransUnion | A downstream field repeated, or could imply, superseded wording. | The original study discloses selection as well as financial interest. [26] | structural | TransUnion commissioned Forrester and supplied customer names. It did not participate in the interviews. | The original study discloses selection as well as financial interest. | evidence/research/C.json, sources.10.funding_or_interest |
| X75 | actual | A downstream field repeated, or could imply, superseded wording. | Three-year composite model; consistent with the forecast labeling in both research data tables. [26] | structural | forecast | Three-year composite model; consistent with the forecast labeling in both research data tables. | evidence/research/C.json, findings.9.claim_status |
| X76 | Four customer decision-makers interviewed; experiences aggregated to a hypothetical enterprise. No case-level observed numerator, denominator, control or US-only rate published in the cited summary. | A downstream field repeated, or could imply, superseded wording. | The method includes sponsor participation in recruiting. [26] | structural | Forrester interviewed four decision-makers from US-headquartered customers whose names TransUnion supplied. The three-year composite uses modeled rates of 60% and 80%. No tested file is disclosed. | The method includes sponsor participation in recruiting. | evidence/research/C.json, findings.9.study_design |
| X77 | full | A downstream field repeated, or could imply, superseded wording. | Do not describe the live link as accessible at this verification. [414] | structural | Full issuer PDF read through the indexed original. The direct issuer URL returned 404. | Do not describe the live link as accessible at this verification. | evidence/research/C.json, sources.20.access_limits |
| X78 | technical_doc | A downstream field repeated, or could imply, superseded wording. | Promotional product sheet with no test protocol or implementable resolution method; treat as vendor marketing for scoring. [414] | structural | vendor | Promotional product sheet with no test protocol or implementable resolution method; treat as vendor marketing for scoring. | evidence/research/C.json, sources.20.source_type |
| X79 | General technical product sheet on issuer domain. Broader than the grocery transaction matching service. Publication day not stated in the opened record; page observed on the cutoff date. | A downstream field repeated, or could imply, superseded wording. | The specific pass-three description overrides general deterministic marketing language. [12, 414] | structural | General Deep Sync marketing sheet recovered from indexed issuer PDF. It describes identity linkage on shared information; it cannot establish the design of the grocery transaction-matching service. | The specific pass-three description overrides general deterministic marketing language. | evidence/research/C.json, sources.20.reliability_notes |
| X80 | Seller also describes roughly 70% of anonymous transactions, 64–73% of monthly volume and 73.0% cumulative. Verify full-file vs anonymous base, period and person versus household output before using H1. | A downstream field repeated, or could imply, superseded wording. | Retain the directly observable discrepancy as well as the unit issue. [12] | structural | Monthly POS file and anonymous transactions are described separately. The waterfall ends at 73.0%; the summary gives 64–73% and an approximate 32m total, but the three approximate pass counts sum to 31.0m. No calendar month, exact common numerator or person-level accuracy is established. | Retain the directly observable discrepancy as well as the unit issue. | evidence/research/C.json, findings.15.limitations |
| X81 | Customer is unnamed. UID2 is the measured destination, while Deep Sync is the party publishing the case. The page identifies the upload base but does not disclose detailed receiver eligibility rules or confirm match correctness. | A downstream field repeated, or could imply, superseded wording. | The counts should not be used to reconstruct an exact destination denominator. [31] | structural | The advertiser is unnamed. Deep Sync published the labeled UID2 row; recipient eligibility, precise denominator and percent rounding are unstated. Dividing the submitted nominal file into the listed counts gives about 51.7% and 66.6%. No independent test of match correctness. | The counts should not be used to reconstruct an exact destination denominator. | evidence/research/C.json, findings.25.limitations |
| X82 | Seller records expose several distinct quantities: destination match yield, rows recognized in a graph, relative change, and a financial model’s assumed gain. Deep Sync publishes US cases with named inputs, outputs and matching procedures. Its two case pages give different precision or populations for several headline figures. They are priority H1 counterexample candidates pending product admission and source-led verification. The Experian/MiQ case names both IP inputs and cookieless outputs but does not identify a US-only sample. Forrester’s TransUnion study models a composite enterprise. A LiveRamp rename is documented, with retained legacy keys in its adapter. That is firmer evidence of namespace continuity than it is of unchanged graph edges. The study logs lead searches for every buying-task seed and retains inaccessible or unresolved candidates. No full-frame absence or H8 majority is inferred. | A downstream field repeated, or could imply, superseded wording. | The full insurer PDF materially revises the documented H1 candidate set. [18, 19, 21, 26, 185, 184, 12] | structural | The sellers’ disclosures measure destination match yield, records recognized, relative change or modeled gains. Deep Sync publishes US cases but its transaction waterfall does not reconcile exactly. The complete TransUnion insurer report names a record-to-person join and the field agreement for most links. The acceptance rule for all links and commercial product attribution remain open. Experian’s MiQ case identifies input IPs and output IDs without a US-only sample. Forrester models a TransUnion-funded composite. LiveRamp documents a rename and continued use of legacy adapter keys. Neither a market-wide absence nor a majority retaining earlier graph edges follows from the discovery screen. | The full insurer PDF materially revises the documented H1 candidate set. | evidence/research/C.json, summary |
| X83 | [{"source_id": "IDN-C-S04", "passage": "Match 99.5% of its customers to real, verified identities", "locator": "Case summary, first bullet"}] | A downstream field repeated, or could imply, superseded wording. | The case report is the direct basis for the revised wording. The companion catalog distinguishes the diagnostic product from commercial resolution. [17, 18, 19] | structural | [{"source_id": "IDN-C-S04", "passage": "Match 99.5% of its customers to real, verified identities", "locator": "Case summary, first bullet"}, {"source_id": "IDN-CV-S01", "passage": "TransUnion matched 99.5% of records to a person in the TransUnion offline identity graph", "locator": "p. 2, Solution, first column (99.5% and 97.4%)"}, {"source_id": "IDN-CV-S02", "passage": "The DHA app shows the rate at which TransUnion can resolve your data to a unique individual or household", "locator": "Snowflake Marketplace Offerings: TruAudience Identity – Data Health Assessment; following heading: TruAudience Identity Resolution & Enrichment"}] | The case report is the direct basis for the revised wording. The companion catalog distinguishes the diagnostic product from commercial resolution. | evidence/research/C.json, findings.3.support |
| X84 | ["IDN-C-S01", "IDN-C-S02", "IDN-C-S04", "IDN-C-S05", "IDN-C-S07", "IDN-C-S08", "IDN-C-S09", "IDN-C-S10", "IDN-C-S11", "IDN-C-S19", "IDN-C-S20", "IDN-C-S25", "IDN-C-S27", "IDN-C-S30"] | A downstream field repeated, or could imply, superseded wording. | Include the deciding original PDF and the product-scope companion for the updated insurer row. [18, 19] | structural | ["IDN-C-S01", "IDN-C-S02", "IDN-C-S04", "IDN-C-S05", "IDN-C-S07", "IDN-C-S08", "IDN-C-S09", "IDN-C-S10", "IDN-C-S11", "IDN-C-S19", "IDN-C-S20", "IDN-C-S25", "IDN-C-S27", "IDN-C-S30", "IDN-CV-S01", "IDN-CV-S02"] | Include the deciding original PDF and the product-scope companion for the updated insurer row. | evidence/research/C.json, data_tables.0.source_ids |
| X85 | Not reported | A downstream field repeated, or could imply, superseded wording. | C-S17 is canonical for G-S09 and U-S37. The same issuer announcement distinguishes the date of the notice from projected availability. [398] | structural | Announced 2025-04-15. The release projected availability beginning in May 2025, with rolling partner integrations. | C-S17 is canonical for G-S09 and U-S37. The same issuer announcement distinguishes the date of the notice from projected availability. | evidence/research/C.json, sources.16.measurement_period |
| X86 | Publicis reported closing its Epsilon acquisition on July 2, 2019. [Evidence class: vendor_assertion] | Independent re-read tested the original scope and classification. | Publicis announced completion on July 2. Alliance Data’s item 8.01 says the sellers completed the Epsilon sale to Publicis and its subsidiaries on July 1. The buyer’s release alone does not date the underlying close. [319, 320] | contradicted | Alliance Data’s filing dates the sale of Epsilon to Publicis to July 1, 2019. Publicis announced completion on July 2. [Evidence class: direct_record] [Source IDs added: IDN-DV-S01] | The corrected close date is supplied directly by the seller’s 8-K. The July 2 buyer announcement remains a party release. The original is recorded as a July 2 event. The buyer’s announcement only says the acquisition was completed; it does not report July 2 as the close. The seller’s 8-K affirmatively dates the close to July 1. | data/claims-overlay.json, IDN-D-F07; evidence/verify/D.json |
| X87 | TransUnion and Neustar reported a $3.1 billion closing on December 1, 2021, and expressly excluded Neustar Security Services. | Independent re-read tested the original scope and classification. | The parties’ December 1 release states USD 3.1 billion and excludes Neustar Security Services. TransUnion’s separate filed exhibit confirms the December 1 closing and defines the acquired business as marketing, risk and communications. [323, 324] | supported as stated | TransUnion and Neustar reported a $3.1 billion closing on December 1, 2021, and expressly excluded Neustar Security Services. [Source IDs added: IDN-DV-S03] | Checked transaction parties, timing and units against the reopened record. | data/claims-overlay.json, IDN-D-F12; evidence/verify/D.json |
| X88 | IPG’s October 1, 2018 Acxiom transaction excluded LiveRamp. | Independent re-read tested the original scope and classification. | The October 1 release excludes LiveRamp. Item 2.01 of IPG’s October 2 amended 8-K explains that Acxiom distributed its LiveRamp interests to Holdco and IPG acquired all the equity of the entity holding Marketing Solutions. The filed transaction disclosure substantiates the direct-record label. [325, 326] | supported as stated | IPG’s October 1, 2018 Acxiom transaction excluded LiveRamp. [Source IDs added: IDN-DV-S02] | Checked transaction parties, timing and units against the reopened record. | data/claims-overlay.json, IDN-D-F13; evidence/verify/D.json |
| X89 | At the September 1, 2021 closing, Apollo said Yahoo would operate under Apollo Funds and Verizon retained 10%. | Independent re-read tested the original scope and classification. | Apollo’s September 1 release describes the closing by affiliate-managed funds and the retained 10% stake. Verizon’s financial statements confirm September 1 and identify 10% of fully diluted common limited partnership units in the Apollo affiliate. This is the interest at closing. [329, 330] | supported as stated | At the September 1, 2021 closing, Apollo said Yahoo would operate under Apollo Funds and Verizon retained 10%. [Source IDs added: IDN-DV-S04] | Checked transaction parties, timing and units against the reopened record. | data/claims-overlay.json, IDN-D-F15; evidence/verify/D.json |
| X90 | The Trade Desk’s FY2025 filing defines gross spend using client inventory, value-added services, data and the platform fee. | Independent re-read tested the original scope and classification. | Item 7 defines gross spend as amounts clients spend on inventory, value-added services and data on the platform plus the platform fee. This is the client’s spend on inventory. The company’s separate consolidated FY2025 revenue is USD 2,896,284 thousand. Neither measure is UID2 revenue. [331] | supported with qualification | The Trade Desk defines gross spend in its FY2025 filing as clients’ spending on advertising inventory, value-added services and data on its platform, plus its platform fee. | The unit is client spending on inventory, not inventory belonging to clients. | data/claims-overlay.json, IDN-D-F16; evidence/verify/D.json |
| X91 | {"action": "Publicis reported closing", "event_date": "2019-07-02", "period": "2019-07-02", "source_ids": "IDN-D-S07"} | The table row was checked against the reopened records. | The seller’s filed closing date precedes the dated Publicis announcement by one day. [319, 320] | structural | {"action": "Publicis acquired Epsilon. Publicis announced completion on July 2.", "event_date": "2019-07-01", "period": "2019-07-01 to 2019-07-02", "source_ids": "IDN-D-S07;IDN-DV-S01"} | Keep the table consistent with the independently verified disclosure. | evidence/research/D.json, ownership_events, data row 4 |
| X92 | {"action": "ID5 announced acquisition; later confirmed by seller"} | The table row was checked against the reopened records. | ID5 bought the original business. It sold the audience business in a separate transaction. [132, 217] | structural | {"action": "ID5 announced its TrueData acquisition in an October 30 datelined statement. ID5 subsequently confirmed the completed purchase."} | Keep the table consistent with the independently verified disclosure. | evidence/research/D.json, ownership_events, data row 5 |
| X93 | {"source_ids": "IDN-D-S13;IDN-D-S13"} | The table row was checked against the reopened records. | Remove duplicate and include dated transaction filing. [323, 324] | structural | {"source_ids": "IDN-D-S13;IDN-DV-S03"} | Keep the table consistent with the independently verified disclosure. | evidence/research/D.json, ownership_events, data row 8 |
| X94 | {"source_ids": "IDN-D-S14;IDN-D-S14"} | The table row was checked against the reopened records. | Remove duplicate and add the filed acquisition-perimeter record. [325, 326] | structural | {"source_ids": "IDN-D-S14;IDN-DV-S02"} | Keep the table consistent with the independently verified disclosure. | evidence/research/D.json, ownership_events, data row 9 |
| X95 | {"source_ids": "IDN-D-S17"} | The table row was checked against the reopened records. | The seller’s filing specifies the date and denominator of its retained interest. [329, 330] | structural | {"source_ids": "IDN-D-S17;IDN-DV-S04"} | Keep the table consistent with the independently verified disclosure. | evidence/research/D.json, ownership_events, data row 11 |
| X96 | 2019-07-02 | A downstream field repeated, or could imply, superseded wording. | Underlying closing distinguished from July 2 announcement. [319, 320] | structural | 2019-07-01 | Underlying closing distinguished from July 2 announcement. | evidence/research/D.json, sources.6.event_date |
| X97 | 2019-07-02 | A downstream field repeated, or could imply, superseded wording. | Separate the two dates. [319, 320] | structural | 2019-07-01 close; 2019-07-02 announcement | Separate the two dates. | evidence/research/D.json, findings.6.scope_time |
| X98 | Annual report describes board approval and extent of ownership; December closing statement is separate. | A downstream field repeated, or could imply, superseded wording. | The financial statement note supplies completion as well as the board section’s authorized perimeter. [322] | structural | The board approved the 100% acquisition in November 2024 (p. 111). Note 41 independently dates the purchase of all Predictive Pop, Inc. (Audigent) shares to December 4, 2024. | The financial statement note supplies completion as well as the board section’s authorized perimeter. | evidence/research/D.json, sources.11.reliability_notes |
| X99 | paragraph 2 | A downstream field repeated, or could imply, superseded wording. | Both cited phrases occur in the first substantive paragraph of the press release. [325] | structural | Opening paragraph, final sentence (October 1, 2018) | Both cited phrases occur in the first substantive paragraph of the press release. | evidence/research/D.json, findings.12.support.1.locator |
| X100 | Chromium’s July 14 Protected Audience change replaced renderer behavior, including runAdAuction, with rejected promises or default values. | Independent re-read tested the original scope and classification. | The July 14, 2026 Chromium change replaced Protected Audience renderer functions including runAdAuction with rejections or default responses. A later October 30 main-branch commit removed remaining stubbed bindings and runtime features. Neither commit dates completion on all deployed browsers. [268, 5] | supported as stated | Chromium’s July 14 Protected Audience change replaced renderer behavior, including runAdAuction, with rejected promises or default values. [Source IDs added: IDN-EV-S01] | The July finding is historically correct. The October commit advances the repository status without proving a universal Stable rollout. | data/claims-overlay.json, IDN-E-F05; evidence/verify/E.json |
| X101 | In a September project update, the Topics engineers reported that the stable stub-removal trial had reached 1% on milestone 153 and later. | Independent re-read tested the original scope and classification. | The September 24 Topics owner post reports a 1% Stable stub-removal trial on Chrome M153+ and requests further stages, with a target after October 15 for 100% Stable. It separately cites a usage metric. The 1% is trial rollout on the named Stable versions, not API adoption or all web traffic. [270, 267] | supported as stated | In a September project update, the Topics engineers reported that the stable stub-removal trial had reached 1% on milestone 153 and later. [Source IDs added: IDN-E-S05] | One percent is a Stable field-trial share on M153+. The thread has no device count or geographical breakdown. | data/claims-overlay.json, IDN-E-F07; evidence/verify/E.json |
| X102 | In October, the Attribution Reporting team distinguished a window-opening compatibility fix already deployed in Chrome 155 Stable from a proposed later rollout to all Stable users. | Independent re-read tested the original scope and classification. | The October 8 author separates a fix for window.open with attributionsrc, deployed in M155 Stable, from the remaining staged Attribution Reporting rollout. Ten percent Stable was to be held through October 24, with 100% targeted later. The thread records approval, without a completion report. [271, 269] | supported with qualification | On October 8, 2026, the Attribution Reporting team reported that its window.open compatibility fix had reached Chrome 155 Stable. It proposed a staged rollout of the remaining API removal to 100% of Stable after October 24. [Source IDs added: IDN-E-S07] | Three later approvals in the thread cover the proposal. No completion percentage is inferred. | data/claims-overlay.json, IDN-E-F08; evidence/verify/E.json |
| X103 | Google’s mobile-ads SDK guidance dates the Android Privacy Sandbox deprecation to October 17, 2025. | Independent re-read tested the original scope and classification. | The Android Ad Manager SDK page dates deprecation to October 17, 2025; its current footer says November 9, 2026. It still lists APIs and SDK permissions. Deprecation does not date rejection on every Android installation. [272, 220] | supported as stated | Google’s mobile-ads SDK guidance dates the Android Privacy Sandbox deprecation to October 17, 2025. [Source IDs added: IDN-E-S02] | The recorded deprecation event remains October 17, 2025. The current documentation footer reads November 9, 2026, superseding the stored November 4 date. | data/claims-overlay.json, IDN-E-F09; evidence/verify/E.json |
| X104 | Google scoped the IP Protection design to third-party requests to listed domains in Incognito. | Independent re-read tested the original scope and classification. | The design page limits IP masking to eligible third-party requests in Chrome Incognito for domains on the Masked Domain List. The scheme excludes first-party requests even for listed domains and uses two proxies. The October retirement decision covers this technology. [415, 220] | supported as stated | Google scoped the IP Protection design to third-party requests to listed domains in Incognito. [Source IDs added: IDN-E-S02] | The design page limits IP masking to eligible third-party requests in Chrome Incognito for domains on the Masked Domain List. The scheme excludes first-party requests even for listed domains and uses two proxies. The October retirement decision covers this technology. | data/claims-overlay.json, IDN-E-F11; evidence/verify/E.json |
| X105 | In its OS 27 releases, WebKit says Safari and WebKit block network access to identity-resolution infrastructure at domain level. | Independent re-read tested the original scope and classification. | WebKit’s October 19, 2026 post explicitly names iOS, iPadOS, macOS and visionOS 27. It describes domain-level network blocking of third-party identity-resolution infrastructure serving cross-site tracking. It does not supply a full list of affected companies or describe all first-party personalization as blocked. [151, 152] | supported with qualification | On October 19, 2026, WebKit reported that Safari and WebKit on iOS, iPadOS, macOS and visionOS 27 block network requests at the domain level to third-party identity-resolution infrastructure used for cross-site tracking. [Source IDs added: IDN-E-S16] | WebKit’s October 19, 2026 post explicitly names iOS, iPadOS, macOS and visionOS 27. It describes domain-level network blocking of third-party identity-resolution infrastructure serving cross-site tracking. It does not supply a full list of affected companies or describe all first-party personalization as blocked. | data/claims-overlay.json, IDN-E-F15; evidence/verify/E.json |
| X106 | UID2’s publisher guide identifies client-side requests to uidapi.com as blocked on the specified iOS 27 and macOS 27 browser paths. | Independent re-read tested the original scope and classification. | The UID2 Prebid.js client-side guide identifies every browser on iOS 27 or later and Safari on macOS 27 or later. It says requests to uidapi.com, client-side token generation and refresh are affected and gives server-side alternatives. The reopened page’s footer is November 11. WebKit’s own post confirms domain-level blocking but does not name uidapi.com. [152, 151] | supported with qualification | UID2’s Prebid.js guide says that client-side requests to uidapi.com are blocked in all browsers on iOS 27 or later and in Safari on macOS 27 or later. It directs affected publishers to a server-side integration. [Source IDs added: IDN-E-S14] | The live footer says November 11, 2026, not November 5. These are vendor integration instructions, not tests in every affected browser. | data/claims-overlay.json, IDN-E-F16; evidence/verify/E.json |
| X107 | ID5 documents an integration that synchronizes with ad-tech vendors while third-party cookies are available. | Independent re-read tested the original scope and classification. | ID5’s April 24, 2026 instructions describe syncing with partners through a redirect pixel when third-party cookies are available. The page also describes automatic sync with its JavaScript API and a configurable cascade limit. No path-level traffic, policy-induced increase or market share is measured. [221, 179, 180] | supported as stated | ID5 documents an integration that synchronizes with ad-tech vendors while third-party cookies are available. [Source IDs added: IDN-E-S21, IDN-E-S23] | ID5’s April 24, 2026 instructions describe syncing with partners through a redirect pixel when third-party cookies are available. The page also describes automatic sync with its JavaScript API and a configurable cascade limit. No path-level traffic, policy-induced increase or market share is measured. | data/claims-overlay.json, IDN-E-F18; evidence/verify/E.json |
| X108 | Three versioned ID5 publisher SDK guides spanning both 2025 Google announcements give the same maxCascades default of eight cookie syncs and the same disable value of minus one. | Independent re-read tested the original scope and classification. | Each pinned ID5 README defines maxCascades as a ceiling when user sync is required, default 8 and disabled at −1. Repository commit timestamps are April 18, 2025, October 15, 2025 and November 6, 2026. These are the same path at the three boundaries. The parameter does not report how many syncs occur or establish primacy for US Chrome. [179, 223, 180] | supported with qualification | In ID5 publisher SDK guides dated April 18, 2025, October 15, 2025 and November 6, 2026, maxCascades caps cookie syncs at eight by default when user syncing is needed. A value of −1 disables syncing. | The ceiling is conditional. Repository dates do not measure browser traffic or identify the integration as the main US buying route. The research confidence rationale refers to two editions, although there are three. | data/claims-overlay.json, IDN-E-F20; evidence/verify/E.json |
| X109 | Three versions of UID2’s client-side JavaScript guide tell publishers to inspect browser local storage for the generated token. | Independent re-read tested the original scope and classification. | The three JavaScript guide revisions dated March 10 and July 8, 2025 and October 13, 2026 describe inspecting local storage as a token-generation verification step for the client-side workflow. The October version separately directs affected OS 27 browsers to server-side generation. A stored token provides no evidence of third-party-cookie syncing. [224, 225, 226] | supported with qualification | All three UID2 JavaScript guide revisions tell publishers using a supported client-side integration to look for the generated token in local storage. The October 2026 revision directs publishers to generate and refresh tokens server-side in affected OS 27 browsers. | The research rationale refers to two versions, although three were checked. Local storage is not third-party-cookie sync; the newest guide is explicitly unavailable in affected browsers. | data/claims-overlay.json, IDN-E-F21; evidence/verify/E.json |
| X110 | Prebid’s March 2025 and June 2026 Utiq module editions both discuss local-storage consent controls. | Independent re-read tested the original scope and classification. | The March 17, 2025 and June 23, 2026 Utiq documents describe configuring TCF Control around Utiq’s own consent check for local device storage. An additional July 7, 2025 revision, the latest file before the October retirement announcement, has the same language. None of these versions assigns third-party-cookie sync a required role. [416, 417, 418] | supported as stated | Prebid’s March 2025 and June 2026 Utiq module editions both discuss local-storage consent controls. [Source IDs added: IDN-EV-S02] | The extra July 2025 edition dates the guide before the secondary October event. Utiq US eligibility and comparable primary route remain open. | data/claims-overlay.json, IDN-E-F27; evidence/verify/E.json |
| X111 | {"documented_stage": "window.open compatibility fix in Stable; all-Stable rollout target described"} | The table row was checked against the reopened records. | Identify which change was delivered and which was only proposed. [271] | structural | {"documented_stage": "M155 Stable received the window.open fix. The team proposed a later rollout of the remaining API removal to all Stable users."} | Keep the table consistent with the independently verified disclosure. | evidence/research/E.json, browser_os_event_register, data row 11 |
| X112 | {"period": "Design 2025-10-27"} | The table row was checked against the reopened records. | The available date is the page revision, not evidence that the design began or was deployed on October 27. [415] | structural | {"period": "Design page last updated 2025-10-27"} | Keep the table consistent with the independently verified disclosure. | evidence/research/E.json, browser_os_event_register, data row 14 |
| X113 | {"period": "2026-11-05 guide"} | The table row was checked against the reopened records. | The current issuer footer shows November 11. [152] | structural | {"period": "Guide last updated 2026-11-11; read 2026-11-11"} | Keep the table consistent with the independently verified disclosure. | evidence/research/E.json, browser_os_event_register, data row 20 |
| X114 | {"observation": "Publisher inspects local storage"} | The table row was checked against the reopened records. | The final revision expressly restricts its scope. [226] | structural | {"observation": "Publishers on supported client-side paths inspect local storage. The guide directs affected OS 27 browsers to a server-side integration."} | Keep the table consistent with the independently verified disclosure. | evidence/research/E.json, browser_document_pairs, data row 6 |
| X115 | {"period": "Read 2026-11-11", "comparison_use": "Current workflow only; a like-for-like pre-policy copy is unresolved."} | The table row was checked against the reopened records. | The page and relevant cookie-file section contain an issuer date.; A dated page is evidence for one workflow; repeat access does not create a second dated edition. [222] | structural | {"period": "Last modified 2025-10-15; read 2026-11-11", "comparison_use": "Guide last modified before the October 2025 announcement. No comparable pre-April edition or distinct post-October edition was established."} | Keep the table consistent with the independently verified disclosure. | evidence/research/E.json, browser_document_pairs, data row 7 |
| X116 | 2026-11-04 | A downstream field repeated, or could imply, superseded wording. | The Google SDK page footer now gives November 9; this is the documentation revision, not the deprecation event. [272] | structural | [Field cleared; see reason] | The Google SDK page footer now gives November 9; this is the documentation revision, not the deprecation event. | evidence/research/E.json, sources.7.pub_date |
| X117 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | The Google SDK page footer now gives November 9; this is the documentation revision, not the deprecation event. [272] | structural | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. Last updated November 9, 2026. The deprecation date is October 17, 2025. | The Google SDK page footer now gives November 9; this is the documentation revision, not the deprecation event. | evidence/research/E.json, sources.7.reliability_notes |
| X118 | 2026-08-03 | A downstream field repeated, or could imply, superseded wording. | The footer identifies this date as the current page revision. [273] | structural | [Field cleared; see reason] | The footer identifies this date as the current page revision. | evidence/research/E.json, sources.8.pub_date |
| X119 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | The footer identifies this date as the current page revision. [273] | structural | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. Last updated August 3, 2026. Android reference marks the classes deprecated at API level 37. | The footer identifies this date as the current page revision. | evidence/research/E.json, sources.8.reliability_notes |
| X120 | 2025-10-27 | A downstream field repeated, or could imply, superseded wording. | The Google page gives a documentation revision date, not a rollout date. [415] | structural | [Field cleared; see reason] | The Google page gives a documentation revision date, not a rollout date. | evidence/research/E.json, sources.9.pub_date |
| X121 | Design and scope description. The page alone does not establish general release or removal. | A downstream field repeated, or could imply, superseded wording. | The Google page gives a documentation revision date, not a rollout date. [415] | structural | Design and scope description. The page alone does not establish general release or removal. Last updated October 27, 2025. | The Google page gives a documentation revision date, not a rollout date. | evidence/research/E.json, sources.9.reliability_notes |
| X122 | 2026-11-05 | A downstream field repeated, or could imply, superseded wording. | The current issuer HTML explicitly labels this dateModified. [152] | structural | [Field cleared; see reason] | The current issuer HTML explicitly labels this dateModified. | evidence/research/E.json, sources.15.pub_date |
| X123 | This is the issuer’s client-side Prebid guidance. Applies only to the named route and listed browser/OS configurations. | A downstream field repeated, or could imply, superseded wording. | The current issuer HTML explicitly labels this dateModified. [152] | structural | This is the issuer’s client-side Prebid guidance. Applies only to the named route and listed browser/OS configurations. Last updated November 11, 2026. | The current issuer HTML explicitly labels this dateModified. | evidence/research/E.json, sources.15.reliability_notes |
| X124 | 2026-04-24 | A downstream field repeated, or could imply, superseded wording. | The article provides a timestamp but does not identify it as original publication. [221] | structural | [Field cleared; see reason] | The article provides a timestamp but does not identify it as original publication. | evidence/research/E.json, sources.16.pub_date |
| X125 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | The article provides a timestamp but does not identify it as original publication. [221] | structural | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. Article header displays April 24, 2026 at 5:21 a.m. EDT, without specifying the date type. | The article provides a timestamp but does not identify it as original publication. | evidence/research/E.json, sources.16.reliability_notes |
| X126 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | LiveRamp displays a revision date preceding the second of the two Google announcements. Its publication-date field is blank. [222] | structural | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. No publication date established; text as accessed at the cutoff. Main page and Mapping File Types section last modified October 15, 2025. | LiveRamp displays a revision date preceding the second of the two Google announcements. Its publication-date field is blank. | evidence/research/E.json, sources.17.reliability_notes |
| X127 | Chrome 152, 2026-08-25 | A downstream field repeated, or could imply, superseded wording. | The finding compares both milestone release records. [265, 266] | structural | Chrome 144: 2026-01-13; Chrome 152: 2026-08-25 | The finding compares both milestone release records. | evidence/research/E.json, findings.2.scope_time |
| X128 | Guide updated 2026-11-05 | A downstream field repeated, or could imply, superseded wording. | The reopened document footer supersedes the source-freeze value of November 5. [152] | structural | Guide last updated 2026-11-11; accessed 2026-11-11 | The reopened document footer supersedes the source-freeze value of November 5. | evidence/research/E.json, findings.15.scope_time |
| X129 | The two dated versions document the same named setting or workflow; deployment and population remain unobserved. | A downstream field repeated, or could imply, superseded wording. | The source set contains three distinct revisions. [179, 223, 180] | structural | Three dated revisions of the same SDK guide document the same conditional sync ceiling. Deployment, frequency and the main US integration remain unmeasured. | The source set contains three distinct revisions. | evidence/research/E.json, findings.19.confidence_reason |
| X130 | The two dated versions document the same named setting or workflow; deployment and population remain unobserved. | A downstream field repeated, or could imply, superseded wording. | The source set contains three revisions and the newest includes an OS 27 exception. [224, 225, 226] | structural | Three dated revisions retain the same local-storage inspection step. The latest restricts the client-side workflow in affected browsers. | The source set contains three revisions and the newest includes an OS 27 exception. | evidence/research/E.json, findings.20.confidence_reason |
| X131 | Google kept Chrome’s existing cookie choice and subsequently selected named Privacy Sandbox features for withdrawal. Implementation followed different clocks for source-code work, release notes, public API surfaces and Android classes. WebKit and UID2 each documented a specific 2026 client-side constraint. ID5’s versioned publisher SDK retained the same conditional cookie-sync setting in a pair of documents bracketing the April decision. The dated comparisons do not establish cross-product changes in traffic. | A downstream field repeated, or could imply, superseded wording. | The summary now reflects all three revisions while retaining the original scope. [4, 220, 266, 273, 151, 152, 179, 223, 180] | structural | Google retained Chrome’s third-party-cookie choice in April 2025 and selected named Privacy Sandbox technologies for retirement that October. Chrome release notes, Chromium code changes and Android deprecation notices record different stages. WebKit and UID2 identify specific OS 27 browser restrictions. Three versions of ID5’s publisher SDK guide retained the same conditional cookie-sync ceiling. These documents do not measure changes in traffic or identify a shift for the population of US products. | The summary now reflects all three revisions while retaining the original scope. | evidence/research/E.json, summary |
| X132 | For matched email or phone inputs, Google Ads says it adds the corresponding Google Account to the Customer Match segment. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Google’s email and phone subsection describes comparison of customer and account hashes. On a match Google adds the corresponding account to the customer list. The address subsection uses a separate composite key. This is Google’s account-level procedure. [213] | supported as stated | For matched email or phone inputs, Google Ads says it adds the corresponding Google Account to the Customer Match segment. [Evidence class: vendor_assertion] | Google’s product instructions describe its own matching feature. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F01; evidence/verify/F.json |
| X133 | Google says API uploads need at least 100 matched unique users for the match rate to display. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | The visibility rule applies to API uploads and counts unique users who matched. A list with 100 uploaded records need not have 100 distinct matches. The number is a display threshold, with no accuracy or delivery denominator. [300] | supported as stated | Google says API uploads need at least 100 matched unique users for the match rate to display. [Evidence class: vendor_assertion] | The threshold comes from Google’s product documentation. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F02; evidence/verify/F.json |
| X134 | Google defines its match rate against the uploaded data connected to logged-in Google users and reports a 29%–62% range for most advertisers. | Independent re-read tested the original scope and classification. | Google’s upload denominator is the advertiser’s submitted data. The numerator is the part linked to logged-in Google users. The following tip gives a range for most advertisers. The source gives no number of advertisers, period, geography or distribution method for this benchmark. The footnote on global conversion lift elsewhere on the page belongs to a different statistic. [287] | supported with qualification | Google defines Customer Match rate as the share of an upload it can link to logged-in Google users. Google says most advertisers fall between 29% and 62%. It gives no period, geography or advertiser sample for that range. | Google’s upload denominator is the advertiser’s submitted data. The numerator is the part linked to logged-in Google users. The following tip gives a range for most advertisers. The source gives no number of advertisers, period, geography or distribution method for this benchmark. The footnote on global conversion lift elsewhere on the page belongs to a different statistic. | data/claims-overlay.json, IDN-F-F03; evidence/verify/F.json |
| X135 | From April 1, 2026 Google directed most new Customer Match integrations to the Data Manager API while retaining a Google Ads API route for eligible existing integrations. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | The guide, last updated October 30, 2026, states both the April 1 restriction on the two named Google Ads API services and the continuation route for eligible integrations. Eligibility depends on prior Customer Match activity associated with the developer token. Most new integrations use the Data Manager API. [288] | supported as stated | From April 1, 2026 Google directed most new Customer Match integrations to the Data Manager API while retaining a Google Ads API route for eligible existing integrations. [Evidence class: vendor_assertion] | The scope is a vendor developer guide about access to its own APIs. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F04; evidence/verify/F.json |
| X136 | Google’s Authorized Buyers policy lets publishers send obfuscated signals in bid requests to approved third-party bidders. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | The bidder instructions distinguish the signal originator, the publisher that authorizes sharing and the third-party bidder that can receive it. Signals must be obfuscated before reaching Google and should be unreadable by Google. Registration and both parties’ settings matter. Some markets have separate preferences for cookies and secure signals. [289] | supported with qualification | Google’s Authorized Buyers guide says publishers can instruct Ad Manager to send obfuscated signals in bid requests to third-party bidders they have approved. The signal must be registered, and bidders select which signals to receive. [Evidence class: vendor_assertion] | The source is operational guidance for bidders and publishers about Google’s product. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F05; evidence/verify/F.json |
| X137 | Amazon’s advertiser-audience agreement specifies hashed data for matching to Amazon IDs and segment identifiers supplied by the advertiser. | Independent re-read tested the original scope and classification. | Section 1(a) of the November 20, 2024 agreement limits the hashed-data route to matching against Amazon IDs and requires one or more associated segment identifiers. The agreement separately defines other routes, including a data provider. The document establishes contractual purpose and advertiser duties. [290] | supported with qualification | Under the hashed-data route in Amazon’s November 20, 2024 advertiser-audience agreement, advertisers supply hashed data for matching to Amazon IDs and provide one or more associated segment identifiers. | Section 1(a) of the November 20, 2024 agreement limits the hashed-data route to matching against Amazon IDs and requires one or more associated segment identifiers. The agreement separately defines other routes, including a data provider. The document establishes contractual purpose and advertiser duties. | data/claims-overlay.json, IDN-F-F06; evidence/verify/F.json |
| X138 | Amazon Marketing Cloud limits the analyst-facing outputs it describes to aggregated anonymous results while directing custom audiences to ad activation. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Amazon says only aggregated, anonymous AMC output is accessible to the analyst, describes thresholded queries on advertiser and Amazon signals, and separately describes direct activation of AMC audiences for sponsored, video and display advertising. These are Amazon’s statements about its service. [301] | supported as stated | Amazon Marketing Cloud limits the analyst-facing outputs it describes to aggregated anonymous results while directing custom audiences to ad activation. [Evidence class: vendor_assertion] | The feature and output restrictions come from Amazon’s own commercial product page. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. Commercial product page. Preserve its marketing provenance and corresponding grade cap if considered in later stages. | data/claims-overlay.json, IDN-F-F08; evidence/verify/F.json |
| X139 | Meta’s Marketing API lists email addresses and phone numbers among the inputs used to build target Custom Audiences from customer information. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Meta’s developer guide, updated July 14, 2026, explicitly lists email and phone among accepted customer fields for building a target audience. It describes normalizing and hashing contact fields. Source coverage ends at documented inputs and matching rules. [292] | supported as stated | Meta’s Marketing API lists email addresses and phone numbers among the inputs used to build target Custom Audiences from customer information. [Evidence class: vendor_assertion] | This is Meta’s technical documentation of its own audience API. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F10; evidence/verify/F.json |
| X140 | Instacart’s clean-room setup guide uses a hashed-email intersection between two parties as an example of an analysis where complete datasets stay hidden. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Instacart’s guide, dated May 14, 2026 in the page markup, gives a two-company hashed-email intersection as an example for restricted analysis. It distinguishes Overlap Analysis, which needs consumer data linking, from listed templates based only on Instacart data. No realized overlap is reported. [297] | supported as stated | Instacart’s clean-room setup guide uses a hashed-email intersection between two parties as an example of an analysis where complete datasets stay hidden. [Evidence class: vendor_assertion] | This example comes from Instacart’s technical guide. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F15; evidence/verify/F.json |
| X141 | For Data Hub activation queries, Instacart directs results to its own account and says partners cannot see the specific matched identifiers. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Instacart’s activation instructions, dated May 14, 2026 in the page markup, route encrypted audience identifiers from the overlap query to Instacart’s Snowflake account. Instacart manages approved platform activation. The partner cannot inspect the matched IDs or results of the activation query. [298] | supported as stated | For Data Hub activation queries, Instacart directs results to its own account and says partners cannot see the specific matched identifiers. [Evidence class: vendor_assertion] | The source is Instacart’s technical workflow documentation. Treat as first-party vendor disclosure. Reserve direct_record for a formal platform policy or agreement, a standard or another qualifying record. | data/claims-overlay.json, IDN-F-F16; evidence/verify/F.json |
| X142 | {"buyer_output": "Bidder receipt according to publisher settings"} | The table row was checked against the reopened records. | Settings and documented eligibility are supported. No bidder receipt was inspected. [289] | structural | {"buyer_output": "Signal eligible for forwarding to approved bidders that opted to receive it"} | Keep the table consistent with the independently verified disclosure. | evidence/research/F.json, destination_identity_handoffs, data row 2 |
| X143 | {"matching_or_transport": "Intersection and activation handled in clean room"} | The table row was checked against the reopened records. | Instacart distinguishes the clean-room query from its later off-platform activation step. [296, 297, 298] | structural | {"matching_or_transport": "Partner and Instacart tables intersect in the clean room. Instacart handles onward activation on approved external platforms."} | Keep the table consistent with the independently verified disclosure. | evidence/research/F.json, destination_identity_handoffs, data row 9 |
| X144 | Routing rule for most new integrations; the text expressly allows eligible existing integrations to remain on the Google Ads API. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | Google labels October 30 as the guide update. April 1 is the effective date of the restriction. [288] | structural | The API guide’s footer dates the revision to 2026-10-30 UTC. The stated access restriction took effect on 2026-04-01. The guide allows eligible existing integrations to continue. First publication is undated. | Google labels October 30 as the guide update. April 1 is the effective date of the restriction. | evidence/research/F.json, sources.3.reliability_notes |
| X145 | Rule effective 2026-04-01 | A downstream field repeated, or could imply, superseded wording. | Separates rule date, edition and read date. [288] | structural | Rule effective 2026-04-01. Guide revised 2026-10-30 UTC and read 2026-11-11. | Separates rule date, edition and read date. | evidence/research/F.json, findings.3.scope_time |
| X146 | 2024-11-20 | A downstream field repeated, or could imply, superseded wording. | The issuer explicitly labels the date last updated. [290] | structural | [Field cleared; see reason] | The issuer explicitly labels the date last updated. | evidence/research/F.json, sources.5.pub_date |
| X147 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The issuer explicitly labels the date last updated. [290] | structural | 2024-11-20 | The issuer explicitly labels the date last updated. | evidence/research/F.json, sources.5.event_date |
| X148 | US-domain agreement retrieved by direct HTTP GET. The governing agreement distinguishes hashed-data, service-provider, and events-manager routes. It defines an identifier scope, not a realized match rate. | A downstream field repeated, or could imply, superseded wording. | The issuer explicitly labels the date last updated. [290] | structural | Agreement marked last updated 2024-11-20. First publication and a separate effective date are not established. Section 1 covers hashed inputs and associated segment identifiers. Sections 2 and 3 cover data providers and Events Manager. | The issuer explicitly labels the date last updated. | evidence/research/F.json, sources.5.reliability_notes |
| X149 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The page separately marks datePublished and dateModified for this guide. [291] | structural | 2017-06-12 | The page separately marks datePublished and dateModified for this guide. | evidence/research/F.json, sources.6.pub_date |
| X150 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The page separately marks datePublished and dateModified for this guide. [291] | structural | 2026-10-08 | The page separately marks datePublished and dateModified for this guide. | evidence/research/F.json, sources.6.event_date |
| X151 | Advertiser-facing self-service account. No US-specific measured advertiser benchmark is attributed. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | The page separately marks datePublished and dateModified for this guide. [291] | structural | Amazon’s Article metadata gives 2017-06-12 for publication and 2026-10-08 for revision. The instructions describe Amazon’s own advertiser workflow. | The page separately marks datePublished and dateModified for this guide. | evidence/research/F.json, sources.6.reliability_notes |
| X152 | Read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Edition of the advertiser workflow used as evidence. [291] | structural | Guide published 2017-06-12, revised 2026-10-08 and read 2026-11-11. | Edition of the advertiser workflow used as evidence. | evidence/research/F.json, findings.6.scope_time |
| X153 | technical_doc | A downstream field repeated, or could imply, superseded wording. | The commercial AMC product page is a vendor feature disclosure, rather than a formal platform policy. [301] | structural | vendor | The commercial AMC product page is a vendor feature disclosure, rather than a formal platform policy. | evidence/research/F.json, sources.7.source_type |
| X154 | AMC product policy describes audience activation separately from analyst query results. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | The commercial AMC product page is a vendor feature disclosure, rather than a formal platform policy. [301] | structural | Commercial product overview. Amazon describes separate access rules for analyst-facing outputs and activation audiences. It provides no independent observation of either workflow. | The commercial AMC product page is a vendor feature disclosure, rather than a formal platform policy. | evidence/research/F.json, sources.7.reliability_notes |
| X155 | Marketing API / Custom Audiences / Custom Audiences from a Customer List | A downstream field repeated, or could imply, superseded wording. | The live guide supplies an explicit title and revision date. [292] | structural | Customer File Custom Audiences \| Meta for Developers | The live guide supplies an explicit title and revision date. | evidence/research/F.json, sources.9.title |
| X156 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The live guide supplies an explicit title and revision date. [292] | structural | 2026-07-14 | The live guide supplies an explicit title and revision date. | evidence/research/F.json, sources.9.event_date |
| X157 | Public developer manual retrieved directly with HTTP GET. Identifiers listed are accepted inputs for this API, not proxies for person-level truth. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | The live guide supplies an explicit title and revision date. [292] | structural | The technical guide is headed Customer File Custom Audiences and marked updated 2026-07-14. First publication is undated. Its input list documents the interface, without a per-person accuracy test. | The live guide supplies an explicit title and revision date. | evidence/research/F.json, sources.9.reliability_notes |
| X158 | Read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Distinguishes the revision from the cutoff. [292] | structural | Developer guide revised 2026-07-14 and read 2026-11-11. | Distinguishes the revision from the cutoff. | evidence/research/F.json, findings.9.scope_time |
| X159 | technical_doc | A downstream field repeated, or could imply, superseded wording. | A seller announcement supplies the launch claim. Its later web revision does not change the date of the event. [293] | structural | vendor | A seller announcement supplies the launch claim. Its later web revision does not change the date of the event. | evidence/research/F.json, sources.10.source_type |
| X160 | Seller describes launch partners and starting inventory. No assumption of general availability on other partners or access to raw retailer identity. | A downstream field repeated, or could imply, superseded wording. | A seller announcement supplies the launch claim. Its later web revision does not change the date of the event. [293] | structural | Walmart Connect’s announcement is dated 2026-05-28. Publisher metadata marks a page revision on 2026-08-20. The release names the launch partners and starting inventory. Other partnerships were planned. | A seller announcement supplies the launch claim. Its later web revision does not change the date of the event. | evidence/research/F.json, sources.10.reliability_notes |
| X161 | technical_doc | A downstream field repeated, or could imply, superseded wording. | The retailer’s page metadata explicitly distinguishes first publication from modification. [294] | structural | vendor | The retailer’s page metadata explicitly distinguishes first publication from modification. | evidence/research/F.json, sources.11.source_type |
| X162 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The retailer’s page metadata explicitly distinguishes first publication from modification. [294] | structural | 2023-05-03 | The retailer’s page metadata explicitly distinguishes first publication from modification. | evidence/research/F.json, sources.11.pub_date |
| X163 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The retailer’s page metadata explicitly distinguishes first publication from modification. [294] | structural | 2023-06-07 | The retailer’s page metadata explicitly distinguishes first publication from modification. | evidence/research/F.json, sources.11.event_date |
| X164 | Product guide documents use of Target shopper audiences and Google managed advertising, not a CRM ingress interface for the advertiser. No publication date established; text as accessed at the cutoff. | A downstream field repeated, or could imply, superseded wording. | The retailer’s page metadata explicitly distinguishes first publication from modification. [294] | structural | Roundel’s WebPage metadata gives 2023-05-03 for publication and 2023-06-07 for revision. Its commercial product page describes Target audiences and managed Google advertising. | The retailer’s page metadata explicitly distinguishes first publication from modification. | evidence/research/F.json, sources.11.reliability_notes |
| X165 | Read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Reports the edition. The May 2023 verification footnote applies to a separate performance claim. [294] | structural | Product page published 2023-05-03, revised 2023-06-07 and read 2026-11-11. | Reports the edition. The May 2023 verification footnote applies to a separate performance claim. | evidence/research/F.json, findings.11.scope_time |
| X166 | technical_doc | A downstream field repeated, or could imply, superseded wording. | Kroger issued a release reporting a 2023 launch. [295] | structural | vendor | Kroger issued a release reporting a 2023 launch. | evidence/research/F.json, sources.12.source_type |
| X167 | technical_doc | A downstream field repeated, or could imply, superseded wording. | This is the operator’s launch announcement. [296] | structural | vendor | This is the operator’s launch announcement. | evidence/research/F.json, sources.13.source_type |
| X168 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | This is the operator’s launch announcement. [296] | structural | Instacart issued a January 6, 2026 launch release reporting selected pilots and planned expansion. The release supplies a participant report, without independent measurement of rollout. | This is the operator’s launch announcement. | evidence/research/F.json, sources.13.reliability_notes |
| X169 | 2026-05-14 | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [297] | structural | [Field cleared; see reason] | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.14.pub_date |
| X170 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [297] | structural | 2026-05-14 | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.14.event_date |
| X171 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [297] | structural | Page markup labels 2026-05-14 as dateModified, with the visible footer showing Last updated. No first-publication date was established. The technical guide documents the stated procedure. | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.14.reliability_notes |
| X172 | 2026-05-14 | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [298] | structural | [Field cleared; see reason] | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.15.pub_date |
| X173 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [298] | structural | 2026-05-14 | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.15.event_date |
| X174 | Primary issuer documentation. Confirms the stated workflow or policy, not observed performance. | A downstream field repeated, or could imply, superseded wording. | Instacart gives a page revision, not an initial publication date. [298] | structural | Page markup labels 2026-05-14 as dateModified, with the visible footer showing Last updated. No first-publication date was established. The technical guide documents the stated procedure. | Instacart gives a page revision, not an initial publication date. | evidence/research/F.json, sources.15.reliability_notes |
| X175 | Current Meta developer documentation and the effective customer-list terms were accessible via direct issuer HTTP GET, despite the web reader returning an error or login. The help center endpoint 628750084321697 did not expose readable guidance in the direct response. A historical Facebook product description in the 2013 PwC report is a dated corroboration only. | A downstream field repeated, or could imply, superseded wording. | The secondary help route is separated from the sources that actually supported the findings. [302, 292] | structural | The effective Meta terms and developer documentation were readable from the issuer on November 11, 2026. Meta’s separate help article 628750084321697 redirected to login on the verifier’s web reread. The finding relies on the current terms and developer guide. | The secondary help route is separated from the sources that actually supported the findings. | evidence/research/F.json, open_questions.0 |
| X176 | The Walmart DSP anonymous-ID partner portal returned a script shell on web and direct HTTP access. Searches of Walmart Connect, the operator’s other first-party data articles, the issuer PDF search and an Internet Archive request did not establish an advertiser CRM upload contract on this route. The archive returned HTTP 429. The explicit retailer-to-DSP route is reported separately. | A downstream field repeated, or could imply, superseded wording. | Keeps the remaining portal access limit separate from the accessible retailer launch statement. [293] | structural | The separate Walmart DSP partner portal returned no readable lines on the November 11 verifier web reread. Stage 2 also checked the issuer site, PDF search and archive. The May 28 launch release documents Walmart audiences sent to a DSP. The portal result leaves advertiser CRM import through that route unresolved. | Keeps the remaining portal access limit separate from the accessible retailer launch statement. | evidence/research/F.json, open_questions.1 |
| X177 | Amazon’s Ads Data Manager guide and help-center route returned script shells. The US Advertiser Audience Agreement was recovered from Amazon itself and governs the limited-purpose matching claim. The Amazon operator article and AMC product overview provide separate activation and output records. | A downstream field repeated, or could imply, superseded wording. | Updates the access boundary without extending any one source to a separate workflow. [290, 291, 301] | structural | On November 11 the verifier’s web reads of Amazon’s Ads Data Manager guide and help page GKSQZKMHW95GJDB7 exposed no substantive guidance. Amazon’s advertiser-audience agreement, self-service guide and AMC overview were readable and cover their respective workflows. | Updates the access boundary without extending any one source to a separate workflow. | evidence/research/F.json, open_questions.2 |
| X178 | Activation templates / How activation works, paragraph after step 3 | A downstream field repeated, or could imply, superseded wording. | The reopened source gives this section or named function as the passage locator. [298] | structural | Activation templates / How it works, paragraph following the numbered workflow | The reopened source gives this section or named function as the passage locator. | evidence/research/F.json, findings.15.support.0.locator |
| X179 | Viant says its household assignment screens IP addresses by residential ISP, device count and time of association before linking observations to a registered address. | Independent re-read tested the original scope and classification. | Viant’s July 10, 2026 page says it filters IPs by residential ISP, device count and the times of day associated devices appear. It then uses location coordinates from geolocation-enabled devices to match to a registered physical address. [304] | supported with qualification | Viant says it screens IP addresses by residential ISP, device count and time of day, then uses coordinates from location-enabled devices to match them to a registered home address. | Time of day is an input feature, not time elapsed since the link was seen. Viant calls the result deterministic, but this marketing description does not independently validate assignment precision. | data/claims-overlay.json, IDN-G-F04; evidence/verify/G.json |
| X180 | Blockgraph and Cadent described an integration through which Aperture customers could activate TV audiences within participating distributors’ addressable household footprint. | Independent re-read tested the original scope and classification. | On April 27, 2022, Blockgraph announced that its agreement with Cadent would allow Aperture customers to match and activate audiences using aggregated, de-identified data within participating distributors’ household footprints. [419] | supported with qualification | In April 2022, Blockgraph and Cadent announced that Aperture customers would be able to match and activate TV audiences within participating distributors’ addressable household footprints. | The issuer uses prospective language for this integration. The announcement is not a separate production observation and does not establish the scope of a distributor operating today. | data/claims-overlay.json, IDN-G-F06; evidence/verify/G.json |
| X181 | FreeWheel describes its embedded Identity Network as using Comcast signals, publisher partner IDs and the buyer’s first-party data. | Independent re-read tested the original scope and classification. | FreeWheel describes embedded access to Comcast’s deterministic signals and publishers’ proprietary partner IDs. It also offers buyers the option of connecting their own first-party data. [305] | supported with qualification | FreeWheel says its embedded Identity Network provides access to Comcast’s deterministic signals and publisher partner IDs. Buyers can also connect their own first-party data. | The buyer input is optional. The page does not show that every customer provides it or that every partner uses the same assignment method. | data/claims-overlay.json, IDN-G-F08; evidence/verify/G.json |
| X182 | OpenAP describes Open Identity as an evolution of OpenID that enables direct matching across parties without a central identity spine. | Independent re-read tested the original scope and classification. | The April 15, 2025 release describes Open Identity as a federated evolution of OpenID that enables direct matches without a central spine. It identifies providers at launch and others undergoing integration, and says the technology would be live beginning in May. [398] | supported with qualification | In April 2025, OpenAP described Open Identity as a successor to OpenID for direct matching among participating parties without a central identity spine. It announced availability beginning in May 2025. | The source’s April event is the announcement, not proof of April deployment. The publisher and advertiser choose participating identity providers and householding business logic. | data/claims-overlay.json, IDN-G-F09; evidence/verify/G.json |
| X183 | Factori describes its cross-device offering as linking encrypted mobile ad IDs and IP addresses with daily server-to-server feed delivery. | Independent re-read tested the original scope and classification. | Factori’s graph documentation ties encrypted mobile ad IDs and IP addresses to its cross-device data and calls the S2S delivery daily. A separate seller AWS listing offers a distinct subscription to the dataset and describes this same cadence. [308, 309] | supported as stated | Factori describes its cross-device offering as linking encrypted mobile ad IDs and IP addresses with daily server-to-server feed delivery. [Source IDs added: IDN-GV-S04] | Both descriptions are vendor-origin. The AWS listing identifies a purchasable cross-device dataset with global coverage claimed, without a US-specific contract or validated household-link expiry. Daily delivery, event capture and historic file availability are different clocks. | data/claims-overlay.json, IDN-G-F12; evidence/verify/G.json |
| X184 | LtvAdx documents an IP-based path for joining devices to a household and a 30-day lifetime after last observation for a device–household link. [Evidence class: synthesis] | Independent re-read tested the original scope and classification. | The API first recognizes a previously linked device ID, then tries a previously seen IP as a probabilistic match. Its retention guide gives 30 days since the device's last sighting for a device link and 90 for a household profile. The US site and pricing plan bundle HouseholdID targeting with campaign buying. [227, 228, 229, 230, 231] | supported with qualification | LtvAdx documents a shared-IP fallback for assigning a device to a household and removes a device link 30 days after the device was last seen. It offers HouseholdID targeting in its US advertising workflow. [Evidence class: vendor_assertion] [Source IDs added: IDN-GV-S01, IDN-GV-S02, IDN-GV-S03] | The operator itself directly describes both rules, and its own commercial pages describe the US route. Combining these pages does not supply an independent measurement or require a cross-source synthesis class. Both technical documents come from the same operator. The US exchange buying route is documented. A separately licensable resolution product, actual enforcement, and independent calibration of the API’s 0.75 confidence code remain unestablished. | data/claims-overlay.json, IDN-G-F13; evidence/verify/G.json |
| X185 | In January 2020, Skydeo described a US mobile-data integration with the then-named Tru Optik Household Graph. | Independent re-read tested the original scope and classification. | The January 23, 2020 release announces an agreement whereby Skydeo’s mobile data would be mapped to the existing Tru Optik Household Graph. Its US device and US home totals are separate seller assertions. [310] | supported with qualification | In January 2020, Skydeo announced an agreement to map its mobile data to the then-named Tru Optik Household Graph in the United States. | A planned mapping at announcement is not proof it was deployed, that the historic graph is licensed separately today, or that its claimed footprint was validated. | data/claims-overlay.json, IDN-G-F15; evidence/verify/G.json |
| X186 | Viant’s household-identification patent gives 45 days as an example minimum IP assignment age for further processing. | Independent re-read tested the original scope and classification. | Specification step 508 and Figure 5 describe checking whether assignment age exceeds a configurable threshold. Forty-five days is an example for deciding whether an IP address is stable enough for further processing. [311] | supported with qualification | Viant’s US patent gives 45 days as an example threshold that an IP assignment’s age must exceed to qualify for further processing. | The comparison uses greater than, not greater than or equal to. This is a patent embodiment, not an expiry after the last sighting or evidence that a live product uses 45 days. | data/claims-overlay.json, IDN-G-F17; evidence/verify/G.json |
| X187 | {"documented_service": "Household postal address; GraphPort maps third-party IDs into OS; US distributor route via Cadent."} | The table row was checked against the reopened records. | The 2022 release is prospective and does not verify 2026 use. [166, 419, 420] | structural | {"documented_service": "Postal-address household anchor; GraphPort maps third-party IDs into IDoS. The 2022 Cadent agreement described planned activation within participating distributors’ footprints."} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 2 |
| X188 | {"period": "Sources as read 2026-11-11; individual dates in source register"} | The table row was checked against the reopened records. | Distinguish the announcement from the advertised availability window. [398] | structural | {"period": "2025-04-15 announcement; availability planned beginning 2025-05, with rolling partner integrations. Read 2026-11-11."} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 4 |
| X189 | {"documented_service": "Comcast, publisher and customer IDs connected within host platform", "ip_role": "Mapping stage/inputs vary by partner"} | The table row was checked against the reopened records. | Buyer data is an option.; Varying partner-specific IP roles are not documented on the cited page. [305] | structural | {"documented_service": "Embedded access to Comcast deterministic signals and publisher partner IDs, with optional buyer first-party data integration.", "ip_role": "IP-to-household mapping is unspecified in the FreeWheel network description."} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 5 |
| X190 | {"documented_clock": "Six-month observation lookback; daily/weekly input files recommended"} | The table row was checked against the reopened records. | The study table should retain the upper-bound and distinguish initial and periodic deliveries. [307] | structural | {"documented_clock": "Up to six months of input observations; at least two months of recent records recommended initially, followed by daily or weekly files."} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 7 |
| X191 | {"scope_or_unresolved": "US commercial route not established by API guide; keep provisional.", "source_ids": "IDN-G-S12"} | The table row was checked against the reopened records. | Seller marketplace listing resolves the separate purchasing route, while leaving US-specific eligibility open.; Add the named purchasable dataset. [308, 309] | structural | {"scope_or_unresolved": "Factori Technologies Pte. Ltd. lists a distinct CTV/Cross Device Data subscription on AWS Data Exchange. The global listing does not specify US coverage or an edge-expiry policy.", "source_ids": "IDN-G-S12; IDN-GV-S04"} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 8 |
| X192 | {"scope_or_unresolved": "Joint documentary case. Establish US service availability and distinguish independently purchased capability from exchange feature. No network testing conducted.", "source_ids": "IDN-G-S13; IDN-G-S14"} | The table row was checked against the reopened records. | US platform availability is established; separate CTV graph eligibility and enforcement remain unverified.; Add US commercial offer, platform plan and product boundary. [227, 228, 229, 230, 231] | structural | {"scope_or_unresolved": "LtvAdx offers HouseholdID targeting to US advertisers within its impression-based exchange plan. The reviewed product, pricing and US offer do not identify a separately licensable household-resolution service. No network testing conducted.", "source_ids": "IDN-G-S13; IDN-G-S14; IDN-GV-S01; IDN-GV-S02; IDN-GV-S03"} | Keep the table consistent with the independently verified disclosure. | evidence/research/G.json, ctv_household_method_screen, data row 9 |
| X193 | A TV workflow can match first-party records against a general graph, import another operator’s graph, or resolve a household itself. The extracted records identify which of these functions the named products describe. Refreshing a data file, waiting for an attribution event and expiring a device-to-household link use different clocks. One discovery lead documents both an IP-based match and the elapsed-time rule for the link; US availability and separate-service eligibility still require confirmation. | A downstream field repeated, or could imply, superseded wording. | US advertiser route confirmed on operator offer and pricing pages; deployment and standalone licensing remain distinct. [227, 228, 229, 230, 231] | structural | CTV workflows may construct household assignments, import a partner graph or use an existing graph to match audiences. LtvAdx publishes a shared-IP fallback and a 30-day device-link inactivity limit. Its US exchange offers HouseholdID targeting within advertising plans. A separate resolution licence remains unconfirmed. File refresh, link expiry and post-view attribution use different clocks. | US advertiser route confirmed on operator offer and pricing pages; deployment and standalone licensing remain distinct. | evidence/research/G.json, summary |
| X194 | For H5, distinguish a positive documentary edge-and-clock pair from service eligibility. The named LtvAdx API and device-link retention page meet the documentary pair for this route. Its policy references possible US processing, not evidence of a named US customer or a separately available US household graph. No eligibility or majority verdict is drawn. | A downstream field repeated, or could imply, superseded wording. | The US offer and plan support commercial availability for this bundled use case. [227, 228, 229, 230, 231] | structural | For H5, LtvAdx documents both an IP assignment path and 30-day device-link expiry. Its US exchange offers HouseholdID within commercial campaign plans. Whether buyers can purchase the household-resolution service as a distinct buying task remains open; the records do not identify a standalone licence or a confirmed eligible cohort row. | The US offer and plan support commercial availability for this bundled use case. | evidence/research/G.json, open_questions.1 |
| X195 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Viant page displays July 10, 2026. [304] | structural | 2026-07-10 | Viant page displays July 10, 2026. | evidence/research/G.json, sources.3.pub_date |
| X196 | 2025-04-15 product launch | A downstream field repeated, or could imply, superseded wording. | The release distinguishes announcement from planned availability. [398] | structural | 2025-04-15 announcement; stated availability beginning May 2025. | The release distinguishes announcement from planned availability. | evidence/research/G.json, sources.8.measurement_period |
| X197 | [field absent] | A downstream field repeated, or could imply, superseded wording. | Guide footer supplies the update date. [307] | structural | Last modified May 25, 2021; reopened 2026-11-11. | Guide footer supplies the update date. | evidence/research/G.json, sources.10.version |
| X198 | 2020-01 | A downstream field repeated, or could imply, superseded wording. | Partner page displays January 23, 2020. [310] | structural | 2020-01-23 | Partner page displays January 23, 2020. | evidence/research/G.json, sources.15.pub_date |
| X199 | 2022 | A downstream field repeated, or could imply, superseded wording. | The TransUnion release gives June 7, 2022. [245] | structural | 2022-06-07 | The TransUnion release gives June 7, 2022. | evidence/research/G.json, sources.16.pub_date |
| X200 | Documentation read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Publisher supplies date. [304] | structural | Viant product description dated 2026-07-10; read 2026-11-11. | Publisher supplies date. | evidence/research/G.json, findings.3.scope_time |
| X201 | 2025-04-15 product launch | A downstream field repeated, or could imply, superseded wording. | Issuer uses prospective language for go-live. [398] | structural | 2025-04-15 announcement; service availability projected beginning May 2025. | Issuer uses prospective language for go-live. | evidence/research/G.json, findings.8.scope_time |
| X202 | Multiple countries; US contract coverage unconfirmed here | A downstream field repeated, or could imply, superseded wording. | Global dataset purchase is documented separately from tested US geography. [308, 309] | structural | Global seller listing; the cited Factori documents do not identify a US-specific licence or participating US households. | Global dataset purchase is documented separately from tested US geography. | evidence/research/G.json, findings.11.scope_geo |
| X203 | New candidate. US buying route and a timed rule for the actual linkage need further evidence. | A downstream field repeated, or could imply, superseded wording. | The seller marketplace page establishes a dataset purchase route. [308, 309] | structural | Factori lists a distinct CTV/Cross Device Data subscription on AWS Data Exchange. US-specific coverage and a timed rule governing the household link remain unconfirmed in these two records. | The seller marketplace page establishes a dataset purchase route. | evidence/research/G.json, findings.11.limitations |
| X204 | New candidate. US buying route and a timed rule for the actual linkage need further evidence. | A downstream field repeated, or could imply, superseded wording. | The seller marketplace page establishes a dataset purchase route. [308, 309] | structural | Factori lists a distinct CTV/Cross Device Data subscription on AWS Data Exchange. US-specific coverage and a timed rule governing the household link remain unconfirmed in these two records. | The seller marketplace page establishes a dataset purchase route. | evidence/research/G.json, findings.11.counterevidence |
| X205 | Product route; US sales unconfirmed | A downstream field repeated, or could imply, superseded wording. | Read the operator’s dedicated US offer alongside published pricing plans. [227, 228, 229, 230, 231] | structural | US advertiser platform offering documented; separate household-resolution licence unconfirmed. | Read the operator’s dedicated US offer alongside published pricing plans. | evidence/research/G.json, findings.12.scope_geo |
| X206 | Both records are from the same operator. Together they describe a rule, not observed enforcement. Hold the product outside the confirmed US cohort until a US commercial route and separate service eligibility are established. | A downstream field repeated, or could imply, superseded wording. | Distinguish the documented US commercial bundle from a separate product and independent enforcement. [227, 228, 229, 230, 231] | structural | The API and retention rule come from the operator and have no independent enforcement test here. Its US exchange advertiser plan includes HouseholdID targeting; the sources do not establish a separate household-resolution purchase. | Distinguish the documented US commercial bundle from a separate product and independent enforcement. | evidence/research/G.json, findings.12.limitations |
| X207 | Both records are from the same operator. Together they describe a rule, not observed enforcement. Hold the product outside the confirmed US cohort until a US commercial route and separate service eligibility are established. | A downstream field repeated, or could imply, superseded wording. | Distinguish the documented US commercial bundle from a separate product and independent enforcement. [227, 228, 229, 230, 231] | structural | The API and retention rule come from the operator and have no independent enforcement test here. Its US exchange advertiser plan includes HouseholdID targeting; the sources do not establish a separate household-resolution purchase. | Distinguish the documented US commercial bundle from a separate product and independent enforcement. | evidence/research/G.json, findings.12.counterevidence |
| X208 | ["IDN-G-S13", "IDN-G-S14"] | A downstream field repeated, or could imply, superseded wording. | The verified wording also cites the US buying route and buyer-facing product boundary. [227, 228, 229, 230, 231] | structural | ["IDN-G-S13", "IDN-G-S14", "IDN-GV-S01", "IDN-GV-S02", "IDN-GV-S03"] | The verified wording also cites the US buying route and buyer-facing product boundary. | evidence/research/G.json, findings.12.source_ids |
| X209 | [{"source_id": "IDN-G-S13", "passage": "two devices are joined to one household only through a shared device id or IP", "locator": "Identity API: matching order, text following the priority table"}, {"source_id": "IDN-G-S14", "passage": "30 days after the device was last seen.", "locator": "What is kept, and for how long: A device linked to a household"}] | A downstream field repeated, or could imply, superseded wording. | Adds direct issuer passages for US availability and plan/product scope. [227, 228, 229, 230, 231] | structural | [{"source_id": "IDN-G-S13", "passage": "two devices are joined to one household only through a shared device id or IP", "locator": "Identity API: matching order, text following the priority table"}, {"source_id": "IDN-G-S14", "passage": "30 days after the device was last seen.", "locator": "What is kept, and for how long: A device linked to a household"}, {"source_id": "IDN-GV-S01", "passage": "Reach US households on the big screen", "locator": "For Advertisers, US market offer"}, {"source_id": "IDN-GV-S02", "passage": "HouseholdID & ACR segment targeting", "locator": "Advertiser plan, included capabilities"}, {"source_id": "IDN-GV-S03", "passage": "Buyer-facing segment IDs, not raw graphs", "locator": "How it works on your stack, Stage 01"}] | Adds direct issuer passages for US availability and plan/product scope. | evidence/research/G.json, findings.12.support |
| X210 | Identifies TV-specific input and delivery path and recommended ingestion window. File delivery and input lookback differ from link expiry. | A downstream field repeated, or could imply, superseded wording. | The dated guide is canonical for the equivalent U-S58 source. The kit retains reliability_notes but drops version. [307] | structural | Identifies TV-specific inputs and the recommended file-delivery window. The guide was last modified on May 25, 2021. The original publication date is not reported. File delivery and input lookback do not establish link expiry. | The dated guide is canonical for the equivalent U-S58 source. The kit retains reliability_notes but drops version. | evidence/research/G.json, sources.10.reliability_notes |
| X211 | Includes example device and IP fields, event time and feed frequency. A per-edge expiry is a separate evidence request. | A downstream field repeated, or could imply, superseded wording. | The G record is canonical for U-S59. The body of the exact implementation guide dates this revision. [308] | structural | The document body says it was last updated in September 2026. Original publication date is unknown. Example device and IP fields, event time and feed frequency do not establish individual link expiry. | The G record is canonical for U-S59. The body of the exact implementation guide dates this revision. | evidence/research/G.json, sources.11.reliability_notes |
| X212 | Mobilewalla reported to investors in August 2026 that its US operations applied a third-party sensitive-location filter to incoming signals and outgoing deliveries after its FTC order. | Independent re-read tested the original scope and classification. | The August 12 S-4 describes, as Mobilewalla’s own US implementation after the order, a third-party sensitive-location filter for all inbound signals and outbound deliveries. The October 16 amended filing repeats this description and six-month effectiveness checks. Both are the same business’s statements. [72, 73] | supported as stated | Mobilewalla reported to investors in August 2026 that its US operations applied a third-party sensitive-location filter to incoming signals and outgoing deliveries after its FTC order. [Source IDs added: IDN-HV-S02] | Passages, locators, date, named party and scope checked against the controlling record. | data/claims-overlay.json, IDN-H-F10; evidence/verify/H.json |
| X213 | California’s February 24, 2026 Disney judgment requires a logged-in consumer’s opt-out to apply across streaming services associated with the same Disney account. | Independent re-read tested the original scope and classification. | The signed page 10 is dated February 11, 2026, and bears Judge Daniel M. Crowley’s signature. Paragraph 26(a) applies a logged-in opt-out across associated Disney streaming services. Paragraph 26(b) separately treats consumers who stay logged out or provide no additional information. The AG announced this settlement on February 11 as well. [74, 75] | contradicted | California’s February 11, 2026 judgment requires Disney DTC and ABC Enterprises to honor a logged-in consumer’s opt-out across the Disney streaming services associated with that account. [Source IDs added: IDN-H-S12] | The original February 24 date is contradicted by the judge’s signed date. | data/claims-overlay.json, IDN-H-F11; evidence/verify/H.json |
| X214 | The California attorney general announced the Disney opt-out settlement on February 11, 2026. | Independent re-read tested the original scope and classification. | The California Attorney General’s February 11, 2026 release describes a settlement with Disney and USD 2.75 million in civil penalties. It describes gaps uncovered in the investigation. The Los Angeles court judgment independently supplies the binding terms. [75, 74] | supported as stated | The California attorney general announced the Disney opt-out settlement on February 11, 2026. [Source IDs added: IDN-H-S11] | Passages, locators, date, named party and scope checked against the controlling record. | data/claims-overlay.json, IDN-H-F12; evidence/verify/H.json |
| X215 | Texas’s December 2025 petition against Sony cited Samba Interactive TV when describing the opt-out path on BRAVIA televisions. | Independent re-read tested the original scope and classification. | The original Texas pleading dated December 15, 2025 names Sony Corporation of America and Sony Electronics, Inc. Figure 5 and paras 57–63 refer to Samba Interactive TV on BRAVIA models. The caption does not name Samba as a defendant and the petition does not decide liability. The AG still described the distinct Sony case as ongoing on May 11, 2026. [76, 77] | supported as stated | Texas’s December 2025 petition against Sony cited Samba Interactive TV when describing the opt-out path on BRAVIA televisions. [Source IDs added: IDN-HV-S01] | Passages, locators, date, named party and scope checked against the controlling record. | data/claims-overlay.json, IDN-H-F13; evidence/verify/H.json |
| X216 | Texas’s April 2026 agreed judgment against LG requires affirmative express consent for covered viewing-data use after its 180-day transition. | Independent re-read tested the original scope and classification. | In Bell County cause 25DCV358507, the state’s and LG’s posted agreed text makes covered viewing-data practices subject to affirmative express consent beginning 180 days after judicial signature and entry. The date and judge’s signature are both blank on page 9. The Texas AG confirms the settlement on May 11, 2026. The posted evidence does not establish whether or when a judge signed a different copy. [78, 77] | supported with qualification | On May 11, 2026, Texas announced a settlement with LG Electronics U.S.A. The agreed terms call for affirmative express consent for covered viewing-data practices 180 days after judicial signature and entry. The posted copy leaves the judge’s signature and date blank. [Source IDs added: IDN-HV-S01] | A settlement is confirmed. The April judgment date and onset of the 180-day period are unverified. | data/claims-overlay.json, IDN-H-F14; evidence/verify/H.json |
| X217 | In September 2026 the CNIL closed the Google injunction after confirming a disclosure change to the account-creation flow. | Independent re-read tested the original scope and classification. | The CNIL’s October 14 announcement reports its September 23, 2026 decision closing the injunction after verification of changed account-creation information and a change in the Gmail ad format. The closure concerns that order only; the panel expressly reserved broader Article 82 assessment. [80, 79] | supported with qualification | On September 23, 2026, the CNIL closed its injunction after Google changed its French account-creation disclosure and the format of Gmail ads. The closure applied to those two requirements. [Source IDs added: IDN-H-S15] | The closure of the whole injunction rested on both limbs. | data/claims-overlay.json, IDN-H-F16; evidence/verify/H.json |
| X218 | {"effective": "2026-02-24 signed judgment", "record_dates": "2026-02-11 announcement; 2026-02-24 judge-dated"} | The table row was checked against the reopened records. | The signed final page and the California AG dateline both say February 11. [74, 75] | structural | {"effective": "2026-02-11 judge-dated; the order takes effect upon entry", "record_dates": "2026-02-11 AG announcement and judge-dated judgment"} | Keep the table consistent with the independently verified disclosure. | evidence/research/H.json, authority_action_and_implementation, data row 5 |
| X219 | {"instrument": "Texas v LG Electronics U.S.A., 25DCV358507", "requires": "Paragraph 16 requires affirmative express consent for covered viewing-data practices from day 180.", "effective": "2026-04-17 judge-dated; paragraph 16 sets its own implementation window", "kind": "Agreed final judgment", "source_ids": "IDN-H-S14", "implementation_record": "Published judgment sets two different deadlines, including the 30-day update provision in paragraph 17. No completed product-specific rollout evidenced by the instrument.", "response_in_record": "Agreed judgment.", "record_dates": "2026-04-17"} | The table row was checked against the reopened records. | AG confirmed a settlement. Pages 1 and 9 of the posted instrument identify Bell County and show an empty judicial signature/date block. No entry date is established. [78, 77] | structural | {"instrument": "Texas v LG Electronics U.S.A., Inc., Bell County, 169th District Court, 25DCV358507", "requires": "The agreed terms call for affirmative express consent for covered viewing-data practices beginning 180 days after judicial signature and entry.", "effective": "2026-05-11 settlement announced. Judicial signature and entry date unverified.", "kind": "AG-announced settlement and published agreed judgment terms", "source_ids": "IDN-H-S14;IDN-HV-S01", "implementation_record": "The agreement specifies a 180-day consent requirement and distinct 30-day updates. Neither court entry nor a product rollout is shown in the posted copy.", "response_in_record": "LG agreed to settlement terms; the Texas AG announced the agreement.", "record_dates": "2026-05-11 AG announcement; judicial date unverified"} | Keep the table consistent with the independently verified disclosure. | evidence/research/H.json, authority_action_and_implementation, data row 7 |
| X220 | {"implementation_record": "In the Aug 12, 2026 S-4 the company attributed its US screening to the order. No qualifying independent audit or filed sworn operational certification identified in the logged search.", "source_ids": "IDN-H-S09;IDN-H-S10", "record_dates": "2025-01-13; 2026-08-12"} | The table row was checked against the reopened records. | The additional issuer filing repeats the US implementation statement and does not independently certify it. [71, 72, 73] | structural | {"implementation_record": "The August 12 S-4 and October 16 S-4/A describe the same US screening program. Both are issuer accounts. Neither is independent testing or a sworn compliance submission to the FTC.", "source_ids": "IDN-H-S09;IDN-H-S10;IDN-HV-S02", "record_dates": "2025-01-13 FTC order; 2026-08-12 S-4; 2026-10-16 S-4/A"} | Keep the table consistent with the independently verified disclosure. | evidence/research/H.json, authority_action_and_implementation, data row 4 |
| X221 | {"scope": "Sony is the respondent; Samba Interactive TV is named in the pleading. Samba household graph IDN-P030 is an author interest.", "implementation_record": "A complaint is not evidence of a product change. Search for a later Texas Sony disposition did not settle the record.", "source_ids": "IDN-H-S13"} | The table row was checked against the reopened records. | The caption names two Sony defendants. Samba is not named as a defendant.; The LG announcement provides a distinct dated status for the Sony action.; Add intervening AG status report; no later conclusion assumed. [76, 77] | structural | {"scope": "Sony Corporation of America and Sony Electronics, Inc., were named defendants in the Nueces County petition. The pleading mentions Samba Interactive TV. The Samba household graph, IDN-P030, is an author interest.", "implementation_record": "The petition describes allegations. Texas reported the Sony case ongoing on May 11, 2026. The reviewed records do not establish disposition by the cutoff or a product change.", "source_ids": "IDN-H-S13;IDN-HV-S01"} | Keep the table consistent with the independently verified disclosure. | evidence/research/H.json, authority_action_and_implementation, data row 6 |
| X222 | {"implementation_record": "The regulator described the added disclosure following the companies’ response. Scope of closure is expressly limited."} | The table row was checked against the reopened records. | The closure decision assessed both specified changes. [79, 80] | structural | {"implementation_record": "The regulator describes both the new account-creation disclosure and the revised format of Gmail ads. Its closure is limited to the two limbs of the injunction."} | Keep the table consistent with the independently verified disclosure. | evidence/research/H.json, authority_action_and_implementation, data row 8 |
| X223 | 2023-06-15 | A downstream field repeated, or could imply, superseded wording. | The Légifrance masthead dates publication June 22; June 15 was the decision. [63] | structural | 2023-06-22 | The Légifrance masthead dates publication June 22; June 15 was the decision. | evidence/research/H.json, sources.0.pub_date |
| X224 | 2026-02-24 | A downstream field repeated, or could imply, superseded wording. | Court signature and dated registered copy. [74, 75] | structural | 2026-02-11 | Court signature and dated registered copy. | evidence/research/H.json, sources.10.event_date |
| X225 | 2026-02-24 | A downstream field repeated, or could imply, superseded wording. | The order is dated February 11; the PDF does not separately timestamp online posting. [74, 75] | structural | 2026-02 | The order is dated February 11; the PDF does not separately timestamp online posting. | evidence/research/H.json, sources.10.pub_date |
| X226 | Judge date stamp inspected on last page. The form footer retained the word proposed. AG announcement was on February 11. | A downstream field repeated, or could imply, superseded wording. | Visual inspection of signed page 10. [74, 75] | structural | The proposed-form footer remained on the signed copy. The final page bears Judge Daniel M. Crowley’s signature dated February 11, 2026. The AG separately announced the settlement that day. | Visual inspection of signed page 10. | evidence/research/H.json, sources.10.reliability_notes |
| X227 | 2026-02-24; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | The signed page supplies the date. [74, 75] | structural | 2026-02-11; read 2026-11-11 | The signed page supplies the date. | evidence/research/H.json, findings.10.scope_time |
| X228 | Paragraph 26(a), printed p. 6; judge’s date stamp p. 10 | A downstream field repeated, or could imply, superseded wording. | The date is entered in the signature block. [74, 75] | structural | Paragraph 26(a), printed p. 6; signed judicial date on p. 10 | The date is entered in the signature block. | evidence/research/H.json, findings.10.support.0.locator |
| X229 | District Court, Fort Bend County, Texas | A downstream field repeated, or could imply, superseded wording. | The caption specifically identifies Bell County and the 169th Judicial District; Fort Bend is incorrect. [78, 77] | structural | State of Texas and LG Electronics U.S.A., Inc.; agreed filing for the 169th District Court, Bell County, Texas | The caption specifically identifies Bell County and the 169th Judicial District; Fort Bend is incorrect. | evidence/research/H.json, sources.13.org_or_author |
| X230 | State of Texas v LG Electronics U.S.A., agreed final judgment and permanent injunction, 25DCV358507 | A downstream field repeated, or could imply, superseded wording. | Court/date lines blank on page 9. [78, 77] | structural | State of Texas v LG Electronics U.S.A., Inc., agreed final judgment and permanent injunction, posted copy without judicial signature, 25DCV358507 | Court/date lines blank on page 9. | evidence/research/H.json, sources.13.title |
| X231 | 2026-04-17 | A downstream field repeated, or could imply, superseded wording. | The AG linked the text in its May 11 announcement, but exact first publication of the PDF is not attested. [78, 77] | structural | 2026 | The AG linked the text in its May 11 announcement, but exact first publication of the PDF is not attested. | evidence/research/H.json, sources.13.pub_date |
| X232 | 2026-04-17 | A downstream field repeated, or could imply, superseded wording. | The date of judicial signature and entry is unknown. The announcement date is recorded against the separate AG release. [78, 77] | structural | [Field cleared; see reason] | The date of judicial signature and entry is unknown. The announcement date is recorded against the separate AG release. | evidence/research/H.json, sources.13.event_date |
| X233 | Date verified from judicial signature on printed page 8. Announcement followed on May 11. | A downstream field repeated, or could imply, superseded wording. | The copy cannot corroborate an April 17 signature or start the 180-day clock. [78, 77] | structural | Agreed terms for Bell County, 169th Judicial District. The judge’s signature and date lines on p. 9 are blank. Counsel approvals follow. Texas separately announced the settlement on May 11, 2026. | The copy cannot corroborate an April 17 signature or start the 180-day clock. | evidence/research/H.json, sources.13.reliability_notes |
| X234 | 2026-04-17; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | The separately dated AG announcement establishes the settlement. [78, 77] | structural | 2026-05-11 AG announcement; judicial effective date unverified; read 2026-11-11 | The separately dated AG announcement establishes the settlement. | evidence/research/H.json, findings.13.scope_time |
| X235 | Paragraph 16 and subparagraph (a), printed p. 5; judge’s signature printed p. 8 | A downstream field repeated, or could imply, superseded wording. | Page 8 does not contain a judicial signature. [78, 77] | structural | Paragraph 16, printed p. 5; effective-date definition, p. 2; blank judicial signature/date lines, p. 9 | Page 8 does not contain a judicial signature. | evidence/research/H.json, findings.13.support.0.locator |
| X236 | Date is April 17 on the signed judgment. This procurement boundary concerns viewing-data collection; it does not establish an external household identity product or an implemented graph change. | A downstream field repeated, or could imply, superseded wording. | Removes the unsupported claim of an April 17 judicial signature. [78, 77] | structural | Public copies confirm a settlement and its agreed terms. The judge’s signature/date lines in the posted text are blank. A start date, implemented graph change and applicability to an external household identity service remain unverified. | Removes the unsupported claim of an April 17 judicial signature. | evidence/research/H.json, findings.13.limitations |
| X237 | Date is April 17 on the signed judgment. This procurement boundary concerns viewing-data collection; it does not establish an external household identity product or an implemented graph change. | A downstream field repeated, or could imply, superseded wording. | Removes the unsupported claim of an April 17 judicial signature. [78, 77] | structural | Public copies confirm a settlement and its agreed terms. The judge’s signature/date lines in the posted text are blank. A start date, implemented graph change and applicability to an external household identity service remain unverified. | Removes the unsupported claim of an April 17 judicial signature. | evidence/research/H.json, findings.13.counterevidence |
| X238 | Criteo’s French appeal, Experian’s UK tribunal proceeding, and US federal and state records show how easily a product claim can outrun an order. Criteo’s partner audits were already taking place before its fine. A US filing gives a concrete account of Mobilewalla’s post-order filters, but the filing is the company’s own account. California’s signed Disney judgment carries a later date than its announcement and a distinct rule for logged-in accounts. The French regulator separately confirmed a change to Google’s account-creation notice, a platform boundary case. The Texas Sony complaint identifies Samba Interactive TV in an opt-out exhibit. It does not adjudicate Samba’s practices. The register reports each action’s stage, named party, obligation, response and evidence of implementation. Legal findings stay outside the IQ. | A downstream field repeated, or could imply, superseded wording. | The announcement and judicial signature are dated the same day. [74, 75] | structural | Criteo’s French appeal, Experian’s UK tribunal proceeding, and US federal and state records show how easily a product claim can outrun an order. Criteo’s partner audits were already taking place before its fine. A US filing gives a concrete account of Mobilewalla’s post-order filters, but the filing is the company’s own account. California’s signed Disney judgment and the state announcement both date to February 11, 2026. The judgment sets a distinct rule for logged-in accounts. The French regulator separately confirmed a change to Google’s account-creation notice, a platform boundary case. The Texas Sony complaint identifies Samba Interactive TV in an opt-out exhibit. It does not adjudicate Samba’s practices. The register reports each action’s stage, named party, obligation, response and evidence of implementation. Legal findings stay outside the IQ. | The announcement and judicial signature are dated the same day. | evidence/research/H.json, summary |
| X239 | Stage 3 should seek a filed compliance certification, regulator follow-up or independent assessment for Mobilewalla and establish a sold in-scope service to which that evidence actually applies. The October S-4/A full text exceeded the retrieval limit; its EDGAR index was opened and no issuer PDF counterpart was located in the logged query. The archival route also failed. The August S-4 text and filing index were opened. | A downstream field repeated, or could imply, superseded wording. | Later evidence resolves the feasibility of opening the amendment without supplying independent evidence of execution. [71, 72, 73] | structural | The August 12 S-4 and October 16 S-4/A each set out Mobilewalla’s US screening measures. The amendment was retrieved from EDGAR during verification. Neither provides independent attestation or a separate sworn compliance submission, and attribution to a specific sold identity service remains open. | Later evidence resolves the feasibility of opening the amendment without supplying independent evidence of execution. | evidence/research/H.json, open_questions.0 |
| X240 | The announcement date differs from the February 24 date on the signed judgment. Allegations in this announcement remain attributed. | A downstream field repeated, or could imply, superseded wording. | The court’s visual signature and the release dateline agree. [74, 75] | structural | The California AG announced the settlement on February 11, 2026, the same day the judge signed the order. The AG account describes allegations rather than adjudicated violations. | The court’s visual signature and the release dateline agree. | evidence/research/H.json, findings.11.limitations |
| X241 | The announcement date differs from the February 24 date on the signed judgment. Allegations in this announcement remain attributed. | A downstream field repeated, or could imply, superseded wording. | The court’s visual signature and the release dateline agree. [74, 75] | structural | The California AG announced the settlement on February 11, 2026, the same day the judge signed the order. The AG account describes allegations rather than adjudicated violations. | The court’s visual signature and the release dateline agree. | evidence/research/H.json, findings.11.counterevidence |
| X242 | CIMM’s buyer guide asks how many matched records can actually be addressed on the buyer’s primary activation platforms. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Printed page 17 of CIMM’s April 2026 guide separates supplied and matched records from the matched records that remain addressable on primary activation platforms. It frames this as a buyer’s evaluation question. No deployment or observed survival rate is reported. [259] | supported with qualification | CIMM’s April 2026 buyer guide recommends checking how many matched records can be addressed on the buyer’s primary activation platforms. [Evidence class: synthesis] | CIMM describes mixed-method guidance drawn from interviews and references. Its guide is a source of original buyer advice, not a formal standard, regulator record or independent product test. | data/claims-overlay.json, IDN-I-F01; evidence/verify/I.json |
| X243 | CIMM’s buyer guide calls for a truth set when evaluating matching accuracy. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | On printed page 20, CIMM advises building a reference truth set and gives precision, recall and accuracy concepts. The paper cites industry material and interviews with more than 40 contributors to date, some more than once. It supplies no validation result for a named product. [259] | supported with qualification | CIMM’s April 2026 buyer guide recommends establishing a truth set to evaluate match accuracy and distinguishes precision from recall. [Evidence class: synthesis] | CIMM describes mixed-method guidance drawn from interviews and references. Its guide is a source of original buyer advice, not a formal standard, regulator record or independent product test. | data/claims-overlay.json, IDN-I-F02; evidence/verify/I.json |
| X244 | In April 2021 Mediavine said its Grow.me toolkit had active integrations with LiveRamp ATS, UID2 and ID5 Universal ID. | Independent re-read tested the original scope and classification. | Mediavine specifically says Grow.me is integrated with ID5 Universal ID. The next sentence attributes three active integrations, including ATS and UID2, to Mediavine. The announcement does not expressly assign all three integrations to Grow.me or confirm use by every publisher. [280] | supported with qualification | In April 2021 Mediavine said Grow.me was integrated with ID5 Universal ID. It also reported that Mediavine had active integrations with LiveRamp ATS, UID2 and ID5. | Mediavine specifically says Grow.me is integrated with ID5 Universal ID. The next sentence attributes three active integrations, including ATS and UID2, to Mediavine. The announcement does not expressly assign all three integrations to Grow.me or confirm use by every publisher. | data/claims-overlay.json, IDN-I-F11; evidence/verify/I.json |
| X245 | Roku announced that its inventory would be available for encrypted identity matching through Google Display & Video 360 as a Confidential Publisher Match launch partner. | Independent re-read tested the original scope and classification. | Roku’s 23 March 2026 article reports it joined Google’s deterministic launch and describes encrypted matching between advertisers’ first-party data and Roku Media inventory via DV360. Availability is framed prospectively for the launch partnership. No numeric gain, completed match or availability across all inventory is established. [299] | supported with qualification | On 23 March 2026 Roku announced it was a launch partner for Google’s Confidential Publisher Match and described a route for DV360 advertisers to match first-party data to Roku Media inventory. | Roku’s 23 March 2026 article reports it joined Google’s deterministic launch and describes encrypted matching between advertisers’ first-party data and Roku Media inventory via DV360. Availability is framed prospectively for the launch partnership. No numeric gain, completed match or availability across all inventory is established. | data/claims-overlay.json, IDN-I-F16; evidence/verify/I.json |
| X246 | {"what_a_buyer_can_establish": "Ask how many IDs survive delivery to the platform and assess quality against truth labels."} | The table row was checked against the reopened records. | CIMM’s denominator is matched records, not necessarily distinct IDs, and it recommends an evaluation rather than reporting test findings. [259] | structural | {"what_a_buyer_can_establish": "Ask how many matched records remain addressable on primary platforms and evaluate matching against truth labels."} | Keep the table consistent with the independently verified disclosure. | evidence/research/I.json, buyer_operator_evidence, data row 1 |
| X247 | {"product_or_route": "Grow.me, ATS, UID2, ID5 ID"} | The table row was checked against the reopened records. | The press release names the toolkit for ID5 and the company for its three integrations. [280] | structural | {"product_or_route": "Grow.me integrates with ID5; Mediavine names ATS, UID2 and ID5 among its active integrations."} | Keep the table consistent with the independently verified disclosure. | evidence/research/I.json, buyer_operator_evidence, data row 9 |
| X248 | {"period": "As read 2026-11-11"} | The table row was checked against the reopened records. | The publisher supplies a dated policy header. [281] | structural | {"period": "Policy updated 2026-07-14; read 2026-11-11"} | Keep the table consistent with the independently verified disclosure. | evidence/research/I.json, buyer_operator_evidence, data row 12 |
| X249 | {"period": "As read 2026-11-11"} | The table row was checked against the reopened records. | The issuer supplies a dated list header. [421] | structural | {"period": "Participant list updated 2023-12-01; read 2026-11-11"} | Keep the table consistent with the independently verified disclosure. | evidence/research/I.json, buyer_operator_evidence, data row 13 |
| X250 | {"what_a_buyer_can_establish": "Second, separately named buying path on Roku inventory."} | The table row was checked against the reopened records. | Scope the route to Roku Media and attribute it to the launch announcement. [299] | structural | {"what_a_buyer_can_establish": "Roku announced encrypted first-party matching on Roku Media through DV360 as a launch partner."} | Keep the table consistent with the independently verified disclosure. | evidence/research/I.json, buyer_operator_evidence, data row 14 |
| X251 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | This is the date the issuer labels last updated. [281] | structural | 2026-07-14 | This is the date the issuer labels last updated. | evidence/research/I.json, sources.11.event_date |
| X252 | First-party record. Interpret only within the named product, party and period. Publication day not stated in the opened record; page observed on the cutoff date. | A downstream field repeated, or could imply, superseded wording. | Version was verified on the issuer’s page. [281] | structural | First-party privacy policy, revised 2026-07-14; dates and scopes refer to the issuer’s disclosure. No processing or opt-out outcome was independently observed. | Version was verified on the issuer’s page. | evidence/research/I.json, sources.11.reliability_notes |
| X253 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | This is the date the issuer labels last updated. [421] | structural | 2023-12-01 | This is the date the issuer labels last updated. | evidence/research/I.json, sources.12.event_date |
| X254 | First-party record. Interpret only within the named product, party and period. Publication day not stated in the opened record; page observed on the cutoff date. | A downstream field repeated, or could imply, superseded wording. | Version was verified on the issuer’s page. [421] | structural | First-party participant list, revised 2023-12-01; dates and scopes refer to the issuer’s disclosure. No processing or opt-out outcome was independently observed. | Version was verified on the issuer’s page. | evidence/research/I.json, sources.12.reliability_notes |
| X255 | Page read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Dated issuer header. [281] | structural | Privacy policy updated 2026-07-14; read 2026-11-11. | Dated issuer header. | evidence/research/I.json, findings.13.scope_time |
| X256 | Page read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Dated issuer header. [421] | structural | Participant list updated 2023-12-01; read 2026-11-11. | Dated issuer header. | evidence/research/I.json, findings.14.scope_time |
| X257 | AdCP’s 3.2.8 Trusted Match specification defines a publisher-to-buyer identity request carrying opaque tokens while prohibiting page context on that path. | Independent re-read tested the original scope and classification. | The IdentityMatchRequest section in the 3.2.8 documentation specifies the publisher as caller, routing via the TMP router to buyer agents, one or more opaque identity tokens, and no page context. The page marks Trusted Match experimental. Its consent object includes regulated-jurisdiction handling. This is schema-level evidence. [91] | supported with qualification | The experimental Trusted Match specification in AdCP 3.2.8 defines an identity request from a publisher via a router to buyer agents. The request carries one or more opaque identity tokens and must omit page context. | The IdentityMatchRequest section in the 3.2.8 documentation specifies the publisher as caller, routing via the TMP router to buyer agents, one or more opaque identity tokens, and no page context. The page marks Trusted Match experimental. Its consent object includes regulated-jurisdiction handling. This is schema-level evidence. | data/claims-overlay.json, IDN-J-F01; evidence/verify/J.json |
| X258 | The September 11 AdCP Go publisher client builds a typed request from a user token and its namespace type and posts the request to its Trusted Match identity route. | Independent re-read tested the original scope and classification. | At the September 11, 2026 commit, Activate constructs identities with UserToken and UIDType, carries consent, validates, and invokes IdentityMatch; the latter posts to the configured router at /tmp/identity. Activate joins the context and identity results locally. The same commit’s test file describes a local mock server with a typed test token and an expected eligible package. [92, 93, 94] | supported as stated | The September 11 AdCP Go publisher client builds a typed request from a user token and its namespace type and posts the request to its Trusted Match identity route. [Source IDs added: IDN-JV-S01, IDN-JV-S02] | The commit record dates the source snapshot to 2026-09-11. The commit date is distinct from the date the verifier read the 3.2.8 documentation. | data/claims-overlay.json, IDN-J-F03; evidence/verify/J.json |
| X259 | The AdCP reference identity engine builds lookups for request identities and target audience IDs and submits them to the audience service’s membership check. | Independent re-read tested the original scope and classification. | The pinned engine takes the Cartesian product of identities and target audience IDs and calls IsMemberBatch, then uses returned memberships to evaluate package eligibility. The function returns before any lookup when the audience service, identities or target segments are missing. This is source-code evidence of an implemented identity check. [95, 93] | supported with qualification | At the September 11, 2026 adcp-go commit, the reference identity engine submits request tokens and target audience IDs for a batch membership check when an audience service, identities and target segments are present. [Source IDs added: IDN-JV-S01] | The commit record dates the source snapshot to 2026-09-11. The commit date is distinct from the date the verifier read the 3.2.8 documentation. | data/claims-overlay.json, IDN-J-F04; evidence/verify/J.json |
| X260 | The AdCP reference HTTP identity handler can pass a decoded request to its service for evaluation and constructs a response carrying eligible package IDs. | Independent re-read tested the original scope and classification. | The handler parses the HTTP request, validates it and sends the prepared request to Evaluate or EvaluateWithDecode, depending on canonicalizer configuration. On the normal path it filters eligible packages into EligiblePackageIDs. Terminal timeout or provider failure yields a typed error. The code does not establish a successful external exchange. [108, 93] | supported as stated | The AdCP reference HTTP identity handler can pass a decoded request to its service for evaluation and constructs a response carrying eligible package IDs. [Source IDs added: IDN-JV-S01] | The commit record dates the source snapshot to 2026-09-11. The commit date is distinct from the date the verifier read the 3.2.8 documentation. The executable error path takes precedence over a stale file-header comment on timeouts. | data/claims-overlay.json, IDN-J-F05; evidence/verify/J.json |
| X261 | The September 11 adcp-go code includes UID2 and EUID decoders that call the configured client to decrypt an incoming advertising token. | Independent re-read tested the original scope and classification. | The pinned UID2 and EUID Decode methods both delegate to uid2Lookup, which checks for a configured client and calls Decrypt on the token. The implemented switch maps invalid, expired, missing-key, unsupported-version and scope-mismatch errors to a dropped entry. Other errors propagate. The file contains no evidence of a live operator transaction. [96, 93] | supported as stated | The September 11 adcp-go code includes UID2 and EUID decoders that call the configured client to decrypt an incoming advertising token. [Source IDs added: IDN-JV-S01] | The commit record dates the source snapshot to 2026-09-11. The commit date is distinct from the date the verifier read the 3.2.8 documentation. In the switch, five explicit error conditions drop entries. The file’s lead comment is insufficient evidence of operator opt-out processing. | data/claims-overlay.json, IDN-J-F06; evidence/verify/J.json |
| X262 | The Prebid Agentic Audiences RTD implementation copies an input entry’s type and vector into the outbound OpenRTB user-data segment, according to its merged code and module documentation. | Independent re-read tested the original scope and classification. | The merged module copies vector and type into ext.aa. When qualifying stored entries are present it merges the segment list into ortb2Fragments.global.user.data. The PR includes isolated tests using stubbed storage. The observed code boundary is construction of the shared auction fragment; actual buyer delivery and consumption need separate evidence. [103, 105, 104] | supported with qualification | Prebid’s merged Agentic Audiences module copies a stored entry’s type and vector into the global OpenRTB auction fragment under user.data[].segment[].ext.aa. The code and module guide establish this handoff within Prebid. [Source IDs added: IDN-J-S11] | The merged module copies vector and type into ext.aa. When qualifying stored entries are present it merges the segment list into ortb2Fragments.global.user.data. The PR includes isolated tests using stubbed storage. The observed code boundary is construction of the shared auction fragment; actual buyer delivery and consumption need separate evidence. | data/claims-overlay.json, IDN-J-F10; evidence/verify/J.json |
| X263 | {"status_stated": "Specified in AdCP 3.2.8", "observable": "Specified in AdCP 3.2.8"} | The table row was checked against the reopened records. | The versioned issuer document explicitly marks TMP experimental.; Records the router and the observed level of evidence. [91] | structural | {"status_stated": "Experimental Trusted Match schema in AdCP 3.2.8", "observable": "Specified request from publisher through router to buyer agents"} | Keep the table consistent with the independently verified disclosure. | evidence/research/J.json, agent_identity_maturity, data row 1 |
| X264 | {"note": "OpenRTB route is not inherently agent-mediated."} | The table row was checked against the reopened records. | The diff establishes fragment construction. The merge date is visible on the PR header, and receipt is a separate observation. [103] | structural | {"note": "Prebid records a merge into master on April 14, 2026. The module constructs the auction fragment. A buyer receipt and agent mediation remain unverified."} | Keep the table consistent with the independently verified disclosure. | evidence/research/J.json, agent_identity_maturity, data row 8 |
| X265 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The versioned page gives dateModified. Initial publication remains undated. [91] | structural | 2026-10-29 | The versioned page gives dateModified. Initial publication remains undated. | evidence/research/J.json, sources.0.event_date |
| X266 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The versioned page gives dateModified. Initial publication remains undated. [91] | structural | The page’s JSON-LD marks 2026-10-29 as the revision of the versioned 3.2.8 document. The Trusted Match surface is experimental. The date does not establish publication of the entire protocol release. | The versioned page gives dateModified. Initial publication remains undated. | evidence/research/J.json, sources.0.reliability_notes |
| X267 | AdCP 3.2.8 documentation | A downstream field repeated, or could imply, superseded wording. | States which edition supplied the schema. [91] | structural | AdCP 3.2.8 Trusted Match documentation revised 2026-10-29 and read 2026-11-11. | States which edition supplied the schema. | evidence/research/J.json, findings.0.scope_time |
| X268 | 2026-09-11 | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [92, 93] | structural | [Field cleared; see reason] | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.2.pub_date |
| X269 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [92, 93] | structural | 2026-09-11 | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.2.event_date |
| X270 | Full file read from the GitHub contents API at immutable SHA a6c741b2b7387412bf72e2992c31b3402acf8986; commit metadata dated 2026-09-11. | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [92, 93] | structural | Immutable adcp-go snapshot a6c741b2b7387412bf72e2992c31b3402acf8986. GitHub records both authoring and commit at 2026-09-11T14:56:32+02:00, equivalent to 12:56:32 UTC and 08:56:32 in New York on September 11. The event date is the commit date, independent of when the file was first introduced. Full pinned source read on raw.githubusercontent.com. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.2.reliability_notes |
| X271 | 2026-09-11 | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [108, 93] | structural | [Field cleared; see reason] | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.3.pub_date |
| X272 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [108, 93] | structural | 2026-09-11 | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.3.event_date |
| X273 | Full file read from the GitHub contents API at immutable SHA a6c741b2b7387412bf72e2992c31b3402acf8986; commit metadata dated 2026-09-11. | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [108, 93] | structural | Immutable adcp-go snapshot a6c741b2b7387412bf72e2992c31b3402acf8986. GitHub records both authoring and commit at 2026-09-11T14:56:32+02:00, equivalent to 12:56:32 UTC and 08:56:32 in New York on September 11. The event date is the commit date, independent of when the file was first introduced. Full pinned source read on raw.githubusercontent.com. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.3.reliability_notes |
| X274 | 2026-09-11 | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [95, 93] | structural | [Field cleared; see reason] | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.4.pub_date |
| X275 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [95, 93] | structural | 2026-09-11 | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.4.event_date |
| X276 | Full file read from the GitHub contents API at immutable SHA a6c741b2b7387412bf72e2992c31b3402acf8986; commit metadata dated 2026-09-11. | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [95, 93] | structural | Immutable adcp-go snapshot a6c741b2b7387412bf72e2992c31b3402acf8986. GitHub records both authoring and commit at 2026-09-11T14:56:32+02:00, equivalent to 12:56:32 UTC and 08:56:32 in New York on September 11. The event date is the commit date, independent of when the file was first introduced. Full pinned source read on raw.githubusercontent.com. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.4.reliability_notes |
| X277 | 2026-09-11 | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [96, 93] | structural | [Field cleared; see reason] | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.5.pub_date |
| X278 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [96, 93] | structural | 2026-09-11 | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.5.event_date |
| X279 | Full file read from the GitHub contents API at immutable SHA a6c741b2b7387412bf72e2992c31b3402acf8986; commit metadata dated 2026-09-11. | A downstream field repeated, or could imply, superseded wording. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. [96, 93] | structural | Immutable adcp-go snapshot a6c741b2b7387412bf72e2992c31b3402acf8986. GitHub records both authoring and commit at 2026-09-11T14:56:32+02:00, equivalent to 12:56:32 UTC and 08:56:32 in New York on September 11. The event date is the commit date, independent of when the file was first introduced. Full pinned source read on raw.githubusercontent.com. | The code is pinned at a verified commit. The timestamp dates a repository revision and does not date the first publication of the individual file. | evidence/research/J.json, sources.5.reliability_notes |
| X280 | 2026-04-14 | A downstream field repeated, or could imply, superseded wording. | The date in the original publication slot described the merge event. [103, 104] | structural | [Field cleared; see reason] | The date in the original publication slot described the merge event. | evidence/research/J.json, sources.9.pub_date |
| X281 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The date in the original publication slot described the merge event. [103, 104] | structural | 2026-04-14 | The date in the original publication slot described the merge event. | evidence/research/J.json, sources.9.event_date |
| X282 | Files changed in merged public PR. Tests and code opened. No independent execution or third-party traffic measured here. | A downstream field repeated, or could imply, superseded wording. | The date in the original publication slot described the merge event. [103, 104] | structural | The final PR diff shows source and tests for the change merged 2026-04-14. The PR was created 2026-03-22 UTC, but that does not establish when these final file versions first appeared. | The date in the original publication slot described the merge event. | evidence/research/J.json, sources.9.reliability_notes |
| X283 | 2026-04-14 | A downstream field repeated, or could imply, superseded wording. | The PR record distinguishes its UTC creation from the subsequent merge. [104] | structural | 2026-03-22 | The PR record distinguishes its UTC creation from the subsequent merge. | evidence/research/J.json, sources.10.pub_date |
| X284 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The PR record distinguishes its UTC creation from the subsequent merge. [104] | structural | 2026-04-14 | The PR record distinguishes its UTC creation from the subsequent merge. | evidence/research/J.json, sources.10.event_date |
| X285 | Repository merge date from PR; development start and release-tag dates are separate. | A downstream field repeated, or could imply, superseded wording. | The PR record distinguishes its UTC creation from the subsequent merge. [104] | structural | GitHub’s PR metadata gives creation at 2026-03-22T00:10:37Z. The merge event into prebid:master occurred on 2026-04-14. The dates do not identify a commercial deployment or a package release. | The PR record distinguishes its UTC creation from the subsequent merge. | evidence/research/J.json, sources.10.reliability_notes |
| X286 | PubMatic and Butler/Till launch AdCP-enabled agentic AI campaign across CTV | A downstream field repeated, or could imply, superseded wording. | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. [106] | structural | PubMatic and Butler/Till Launch AdCP-Enabled Agentic AI Campaign Across CTV | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. | evidence/research/J.json, sources.12.title |
| X287 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. [106] | structural | 2025-12-15 | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. | evidence/research/J.json, sources.12.event_date |
| X288 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. [106] | structural | The release and its issuer-hosted PDF display December 15, 2025. The IR webpage’s machine metadata carries a later date of January 7, 2026 for that webpage. The finding dates the original press announcement. | The issuer’s dated release and PDF agree on the announcement date. Later webpage metadata is distinct. | evidence/research/J.json, sources.12.reliability_notes |
| X289 | Optable and PubMatic partner on audience discovery and activation | A downstream field repeated, or could imply, superseded wording. | The issuer page supplies the full release title and March 12 dateline. [107] | structural | Optable and PubMatic Partner to Advance Agentic Audience Discovery and Real-Time Activation Across PubMatic AgenticOS | The issuer page supplies the full release title and March 12 dateline. | evidence/research/J.json, sources.13.title |
| X290 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The issuer page supplies the full release title and March 12 dateline. [107] | structural | 2026-03-12 | The issuer page supplies the full release title and March 12 dateline. | evidence/research/J.json, sources.13.event_date |
| X291 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. [99] | structural | 2026-02-26 | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. | evidence/research/J.json, sources.14.pub_date |
| X292 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. [99] | structural | 2026-10-27 | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. | evidence/research/J.json, sources.14.event_date |
| X293 | Issuer demonstration description, with no independent transaction receipt. | A downstream field repeated, or could imply, superseded wording. | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. [99] | structural | IAB lists February 26, 2026 as datePublished and October 27 as dateModified in WebPage metadata. Its visible last-updated date also reads October 27. The dates identify the overview and its revision. The precise demo date is not established here. | Dates in the issuer metadata apply to the overview page, not to the recorded SDK demo. | evidence/research/J.json, sources.14.reliability_notes |
| X294 | As read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | The overview’s version date is distinct from when the demo was performed. [99] | structural | AAMP overview revised 2026-10-27 and read 2026-11-11. Demonstration date unspecified. | The overview’s version date is distinct from when the demo was performed. | evidence/research/J.json, findings.15.scope_time |
| X295 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The article metadata marks separate original publication and revision dates. [101] | structural | 2026-08-04 | The article metadata marks separate original publication and revision dates. | evidence/research/J.json, sources.15.pub_date |
| X296 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The article metadata marks separate original publication and revision dates. [101] | structural | 2026-08-08 | The article metadata marks separate original publication and revision dates. | evidence/research/J.json, sources.15.event_date |
| X297 | Issuer description of non-agentic applicability; do not infer deployment of the agent protocol. | A downstream field repeated, or could imply, superseded wording. | The article metadata marks separate original publication and revision dates. [101] | structural | The article’s datePublished is 2026-08-04 and its dateModified is 2026-08-08 UTC. Its section on OpenRTB and Prebid describes a programmatic use outside agent-mediated buying. | The article metadata marks separate original publication and revision dates. | evidence/research/J.json, sources.15.reliability_notes |
| X298 | Issuer post read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Locates the issuer’s use-case distinction in the dated edition. [101] | structural | Published 2026-08-04, revised 2026-08-08 and read 2026-11-11. | Locates the issuer’s use-case distinction in the dated edition. | evidence/research/J.json, findings.11.scope_time |
| X299 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Page publication and modification dates must not be used as the release dates of the individual protocols. [102] | structural | 2026-01-07 | Page publication and modification dates must not be used as the release dates of the individual protocols. | evidence/research/J.json, sources.16.pub_date |
| X300 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Page publication and modification dates must not be used as the release dates of the individual protocols. [102] | structural | 2026-09-23 | Page publication and modification dates must not be used as the release dates of the individual protocols. | evidence/research/J.json, sources.16.event_date |
| X301 | Issuer record for documented scope. No independent implementation or performance test. | A downstream field repeated, or could imply, superseded wording. | Page publication and modification dates must not be used as the release dates of the individual protocols. [102] | structural | The IAB listing’s WebPage metadata dates publication to January 7, 2026 and revision to September 23, 2026. Its initiative rows separately give August and September dates for the named releases. | Page publication and modification dates must not be used as the release dates of the individual protocols. | evidence/research/J.json, sources.16.reliability_notes |
| X302 | Issuer version register at cutoff | A downstream field repeated, or could imply, superseded wording. | Separates the page revision from the dates inside the version register. [102] | structural | Issuer listing revised 2026-09-23 and read 2026-11-11. Release dates are those in the individual rows. | Separates the page revision from the dates inside the version register. | evidence/research/J.json, findings.12.scope_time |
| X303 | The browser could not fetch some GitHub commit-pinned pages or guessed release paths. Project source files were recovered via the GitHub contents API at a dated immutable SHA. The failed browser path is not evidence about implementation status. | A downstream field repeated, or could imply, superseded wording. | The surviving limitation concerns live activity, not access to the cited code. [92, 108, 95, 96, 103, 104, 93] | structural | In Stage 2 the GitHub contents API provided the pinned files. On November 11 the verifier read all four files directly from raw.githubusercontent.com and read the commit timestamp in GitHub’s HTML. GitHub REST rate-limited the Prebid PR requests. The public PR page and its first-party raw diff supplied the source, tests and merge record. All cited GitHub records were readable. | The surviving limitation concerns live activity, not access to the cited code. | evidence/research/J.json, open_questions.3 |
| X304 | §5.3 Audience and targeting, buyer_supplied, line 229 | A downstream field repeated, or could imply, superseded wording. | The reopened source gives this section or named function as the passage locator. [97] | structural | OpenProposal v3.0-draft-1 §5.3, Audience reach, buyer_supplied row | The reopened source gives this section or named function as the passage locator. | evidence/research/J.json, findings.7.support.0.locator |
| X305 | PR #14626 Files changed, mapEntryToOpenRtbSegment, line 308; getBidRequestData lines 319–350 | A downstream field repeated, or could imply, superseded wording. | The reopened source gives this section or named function as the passage locator. [103] | structural | PR #14626 diff, modules/agenticAudienceAdapter.js, mapEntryToOpenRtbSegment and getBidRequestData | The reopened source gives this section or named function as the passage locator. | evidence/research/J.json, findings.9.support.0.locator |
| X306 | LiveRamp describes RampID as the label for a record in its identity graph. | Independent re-read tested the original scope and classification. | LiveRamp's API distinguishes individual and household RampIDs, client encodings, and maintained versus derived matches. A returned derived ID persists for equal input even without a graph match. An ID returned is not a validated identity. LiveRamp describes deterministic offline PII merging, online device linking and joining the two, including US-based devices. Correctness and completeness are vendor goals. Its graph is a technology/data asset rather than a second name for every use of RampID. [234, 181] | supported with qualification | LiveRamp describes RampIDs as pseudonymous identifiers. Its API can return a maintained RampID for an identity graph match or a derived RampID when it finds no match. | A blanket reading that all returned IDs label records already in the graph fails on the documented derived-ID exception. | data/claims-overlay.json, IDN-U-F01; evidence/verify/U.json |
| X307 | ID5 describes a publisher-issued common ID and reports a substantial US client base. | Independent re-read tested the original scope and classification. | ID5 publisher instructions distinguish the shared encrypted ID from the optional ID5 Identity Graph. Publishers can pass the ID in bid requests through supported integrations, with authorized partners able to decrypt subject to consent. ID5's October 30, 2025 release reports acquiring the US graph provider TrueData and describes an existing US business. A unified offering was put in future tense. The acquired graph and publisher ID have distinct functions. Partner agreement implementation requirements, section 2 explicitly applies the ID5 ID Service to US-originating traffic. It does not identify a particular publisher deployment. [369, 132, 177] | supported with qualification | ID5 says publishers can generate and distribute its common first-party identifier. Its publisher requirements cover US-originating traffic. Its 2025 acquisition release describes a US client base for ID5 as a business. [Source IDs added: IDN-UV-S01] | Publisher generation through ID5 is an integration role. It does not make each publisher the issuer of a new namespace or attribute company-wide US clients to a particular integration. | data/claims-overlay.json, IDN-U-F04; evidence/verify/U.json |
| X308 | Yahoo documents a server-to-server ConnectID lookup using a hashed email and a publisher ID, and the joint issuer release describes a US user base. | Independent re-read tested the original scope and classification. | Yahoo describes server-to-server ConnectID generation using credentials, a required hashed email and publisher ID, with an optional device signal, opt-out returning an empty response, and OpenRTB 2.5/2.6 transmission through different EID paths. A joint Yahoo/LiveRamp announcement in 2023 says ATS publishers can integrate ConnectID and cites Yahoo's own estimate of nearly 200 million authenticated US users across its relationships. A joint announcement is not an independent performance check. Issuer-distributed PDF dates the original joint communication. Same evidence origin as IDN-U-S09; no additional reach corroboration. [165, 371, 372] | supported with qualification | Yahoo documents a server-to-server ConnectID lookup using a hashed email and publisher ID. In an October 16, 2023 joint release, Yahoo and LiveRamp attributed ConnectID’s US base to Yahoo’s authenticated consumer relationships. [Source IDs added: IDN-UV-S07] | The historical joint release dates the scope. The same communication was reopened from LiveRamp’s PDF after Yahoo’s direct version was blocked. | data/claims-overlay.json, IDN-U-F05; evidence/verify/U.json |
| X309 | Prebid names a Criteo ID for Exchanges module and describes its use by bidders. | Independent re-read tested the original scope and classification. | Prebid lists Criteo ID for Exchanges as a module for publisher exchange inventory and Criteo-specific demand. The Direct Bidder is another route to Criteo demand. Integration instructions alone do not prove US use or cross-exchange recognition. [250] | supported with qualification | Prebid documents a Criteo ID for Exchanges module that bidders can use in their adapters to access Criteo demand on exchange inventory. | The conditional instruction refers to what bidders can implement. It is not observed adapter use, an equivalence with OneTag, or evidence that unrelated demand receives the ID. | data/claims-overlay.json, IDN-U-F07; evidence/verify/U.json |
| X310 | LiveRamp documents a known-identity service that maps consumer PII into AbiliTec Links or Known IDs. | Independent re-read tested the original scope and classification. | LiveRamp known resolution returns AbiliTec Links via API or LiveRamp Known IDs in cloud environments. The product family groups distinct outputs. RampID circulation is a different task. [235] | supported with qualification | LiveRamp documents two known-identity resolution routes. Its API takes consumer PII and returns AbiliTec Links. Its cloud service takes PII or email addresses and returns Known IDs. | The two choices have named route-specific outputs. The singular wording could attribute both output types to one interface. | data/claims-overlay.json, IDN-U-F12; evidence/verify/U.json |
| X311 | Permutive documents cohort activation with publishers on its platform. | Independent re-read tested the original scope and classification. | Permutive's dated advertiser guide requires selecting connected publishers, creating a cohort, and a publisher deal setup before DSP activation. The route is a cohort/deal, not a universal ID. Permutive's own explainer links first-party data to targetable cohorts rather than one-to-one targeting. The article shown is a brief teaser, with no geographic evidence or independent test. Partner case: TelevisaUnivision transmitted Permutive standard cohorts through Ad Manager on its US inventory in a publisher-provided signals beta. The partner release describes access for US marketers to publisher audiences via OpenX and Permutive without third-party cookies or device IDs. [205, 206, 207, 208] | supported with qualification | Permutive documents activating publisher cohorts through deal setup. In a Google Ad Manager case study, TelevisaUnivision used Permutive’s standard cohorts on US inventory. OpenX separately announced a US buyer route for Permutive audiences without third-party cookies or device IDs. [Source IDs added: IDN-UV-S02, IDN-UV-S03] | The US deployment pertains to publisher-provided signals in the named partner test. The separate OpenX buy route was announced in 2022. No market-share or general availability claim follows. | data/claims-overlay.json, IDN-U-F19; evidence/verify/U.json |
| X312 | Audigent describes a cookieless media-curation route using TransUnion audiences. | Independent re-read tested the original scope and classification. | Audigent's partner release offers TruAudience data combined with publisher audiences or contextual data in SmartPMPs and ContextualPMPs. It expressly sells cookieless media/data curation; Hadron appears separately in navigation. US audience statements describe TransUnion data. Implementation reference documents a separate Hadron ID namespace and module. SmartPMP is a media deal, not its alias. The beta curation partner table lists Hadron ID by Audigent as a secure signal for Experian – Audience. Evidence of a named receiving route, not a nationwide production census. Prebid lists Fabrick ID by Neustar separately, another discovery lead, without proving current operation. [236, 237, 238, 239] | supported as stated | Audigent describes a cookieless media-curation route using TransUnion audiences. [Source IDs added: IDN-UV-S05, IDN-UV-S06, IDN-A-S06] | A separately named identity module exists. Do not move its integration or audience identifier into this SmartPMP row. | data/claims-overlay.json, IDN-U-F20; evidence/verify/U.json |
| X313 | Ogury says the airline campaign used Personified Advertising to reach travelers without relying on their advertising IDs. | Independent re-read tested the original scope and classification. | Ogury reports a Jan 17–Feb 15, 2025 campaign reaching inbound travelers in Japan, including travelers from the US. Ogury's personas use survey signals, without targeting on individual ad IDs. Nationality/source market does not prove US campaign delivery. [248] | supported with qualification | Ogury says it used Personified Advertising to target travelers arriving in Japan without relying on their advertising IDs during an All Nippon Airways campaign from January 17 to February 15, 2025. The travelers included visitors from the United States. | Travelers from the US were reached in Japan. This is not a domestic US campaign. | data/claims-overlay.json, IDN-U-F21; evidence/verify/U.json |
| X314 | TransUnion describes Epsilon audiences mapped across its US connected-home graph through the TruAudience Data Marketplace. | Independent re-read tested the original scope and classification. | TransUnion reports that Epsilon audiences became available through its marketplace, matching a vendor-stated 250 million US consumers against a TransUnion household graph covering over 80 million US connected homes. The data, graph and distribution channel differ. TransUnion's launch names three offerings, Identity, Data Marketplace (formerly Tru Optik), and Platform. It rolls up Tru Optik, Signal Digital and TruSignal capabilities across the suite. That sentence does not make the Tru Optik marketplace an alias of the household graph. Comcast and TransUnion’s June 6, 2024 release documents TV activation via Comcast using TransUnion’s graph, rather than a right to acquire a separate streaming graph. [245, 246, 247] | supported with qualification | TransUnion said Epsilon audiences were available through the TruAudience Data Marketplace using its US connected-home graph. A separate Comcast agreement described using TransUnion’s graph to build segments for TV campaigns on Comcast. Both records describe audience activation. [Source IDs added: IDN-G-S02] | Tru Optik is named as the former Data Marketplace. Do not convert that fact into a graph alias, second household SKU or second score. | data/claims-overlay.json, IDN-U-F24; evidence/verify/U.json |
| X315 | Blockgraph announced a FullThrottle TV integration in which customers could use its Identity Operating System to match and activate data. | Independent re-read tested the original scope and classification. | Blockgraph's June 9, 2022 release announces FullThrottle's agreement to match/activate household data via the Identity Operating System and participating partners. It discusses future operations. Partnership geography and issuer location do not establish current availability. [396] | supported with qualification | In June 2022, Blockgraph announced an agreement under which FullThrottle customers would be able to match and activate household data through Blockgraph’s Identity Operating System with participating partners. | Future tense is explicit in the partnership release: planned customer capability is not independently established deployment. | data/claims-overlay.json, IDN-U-F25; evidence/verify/U.json |
| X316 | OpenAP defined OpenID as a person-level audience construct across linear and digital video. | Independent re-read tested the original scope and classification. | OpenAP's April 28, 2021 release calls OpenID a person-level common TV identifier and names US publishers able to accept it for cross-platform audiences. Its unit is not automatically a household. OpenAP's April 15, 2025 release calls Open Identity an evolution of OpenID and a federated interoperability infrastructure without a central identity spine. Several vendors are named as participating data suppliers. This is not evidence that both names denote an unchanged identifier or one can buy their vendors' products as part of it. [397, 398] | supported with qualification | OpenAP described OpenID as a person-level identifier for linear and digital TV in 2021. In 2025 it described Open Identity as an evolution of OpenID, with federated matching that does not require a central identity spine. | The 2025 issuer explicitly supplies a lineage relationship, not a statement of identical product function or household assignment. | data/claims-overlay.json, IDN-U-F27; evidence/verify/U.json |
| X317 | Cadent documents an IP-to-household Viewer Graph and describes the integration of its graph in a Premion partnership. | Independent re-read tested the original scope and classification. | Cadent's 2024 Aperture product PDF says audience targeting can use its IP-to-household Viewer Graph alongside contextual targeting and also independently. Aperture is the platform; Viewer Graph is the graph. In its February 2, 2021 Premion announcement, Cadent describes a graph that connects TV devices to households from first- and third-party data, claiming a 104-million-household graph. An earlier partner project is not proof of the current census, market or rate. The current Total Video offer names the proprietary Viewer Graph and markets US CTV buying. Only the integrated media route is established. [306, 314, 315] | supported with qualification | Cadent identified its IP-to-household Viewer Graph in a 2024 Aperture product sheet. A 2021 release announced its planned use with Premion. As of November 11, 2026, Cadent also names Viewer Graph in its US CTV media offer. [Source IDs added: IDN-UV-S04] | The historical Premion release speaks in future tense. Direct live PDF fetches redirected unsuccessfully; search-indexed copies of both exact issuer PDFs reopened with the quoted text. Cadent’s current offering is a media buy. | data/claims-overlay.json, IDN-U-F29; evidence/verify/U.json |
| X318 | AWS describes a Clean Rooms query as accessing and analyzing configured tables in a collaboration. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | AWS Clean Rooms' analysis guide describes SQL queries or approved PySpark jobs on configured tables within a collaboration. It does not market a cross-site identifier by that name. [240] | supported as stated | AWS describes a Clean Rooms query as accessing and analyzing configured tables in a collaboration. [Evidence class: vendor_assertion] | Vendor technical documentation (grade C), not direct_record. The company’s user guide establishes a function of its own product and supplies no regulatory, independent, or standards-body finding. | data/claims-overlay.json, IDN-U-F31; evidence/verify/U.json |
| X319 | Microsoft documents Curate as enabling curators to enhance a seller’s inventory with their own assets. [Evidence class: direct_record] | Independent re-read tested the original scope and classification. | Microsoft Curate documentation lets curators enrich sellers' inventory and sell deals. The October 2025 page describes a curation route, not an identity product; it does not establish 2026 operation of Microsoft Invest. [241] | supported as stated | Microsoft documents Curate as enabling curators to enhance a seller’s inventory with their own assets. [Evidence class: vendor_assertion] | Vendor technical documentation (grade C), not direct_record. The October 2025 guide supports the curation function. It is corporate product documentation and cannot alone establish the present status of a named buying platform. | data/claims-overlay.json, IDN-U-F32; evidence/verify/U.json |
| X320 | Prebid documents a 33Across User ID module that uses Lexicon technology. | Independent re-read tested the original scope and classification. | Prebid documents 33Across' module, which uses probabilistic Lexicon technology, with stated support for US, Canada, Australia and Japan. A module name and its enabling technology are not two eligible products by default. [401] | supported with qualification | Prebid documents a 33Across User ID module that uses Lexicon technology. Its guide lists support in the US, Canada, Australia and Japan. | The table previously missed explicit US support in the same guide. Manufacturer support and actual US deployment remain separate. | data/claims-overlay.json, IDN-U-F38; evidence/verify/U.json |
| X321 | Deep Sync describes a US grocery case in which it resolved transactions with payment signals and its identity graph. | Independent re-read tested the original scope and classification. | Deep Sync describes one monthly US grocery POS file and three matching passes, ending with probabilistic store-catchment assignment. It mixes volumes processed, resolved, and cumulative rates with conflicting headline labels. The grocery outcome is not interchangeable with retail CRM graph match accuracy. [12] | supported with qualification | Deep Sync describes three matching passes on a US grocer’s transaction file. The second matched payment-card signals against its identity graph. The third used store geography and household data to infer a buyer. | Payment-card matching is only one stage. The publisher’s ~70%/73% and processed/resolved counts do not independently validate identity or reconcile the different denominators. | data/claims-overlay.json, IDN-U-F43; evidence/verify/U.json |
| X322 | LtvAdx documents a household ID lookup or creation route in its Identity API. | Independent re-read tested the original scope and classification. | LtvAdx documents VAST request household resolution, including an optional shared-IP-protected fallback disabled by default, and separate Identity API behavior. It drops device links 30 days and households 90 days after last seen. No US availability or observed calibration is established. LtvAdx offers US household targeting and frequency in an advertiser plan billed on delivered impressions. Its product page describes household segments; no rate or right is given for a standalone graph licence. [227, 229, 230, 231] | supported with qualification | LtvAdx documents an Identity API that returns an existing HouseholdID or creates a new one. It markets household targeting in US TV campaigns and includes HouseholdID targeting in a plan billed on delivered impressions. The plan does not establish a separately licensed graph. [Source IDs added: IDN-GV-S01, IDN-GV-S02, IDN-GV-S03] | The old quoted sentence is absent in the reopened revision, but the resolving table confirms the function. Its quoted replacement is “nothing matched — a new household is created” under Resolving a HouseholdID, table row 3. VAST serving fallback and the Identity API are different interfaces. | data/claims-overlay.json, IDN-U-F45; evidence/verify/U.json |
| X323 | Factori describes its CTV graphs as identifying household users across devices. | Independent re-read tested the original scope and classification. | Factori describes encrypted MAID/IP-linked devices, household CTV graph and daily S2S data delivery. Its three-billion figure is records, not unique US homes; daily update/feed is not a link decay term. Factori’s AWS Marketplace listing offers a contract for its CTV/cross-device dataset. The listing describes global scope without a US data breakout. [308, 309] | supported with qualification | Factori’s cross-device and CTV graph guide, marked last updated in September 2026, describes identifying household users across devices and delivering a daily data feed. Its AWS Marketplace listing offers a contract for the dataset and describes global coverage without a US breakout. [Source IDs added: IDN-GV-S04] | The reopened page explicitly says “Last Updated on September, 2026”; January 2026 is not the version date shown. Daily feed delivery is not link decay. | data/claims-overlay.json, IDN-U-F47; evidence/verify/U.json |
| X324 | {"source_ids": ["IDN-U-S06", "IDN-U-S07"], "geo": "US business footprint described by the supplier; confirm ID-specific US receiving route.", "note": "Use the interoperable ID for the profile. A graph service requires a separate task and lineage; the US footprint alone does not fix the geography of a particular integration.", "period": "Dated records 2025-10-30; read 2026-11-11"} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Separate the documented service, status, unit and boundary from an unresolved hypothesis.; Include the US publisher requirements record in the period. [369, 132, 177] | structural | {"source_ids": ["IDN-U-S06", "IDN-U-S07", "IDN-UV-S01"], "geo": "The ID5 ID Service specifies publisher requirements for traffic originating in the United States. ID5 also reports a US client base across the business.", "note": "The issuer terms cover the US publisher side of the ID5 ID. Verify a named receiving deployment separately. The acquired TrueData graph is not an alias of the ID.", "period": "October 30, 2025 acquisition announcement; February 1, 2026 publisher requirements; read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 4 |
| X325 | {"source_ids": ["IDN-U-S08", "IDN-U-S09"], "period": "Public documentation read 2026-11-11; see per-source dates", "geo": "US scope explicitly described in joint issuer release."} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Use the documented date and distinguish document update, publication and event.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [165, 371, 372] | structural | {"source_ids": ["IDN-U-S08", "IDN-U-S09", "IDN-UV-S07"], "period": "Joint release October 16, 2023; technical documentation read November 11, 2026.", "geo": "The 2023 joint release describes US authenticated consumer relationships behind Yahoo ConnectID. It does not measure current reachable users."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 5 |
| X326 | {"source_ids": ["IDN-U-S25", "IDN-U-S26"], "geo": "US schedule or client scope unresolved from documentation.", "note": "Check the separately purchased cohort output and an explicit cookie or ID substitution pitch for the same service and market.", "period": "Dated records 2023-01-13; read 2026-11-11"} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Separate the documented service, status, unit and boundary from an unresolved hypothesis.; Use the documented date and distinguish document update, publication and event. [205, 206, 207, 208] | structural | {"source_ids": ["IDN-U-S25", "IDN-U-S26", "IDN-UV-S02", "IDN-UV-S03"], "geo": "Permutive standard cohorts are documented on TelevisaUnivision US inventory through Google Ad Manager. OpenX separately advertised a US buyer route in 2022.", "note": "Google’s case concerns a partner deployment of standard interest cohorts. OpenX described an addressable buying option without third-party cookies or device IDs. Preserve the actual US deployment separately from current commercial terms.", "period": "OpenX announcement November 3, 2022; advertiser guide January 13, 2023; undated Google partner case read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 19 |
| X327 | {"source_ids": ["IDN-U-S31", "IDN-U-S32"], "note": "Keep the scheduled profile as an open boundary question. The opened launch, Epsilon and Comcast records support a TV route and an omnichannel identity service, without settling whether there is a separately purchasable US streaming household resolution SKU.", "routes": "Streaming data marketplace", "period": "Dated records 2022-06-07, 2021-08-24; read 2026-11-11"} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Separate the documented service, status, unit and boundary from an unresolved hypothesis.; The two partner announcements describe different TV routes.; Use the documented date and distinguish document update, publication and event. [245, 246, 247] | structural | {"source_ids": ["IDN-U-S31", "IDN-U-S32", "IDN-G-S02"], "note": "Epsilon via the TruAudience Data Marketplace and Comcast addressable campaigns establish TV activation routes. Tru Optik is the former marketplace name. Neither release confirms a separately purchasable household resolution service distinct from TruAudience Identity.", "routes": "TruAudience Data Marketplace; partner activation on Comcast platforms", "period": "TruAudience launch August 24, 2021; Epsilon June 7, 2022; Comcast June 6, 2024. Read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 24 |
| X328 | {"source_ids": ["IDN-U-S40", "IDN-U-S41"], "name": "Cadent Aperture Viewer Graph", "geo": "Geographic scope for the current service needs confirmation.", "note": "Test the present-day service name and separate buyability, and map the difference between Aperture and the specific viewer graph.", "period": "Dated records 2024, 2021-02-02; read 2026-11-11"} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Separate the documented service, status, unit and boundary from an unresolved hypothesis.; Distinguish dated announcement and product sheet from the current offer. [306, 314, 315] | structural | {"source_ids": ["IDN-U-S40", "IDN-U-S41", "IDN-UV-S04"], "name": "Cadent Viewer Graph", "geo": "Cadent markets US CTV buying using its Viewer Graph within Total Video. Separate graph access or licensing is unconfirmed.", "note": "Viewer Graph is the named graph. Aperture’s 2024 targeting sheet and the current Total Video offer are platforms or routes using it. Premion’s 2021 announcement described planned integration.", "period": "Premion agreement February 2, 2021; Aperture sales sheet 2024; Total Video offer read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 29 |
| X329 | {"source_ids": ["IDN-U-S57"], "geo": "US purchase or deployment unconfirmed.", "note": "Check licensing, deployment geography and what part of the procedure uses an IP edge. See G for TTL scope."} | The table row was checked against the reopened records. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Separate the documented service, status, unit and boundary from an unresolved hypothesis. [227, 229, 230, 231] | structural | {"source_ids": ["IDN-U-S57", "IDN-GV-S01", "IDN-GV-S02", "IDN-GV-S03"], "geo": "LtvAdx markets a US TV campaign route with HouseholdID targeting in a paid-impression plan. Separate licensing of household resolution is unconfirmed.", "note": "Identity API documents household creation and lookup. The US page and pricing evidence establish paid media access to HouseholdID targeting, without a separately priced graph licence. Distinguish device-link and household retention from VAST serving fallback."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 45 |
| X330 | {"name": "Criteo OneTag identity service", "note": "Replace provisional OneTag label only after verifying commercial identity and open cross-party use. A Criteo demand path could be a closed integration."} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; Separate the documented service, status, unit and boundary from an unresolved hypothesis. [250] | structural | {"name": "Criteo ID for Exchanges (OneTag relationship unconfirmed)", "note": "Prebid documents Criteo ID for Exchanges as a route to Criteo demand. OneTag was the provisional seed label. Neither an alias nor independent downstream use is established."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 7 |
| X331 | {"name": "Amazon Publisher Services publisher ID", "routes": "Publisher reporting tool"} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; A reporting tool cannot be presented as the route of an unidentified product. [378] | structural | {"name": "Amazon Publisher Services publisher ID (unverified seed)", "routes": "No proprietary publisher-ID route established; Signal IQ measures bidstream signals."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 8 |
| X332 | {"name": "LiveRamp identity resolution", "note": "Keep the known-identity resolution service distinct from RampID only to the extent the deliverable is separately obtainable."} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; Separate the documented service, status, unit and boundary from an unresolved hypothesis. [235] | structural | {"name": "LiveRamp Known Identity Resolution (AbiliTec API and cloud routes)", "note": "Known Identity Resolution is the service family. Attribute AbiliTec Links to the API and Known IDs to the cloud route. Confirm the exact purchased output before any score."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 12 |
| X333 | {"name": "Experian identity resolution"} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [117] | structural | {"name": "Experian Digital Graph"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 13 |
| X334 | {"name": "Acxiom Real Identity", "geo": "US-targeted distributor/terms still to be established."} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [387, 388] | structural | {"name": "Acxiom Real ID in the cloud", "geo": "A US retailer is named for Digital Recognition powered by Real ID; a standalone US sale of the cloud app is unconfirmed."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 16 |
| X335 | {"group": "Cohort C discovery lead", "note": "A SmartPMP deal is inventory curation unless a separate buyer-selectable identity-replacement deliverable is documented.", "source_ids": ["IDN-U-S27"]} | The table row was checked against the reopened records. | The discovery instrument must reflect the evidence for the buying task, separately from eligibility or admission.; Separate the documented service, status, unit and boundary from an unresolved hypothesis.; Add sources for the distinction between the media deal and Hadron ID. [236, 237, 238, 239] | structural | {"group": "Listed curation boundary", "note": "SmartPMP is a curated media deal. Hadron ID is a separately documented identity module. Do not transfer its features to SmartPMP or count the two names as aliases.", "source_ids": ["IDN-U-S27", "IDN-UV-S05", "IDN-UV-S06", "IDN-A-S06"]} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 20 |
| X336 | {"geo": "US travelers in an international inbound campaign; a US ad market is unconfirmed.", "period": "Public documentation read 2026-11-11; see per-source dates"} | The table row was checked against the reopened records. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Use the documented date and distinguish document update, publication and event. [248] | structural | {"geo": "The Jan 17–Feb 15, 2025 campaign reached inbound travelers in Japan, including visitors from the US. It does not establish US ad delivery.", "period": "Campaign January 17–February 15, 2025; read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 21 |
| X337 | {"routes": "IDoS partner integration"} | The table row was checked against the reopened records. | Future-tense agreement is evidence of plans. [396] | structural | {"routes": "Announced IDoS integration with FullThrottle (2022)"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 25 |
| X338 | {"note": "Check whether the household output can be separately obtained or only used in the DSP. Keep household counts as vendor assertions and investigate the mapping clock."} | The table row was checked against the reopened records. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [312, 313] | structural | {"note": "Viant says its DSP recognizes a Household ID profile from bid request inputs, rather than receiving the Household ID in the bidstream. The US adult figure is a vendor reach assertion. Verify a separate household deliverable."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 26 |
| X339 | {"note": "Test whether a separable household assignment service exists. Keep the later Open Identity initiative distinct until its relationship to OpenID is shown."} | The table row was checked against the reopened records. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [397, 398] | structural | {"note": "OpenAP called Open Identity a 2025 evolution of its 2021 person-level OpenID. The later product is federated infrastructure, not evidence that OpenID assigns a household. Keep a separate household purchase unconfirmed."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 27 |
| X340 | {"name": "IAB Tech Lab Seller Defined Audiences"} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [251] | structural | {"name": "IAB Tech Lab Curated Audiences (formerly Seller Defined Audiences)"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 34 |
| X341 | {"status": "unconfirmed"} | The table row was checked against the reopened records. | Google expressly described FLoC as a proposal replaced by Topics in 2022. The global unconfirmed default obscures this history. [284] | structural | {"status": "historical"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 37 |
| X342 | {"geo": "US eligibility unconfirmed.", "note": "Check the current commercial operator, US availability and recipient evidence. No reach claim from the blocked launch page was used."} | The table row was checked against the reopened records. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence.; Separate the documented service, status, unit and boundary from an unresolved hypothesis. [401] | structural | {"geo": "Prebid expressly lists United States, Canada, Australia and Japan as Lexicon support markets. Current commercial terms and named US deployments are unverified.", "note": "The documented module uses Lexicon technology with US support. Check named receiving deployments and present commercial terms. No reach claim is taken from the blocked launch page."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 38 |
| X343 | {"group": "Cohort A discovery lead"} | The table row was checked against the reopened records. | The discovery instrument must reflect the evidence for the buying task, separately from eligibility or admission. [405] | structural | {"group": "Listed first-party activation workflow"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 41 |
| X344 | {"name": "Kochava household mapping / Kochava Collective lead", "geo": "US availability of the graph as a standalone product unresolved."} | The table row was checked against the reopened records. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID.; Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [35] | structural | {"name": "Kochava advanced measurement household graph (QSR case)", "geo": "Kochava describes a US QSR CTV campaign using household mapping. The source does not establish a Kochava Collective product or standalone graph licence."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 44 |
| X345 | {"period": "Dated records 2026-01; read 2026-11-11", "source_ids": ["IDN-U-S59"], "geo": "US purchase or deployment unconfirmed.", "note": "Seek a specific US buyer route, link time horizon, and criteria for household membership. A data feed update cannot establish link expiry.", "routes": "CTV graph feed"} | The table row was checked against the reopened records. | Use the documented date and distinguish document update, publication and event.; The AWS catalog and pricing table confirm a dataset offered on contract; S59 describes the feed.; Country-specific coverage is distinct from a global commercial listing or US financing boilerplate.; The AWS subscription answers separate buyability but not the country or edge-expiry questions.; Product-specific purchasing route confirmed. [308, 309] | structural | {"period": "Documentation explicitly last updated September 2026; read November 11, 2026.", "source_ids": ["IDN-U-S59", "IDN-GV-S04"], "geo": "Factori markets the CTV/cross-device dataset globally through AWS Marketplace. The listing does not identify US dataset coverage or a US deployment.", "note": "A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule.", "routes": "Daily CTV/cross-device data feed; AWS Marketplace contract subscription."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 47 |
| X346 | {"note": "Investigate the commissioned output and whether a current US buyer can purchase it. Graphs behind an alternative do not by themselves disqualify it from cohort C.", "group": "Cohort C discovery lead"} | The table row was checked against the reopened records. | Separate the documented service, status, unit and boundary from an unresolved hypothesis.; The discovery instrument must reflect the evidence for the buying task, separately from eligibility or admission. [406] | structural | {"note": "Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task.", "group": "Limited 1:1 audience targeting; cohort C fit unconfirmed"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 48 |
| X347 | {"period": "Dated records 2021-05-07; read 2026-11-11"} | The table row was checked against the reopened records. | Use the documented date and distinguish document update, publication and event. [407] | structural | {"period": "Public-comment draft March 2021; comment deadline May 7, 2021. Read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 49 |
| X348 | {"owner": "Not established here"} | The table row was checked against the reopened records. | The SharedID implementation guide explicitly identifies this first-party ID as Prebid-owned; this is product ownership, not a separate corporate acquisition finding. [286] | structural | {"owner": "Prebid (described by Prebid as owner of SharedID)"} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 9 |
| X349 | {"period": "Dated records 2026-11-05; read 2026-11-11"} | The table row was checked against the reopened records. | The re-opened UID2 operator and environment guide footers date this revision to November 11. [203, 153] | structural | {"period": "Documentation last updated and read November 11, 2026."} | Keep the table consistent with the independently verified disclosure. | evidence/research/U.json, identity_product_discovery, data row 2 |
| X350 | ["IDN-U-S06", "IDN-U-S07"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [369, 132, 177] | structural | ["IDN-U-S06", "IDN-U-S07", "IDN-UV-S01"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P004/source_ids |
| X351 | ["IDN-U-S08", "IDN-U-S09"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [165, 371, 372] | structural | ["IDN-U-S08", "IDN-U-S09", "IDN-UV-S07"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P005/source_ids |
| X352 | ["IDN-U-S25", "IDN-U-S26"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [205, 206, 207, 208] | structural | ["IDN-U-S25", "IDN-U-S26", "IDN-UV-S02", "IDN-UV-S03"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P019/source_ids |
| X353 | ["IDN-U-S31", "IDN-U-S32"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [245, 246, 247] | structural | ["IDN-U-S31", "IDN-U-S32", "IDN-G-S02"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P024/source_ids |
| X354 | ["IDN-U-S40", "IDN-U-S41"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [306, 314, 315] | structural | ["IDN-U-S40", "IDN-U-S41", "IDN-UV-S04"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P029/source_ids |
| X355 | ["IDN-U-S57"] | Product-level buying task, evidence route or disposition was rechecked. | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. [227, 229, 230, 231] | structural | ["IDN-U-S57", "IDN-GV-S01", "IDN-GV-S02", "IDN-GV-S03"] | Preserve the original source IDs and append the reopened product or partner records that settle this row’s narrower scope. | evidence/research/U.json, register_proposals/IDN-P045/source_ids |
| X356 | US business footprint described by the supplier; confirm ID-specific US receiving route. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [369, 132, 177] | structural | The ID5 ID Service specifies publisher requirements for traffic originating in the United States. ID5 also reports a US client base across the business. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P004/us_scope |
| X357 | Use the interoperable ID for the profile. A graph service requires a separate task and lineage; the US footprint alone does not fix the geography of a particular integration. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [369, 132, 177] | structural | The issuer terms cover the US publisher side of the ID5 ID. Verify a named receiving deployment separately. The acquired TrueData graph is not an alias of the ID. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P004/scope_test |
| X358 | US scope explicitly described in joint issuer release. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [165, 371, 372] | structural | The 2023 joint release describes US authenticated consumer relationships behind Yahoo ConnectID. It does not measure current reachable users. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P005/us_scope |
| X359 | Criteo OneTag identity service | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [250] | structural | Criteo ID for Exchanges (OneTag relationship unconfirmed) | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P007/product |
| X360 | Replace provisional OneTag label only after verifying commercial identity and open cross-party use. A Criteo demand path could be a closed integration. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [250] | structural | Prebid documents Criteo ID for Exchanges as a route to Criteo demand. OneTag was the provisional seed label. Neither an alias nor independent downstream use is established. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P007/scope_test |
| X361 | Amazon Publisher Services publisher ID | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [378] | structural | Amazon Publisher Services publisher ID (unverified seed) | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P008/product |
| X362 | LiveRamp identity resolution | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [235] | structural | LiveRamp Known Identity Resolution (AbiliTec API and cloud routes) | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P012/product |
| X363 | Keep the known-identity resolution service distinct from RampID only to the extent the deliverable is separately obtainable. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [235] | structural | Known Identity Resolution is the service family. Attribute AbiliTec Links to the API and Known IDs to the cloud route. Confirm the exact purchased output before any score. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P012/scope_test |
| X364 | Experian identity resolution | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [117] | structural | Experian Digital Graph | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P013/product |
| X365 | Acxiom Real Identity | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [387, 388] | structural | Acxiom Real ID in the cloud | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P016/product |
| X366 | US-targeted distributor/terms still to be established. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [387, 388] | structural | A US retailer is named for Digital Recognition powered by Real ID; a standalone US sale of the cloud app is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P016/us_scope |
| X367 | US schedule or client scope unresolved from documentation. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [205, 206, 207, 208] | structural | Permutive standard cohorts are documented on TelevisaUnivision US inventory through Google Ad Manager. OpenX separately advertised a US buyer route in 2022. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P019/us_scope |
| X368 | Check the separately purchased cohort output and an explicit cookie or ID substitution pitch for the same service and market. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [205, 206, 207, 208] | structural | Google’s case concerns a partner deployment of standard interest cohorts. OpenX described an addressable buying option without third-party cookies or device IDs. Preserve the actual US deployment separately from current commercial terms. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P019/scope_test |
| X369 | Retain subject to substitution screen | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [205, 206, 207, 208] | structural | Retain documented US publisher cohorts; verify current purchase terms | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P019/screen_disposition |
| X370 | A SmartPMP deal is inventory curation unless a separate buyer-selectable identity-replacement deliverable is documented. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [236] | structural | SmartPMP is a curated media deal. Hadron ID is a separately documented identity module. Do not transfer its features to SmartPMP or count the two names as aliases. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P020/scope_test |
| X371 | Curation boundary | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [236] | structural | Listed curation boundary; screen Hadron ID separately | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P020/screen_disposition |
| X372 | C | Product-level buying task, evidence route or disposition was rechecked. | SmartPMP itself is a curation offer. The cohort A identifier lead belongs in a different record. [236] | structural | null | SmartPMP itself is a curation offer. The cohort A identifier lead belongs in a different record. | evidence/research/U.json, register_proposals/IDN-P020/candidate_cohort |
| X373 | ["IDN-U-S27"] | Product-level buying task, evidence route or disposition was rechecked. | Add sources for the distinction between the media deal and Hadron ID. [236, 237, 238, 239] | structural | ["IDN-U-S27", "IDN-UV-S05", "IDN-UV-S06", "IDN-A-S06"] | Add sources for the distinction between the media deal and Hadron ID. | evidence/research/U.json, register_proposals/IDN-P020/source_ids |
| X374 | US travelers in an international inbound campaign; a US ad market is unconfirmed. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [248] | structural | The Jan 17–Feb 15, 2025 campaign reached inbound travelers in Japan, including visitors from the US. It does not establish US ad delivery. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P021/us_scope |
| X375 | Keep the scheduled profile as an open boundary question. The opened launch, Epsilon and Comcast records support a TV route and an omnichannel identity service, without settling whether there is a separately purchasable US streaming household resolution SKU. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [245, 246, 247] | structural | Epsilon via the TruAudience Data Marketplace and Comcast addressable campaigns establish TV activation routes. Tru Optik is the former marketplace name. Neither release confirms a separately purchasable household resolution service distinct from TruAudience Identity. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P024/scope_test |
| X376 | Retain | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [396] | structural | Historical partner announcement; deployment and standalone TV purchase unresolved | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P025/screen_disposition |
| X377 | Check whether the household output can be separately obtained or only used in the DSP. Keep household counts as vendor assertions and investigate the mapping clock. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [312, 313] | structural | Viant says its DSP recognizes a Household ID profile from bid request inputs, rather than receiving the Household ID in the bidstream. The US adult figure is a vendor reach assertion. Verify a separate household deliverable. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P026/scope_test |
| X378 | Test whether a separable household assignment service exists. Keep the later Open Identity initiative distinct until its relationship to OpenID is shown. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [397, 398] | structural | OpenAP called Open Identity a 2025 evolution of its 2021 person-level OpenID. The later product is federated infrastructure, not evidence that OpenID assigns a household. Keep a separate household purchase unconfirmed. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P027/scope_test |
| X379 | Cadent Aperture Viewer Graph | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [306, 314, 315] | structural | Cadent Viewer Graph | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P029/product |
| X380 | Geographic scope for the current service needs confirmation. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [306, 314, 315] | structural | Cadent markets US CTV buying using its Viewer Graph within Total Video. Separate graph access or licensing is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P029/us_scope |
| X381 | Test the present-day service name and separate buyability, and map the difference between Aperture and the specific viewer graph. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [306, 314, 315] | structural | Viewer Graph is the named graph. Aperture’s 2024 targeting sheet and the current Total Video offer are platforms or routes using it. Premion’s 2021 announcement described planned integration. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P029/scope_test |
| X382 | Retain; partner deployment | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [306, 314, 315] | structural | Retain current US media route; separately purchased household resolution unresolved | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P029/screen_disposition |
| X383 | IAB Tech Lab Seller Defined Audiences | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [251] | structural | IAB Tech Lab Curated Audiences (formerly Seller Defined Audiences) | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P034/product |
| X384 | US eligibility unconfirmed. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [401] | structural | Prebid expressly lists United States, Canada, Australia and Japan as Lexicon support markets. Current commercial terms and named US deployments are unverified. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P038/us_scope |
| X385 | Check the current commercial operator, US availability and recipient evidence. No reach claim from the blocked launch page was used. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [401] | structural | The documented module uses Lexicon technology with US support. Check named receiving deployments and present commercial terms. No reach claim is taken from the blocked launch page. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P038/scope_test |
| X386 | Discovery lead | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [401] | structural | US-supported Prebid module; named deployment and commercial terms unconfirmed | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P038/screen_disposition |
| X387 | A | Product-level buying task, evidence route or disposition was rechecked. | Galileo uses UID2; no second common ID is documented. [405] | structural | null | Galileo uses UID2; no second common ID is documented. | evidence/research/U.json, register_proposals/IDN-P041/candidate_cohort |
| X388 | Discovery boundary | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [405] | structural | Listed activation workflow; no separate interoperable ID evidenced | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P041/screen_disposition |
| X389 | Kochava household mapping / Kochava Collective lead | Product-level buying task, evidence route or disposition was rechecked. | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. [35] | structural | Kochava advanced measurement household graph (QSR case) | Use the evidenced product name or mark the seed as unverified. Preserve the Stage 1 seed and its stable ID. | evidence/research/U.json, register_proposals/IDN-P044/product |
| X390 | US availability of the graph as a standalone product unresolved. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [35] | structural | Kochava describes a US QSR CTV campaign using household mapping. The source does not establish a Kochava Collective product or standalone graph licence. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P044/us_scope |
| X391 | US purchase or deployment unconfirmed. | Product-level buying task, evidence route or disposition was rechecked. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [227, 229, 230, 231] | structural | LtvAdx markets a US TV campaign route with HouseholdID targeting in a paid-impression plan. Separate licensing of household resolution is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, register_proposals/IDN-P045/us_scope |
| X392 | Check licensing, deployment geography and what part of the procedure uses an IP edge. See G for TTL scope. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [227, 229, 230, 231] | structural | Identity API documents household creation and lookup. The US page and pricing evidence establish paid media access to HouseholdID targeting, without a separately priced graph licence. Distinguish device-link and household retention from VAST serving fallback. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P045/scope_test |
| X393 | Discovery lead | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [227, 229, 230, 231] | structural | US integrated household targeting offered; separate household resolution purchase unresolved. | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P045/screen_disposition |
| X394 | Investigate the commissioned output and whether a current US buyer can purchase it. Graphs behind an alternative do not by themselves disqualify it from cohort C. | Product-level buying task, evidence route or disposition was rechecked. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [406] | structural | Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, register_proposals/IDN-P048/scope_test |
| X395 | Discovery lead; limited release | Product-level buying task, evidence route or disposition was rechecked. | Carry the verified boundary/status into the screening disposition. [406] | structural | Hold at buying-task boundary: limited 1:1 targeting, C deliverable unconfirmed | Carry the verified boundary/status into the screening disposition. | evidence/research/U.json, register_proposals/IDN-P048/screen_disposition |
| X396 | ["IDN-U-S59"] | Product-level buying task, evidence route or disposition was rechecked. | The AWS catalog and pricing table confirm a dataset offered on contract; S59 describes the feed. [308, 309] | structural | ["IDN-U-S59", "IDN-GV-S04"] | The AWS catalog and pricing table confirm a dataset offered on contract; S59 describes the feed. | evidence/research/U.json, register_proposals/IDN-P047/source_ids |
| X397 | US purchase or deployment unconfirmed. | Product-level buying task, evidence route or disposition was rechecked. | Country-specific coverage is distinct from a global commercial listing or US financing boilerplate. [308, 309] | structural | Factori markets the CTV/cross-device dataset globally through AWS Marketplace. The listing does not identify US dataset coverage or a US deployment. | Country-specific coverage is distinct from a global commercial listing or US financing boilerplate. | evidence/research/U.json, register_proposals/IDN-P047/us_scope |
| X398 | Seek a specific US buyer route, link time horizon, and criteria for household membership. A data feed update cannot establish link expiry. | Product-level buying task, evidence route or disposition was rechecked. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. [308, 309] | structural | A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. | evidence/research/U.json, register_proposals/IDN-P047/scope_test |
| X399 | Discovery lead | Product-level buying task, evidence route or disposition was rechecked. | The catalog establishes a contract route for the dataset. [308, 309] | structural | Retain commercially offered CTV dataset; US eligibility unresolved. | The catalog establishes a contract route for the dataset. | evidence/research/U.json, register_proposals/IDN-P047/screen_disposition |
| X400 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Issuer page shows 12/04/2025 12:00 pm EST. [369] | structural | 2025-12-04 | Issuer page shows 12/04/2025 12:00 pm EST. | evidence/research/U.json, sources.5.pub_date |
| X401 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | The issuer PDF of the same release prints 10/16/2023. [371, 372] | structural | 2023-10-16 | The issuer PDF of the same release prints 10/16/2023. | evidence/research/U.json, sources.8.pub_date |
| X402 | [Previously blank] | A downstream field repeated, or could imply, superseded wording. | Joint announcement date in issuer PDF. [371, 372] | structural | 2023-10-16 | Joint announcement date in issuer PDF. | evidence/research/U.json, sources.8.event_date |
| X403 | technical_doc | A downstream field repeated, or could imply, superseded wording. | Product marketing page describing a commercial offering. Do not assign an implementation-document grade solely from its URL. [384] | structural | vendor | Product marketing page describing a commercial offering. Do not assign an implementation-document grade solely from its URL. | evidence/research/U.json, sources.17.source_type |
| X404 | Territory not established for the exact service, or platform-wide documentation | A downstream field repeated, or could imply, superseded wording. | Domestic customer evidence is attached to the powered-by product, not a separate Real ID cloud purchase. [388] | structural | US case: Guitar Center, early adopter of Digital Recognition powered by Real ID. | Domestic customer evidence is attached to the powered-by product, not a separate Real ID cloud purchase. | evidence/research/U.json, sources.21.geography |
| X405 | January 17–February 15, 2025 | A downstream field repeated, or could imply, superseded wording. | Ogury supplies the campaign dates. [248] | structural | 2025-01-17 through 2025-02-15 | Ogury supplies the campaign dates. | evidence/research/U.json, sources.27.measurement_period |
| X406 | Territory not established for the exact service, or platform-wide documentation | A downstream field repeated, or could imply, superseded wording. | Explicit support territories appear in the implementation guide. [401] | structural | US, Canada, Australia, Japan (Lexicon support per Prebid). | Explicit support territories appear in the implementation guide. | evidence/research/U.json, sources.49.geography |
| X407 | Territory not established for the exact service, or platform-wide documentation | A downstream field repeated, or could imply, superseded wording. | The case explicitly identifies the restaurant chain and its US campaigns. [35] | structural | US QSR CTV campaign; standalone product geography unconfirmed. | The case explicitly identifies the restaurant chain and its US campaigns. | evidence/research/U.json, sources.55.geography |
| X408 | 2026-01 | A downstream field repeated, or could imply, superseded wording. | The page supplies a revision month only. The original publication date is unknown. [308] | structural | [Field cleared; see reason] | The page supplies a revision month only. The original publication date is unknown. | evidence/research/U.json, sources.58.pub_date |
| X409 | 2021 | A downstream field repeated, or could imply, superseded wording. | Month of draft printed on the cover. [407] | structural | 2021-03 | Month of draft printed on the cover. | evidence/research/U.json, sources.60.pub_date |
| X410 | 2025-08-20 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [181] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.1.pub_date |
| X411 | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [181] | structural | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. Last modified August 20, 2025. No dated original publication is shown. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.1.reliability_notes |
| X412 | 2026-11-05 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [203] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.2.pub_date |
| X413 | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [203] | structural | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. Footer observed on November 11, 2026: last updated November 11, 2026. This dates the documentation revision, not the first public availability of the operator. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.2.reliability_notes |
| X414 | 2026-11-05 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [153] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.3.pub_date |
| X415 | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [153] | structural | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. Footer observed on November 11, 2026: last updated November 11, 2026. This dates the documentation revision, not the first public availability of the operator. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.3.reliability_notes |
| X416 | 2026-11-02 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [211] | structural | 2024-07-18 | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.4.pub_date |
| X417 | The list is a jurisdiction screen; publication and last-updated stamps have different meanings. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [211] | structural | The list is a jurisdiction screen; publication and last-updated stamps have different meanings. The text dates geographic coverage to publication on July 18, 2024. The page footer says last updated November 2, 2026. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.4.reliability_notes |
| X418 | 2026-03-19 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [235] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.15.pub_date |
| X419 | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [235] | structural | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. Last modified March 19, 2026. The publication-date field is empty. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.15.reliability_notes |
| X420 | 2025-10-21 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [241] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.43.pub_date |
| X421 | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [241] | structural | Primary issuer or technical-maintainer record of the feature or initiative described. No independent effectiveness or market-wide adoption test. Microsoft Learn marks October 21, 2025 as its last update. The page gives no initial publication date. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.43.reliability_notes |
| X422 | 2025-08-25 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [251] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.45.pub_date |
| X423 | Issuing body records a renaming; naming changes do not imply withdrawal of the specification. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [251] | structural | Issuing body records a renaming; naming changes do not imply withdrawal of the specification. The page was last updated August 25, 2025; the rename occurred in December 2024. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.45.reliability_notes |
| X424 | 2026-09-08 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [283] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.47.pub_date |
| X425 | The heading says deprecate and remove; the body uses prospective wording. Release/version enforcement is for stream E. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [283] | structural | The heading says deprecate and remove; the body uses prospective wording. Release/version enforcement is for stream E. September 8, 2026 is the stable Chrome 153 release date and the page update date. No separate publication date is stated. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.47.reliability_notes |
| X426 | 2021-05-25 | A downstream field repeated, or could imply, superseded wording. | The source distinguishes published information from a later revision or event date. [307] | structural | [Field cleared; see reason] | The source distinguishes published information from a later revision or event date. | evidence/research/U.json, sources.57.pub_date |
| X427 | 2021 developer guide; product boundary and availability need confirmation at the cutoff. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [307] | structural | 2021 developer guide; product boundary and availability need confirmation at the cutoff. Last modified May 25, 2021. The publication-date field is empty. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.57.reliability_notes |
| X428 | Documentation includes multiple update descriptions. A dataset refresh is not an IP-edge TTL. | A downstream field repeated, or could imply, superseded wording. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. [308] | structural | Documentation includes multiple update descriptions. A dataset refresh is not an IP-edge TTL. The guide explicitly marks its last update as September 2026, not January 2026. It separately states a daily feed cadence. | Preserve the dated revision in the notes rather than treating an update stamp as first publication. | evidence/research/U.json, sources.58.reliability_notes |
| X429 | Parrable ID for Header Bidding, rendered line 104 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | Current rendering moved the text from the previously recorded line. [249] | structural | Parrable ID, module description, current HTML line 109 | Current rendering moved the text from the previously recorded line. | evidence/research/U.json, findings.10.support.0.locator |
| X430 | Verve Group launches Moments.AI™ (formerly Beemray), expanding contextual targeting capabilities to North America, rendered line 8 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | Use the literal trademark glyph from the publisher. [392] | structural | March 8, 2022 news release, opening paragraph. | Use the literal trademark glyph from the publisher. | evidence/research/U.json, findings.21.support.0.locator |
| X431 | launch of Moments.AITM (formerly Beemray) | A downstream field repeated, or could imply, superseded wording. | Use the literal trademark glyph from the publisher. [392] | structural | launch of Moments.AI™ (formerly Beemray) | Use the literal trademark glyph from the publisher. | evidence/research/U.json, findings.21.support.0.passage |
| X432 | Samba TV Identity, rendered line 5 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | Quoted words run across a PDF line break. [400] | structural | Samba TV Identity, first page, opening paragraph, lines 4–5 | Quoted words run across a PDF line break. | evidence/research/U.json, findings.29.support.0.locator |
| X433 | 3. ZEOTAP’S UNIVERSAL ID: ID+, rendered line 118 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | PDF body line in current extraction is 190, not 118. [402] | structural | Section 3, PDF page 8, principle 2 (Simplicity). | PDF body line in current extraction is 190, not 118. | evidence/research/U.json, findings.38.support.0.locator |
| X434 | 3. ZEOTAP’S UNIVERSAL ID: ID+, rendered line 114 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | PDF body line in current extraction is 184, not 114. [402] | structural | Section 3, PDF page 8, introductory paragraph. | PDF body line in current extraction is 184, not 114. | evidence/research/U.json, findings.38.support.1.locator |
| X435 | QSR Proves Incremental Lift from Connected TV Ad Campaign, rendered line 152 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | The original sentence is present under the solution heading. [35] | structural | Kochava QSR case, The solution, paragraph 1, current rendered line 154. | The original sentence is present under the solution heading. | evidence/research/U.json, findings.43.support.0.locator |
| X436 | Identity API & HouseholdID, rendered line 29 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | Old excerpt is absent from the reopened page. The lookup/create behavior remains documented in rows 1–3. [227] | structural | Resolving a HouseholdID, match-order table, row 3. | Old excerpt is absent from the reopened page. The lookup/create behavior remains documented in rows 1–3. | evidence/research/U.json, findings.44.support.0.locator |
| X437 | ask the Identity API to create or look one up | A downstream field repeated, or could imply, superseded wording. | Old excerpt is absent from the reopened page. The lookup/create behavior remains documented in rows 1–3. [227] | structural | nothing matched — a new household is created | Old excerpt is absent from the reopened page. The lookup/create behavior remains documented in rows 1–3. | evidence/research/U.json, findings.44.support.0.passage |
| X438 | Accountability Platform, rendered line 14 (locate by quoted text) | A downstream field repeated, or could imply, superseded wording. | The stable locator is on the cover and avoids changing extraction lines. [407] | structural | Cover, PDF page 1, draft/public-comment line. | The stable locator is on the cover and avoids changing extraction lines. | evidence/research/U.json, findings.48.support.0.locator |
| X439 | Every preregistered product was screened. The sweep distinguishes issuer evidence of a service from its US scope, separate buyability and eligibility for an IQ row. The six profile assignments have product-level evidence. The TransUnion streaming case requires a separate-purchase check. Discovered candidates include two distinct US Deep Sync services, developer-documented TV workflows and ID-replacement audience products. Publisher-scoped IDs, demand-specific extensions, marketplaces, clean rooms, consent systems and standards require different dispositions. The register proposals preserve the original IDs and report geography and unresolved questions on each row. No row has been admitted or scored. | A downstream field repeated, or could imply, superseded wording. | Distinguish a case or activation route from a separately purchasable identity product. Recounted the preserved 37 seeds and 12 proposals. [245, 246, 247, 27, 12] | structural | All 37 registered seeds were screened, and 12 Stage 2 discovery additions were proposed. The record contains US use cases for two separately named Deep Sync offerings, but a case does not by itself establish a separately purchasable service. TransUnion has a documented CTV audience activation route; a separately purchased streaming household-resolution product remains unconfirmed. No discovery row has been admitted or scored. | Distinguish a case or activation route from a separately purchasable identity product. Recounted the preserved 37 seeds and 12 proposals. | evidence/research/U.json, summary |
| X440 | US business footprint described by the supplier; confirm ID-specific US receiving route. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [369, 132, 177] | structural | The ID5 ID Service specifies publisher requirements for traffic originating in the United States. ID5 also reports a US client base across the business. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.3.scope_geo |
| X441 | Public documentation read 2026-11-11; see per-source dates | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [165, 371, 372] | structural | Joint release October 16, 2023; technical documentation read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.4.scope_time |
| X442 | US scope explicitly described in joint issuer release. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [165, 371, 372] | structural | The 2023 joint release describes US authenticated consumer relationships behind Yahoo ConnectID. It does not measure current reachable users. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.4.scope_geo |
| X443 | US-targeted distributor/terms still to be established. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [387, 388] | structural | A US retailer is named for Digital Recognition powered by Real ID; a standalone US sale of the cloud app is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.15.scope_geo |
| X444 | US schedule or client scope unresolved from documentation. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [205, 206, 207, 208] | structural | Permutive standard cohorts are documented on TelevisaUnivision US inventory through Google Ad Manager. OpenX separately advertised a US buyer route in 2022. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.18.scope_geo |
| X445 | Dated records 2023-01-13; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [205, 206, 207, 208] | structural | OpenX announcement November 3, 2022; advertiser guide January 13, 2023; undated Google partner case read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.18.scope_time |
| X446 | US travelers in an international inbound campaign; a US ad market is unconfirmed. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [248] | structural | The Jan 17–Feb 15, 2025 campaign reached inbound travelers in Japan, including visitors from the US. It does not establish US ad delivery. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.20.scope_geo |
| X447 | Public documentation read 2026-11-11; see per-source dates | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [248] | structural | Campaign January 17–February 15, 2025; read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.20.scope_time |
| X448 | Dated records 2022-06-07, 2021-08-24; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [245, 246, 247] | structural | TruAudience launch August 24, 2021; Epsilon June 7, 2022; Comcast June 6, 2024. Read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.23.scope_time |
| X449 | Geographic scope for the current service needs confirmation. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [306, 314, 315] | structural | Cadent markets US CTV buying using its Viewer Graph within Total Video. Separate graph access or licensing is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.28.scope_geo |
| X450 | US eligibility unconfirmed. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [401] | structural | Prebid expressly lists United States, Canada, Australia and Japan as Lexicon support markets. Current commercial terms and named US deployments are unverified. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.37.scope_geo |
| X451 | US availability of the graph as a standalone product unresolved. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [35] | structural | Kochava describes a US QSR CTV campaign using household mapping. The source does not establish a Kochava Collective product or standalone graph licence. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.43.scope_geo |
| X452 | US purchase or deployment unconfirmed. | A downstream field repeated, or could imply, superseded wording. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. [227, 229, 230, 231] | structural | LtvAdx markets a US TV campaign route with HouseholdID targeting in a paid-impression plan. Separate licensing of household resolution is unconfirmed. | Use the verified geographic scope of the exact route, distinct from corporate footprint or a separate licence. | evidence/research/U.json, findings.44.scope_geo |
| X453 | Dated records 2026-01; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [308, 309] | structural | Documentation explicitly last updated September 2026; read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.46.scope_time |
| X454 | Dated records 2021-05-07; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | Use the documented date and distinguish document update, publication and event. [407] | structural | Public-comment draft March 2021; comment deadline May 7, 2021. Read November 11, 2026. | Use the documented date and distinguish document update, publication and event. | evidence/research/U.json, findings.48.scope_time |
| X455 | US purchase or deployment unconfirmed. | A downstream field repeated, or could imply, superseded wording. | Country-specific coverage is distinct from a global commercial listing or US financing boilerplate. [308, 309] | structural | Factori markets the CTV/cross-device dataset globally through AWS Marketplace. The listing does not identify US dataset coverage or a US deployment. | Country-specific coverage is distinct from a global commercial listing or US financing boilerplate. | evidence/research/U.json, findings.46.scope_geo |
| X456 | Use the interoperable ID for the profile. A graph service requires a separate task and lineage; the US footprint alone does not fix the geography of a particular integration. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [369, 132, 177] | structural | The issuer terms cover the US publisher side of the ID5 ID. Verify a named receiving deployment separately. The acquired TrueData graph is not an alias of the ID. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.3.limitations |
| X457 | Boundary or alternative interpretation: Use the interoperable ID for the profile. A graph service requires a separate task and lineage; the US footprint alone does not fix the geography of a particular integration. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [369, 132, 177] | structural | The issuer terms cover the US publisher side of the ID5 ID. Verify a named receiving deployment separately. The acquired TrueData graph is not an alias of the ID. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.3.counterevidence |
| X458 | Replace provisional OneTag label only after verifying commercial identity and open cross-party use. A Criteo demand path could be a closed integration. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [250] | structural | Prebid documents Criteo ID for Exchanges as a route to Criteo demand. OneTag was the provisional seed label. Neither an alias nor independent downstream use is established. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.6.limitations |
| X459 | Boundary or alternative interpretation: Replace provisional OneTag label only after verifying commercial identity and open cross-party use. A Criteo demand path could be a closed integration. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [250] | structural | Prebid documents Criteo ID for Exchanges as a route to Criteo demand. OneTag was the provisional seed label. Neither an alias nor independent downstream use is established. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.6.counterevidence |
| X460 | Keep the known-identity resolution service distinct from RampID only to the extent the deliverable is separately obtainable. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [235] | structural | Known Identity Resolution is the service family. Attribute AbiliTec Links to the API and Known IDs to the cloud route. Confirm the exact purchased output before any score. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.11.limitations |
| X461 | Boundary or alternative interpretation: Keep the known-identity resolution service distinct from RampID only to the extent the deliverable is separately obtainable. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [235] | structural | Known Identity Resolution is the service family. Attribute AbiliTec Links to the API and Known IDs to the cloud route. Confirm the exact purchased output before any score. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.11.counterevidence |
| X462 | Check the separately purchased cohort output and an explicit cookie or ID substitution pitch for the same service and market. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [205, 206, 207, 208] | structural | Google’s case concerns a partner deployment of standard interest cohorts. OpenX described an addressable buying option without third-party cookies or device IDs. Preserve the actual US deployment separately from current commercial terms. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.18.limitations |
| X463 | Boundary or alternative interpretation: Check the separately purchased cohort output and an explicit cookie or ID substitution pitch for the same service and market. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [205, 206, 207, 208] | structural | Google’s case concerns a partner deployment of standard interest cohorts. OpenX described an addressable buying option without third-party cookies or device IDs. Preserve the actual US deployment separately from current commercial terms. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.18.counterevidence |
| X464 | A SmartPMP deal is inventory curation unless a separate buyer-selectable identity-replacement deliverable is documented. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [236, 237, 238, 239] | structural | SmartPMP is a curated media deal. Hadron ID is a separately documented identity module. Do not transfer its features to SmartPMP or count the two names as aliases. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.19.limitations |
| X465 | Boundary or alternative interpretation: A SmartPMP deal is inventory curation unless a separate buyer-selectable identity-replacement deliverable is documented. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [236, 237, 238, 239] | structural | SmartPMP is a curated media deal. Hadron ID is a separately documented identity module. Do not transfer its features to SmartPMP or count the two names as aliases. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.19.counterevidence |
| X466 | Keep the scheduled profile as an open boundary question. The opened launch, Epsilon and Comcast records support a TV route and an omnichannel identity service, without settling whether there is a separately purchasable US streaming household resolution SKU. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [245, 246, 247] | structural | Epsilon via the TruAudience Data Marketplace and Comcast addressable campaigns establish TV activation routes. Tru Optik is the former marketplace name. Neither release confirms a separately purchasable household resolution service distinct from TruAudience Identity. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.23.limitations |
| X467 | Boundary or alternative interpretation: Keep the scheduled profile as an open boundary question. The opened launch, Epsilon and Comcast records support a TV route and an omnichannel identity service, without settling whether there is a separately purchasable US streaming household resolution SKU. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [245, 246, 247] | structural | Epsilon via the TruAudience Data Marketplace and Comcast addressable campaigns establish TV activation routes. Tru Optik is the former marketplace name. Neither release confirms a separately purchasable household resolution service distinct from TruAudience Identity. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.23.counterevidence |
| X468 | Check whether the household output can be separately obtained or only used in the DSP. Keep household counts as vendor assertions and investigate the mapping clock. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [312, 313] | structural | Viant says its DSP recognizes a Household ID profile from bid request inputs, rather than receiving the Household ID in the bidstream. The US adult figure is a vendor reach assertion. Verify a separate household deliverable. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.25.limitations |
| X469 | Boundary or alternative interpretation: Check whether the household output can be separately obtained or only used in the DSP. Keep household counts as vendor assertions and investigate the mapping clock. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [312, 313] | structural | Viant says its DSP recognizes a Household ID profile from bid request inputs, rather than receiving the Household ID in the bidstream. The US adult figure is a vendor reach assertion. Verify a separate household deliverable. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.25.counterevidence |
| X470 | Test whether a separable household assignment service exists. Keep the later Open Identity initiative distinct until its relationship to OpenID is shown. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [397, 398] | structural | OpenAP called Open Identity a 2025 evolution of its 2021 person-level OpenID. The later product is federated infrastructure, not evidence that OpenID assigns a household. Keep a separate household purchase unconfirmed. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.26.limitations |
| X471 | Boundary or alternative interpretation: Test whether a separable household assignment service exists. Keep the later Open Identity initiative distinct until its relationship to OpenID is shown. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [397, 398] | structural | OpenAP called Open Identity a 2025 evolution of its 2021 person-level OpenID. The later product is federated infrastructure, not evidence that OpenID assigns a household. Keep a separate household purchase unconfirmed. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.26.counterevidence |
| X472 | Test the present-day service name and separate buyability, and map the difference between Aperture and the specific viewer graph. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [306, 314, 315] | structural | Viewer Graph is the named graph. Aperture’s 2024 targeting sheet and the current Total Video offer are platforms or routes using it. Premion’s 2021 announcement described planned integration. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.28.limitations |
| X473 | Boundary or alternative interpretation: Test the present-day service name and separate buyability, and map the difference between Aperture and the specific viewer graph. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [306, 314, 315] | structural | Viewer Graph is the named graph. Aperture’s 2024 targeting sheet and the current Total Video offer are platforms or routes using it. Premion’s 2021 announcement described planned integration. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.28.counterevidence |
| X474 | Check the current commercial operator, US availability and recipient evidence. No reach claim from the blocked launch page was used. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [401] | structural | The documented module uses Lexicon technology with US support. Check named receiving deployments and present commercial terms. No reach claim is taken from the blocked launch page. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.37.limitations |
| X475 | Boundary or alternative interpretation: Check the current commercial operator, US availability and recipient evidence. No reach claim from the blocked launch page was used. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [401] | structural | The documented module uses Lexicon technology with US support. Check named receiving deployments and present commercial terms. No reach claim is taken from the blocked launch page. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.37.counterevidence |
| X476 | Check licensing, deployment geography and what part of the procedure uses an IP edge. See G for TTL scope. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [227, 229, 230, 231] | structural | The US TV media offer includes HouseholdID targeting. A separate resolution licence or independently documented US deployment remains unconfirmed. In VAST serving, IP fallback is off by default and requires exactly one known household at that IP. The Identity API separately allows a probabilistic IP match. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.44.limitations |
| X477 | Boundary or alternative interpretation: Check licensing, deployment geography and what part of the procedure uses an IP edge. See G for TTL scope. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [227, 229, 230, 231] | structural | The US TV media offer includes HouseholdID targeting. A separate resolution licence or independently documented US deployment remains unconfirmed. In VAST serving, IP fallback is off by default and requires exactly one known household at that IP. The Identity API separately allows a probabilistic IP match. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.44.counterevidence |
| X478 | Investigate the commissioned output and whether a current US buyer can purchase it. Graphs behind an alternative do not by themselves disqualify it from cohort C. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [406] | structural | Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.47.limitations |
| X479 | Boundary or alternative interpretation: Investigate the commissioned output and whether a current US buyer can purchase it. Graphs behind an alternative do not by themselves disqualify it from cohort C. | A downstream field repeated, or could imply, superseded wording. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. [406] | structural | Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task. | Separate the documented service, status, unit and boundary from an unresolved hypothesis. | evidence/research/U.json, findings.47.counterevidence |
| X480 | Seek a specific US buyer route, link time horizon, and criteria for household membership. A data feed update cannot establish link expiry. | A downstream field repeated, or could imply, superseded wording. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. [308, 309] | structural | A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. | evidence/research/U.json, findings.46.limitations |
| X481 | Boundary or alternative interpretation: Seek a specific US buyer route, link time horizon, and criteria for household membership. A data feed update cannot establish link expiry. | A downstream field repeated, or could imply, superseded wording. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. [308, 309] | structural | A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule. | The AWS subscription answers separate buyability but not the country or edge-expiry questions. | evidence/research/U.json, findings.46.counterevidence |
| X482 | Dated records 2026-11-05; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | A last-modified stamp dates documentation, not the start of the operator service. [203, 153] | structural | Documentation footer last updated November 11, 2026; read November 11, 2026. | A last-modified stamp dates documentation, not the start of the operator service. | evidence/research/U.json, findings.1.scope_time |
| X483 | Dated records 2025-10-30; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | The company-level acquisition record and later product-specific US publisher document have distinct dates. [369, 132, 177] | structural | Acquisition announced October 30, 2025; publisher requirements page displays February 1, 2026; read November 11, 2026. | The company-level acquisition record and later product-specific US publisher document have distinct dates. | evidence/research/U.json, findings.3.scope_time |
| X484 | Dated records 2024, 2021-02-02; read 2026-11-11 | A downstream field repeated, or could imply, superseded wording. | These documents do not support a uniform rollout date. [306, 314, 315] | structural | Premion agreement announced February 2, 2021; Aperture sales sheet dated 2024; current Total Video offer read November 11, 2026. | These documents do not support a uniform rollout date. | evidence/research/U.json, findings.28.scope_time |
| X485 | [field absent] | A downstream field repeated, or could imply, superseded wording. | Record two dated verifier discoveries separately from the 49 original proposals and the frozen 37-seed denominator. Neither is scored or admitted. [237, 238, 239, 408] | structural | [{"review_date": "2026-11-11", "product": "Audigent Hadron ID", "candidate_cohort": "A", "disposition": "Pending discovery; not an admitted service or a SmartPMP alias.", "documented_route": "Prebid documents a Hadron ID module and namespace. Google Ad Manager lists the signal in a beta partner route.", "us_status": "US commercial scope and a named US deployment were not established in these documents.", "h1_followup": "Confirm current operator, US product availability and independent-party receiving path. If eligible, run the registered product and partner searches for a US match disclosure with a defined population, denominator and method. Retain as pending before any denominator decision.", "source_ids": ["IDN-UV-S05", "IDN-UV-S06", "IDN-A-S06"]}, {"review_date": "2026-11-11", "product": "Fabrick ID by Neustar", "candidate_cohort": "A", "disposition": "Pending discovery; a maintained module page does not prove a current offer or retirement.", "documented_route": "Prebid lists the fabrickId module, Neustar namespace, registration key and optional identifier inputs.", "us_status": "Exact US offering, operator and receiving deployment are unconfirmed.", "h1_followup": "Resolve current operator, relationship to TransUnion services, US availability and receiving path. If eligible, complete the registered product and partner searches for claims with submitted/matched bases and a stated method. Exclude from any confirmed-US denominator pending that evidence.", "source_ids": ["IDN-UV-S08", "IDN-A-S06"]}] | Record two dated verifier discoveries separately from the 49 original proposals and the frozen 37-seed denominator. Neither is scored or admitted. | evidence/research/U.json, verification_discovery_followup |
| X486 | P024 TransUnion TruAudience streaming household identity occupied the sixth profile assignment. P025 Blockgraph Identity Operating System was a listed discovery candidate without a profile. | Identify a US TV-specific household resolution deliverable distinct from general identity matching, segment activation or attribution output. | The 2021 TruAudience launch and the 2024 Comcast agreement document graph infrastructure and audience activation. The reviewed Neustar TV Analytics sheet sells measurement. The US Blockgraph service notice describes selling TV audience resolution. Its GM described network subscriptions in 2020. TVSquared identified IDoS as its converged-TV identity partner in the 2022 NBCUniversal guide. Blockgraph continues to market the household identity capability. [246, 247, 422, 166, 170, 168, 171, 423] | structural | Assign the sixth US cohort D profile to P025 Blockgraph Identity Operating System, slug blockgraph. Keep P024 listed for later eligibility review. The substitution changes the profile roster only. | P024’s distinct purchased household link remained unconfirmed in the reviewed sources. P025 matches the same US buying task, has a named recipient, and comes first by registered product name among remaining US D candidates. Samba is excluded under the declared-interest rule. | Profile assignment P024 to P025; evidence/profiles/blockgraph.json; checks/stage-4.md. The Stage 1 registration remains frozen. |
| X487 | LiveRamp says US ATS.js checks the national GPP signal or a legacy CCPA fallback, and that the UK/EU flow specifies a TCF v2.3 consent check for vendor 97 and purpose 1. | Independent profile review checked the evidence against the named service. | The guide’s November 3 revision labels a UK/EU branch in its diagram. Its prose specifically names v2.3 in the EU and the EEA note, then states vendor 97 and purpose 1 for retrieval. US checks try national GPP, then legacy CCPA with soft enforcement. A missing US string differs from an affirmative opt-in. Passage and scope checked against the reopened original. [182] | supported with qualification | LiveRamp says US ATS.js checks national GPP first and falls back to its legacy CCPA string, whose presence is optional. Its guide groups the UK and EU in one consent branch and specifies TCF v2.3 for the EU/EEA, with vendor 97 and purpose 1 for envelope retrieval. | The guide’s November 3 revision labels a UK/EU branch in its diagram. Its prose specifically names v2.3 in the EU and the EEA note, then states vendor 97 and purpose 1 for retrieval. US checks try national GPP, then legacy CCPA with soft enforcement. A missing US string differs from an affirmative opt-in. Passage and scope checked against the reopened original. | IDN-LR-F05; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X488 | LiveRamp’s Activation documentation makes deletion and opt-out requests available to destinations and describes direct delivery to The Trade Desk for specified audiences and marketplace segments. | Independent profile review checked the evidence against the named service. | The guide covers destination retrieval of files and separately describes delivery to The Trade Desk for advertiser first-party audiences and third-party marketplace segments. It expressly leaves processing to each destination. Making a request available does not establish that recipients have completed it. Passage and scope checked against the reopened original. [139] | supported with qualification | LiveRamp publishes opt-out and deletion requests for activation destinations and describes direct delivery to The Trade Desk for advertiser first-party audiences and third-party marketplace segments. Its guide leaves processing to the destination. | The guide covers destination retrieval of files and separately describes delivery to The Trade Desk for advertiser first-party audiences and third-party marketplace segments. It expressly leaves processing to each destination. Making a request available does not establish that recipients have completed it. Passage and scope checked against the reopened original. | IDN-LR-F07; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X489 | LiveRamp publishes country-specific US terms for SSP Sidecar and RampID use and makes its US DSP/DMP client responsible for handling rights requests within that agreement. | Independent profile review checked the evidence against the named service. | The Sidecar page’s US clause applies subject to an SSP SOW. The DSP/DMP page assigns covered US rights requests to the client and says LiveRamp will give required assistance. These published terms document duties; no processing trace or universal signed contract is shown. The subject of this claim is the content of the contracting party’s own published terms. Passage and scope checked against the reopened original. [183, 140] | supported as stated | LiveRamp publishes country-specific US terms for SSP Sidecar and RampID use and makes its US DSP/DMP client responsible for handling rights requests within that agreement. [Evidence class: direct_record] | The Sidecar page’s US clause applies subject to an SSP SOW. The DSP/DMP page assigns covered US rights requests to the client and says LiveRamp will give required assistance. These published terms document duties; no processing trace or universal signed contract is shown. The subject of this claim is the content of the contracting party’s own published terms. Passage and scope checked against the reopened original. | IDN-LR-F08; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X490 | LiveRamp’s published Activation match-rate formula divides eligible unique records matched to at least one device at a chosen destination by the eligible unique input records. | Independent profile review checked the evidence against the named service. | The numerator is eligible unique identifiable records matched to one or more devices at a specified destination. Non-distributed records are excluded, and the guide expressly has a different Facebook account-based method. This is a measure of file-to-destination addressability. Passage and scope checked against the reopened original. [15] | supported with qualification | For its standard activation rate, LiveRamp divides unique identifiable records matched to one or more online devices at a destination by eligible unique identifiable records, excluding records not being distributed. Facebook uses a separate calculation. | The numerator is eligible unique identifiable records matched to one or more devices at a specified destination. Non-distributed records are excluded, and the guide expressly has a different Facebook account-based method. This is a measure of file-to-destination addressability. Passage and scope checked against the reopened original. | IDN-LR-F11; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X491 | The issuer markets authenticated cookieless inventory via ATS and separately instructs implementers to use a pixel for its cookie-to-RampID mapping-file workflow. | Independent profile review checked the evidence against the named service. | The offer describes cookieless authenticated inventory. The mapping-file page calls for a cookie-sync pixel for cookie mappings. The ATS security guide also expressly describes an optional third-party-cookie endpoint. Neither coexistence nor this endpoint establishes a dated change in Chrome practice. Passage and scope checked against the reopened original. [133, 222, 134] | supported with qualification | LiveRamp describes ATS for authenticated inventory without third-party cookies and also documents an optional third-party-cookie endpoint for ATS. Its separate cookie-to-RampID mapping-file workflow uses a cookie-sync pixel. These sources do not date a change in either workflow. [Additional evidence: IDN-LR-S01] | The offer describes cookieless authenticated inventory. The mapping-file page calls for a cookie-sync pixel for cookie mappings. The ATS security guide also expressly describes an optional third-party-cookie endpoint. Neither coexistence nor this endpoint establishes a dated change in Chrome practice. Passage and scope checked against the reopened original. | IDN-LR-F14; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X492 | LiveRamp’s publisher ATS documents tie service to a statement of work and limit passing the envelope to enabled parties; service fees under its general terms are specified in the order form. | Independent profile review checked the evidence against the named service. | The publisher SOW and its referenced terms govern the licence and recipients. A separate umbrella agreement calls for fees to be identified in the relevant order. On a 2021 earnings call the CEO said the strategy was to drive usage and cross-sales without charging publishers or media-platform partners for ATS. None of these sources fixes a current per-envelope fee. The corrected claim combines written contractual records with a separately attributed historical statement of commercial intent. Passage and scope checked against the reopened original. [352, 353, 345, 354] | supported with qualification | LiveRamp’s publisher ATS form limits envelope use to enabled parties under a statement of work. Its separate general service terms put any associated fees in the order form. In 2021 its CEO said LiveRamp did not intend to charge publishers or media-platform partners for ATS and expected more platform use and cross-sales. [Evidence class: synthesis] [Additional evidence: IDN-VLR-S01] | The publisher SOW and its referenced terms govern the licence and recipients. A separate umbrella agreement calls for fees to be identified in the relevant order. On a 2021 earnings call the CEO said the strategy was to drive usage and cross-sales without charging publishers or media-platform partners for ATS. None of these sources fixes a current per-envelope fee. The corrected claim combines written contractual records with a separately attributed historical statement of commercial intent. Passage and scope checked against the reopened original. | IDN-LR-F16; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X493 | The FTC issued an interim Order to Maintain Assets on October 19 requiring continued access to LiveRamp standalone services, while its separate proposed five-year order remained under public comment at the cutoff. | Independent profile review checked the evidence against the named service. | The Commission actually issued the interim order on October 19. The contemporaneous analysis calls the separate five-year remedy proposed, and the official October 23 Federal Register notice sets November 23 as the deadline. The merger closed November 2, while docket C-4841 still showed pending. Passage and scope checked against the reopened original. [88, 89, 87, 9] | supported as stated | The FTC issued an interim Order to Maintain Assets on October 19 requiring continued access to LiveRamp standalone services, while its separate proposed five-year order remained under public comment at the cutoff. [Additional evidence: IDN-VLR-S02] | The Commission actually issued the interim order on October 19. The contemporaneous analysis calls the separate five-year remedy proposed, and the official October 23 Federal Register notice sets November 23 as the deadline. The merger closed November 2, while docket C-4841 still showed pending. Passage and scope checked against the reopened original. | IDN-LR-F20; data/claims-overlay.json; evidence/verify/liveramp-rampid.json |
| X494 | n/e | Independent profile review checked the evidence against the named service. | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. [15, 134, 59, 136] | structural | (empty) | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. | evidence/profiles/liveramp-rampid.json, capabilities.match_rate_definition.evidence_grade |
| X495 | n/e | Independent profile review checked the evidence against the named service. | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. [136, 59, 181] | structural | (empty) | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. | evidence/profiles/liveramp-rampid.json, capabilities.us_coverage_measure.evidence_grade |
| X496 | n/e | Independent profile review checked the evidence against the named service. | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. [43, 59, 61] | structural | (empty) | No qualifying evidence grade for this unit. The documented adjacent evidence and its grade remain in the qualification. | evidence/profiles/liveramp-rampid.json, capabilities.output_validation.evidence_grade |
| X497 | At the publisher, ATS.js checks US GPP with a legacy fallback. Its UK/EU instructions name TCF v2.3, vendor 97 and purpose 1. US recipient contracts allocate rights-handling duties. The documents screened do not link a specific inbound bid field to an observed DSP refusal decision. | Independent profile review checked the evidence against the named service. | Separate regional heading, explicit version scope, and contract duty. [182, 140, 183] | structural | ATS.js checks US national GPP first and falls back to a legacy CCPA string, whose presence is optional. LiveRamp groups the UK and EU in its consent diagram and specifies v2.3 for the EU/EEA, with vendor 97 and purpose 1 for retrieval. Its US recipient terms assign rights-handling duties. The screened sources show no DSP refusal tied to a particular incoming bid field. | Separate regional heading, explicit version scope, and contract duty. | evidence/profiles/liveramp-rampid.json, capabilities.permission_handling.observed |
| X498 | US country terms explicitly list the USA for SSPs purchasing Sidecar and RampID envelope services. DSP/DMP terms also include US customers. The named NYT and AWS routes establish real integration and a working example respectively; geography of NYT impressions is not broken out. | Independent profile review checked the evidence against the named service. | Separate US offer, reported customer use and illustrated implementation; no price or named AWS deployment inferred. [140, 183, 59, 135] | structural | The SSP Sidecar/RampID and DSP/DMP terms include the United States subject to the applicable agreement. Google describes NYT implementation without reporting US-only impressions. AWS supplies a streaming integration example with no named deployed customer. | Separate US offer, reported customer use and illustrated implementation; no price or named AWS deployment inferred. | evidence/profiles/liveramp-rampid.json, capabilities.us_availability.observed |
| X499 | The publisher-managed method deletes the browser envelope and sets a suppression flag. An adjacent activation route sends opt-outs and deletions to destinations, including The Trade Desk for specified segments. The sources disclose no receipt proving erasure at every ATS recipient. | Independent profile review checked the evidence against the named service. | Design and destination delivery are not universal implemented outcomes. [182, 138, 139, 140] | structural | For a publisher-managed opt-out, LiveRamp instructs publishers to delete the local envelope and keep a suppression flag. Connect separately makes requests available to destinations and describes direct delivery to The Trade Desk for specified first- and third-party segments. Destinations control processing. No screened source documents completed erasure at every ATS recipient. | Design and destination delivery are not universal implemented outcomes. | evidence/profiles/liveramp-rampid.json, capabilities.withdrawal_controls.observed |
| X500 | Guide modified 2026-05-28 | Independent profile review checked the evidence against the named service. | Fresh original footer identifies November 3, not May 28. [182] | structural | Guide last modified November 3, 2026; checked November 11, 2026. | Fresh original footer identifies November 3, not May 28. | evidence/profiles/liveramp-rampid.json, findings.4.scope_time |
| X501 | 2026-05-28 | Independent profile review checked the evidence against the named service. | Original page footer displays Last modified November 3, 2026; no first-publication date is shown. [182] | structural | 2026-11-03 | Original page footer displays Last modified November 3, 2026; no first-publication date is shown. | evidence/profiles/liveramp-rampid.json, sources.2.pub_date |
| X502 | Displayed last-modified date. Issuer describes client hashing, server lookup and Sidecar key handling. Its EU consent version predates the May 2026 privacy guide. | Independent profile review checked the evidence against the named service. | Correct linked reference to the guide’s current version. [134, 182] | structural | Displayed last-modified date. Issuer describes client hashing, server lookup and Sidecar key handling. Its EU consent wording predates the November 3, 2026 privacy-guide revision. | Correct linked reference to the guide’s current version. | evidence/profiles/liveramp-rampid.json, sources.0.reliability_notes |
| X503 | US consent section; UK/EU section states TCF v2.3, vendor and purpose | Independent profile review checked the evidence against the named service. | Separate the exact quote, located in the US paragraph, from the scope of the European text. [182] | structural | For Users in the U.S. (GPP and CCPA), first paragraph; For Users in the UK or EU (GDPR), EU and EEA wording; regional diagram. | Separate the exact quote, located in the US paragraph, from the scope of the European text. | evidence/profiles/liveramp-rampid.json, findings.4.support.0.locator |
| X504 | Furnished Exhibit 99.1, Recent Business Highlights, publisher integrations bullet | Independent profile review checked the evidence against the named service. | The reopened exhibit prints this section title, rather than Recent Business Highlights. [136] | structural | Furnished Exhibit 99.1, p. 3, Additional Business Highlights & Metrics, publisher-domains bullet. | The reopened exhibit prints this section title, rather than Recent Business Highlights. | evidence/profiles/liveramp-rampid.json, findings.8.support.0.locator |
| X505 | RampID Mapping Files, Cookie-Based Mapping Files | Independent profile review checked the evidence against the named service. | The quoted pixel instruction appears in the file-type list. [222] | structural | RampID Mapping Files, Mapping File Types, Cookie file note. | The quoted pixel instruction appears in the file-type list. | evidence/profiles/liveramp-rampid.json, findings.13.support.1.locator |
| X506 | LiveRamp ATS identity resolution use case, Prerequisites; implementation example | Independent profile review checked the evidence against the named service. | Use the exact section heading of the reopened AWS original. [135] | structural | Prerequisites, second bullet; Walkthrough: Enrich ad requests with LiveRamp identity data using MediaTailor Monetization Functions. | Use the exact section heading of the reopened AWS original. | evidence/profiles/liveramp-rampid.json, findings.14.support.0.locator |
| X507 | A US publisher or supply platform obtains the right and interface to issue or handle a RampID envelope and make identity available to an enabled buyer. LiveRamp’s Sidecar turns the wrapped identifier into a DSP-specific RampID. ATS is the implementation path for this bought task. Separately, Known Identity Resolution (IDN-P012) returns AbiliTec Links via API or Known IDs in the cloud. | Independent profile review checked the evidence against the named service. | The API returns the envelope; publishers request and relay it. Explicitly distinguish the SaaS route from resolution. [133, 353, 235] | structural | US publishers can license ATS to request an envelope from LiveRamp, store it and pass it to enabled partners. At an enabled SSP, Sidecar can translate it to a DSP-specific RampID. The separate Known Identity Resolution service (IDN-P012) returns AbiliTec Links by API or Known IDs in the cloud. | The API returns the envelope; publishers request and relay it. Explicitly distinguish the SaaS route from resolution. | evidence/profiles/liveramp-rampid.json, what_is_sold |
| X508 | Publishers license or implement ATS. SSPs contract for Sidecar and envelope handling. DSPs and DMPs have separate country terms. Advertisers can buy enabled inventory through the receiving platforms. New York Times Advertising is a documented publisher user through Google. AWS describes a second integration for a streaming provider with an ATS agreement. | Independent profile review checked the evidence against the named service. | A platform recipe is distinct from a second observed customer integration. An agreement alone does not imply a fee. [133, 353, 140, 183, 59, 135] | structural | Publishers can implement ATS, SSPs can contract for Sidecar, and separate terms cover DSPs and DMPs. Advertisers can buy enabled inventory through those platforms. Google reports that New York Times Advertising used ATS. AWS provides a MediaTailor example requiring an active ATS agreement and a domain placement ID. | A platform recipe is distinct from a second observed customer integration. An agreement alone does not imply a fee. | evidence/profiles/liveramp-rampid.json, who_buys |
| X509 | The publisher SOW licenses limited use of envelopes and requires enabled partners. The US umbrella terms assign fees to the signed order form. The public forms screened show no universal per-envelope rate, buyer CPM or executed customer economics. The issuer reports subscriptions, revenue sharing, usage-based fees and services for the consolidated platform. No standalone RampID revenue is reported in the filings reviewed. | Independent profile review checked the evidence against the named service. | Supply-side participation and the historical commercial strategy must be distinguished from a buyer’s present negotiated price and consolidated results. [353, 345, 352, 7, 344, 354] | structural | In 2021, LiveRamp’s CEO said the company did not intend to charge publishers or media-platform partners for ATS and expected it to drive platform use and cross-sales. The publisher form licenses limited envelope use. General terms put any applicable service fees in the signed order. The reviewed records give no current universal per-envelope price or buyer CPM. They disclose group subscription, revenue-sharing, usage and services income without a standalone RampID line. | Supply-side participation and the historical commercial strategy must be distinguished from a buyer’s present negotiated price and consolidated results. | evidence/profiles/liveramp-rampid.json, how_it_earns |
| X510 | ["IDN-LR-S10", "IDN-LR-S13", "IDN-LR-S15", "IDN-D-S04", "IDN-LR-S16", "IDN-D-S03"] | Independent profile review checked the evidence against the named service. | Cite the newly reviewed CEO transcript on the commercial model. [354] | structural | ["IDN-LR-S10", "IDN-LR-S13", "IDN-LR-S15", "IDN-D-S04", "IDN-LR-S16", "IDN-D-S03", "IDN-VLR-S01"] | Cite the newly reviewed CEO transcript on the commercial model. | evidence/profiles/liveramp-rampid.json, section_evidence.how_it_earns.source_ids |
| X511 | The grant and separate partner agreements still allow approved publishers to monetize eligible inventory. | Independent profile review checked the evidence against the named service. | Identify the pricing counterevidence rather than inferring a paid ATS licence. [353, 345, 352, 354] | structural | In May 2021, the CEO said LiveRamp expected platform usage and cross-sales to benefit without charging publishers or media-platform partners for ATS. That historical position does not set a current order price. | Identify the pricing counterevidence rather than inferring a paid ATS licence. | evidence/profiles/liveramp-rampid.json, findings.15.counterevidence |
| X512 | Provider materials describe both cookieless ATS and cookie-based mapping. Their coexistence does not furnish the matched-version comparison required for H4. | Independent profile review checked the evidence against the named service. | An optional cookie route is documented inside ATS as well as in the separate mapping product. [134, 133, 222] | structural | LiveRamp describes authenticated ATS without third-party cookies, an optional third-party-cookie endpoint within ATS, and a separate cookie-based mapping workflow. The materials do not establish a dated change in the same Chrome path. | An optional cookie route is documented inside ATS as well as in the separate mapping product. | evidence/profiles/liveramp-rampid.json, counterevidence_and_tensions.2 |
| X513 | No eligible before-and-after guide pair for an identical US Chrome workflow identified. Pixel-based mapping and cookieless ATS are separate routes. | Independent profile review checked the evidence against the named service. | Include the vendor-documented ATS cookie endpoint in the H4 screen. [134, 133, 222] | structural | The issuer describes both cookie-independent and optional third-party-cookie routes within ATS, plus a distinct cookie-sync mapping workflow. The screen identified no matched guides documenting a change in the same US Chrome path across the policy event. | Include the vendor-documented ATS cookie endpoint in the H4 screen. | evidence/profiles/liveramp-rampid.json, hypothesis_screen.3.assessment |
| X514 | Connect activation match rate | Independent profile review checked the evidence against the named service. | The adjacent formula has a destination-specific exception for Facebook. [15] | structural | Standard Connect activation match rate | The adjacent formula has a destination-specific exception for Facebook. | evidence/profiles/liveramp-rampid.json, data_tables.0.rows.3.0 |
| X515 | Separate file activation workflow. The formula is not an observed rate for the ATS handoff. | Independent profile review checked the evidence against the named service. | Supply inclusion rule and the named exception. [15] | structural | Unique identifiable records; undistributed records excluded. Facebook has a separate formula. No observed ATS handoff rate. | Supply inclusion rule and the named exception. | evidence/profiles/liveramp-rampid.json, data_tables.0.rows.3.3 |
| X516 | Ownership and reporting clock checked November 11, 2026 New York. Current-status conclusions use the filings. The investor FAQ ticker answer predates the delisting. | Independent profile review checked the evidence against the named service. | The issuer FAQ is undated; only its stale content, not a precise version chronology, is observed. [141, 318, 424] | structural | Ownership and reporting status checked November 11, 2026, New York. Later SEC filings establish delisting. The undated investor FAQ still shows the old ticker. | The issuer FAQ is undated; only its stale content, not a precise version chronology, is observed. | evidence/profiles/liveramp-rampid.json, entity.status_note |
| X517 | AWS documented a streaming workflow using ATS and an active placement agreement. | Independent profile review checked the evidence against the named service. | The prerequisites are separate: an agreement and an identifier, not a placement agreement. [135] | structural | AWS documented a streaming workflow requiring an active ATS agreement and a placement ID for the domain. | The prerequisites are separate: an agreement and an identifier, not a placement agreement. | evidence/profiles/liveramp-rampid.json, timeline.6.event |
| X518 | FTC issued its interim Order to Maintain Assets and released a separate proposed decision for comment. | Independent profile review checked the evidence against the named service. | Verify the official deadline using the statutory notice. [87, 88, 89, 9] | structural | FTC issued an interim Order to Maintain Assets and released a separate proposed decision for comment. The October 23 Federal Register notice set November 23 as the comment deadline. | Verify the official deadline using the statutory notice. | evidence/profiles/liveramp-rampid.json, timeline.8.event |
| X519 | ["IDN-LR-S17", "IDN-LR-S18", "IDN-LR-S19"] | Independent profile review checked the evidence against the named service. | Cite the Federal Register deadline alongside the original FTC documents. [9] | structural | ["IDN-LR-S17", "IDN-LR-S18", "IDN-LR-S19", "IDN-VLR-S02"] | Cite the Federal Register deadline alongside the original FTC documents. | evidence/profiles/liveramp-rampid.json, timeline.8.source_ids |
| X520 | The UID2 FAQ says two participants mapping the same email at the same time obtain the same raw ID. Newly generated advertising tokens differ even for that raw ID. It recommends hourly token refresh and a daily audience check against raw-ID refresh timestamps, with raw IDs changing roughly yearly. | Independent profile review checked the evidence against the named service. | The FAQ matches the same DII at the same instant across participants, differentiates token values by generation, recommends an hourly cadence specifically for manual publisher refresh, and daily audience timestamp checks. It describes approximate yearly raw-ID change. V3 remapping after a timestamp can leave the raw ID unchanged. [144, 146] | supported with qualification | UID2 says participants mapping the same email at the same time get the same raw ID. Each generated advertising token differs. The FAQ recommends hourly refresh for publishers who refresh tokens manually, and daily raw-ID timestamp checks for audience updates. A raw ID changes roughly once a year. [Additional evidence: IDN-UID-S05] | The FAQ matches the same DII at the same instant across participants, differentiates token values by generation, recommends an hourly cadence specifically for manual publisher refresh, and daily audience timestamp checks. It describes approximate yearly raw-ID change. V3 remapping after a timestamp can leave the raw ID unchanged. | IDN-UID-F07; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X521 | The Trade Desk’s UID2 license matrix covers approved API access to generate and maintain identifiers. Under its separate CRM onboarding route, advertisers or their agencies may target within their platform account but do not receive the underlying UID2-to-customer mapping. | Independent profile review checked the evidence against the named service. | The approval matrix distinguishes an API that returns associated identifiers from an accessory CRM service that deposits them into the platform account and does not provide the customer a mapping. The separate operator reference expressly offers public-operator use at no cost to the participant, under contract with credentials. A private operator supplies its own resources. The disclosed public-operator price cannot be extended to TTD media buys or third-party services. [188, 203] | supported with qualification | The Trade Desk’s license matrix distinguishes approved API access from CRM onboarding. CRM onboarding permits targeting in the platform account without returning the IDs or customer mapping. UID2 also says a contracted participant can use a Public Operator at no charge. Private operators provide their own hosting and engineering resources. [Additional evidence: IDN-U-S03] | The approval matrix distinguishes an API that returns associated identifiers from an accessory CRM service that deposits them into the platform account and does not provide the customer a mapping. The separate operator reference expressly offers public-operator use at no cost to the participant, under contract with credentials. A private operator supplies its own resources. The disclosed public-operator price cannot be extended to TTD media buys or third-party services. | IDN-UID-F09; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X522 | UID2 lists production operator endpoints in Ohio and Oregon. Roku describes using The Trade Desk’s DSP to activate against Roku first-party data through Roku Exchange. | Independent profile review checked the evidence against the named service. | US East (Ohio) and US West (Oregon) are marked production separately from the integration endpoint. In its August 2024 account, Roku announced UID2 adoption and described how TTD advertisers could buy Roku Media through Roku Exchange. The article identifies available activation rather than disclosing bid logs for a particular advertiser. [153, 154] | supported with qualification | UID2 lists production operator endpoints in Ohio and Oregon. In August 2024, Roku announced UID2 adoption across its premium streaming inventory and described how The Trade Desk advertisers could activate against Roku first-party data through Roku Exchange. | US East (Ohio) and US West (Oregon) are marked production separately from the integration endpoint. In its August 2024 account, Roku announced UID2 adoption and described how TTD advertisers could buy Roku Media through Roku Exchange. The article identifies available activation rather than disclosing bid logs for a particular advertiser. | IDN-UID-F10; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X523 | A site-level opt-out stops that participant from generating UID2 for the user and is not relayed across UID2. A consumer can use an email address or phone number in the global portal. The operator sends opt-out responses when publishers generate or refresh and distributes a separate DSP webhook. | Independent profile review checked the evidence against the named service. | The US opt-out guide requires a participant to stop generating UID2 after its own opt-out; the local decision is not relayed to UID2. Portal requests apply to one input at a time. The operator describes per-identifier responses on token generate/refresh and DSP webhook dispatch. Implementation and participant compliance are not observed. [149, 150] | supported with qualification | UID2 requires a participant to stop generating the identifier after a site-level opt-out. That request is not relayed to UID2. For ecosystem-wide suppression, a consumer submits each email or phone number separately. The guide describes opt-out responses to publisher generation or refresh requests and a webhook to DSPs. | The US opt-out guide requires a participant to stop generating UID2 after its own opt-out; the local decision is not relayed to UID2. Portal requests apply to one input at a time. The operator describes per-identifier responses on token generate/refresh and DSP webhook dispatch. Implementation and participant compliance are not observed. | IDN-UID-F12; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X524 | After WebKit described its identity-resolution network block on October 19, 2026, UID2’s Prebid guide identified client-side uidapi.com requests as blocked on all iOS 27+ browsers and Safari on macOS 27+. It directs affected publishers to server-side token handling. | Independent profile review checked the evidence against the named service. | The UID2 release log specifically dates its product warning to October 9, 2026. The WebKit announcement is dated October 19. The current Prebid guide confirms the affected browsers and advises server-side generation and refresh. The product documentation predates the cited WebKit announcement. The original text reverses the dates. Neither source measures affected impressions. [151, 152, 145] | contradicted | On October 9, 2026, UID2 documented that browser requests to uidapi.com were blocked in all iOS 27+ browsers and in Safari on macOS 27+. Its Prebid guide directs publishers on those paths to generate and refresh tokens server-side. On October 19, WebKit described its broader domain-level protection. [Additional evidence: IDN-UID-S04] | The UID2 release log specifically dates its product warning to October 9, 2026. The WebKit announcement is dated October 19. The current Prebid guide confirms the affected browsers and advises server-side generation and refresh. The product documentation predates the cited WebKit announcement. The original text reverses the dates. Neither source measures affected impressions. | IDN-UID-F15; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X525 | The UID2 administrator’s privacy notice covers its own role, including sharing an opt-out choice with participants. The publisher/DSP guides describe suppression, but neither the FAQ’s sample bid request nor the DSP guide demonstrates per-impression transfer and enforcement of a TCF or GPP string. | Independent profile review checked the evidence against the named service. | The administrator policy describes opt-out recording and onward notices. The FAQ sample illustrates identifier carriage and the DSP guide specifies an out-of-band withdrawal action. No observed TCF/GPP string evaluation per impression is shown. Separately, the UID2 client DPA assigns processing roles for Company Personal Data, cooperation on rights requests and conditional data destruction. It provides no observed downstream erasure. [150, 144, 186, 187] | supported with qualification | The UID2 administrator says it records opt-outs and shares them with participants. Its US-relevant client addendum separately describes the company and The Trade Desk as controller and processor for covered customer data. These terms and the bidder guide describe obligations. The FAQ’s sample bid request and DSP guide do not demonstrate per-impression TCF or GPP enforcement. [Additional evidence: IDN-VUID-S02] | The administrator policy describes opt-out recording and onward notices. The FAQ sample illustrates identifier carriage and the DSP guide specifies an out-of-band withdrawal action. No observed TCF/GPP string evaluation per impression is shown. Separately, the UID2 client DPA assigns processing roles for Company Personal Data, cooperation on rights requests and conditional data destruction. It provides no observed downstream erasure. | IDN-UID-F23; data/claims-overlay.json; evidence/verify/ttd-uid2.json |
| X526 | Same slide, footnote to Identity Alliance | Independent profile review checked the evidence against the named service. | The graph and internal-resource statements appear on adjoining slides. [425] | structural | Q4 2024 investor presentation, following slide headed “A better foundation for identity”, footnote on Identity Alliance | The graph and internal-resource statements appear on adjoining slides. | evidence/profiles/ttd-uid2.json, findings.2.support.1.locator |
| X527 | Production \| AWS US East (Ohio) | Independent profile review checked the evidence against the named service. | Quote the actual cell without a reconstructed Markdown table delimiter. [153] | structural | AWS US East (Ohio) | Quote the actual cell without a reconstructed Markdown table delimiter. | evidence/profiles/ttd-uid2.json, findings.9.support.0.passage |
| X528 | The UID2 Operator Service distributes information on all opted-out users to DSPs via a webhook provided for the purpose. | Independent profile review checked the evidence against the named service. | Use the minimum exact words from the table, already quoted in the standards stream. [149] | structural | via a webhook provided for the purpose | Use the minimum exact words from the table, already quoted in the standards stream. | evidence/profiles/ttd-uid2.json, findings.11.support.0.passage |
| X529 | WebKit blocks client-side requests to `uidapi.com`. | Independent profile review checked the evidence against the named service. | Avoid adding Markdown backticks to the issuer’s text. The joined code element immediately names the domain. [152] | structural | WebKit blocks client-side requests to | Avoid adding Markdown backticks to the issuer’s text. The joined code element immediately names the domain. | evidence/profiles/ttd-uid2.json, findings.14.support.1.passage |
| X530 | UID2 (The Trade Desk)  \| 73.2M · 51%  \| 94.4M · 66%  \| +15 pts | Independent profile review checked the evidence against the named service. | Use an exact cell from the original table. The locator identifies the baseline and enriched columns. [31] | structural | 94.4M · 66% | Use an exact cell from the original table. The locator identifies the baseline and enriched columns. | evidence/profiles/ttd-uid2.json, findings.15.support.0.passage |
| X531 | 2026-10-29 | Independent profile review checked the evidence against the named service. | The independently reopened page footer dates this version 2026-11-11. This is a revision date. [143] | structural | 2026-11-11 | The independently reopened page footer dates this version 2026-11-11. This is a revision date. | evidence/profiles/ttd-uid2.json, sources.1.pub_date |
| X532 | 2026-11-05 | Independent profile review checked the evidence against the named service. | The independently reopened page footer dates this version 2026-11-11. This is a revision date. [146] | structural | 2026-11-11 | The independently reopened page footer dates this version 2026-11-11. This is a revision date. | evidence/profiles/ttd-uid2.json, sources.4.pub_date |
| X533 | 2026-10-26 | Independent profile review checked the evidence against the named service. | The independently reopened page footer dates this version 2026-11-02. This is a revision date. [368] | structural | 2026-11-02 | The independently reopened page footer dates this version 2026-11-02. This is a revision date. | evidence/profiles/ttd-uid2.json, sources.6.pub_date |
| X534 | Release 2025-07-11. Migration documentation updated 2026-11-05. | Independent profile review checked the evidence against the named service. | The footer of the reopened version differs from the original profile date. [145, 146] | structural | V3 release July 11, 2025. Current v3 guide last updated November 11, 2026. | The footer of the reopened version differs from the original profile date. | evidence/profiles/ttd-uid2.json, findings.7.scope_time |
| X535 | The matrix also describes approved private operator deployment. It gives no fee schedule or executed buyer-specific terms. | Independent profile review checked the evidence against the named service. | A bounded absence from one matrix must not hide the stated public-operator price. [203, 188] | structural | The license matrix provides no fee schedule. The operator guide separately discloses a free public-operator route and participant-funded private infrastructure. Ancillary fees, commercial arrangements and product margin remain undisclosed in these records. | A bounded absence from one matrix must not hide the stated public-operator price. | evidence/profiles/ttd-uid2.json, findings.8.limitations |
| X536 | WebKit 2026-10-19. UID2 integration guide read 2026-11-11. | Independent profile review checked the evidence against the named service. | The dated product notice resolves the event order. [145, 151, 152] | structural | UID2 product notice October 9, 2026. WebKit post October 19, 2026. Prebid guide reopened November 11, 2026. | The dated product notice resolves the event order. | evidence/profiles/ttd-uid2.json, findings.14.scope_time |
| X537 | None found in the cited record for this narrow description. | Independent profile review checked the evidence against the named service. | The dated notice contradicts the original sequence. [145, 151] | structural | UID2 dated its own notice October 9, ten days before the WebKit post. No adoption or lost-impression count was disclosed. | The dated notice contradicts the original sequence. | evidence/profiles/ttd-uid2.json, findings.14.counterevidence |
| X538 | None found in the cited record for this narrow description. | Independent profile review checked the evidence against the named service. | Include the opposing position and procedural limit stated in the filing. [346] | structural | The company disputes the pending claims. The court dismissed the declaratory-relief claim with prejudice, and one plaintiff was later dismissed without prejudice. | Include the opposing position and procedural limit stated in the filing. | evidence/profiles/ttd-uid2.json, findings.20.counterevidence |
| X539 | A publisher may send other consent fields. This finding is limited to the named guide and sample, not a claim about every implementation. | Independent profile review checked the evidence against the named service. | The addendum applies to company personal data, distinct from the administrator’s opt-out records. [187, 150, 186] | structural | The bidder guide and FAQ do not demonstrate purpose-string handling on a US impression. The separate client DPA assigns processing obligations for covered company data. It is not an audit or a promise of erasure throughout the recipient chain. | The addendum applies to company personal data, distinct from the administrator’s opt-out records. | evidence/profiles/ttd-uid2.json, findings.22.limitations |
| X540 | ["IDN-UID-S03"] | Independent profile review checked the evidence against the named service. | Link the decisive comparison or new fact to its source. [146] | structural | ["IDN-UID-S03", "IDN-UID-S05"] | Link the decisive comparison or new fact to its source. | evidence/profiles/ttd-uid2.json, findings.6.source_ids |
| X541 | [{"source_id": "IDN-UID-S03", "passage": "the same raw UID2 in response", "locator": "FAQs, Does the same DII always result in the same raw UID2? and token uniqueness and rotation policy"}, {"source_id": "IDN-UID-S03", "passage": "The recommended refresh interval is hourly.", "locator": "FAQs for publishers, If I choose to manually refresh the token…"}, {"source_id": "IDN-UID-S03", "passage": "roughly once per year", "locator": "FAQs for advertisers and data providers, How often should raw UID2s be refreshed for incremental updates?"}] | Independent profile review checked the evidence against the named service. | Supply the exact supporting evidence for the corrected wording. [146] | structural | [{"source_id": "IDN-UID-S03", "passage": "the same raw UID2 in response", "locator": "FAQs, Does the same DII always result in the same raw UID2? and token uniqueness and rotation policy"}, {"source_id": "IDN-UID-S03", "passage": "The recommended refresh interval is hourly.", "locator": "FAQs for publishers, If I choose to manually refresh the token…"}, {"source_id": "IDN-UID-S03", "passage": "roughly once per year", "locator": "FAQs for advertisers and data providers, How often should raw UID2s be refreshed for incremental updates?"}, {"source_id": "IDN-UID-S05", "passage": "might or might not change", "locator": "Response body elements, r field. The timestamp can advance without a raw-ID change."}] | Supply the exact supporting evidence for the corrected wording. | evidence/profiles/ttd-uid2.json, findings.6.support |
| X542 | ["IDN-UID-S11"] | Independent profile review checked the evidence against the named service. | Link the decisive comparison or new fact to its source. [203] | structural | ["IDN-UID-S11", "IDN-VUID-S01"] | Link the decisive comparison or new fact to its source. | evidence/profiles/ttd-uid2.json, findings.8.source_ids |
| X543 | [{"source_id": "IDN-UID-S11", "passage": "will not receive a mapping of the Company Data to associated UID2s.", "locator": "v10.1.24, UID2 API and CRM Data Onboarding Service columns"}] | Independent profile review checked the evidence against the named service. | Supply the exact supporting evidence for the corrected wording. [203] | structural | [{"source_id": "IDN-UID-S11", "passage": "will not receive a mapping of the Company Data to associated UID2s.", "locator": "v10.1.24, UID2 API and CRM Data Onboarding Service columns"}, {"source_id": "IDN-VUID-S01", "passage": "There is no cost, to the participant, for using a Public Operator.", "locator": "Public Operator: Benefits, then Private Operators and Summary"}] | Supply the exact supporting evidence for the corrected wording. | evidence/profiles/ttd-uid2.json, findings.8.support |
| X544 | ["IDN-E-S14", "IDN-E-S16"] | Independent profile review checked the evidence against the named service. | Link the decisive comparison or new fact to its source. [145] | structural | ["IDN-E-S14", "IDN-E-S16", "IDN-UID-S04"] | Link the decisive comparison or new fact to its source. | evidence/profiles/ttd-uid2.json, findings.14.source_ids |
| X545 | [{"source_id": "IDN-E-S14", "passage": "Safari and WebKit block network access to identity resolution infrastructure to prevent cross-site tracking.", "locator": "Blocking Identity Resolution Infrastructure, October 19, protection scope"}, {"source_id": "IDN-E-S16", "passage": "WebKit blocks client-side requests to", "locator": "Web integration limitations on iOS 27 and macOS 27, affected browsers and Action required"}] | Independent profile review checked the evidence against the named service. | Supply the exact supporting evidence for the corrected wording. [145] | structural | [{"source_id": "IDN-E-S14", "passage": "Safari and WebKit block network access to identity resolution infrastructure to prevent cross-site tracking.", "locator": "Blocking Identity Resolution Infrastructure, October 19, protection scope"}, {"source_id": "IDN-E-S16", "passage": "WebKit blocks client-side requests to", "locator": "Web integration limitations on iOS 27 and macOS 27, affected browsers and Action required"}, {"source_id": "IDN-UID-S04", "passage": "October 9, 2026", "locator": "Documentation updates, Q4 2026, iOS 27 and macOS 27 web integration limitations."}] | Supply the exact supporting evidence for the corrected wording. | evidence/profiles/ttd-uid2.json, findings.14.support |
| X546 | ["IDN-UID-S18", "IDN-UID-S03", "IDN-A-S09"] | Independent profile review checked the evidence against the named service. | Link the decisive comparison or new fact to its source. [187] | structural | ["IDN-UID-S18", "IDN-UID-S03", "IDN-A-S09", "IDN-VUID-S02"] | Link the decisive comparison or new fact to its source. | evidence/profiles/ttd-uid2.json, findings.22.source_ids |
| X547 | [{"source_id": "IDN-UID-S18", "passage": "only in its role as UID2 administrator", "locator": "Introduction, scope and Disclosure of data by UID2 administrator"}, {"source_id": "IDN-UID-S03", "passage": "\"source\": \"uidapi.com\"", "locator": "FAQ, What does a UID2 token look like in the bidstream?, entire example"}, {"source_id": "IDN-A-S09", "passage": "Bidding opt-out logic", "locator": "Bidding opt-out logic and separate opt-out webhook"}] | Independent profile review checked the evidence against the named service. | Supply the exact supporting evidence for the corrected wording. [187] | structural | [{"source_id": "IDN-UID-S18", "passage": "only in its role as UID2 administrator", "locator": "Introduction, scope and Disclosure of data by UID2 administrator"}, {"source_id": "IDN-UID-S03", "passage": "\"source\": \"uidapi.com\"", "locator": "FAQ, What does a UID2 token look like in the bidstream?, entire example"}, {"source_id": "IDN-A-S09", "passage": "Bidding opt-out logic", "locator": "Bidding opt-out logic and separate opt-out webhook"}, {"source_id": "IDN-VUID-S02", "passage": "TD is a Processor or the equivalent of Company Personal Data", "locator": "§2.2 Processor and Controller Responsibilities; §§2.3, 3, 6 and Exhibit A"}] | Supply the exact supporting evidence for the corrected wording. | evidence/profiles/ttd-uid2.json, findings.22.support |
| X548 | A publisher requests an encrypted advertising token based on an email or phone credential. An approved advertiser or data provider can map customer data to raw IDs. A participating DSP can recognize a token at bid time. The service unit is the authorized issuance, transport and use of UID2. Identity Alliance is a separate cross-device graph. [IDN-UID-F03, IDN-UID-F05, IDN-UID-F06, IDN-UID-F09] | Independent profile review checked the evidence against the named service. | Buyers receive different outputs on the API and CRM routes. [142, 143, 425, 188] | structural | Publishers can obtain an encrypted advertising token from an email address or phone number. Approved mappers can receive raw IDs. CRM onboarding instead puts IDs into a Trade Desk account for targeting without returning them to the client. The separately branded Identity Alliance is a cross-device graph. [IDN-UID-F03, IDN-UID-F05, IDN-UID-F06, IDN-UID-F09] | Buyers receive different outputs on the API and CRM routes. | evidence/profiles/ttd-uid2.json, what_is_sold |
| X549 | The Trade Desk says it devoted internal resources to building UID2. Its buying platform earns fees linked to client spend and from services and data. The opened license matrix sets out access routes. Neither that matrix nor the reviewed filings disclose a standalone UID2 price, profit or revenue line. [IDN-UID-F02, IDN-UID-F03, IDN-UID-F09] | Independent profile review checked the evidence against the named service. | The public-operator reference provides the price for one defined service. Company-level revenue does not disclose UID2 earnings. [203, 188, 331, 346] | structural | UID2 says a participant with a contract and credentials can use a Public Operator at no charge. Private operators supply their own infrastructure and engineering. The Trade Desk earns platform fees linked to client spend and earns fees for services and data. Its filings do not disclose UID2 revenue or profit. [IDN-UID-F02, IDN-UID-F03, IDN-UID-F09] | The public-operator reference provides the price for one defined service. Company-level revenue does not disclose UID2 earnings. | evidence/profiles/ttd-uid2.json, how_it_earns |
| X550 | What are the product-level fees, operating cost and financial contribution? | Independent profile review checked the evidence against the named service. | The participant price for using a Public Operator is already documented. [203, 188, 331] | structural | What integration, hosting, CRM onboarding or partner charges apply on the chosen route, and what are the operator’s product-level costs and financial contribution? | The participant price for using a Public Operator is already documented. | evidence/profiles/ttd-uid2.json, open_questions.4 |
| X551 | The issuer needs an adequate input credential, operator keys and salts, and publisher and recipient integrations. Site opt-outs and global suppression reach participants differently. The WebKit change limits browser-side generation on the named Apple paths. The publisher guide directs affected integrations to a server-side option. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F07, IDN-UID-F12, IDN-UID-F13, IDN-UID-F15] | Independent profile review checked the evidence against the named service. | Distinguish the credential and affected route from individual-wide withdrawal. [142, 143, 145, 149, 152] | structural | The operator needs an email address or phone number, or the permitted hash, and current keys and salts. The chosen publisher and recipient must support the route. A site opt-out binds that participant. The global portal requires a separate request for each email or phone. On the affected Apple browser paths, UID2 directs publishers to generate and refresh tokens server-side. [IDN-UID-F05, IDN-UID-F06, IDN-UID-F07, IDN-UID-F12, IDN-UID-F13, IDN-UID-F15] | Distinguish the credential and affected route from individual-wide withdrawal. | evidence/profiles/ttd-uid2.json, supply_and_data_dependencies |
| X552 | A token can identify the same input across integrations. The academic and commercial records reviewed do not test the resulting person link against independent labels. [IDN-UID-F07, IDN-UID-F18, IDN-UID-F22] | Independent profile review checked the evidence against the named service. | Encrypted token values vary even for the same raw identifier. [144, 148] | structural | The raw ID can match the same input across operators at the same time. Individual advertising tokens vary. The reviewed academic and commercial sources do not compare inferred people against independent labels. [IDN-UID-F07, IDN-UID-F18, IDN-UID-F22] | Encrypted token values vary even for the same raw identifier. | evidence/profiles/ttd-uid2.json, limits.0 |
| X553 | A US campaign target and a retailer CRM file give no national UID2 adoption share. Exact retailer eligibility and rounding remain unspecified. [IDN-UID-F16, IDN-UID-F17, IDN-UID-F22] | Independent profile review checked the evidence against the named service. | The edition is US. HP’s campaign geography is not expressly stated in its case. [147, 31] | structural | HP reports reach for one campaign in a US-edition report. Deep Sync reports matches in one US retailer’s CRM. Neither publishes a national UID2 adoption share. Retailer eligibility, deduplication and rounding remain unspecified. [IDN-UID-F16, IDN-UID-F17, IDN-UID-F22] | The edition is US. HP’s campaign geography is not expressly stated in its case. | evidence/profiles/ttd-uid2.json, limits.1 |
| X554 | WebKit describes its identity-resolution network block. UID2 later documents the affected client-side publisher path. | Independent profile review checked the evidence against the named service. | UID2’s dated notice predates the WebKit article. [145, 151, 152] | structural | WebKit describes its broader domain-level block. UID2’s October 9 product notice had already identified the affected client-side paths and advised server-side generation and refresh. | UID2’s dated notice predates the WebKit article. | evidence/profiles/ttd-uid2.json, timeline.9.event |
| X555 | ["IDN-E-S14", "IDN-E-S16"] | Independent profile review checked the evidence against the named service. | The release log is the source of the earlier date. [145] | structural | ["IDN-E-S14", "IDN-E-S16", "IDN-UID-S04"] | The release log is the source of the earlier date. | evidence/profiles/ttd-uid2.json, timeline.9.source_ids |
| X556 | Deep Sync, marketer of CRM enrichment. Advertiser unnamed. | Independent profile review checked the evidence against the named service. | Publication and financial interest are known; an undisclosed funder is not identified. [31] | structural | Deep Sync published the case and sells the enrichment service. The retailer is unnamed. Separate study funding was not disclosed. | Publication and financial interest are known; an undisclosed funder is not identified. | evidence/profiles/ttd-uid2.json, performance_evidence.0.sponsor |
| X557 | The Trade Desk Intelligence, with the participating advertiser HP. | Independent profile review checked the evidence against the named service. | The report identifies its researchers and interviewees, not HP as a financial sponsor. [147] | structural | The Trade Desk Intelligence produced the report with Hypothesis Group. HP supplied the campaign account. Separate campaign-study funding was not disclosed. | The report identifies its researchers and interviewees, not HP as a financial sponsor. | evidence/profiles/ttd-uid2.json, performance_evidence.1.sponsor |
| X558 | Interested platform case report. Device-registration data from HP’s CDP, named publisher destinations, and estimated target household universe. No control group or campaign dates disclosed. | Independent profile review checked the evidence against the named service. | Distinguish the interview study and its fieldwork from one advertiser’s campaign. [147] | structural | HP case in a report based on Q1 2024 interviews with seven brand and agency leaders in the US, UK, Thailand and Hong Kong. The HP campaign used device-registration data. Campaign dates, geography and a control group are unspecified. | Distinguish the interview study and its fieldwork from one advertiser’s campaign. | evidence/profiles/ttd-uid2.json, performance_evidence.1.design |
| X559 | Interested platform case report. Device-registration data from HP’s CDP, named publisher destinations, and estimated target household universe. No control group or campaign dates disclosed. | Independent profile review checked the evidence against the named service. | Correct the funding/method boundary consistently with the performance section. [147] | structural | Platform-published HP case in an interview report by The Trade Desk Intelligence and Hypothesis Group. Q1 2024 fieldwork covered seven brand and agency leaders in four markets. HP’s campaign dates, geography and control group were unspecified. | Correct the funding/method boundary consistently with the performance section. | evidence/profiles/ttd-uid2.json, findings.16.study_design |
| X560 | Media-buying platform and UID2 developer. HP is the featured advertiser. | Independent profile review checked the evidence against the named service. | The report identifies joint research and HP participation, not independent truth measurement or HP financing. [147] | structural | The Trade Desk Intelligence and Hypothesis Group conducted the research. HP is the advertiser featured in this campaign. Separate campaign funding was not stated. | The report identifies joint research and HP participation, not independent truth measurement or HP financing. | evidence/profiles/ttd-uid2.json, sources.11.funding_or_interest |
| X561 | Opened issuer PDF in browser. The identity slide describes internal development resources and labels the separate proprietary cross-device graph. No dollar budget. | Independent profile review checked the evidence against the named service. | The verified reading route and two adjacent slides are more specific. [425] | structural | Read the original Q4 2024 deck through its issuer-published indexed PDF after a direct download returned HTTP 403. The “what and why” slide covers internal investment. The following slide labels the proprietary Identity Alliance graph. Neither discloses an investment amount. | The verified reading route and two adjacent slides are more specific. | evidence/profiles/ttd-uid2.json, sources.9.reliability_notes |
| X562 | The publisher FAQ recommends hourly advertising-token refresh using the indicated time. Advertisers are advised to check raw-ID refresh times daily. Raw IDs change roughly yearly. In v3 the previous raw ID is available for 90 days after rotation. These are distinct timing rules. | Independent profile review checked the evidence against the named service. | A manual publisher refresh recommendation is distinct from expiry, the daily mapper check and an exact time of rotation. The v3 timestamp may advance even when the ID stays the same. [144, 145, 146] | structural | For manual publisher token refresh, UID2 recommends hourly checks of the indicated time. It recommends daily raw-ID refresh timestamp checks for audience updates. Raw IDs change roughly yearly, and may remain unchanged across refresh checks. V3 can return the prior ID for 90 days after rotation. | A manual publisher refresh recommendation is distinct from expiry, the daily mapper check and an exact time of rotation. The v3 timestamp may advance even when the ID stays the same. | evidence/profiles/ttd-uid2.json, capabilities.refresh_expiry.observed |
| X563 | The participant must stop generating UID2 for a site-level opt-out. The global service distributes choices separately. The bidder checks the raw ID at bidding time. The sampled EID example and DSP guide do not establish end-to-end TCF or GPP enforcement for a US impression. | Independent profile review checked the evidence against the named service. | A contract or guide defines responsibilities. Neither establishes execution in any observed auction. TTD is processor for covered Company Personal Data on the cited UID2 client route, separately from administration and the EUID framework. [149, 150, 186, 144, 187] | structural | The UID2 client addendum assigns processing duties for covered customer data. The opt-out guide requires the participant to stop local issuance after a site opt-out. DSPs must screen opted-out raw IDs during bidding. The sample request and DSP guide do not show actual per-impression TCF or GPP enforcement. | A contract or guide defines responsibilities. Neither establishes execution in any observed auction. TTD is processor for covered Company Personal Data on the cited UID2 client route, separately from administration and the EUID framework. | evidence/profiles/ttd-uid2.json, capabilities.permission_handling.observed |
| X564 | ["IDN-A-S08", "IDN-UID-S18", "IDN-A-S09", "IDN-UID-S03"] | Independent profile review checked the evidence against the named service. | A contract or guide defines responsibilities. Neither establishes execution in any observed auction. TTD is processor for covered Company Personal Data on the cited UID2 client route, separately from administration and the EUID framework. [149, 150, 186, 144, 187] | structural | ["IDN-A-S08", "IDN-UID-S18", "IDN-A-S09", "IDN-UID-S03", "IDN-VUID-S02"] | A contract or guide defines responsibilities. Neither establishes execution in any observed auction. TTD is processor for covered Company Personal Data on the cited UID2 client route, separately from administration and the EUID framework. | evidence/profiles/ttd-uid2.json, capabilities.permission_handling.source_ids |
| X565 | The FAQ illustrates user.ext.eids, source uidapi.com and the encrypted token under uids[].id. A bidder-permissioned DSP decrypts the token, checks opt-out and can choose to bid. Approved CRM onboarding users need not receive the underlying mapping. | Independent profile review checked the evidence against the named service. | The guide prescribes screening and the licensing matrix expressly bars returning the underlying IDs or mapping under its CRM onboarding route. [143, 144, 188, 186] | structural | The FAQ shows an encrypted token in user.ext.eids[].uids[].id with source uidapi.com. An approved DSP may decrypt it and must honor opt-outs before bidding on it. The separate CRM onboarding route allows in-account targeting and expressly withholds the IDs and mapping. | The guide prescribes screening and the licensing matrix expressly bars returning the underlying IDs or mapping under its CRM onboarding route. | evidence/profiles/ttd-uid2.json, capabilities.recipient_interface.observed |
| X566 | Roku describes a DSP-to-Roku Exchange activation route. The Trade Desk describes HP using Disney+ and Hulu via its DSP. These are named destinations with concrete uses. The reports do not provide recipient-by-recipient auction logs. | Independent profile review checked the evidence against the named service. | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. [153, 154, 147] | structural | Roku’s August 2024 announcement describes a TTD–Roku Exchange activation route. The 2024 platform report says HP used TTD to reach audiences on Disney+ and Hulu. The reports do not identify those properties as distinct downstream UID2-decrypting operators or quantify cutoff-period use. | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. | evidence/profiles/ttd-uid2.json, capabilities.destination_coverage.observed |
| X567 | C | Independent profile review checked the evidence against the named service. | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. [153, 154, 147] | structural | D | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. | evidence/profiles/ttd-uid2.json, capabilities.destination_coverage.evidence_grade |
| X568 | 2024-08-01 | Independent profile review checked the evidence against the named service. | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. [153, 154, 147] | structural | 2026-11-11 | The decisive evidence is two commercially interested promotional reports. Roku describes an activation route and HP describes a campaign via TTD. Disney+ and Hulu placement does not show two separate operators consumed the UID2. The reports lack cutoff-period recipient logs. | evidence/profiles/ttd-uid2.json, capabilities.destination_coverage.date |
| X569 | The US-edition HP report gives campaign reach of 8% of an estimated 12.5 million target households. The national UID2 eligible base, measurement window and deduplication method were not established in the records reviewed. Report year 2024. Reviewed at the cutoff. | Independent profile review checked the evidence against the named service. | The US edition does not specify the HP campaign geography or measurement period; its target is not a national UID2 eligible base. The article supplies evidence for the attributed case, not a live national measurement capability. [147, 31, 148] | structural | In its US-edition report the platform says HP reached 8% of an estimated 12.5 million campaign target households. The case gives no campaign geography or dates, national UID2 eligible base or deduplication rule. The reviewed sources do not establish coverage of the US product population. | The US edition does not specify the HP campaign geography or measurement period; its target is not a national UID2 eligible base. The article supplies evidence for the attributed case, not a live national measurement capability. | evidence/profiles/ttd-uid2.json, capabilities.us_coverage_measure.observed |
| X570 | live | Independent profile review checked the evidence against the named service. | The US edition does not specify the HP campaign geography or measurement period; its target is not a national UID2 eligible base. The article supplies evidence for the attributed case, not a live national measurement capability. [147, 31, 148] | structural | unconfirmed | The US edition does not specify the HP campaign geography or measurement period; its target is not a national UID2 eligible base. The article supplies evidence for the attributed case, not a live national measurement capability. | evidence/profiles/ttd-uid2.json, capabilities.us_coverage_measure.availability |
| X571 | UID2 lists production operator endpoints in Ohio and Oregon. Its license matrix describes approved access for publishers, buyers and data partners. A US-edition case names HP’s use on Disney+ and Hulu. Production infrastructure alone would not establish campaign reach. | Independent profile review checked the evidence against the named service. | The production region table and approved-access matrix, complemented by the named US retailer case, support a US route. HP’s report edition cannot itself establish campaign geography. [188, 153, 154, 147, 31] | structural | UID2 lists production endpoints in Ohio and Oregon. The license matrix describes approved access. Deep Sync reports UID2 matches for one US retailer, and Roku describes an activation route. None of these establishes the size of the national UID2 population. | The production region table and approved-access matrix, complemented by the named US retailer case, support a US route. HP’s report edition cannot itself establish campaign geography. | evidence/profiles/ttd-uid2.json, capabilities.us_availability.observed |
| X572 | ["IDN-UID-S11", "IDN-U-S04", "IDN-I-S08", "IDN-UID-S12"] | Independent profile review checked the evidence against the named service. | The production region table and approved-access matrix, complemented by the named US retailer case, support a US route. HP’s report edition cannot itself establish campaign geography. [188, 153, 154, 147, 31] | structural | ["IDN-UID-S11", "IDN-U-S04", "IDN-I-S08", "IDN-UID-S12", "IDN-C-S30"] | The production region table and approved-access matrix, complemented by the named US retailer case, support a US route. HP’s report edition cannot itself establish campaign geography. | evidence/profiles/ttd-uid2.json, capabilities.us_availability.source_ids |
| X573 | A global portal accepts an email or phone number. The operator returns opt-out at token generation or refresh and sends a DSP webhook, typically within minutes. A DSP must cease UID2 bidding, may still bid using another identifier, and receives no retrospective recall guarantee in the cited guides. | Independent profile review checked the evidence against the named service. | One input per portal request. The guide gives a typical webhook dispatch time and a bidder requirement. The DPA concerns customer data under a specific agreement, with exceptions; it does not establish receipt, execution or erasure by every recipient. [149, 150, 186, 187] | structural | The portal takes a separate request for each email or phone. The guide describes a publisher opt-out response and a DSP webhook, typically within minutes. The DSP must stop bidding on that raw UID2 and may use a different ID. Recipient deletion and compliance times were not verified. | One input per portal request. The guide gives a typical webhook dispatch time and a bidder requirement. The DPA concerns customer data under a specific agreement, with exceptions; it does not establish receipt, execution or erasure by every recipient. | evidence/profiles/ttd-uid2.json, capabilities.withdrawal_controls.observed |
| X574 | ["IDN-A-S08", "IDN-UID-S18", "IDN-A-S09"] | Independent profile review checked the evidence against the named service. | One input per portal request. The guide gives a typical webhook dispatch time and a bidder requirement. The DPA concerns customer data under a specific agreement, with exceptions; it does not establish receipt, execution or erasure by every recipient. [149, 150, 186, 187] | structural | ["IDN-A-S08", "IDN-UID-S18", "IDN-A-S09", "IDN-VUID-S02"] | One input per portal request. The guide gives a typical webhook dispatch time and a bidder requirement. The DPA concerns customer data under a specific agreement, with exceptions; it does not establish receipt, execution or erasure by every recipient. | evidence/profiles/ttd-uid2.json, capabilities.withdrawal_controls.source_ids |
| X575 | 2026-11-05 | Independent profile review checked the evidence against the named service. | July 11, 2025 dates the launch described in the documentation log. The fresh endpoint guide displays November 11, 2026 as its last update. [145, 146] | structural | 2026-11-11 | July 11, 2025 dates the launch described in the documentation log. The fresh endpoint guide displays November 11, 2026 as its last update. | evidence/profiles/ttd-uid2.json, capabilities.change_traceability.date |
| X576 | The studied retailer reported match counts without an exact destination denominator. The HP case measured one campaign’s target households. | Independent profile review checked the evidence against the named service. | Deep Sync published the figures. HP’s target represents one campaign; attributing the match counts to the unnamed retailer obscures who produced the report. [31, 147, 148] | structural | Deep Sync reported matched CRM records without the exact eligible denominator. The Trade Desk’s HP case measured reach against one campaign’s estimated target households. | Deep Sync published the figures. HP’s target represents one campaign; attributing the match counts to the unnamed retailer obscures who produced the report. | evidence/swot/ttd-uid2.json, weaknesses.0.t |
| X577 | A participant’s local opt-out does not reach the global service. Framework-wide withdrawal requires a separate portal request. | Independent profile review checked the evidence against the named service. | A local preference stays with the participant. Global opt-out is scoped to each submitted credential. [149, 150] | structural | A site opt-out stays with that participant. A consumer must submit a separate portal request for each email address or phone number they want suppressed across UID2. | A local preference stays with the participant. Global opt-out is scoped to each submitted credential. | evidence/swot/ttd-uid2.json, weaknesses.1.t |
| X578 | ["IDN-E-S14", "IDN-E-S16"] | Independent profile review checked the evidence against the named service. | Issuer specifies server-side handling as a documented alternative for affected clients. Evaluating it is an inference; no adoption is claimed. [151, 152, 145] | structural | ["IDN-E-S14", "IDN-E-S16", "IDN-UID-S04"] | Issuer specifies server-side handling as a documented alternative for affected clients. Evaluating it is an inference; no adoption is claimed. | evidence/swot/ttd-uid2.json, opportunities.0.src |
| X579 | WebKit’s block removes client-side generation and refresh from affected iOS and macOS browser paths. Publishers using that integration must change it. | Independent profile review checked the evidence against the named service. | Product notice identifies iOS 27+ and macOS 27+ Safari and prescribes server-side token management. Qualification confines the integration change to publishers continuing UID2 there. [151, 152, 145] | structural | To keep UID2 on affected iOS 27+ browsers and macOS 27+ Safari, publishers using browser-side token generation or refresh must move those steps server-side. | Product notice identifies iOS 27+ and macOS 27+ Safari and prescribes server-side token management. Qualification confines the integration change to publishers continuing UID2 there. | evidence/swot/ttd-uid2.json, threats.1.t |
| X580 | ["IDN-E-S14", "IDN-E-S16"] | Independent profile review checked the evidence against the named service. | Product notice identifies iOS 27+ and macOS 27+ Safari and prescribes server-side token management. Qualification confines the integration change to publishers continuing UID2 there. [151, 152, 145] | structural | ["IDN-E-S14", "IDN-E-S16", "IDN-UID-S04"] | Product notice identifies iOS 27+ and macOS 27+ Safari and prescribes server-side token management. Qualification confines the integration change to publishers continuing UID2 there. | evidence/swot/ttd-uid2.json, threats.1.src |
| X581 | For the year ended 30 June 2025, ID5 Technology Limited reported £702,598 in cash at bank and in hand and elected to omit its profit and loss account from the filed statements. | Independent profile review checked the evidence against the named service. | The UK entity’s unaudited balance sheet shows £702,598 cash in the 2025 column as of 30 June. Its small-company filing election omits the profit and loss account. This is a date-specific stock and not annual cash received. The balance sheet describes one company. It gives no ID5 ID revenue, contract price or US product sales. [340] | supported with qualification | ID5 Technology Limited reported £702,598 in cash at bank and in hand as of 30 June 2025. The UK company elected to exclude its profit and loss account from the unaudited public filing. | The UK entity’s unaudited balance sheet shows £702,598 cash in the 2025 column as of 30 June. Its small-company filing election omits the profit and loss account. This is a date-specific stock and not annual cash received. The balance sheet describes one company. It gives no ID5 ID revenue, contract price or US product sales. | IDN-ID5-F04; data/claims-overlay.json; evidence/verify/id5-id.json |
| X582 | ID5’s JavaScript guide says publishers can register to distribute ID5 ID without charge and that platforms and intermediaries need a commercial license for use beyond the standard publisher integration. | Independent profile review checked the evidence against the named service. | The JS guide requires a commercial licence for platforms, intermediaries, or other commercial use beyond standard publisher integration. The linked agreement limits the no-charge entitlement to encrypted ID delivery and offers decryption, connections and data use for purchase. No enterprise rate card is given. Individual enterprise fees, volume tiers and revenue recognition are not specified in this guide. [56, 351] | supported with qualification | ID5 offers registered publishers encrypted ID delivery without charge. Its guide requires a commercial licence for platforms, intermediaries or commercial uses beyond standard publisher integration. The agreement offers decryption, connections between IDs and use of ID5 data for purchase. [Additional evidence: IDN-VID5-S01] | The JS guide requires a commercial licence for platforms, intermediaries, or other commercial use beyond standard publisher integration. The linked agreement limits the no-charge entitlement to encrypted ID delivery and offers decryption, connections and data use for purchase. No enterprise rate card is given. Individual enterprise fees, volume tiers and revenue recognition are not specified in this guide. | IDN-ID5-F05; data/claims-overlay.json; evidence/verify/id5-id.json |
| X583 | ID5 describes using deterministic signals across domains and devices and combining signals in a model that infers identity; its publisher guide names hashed email and first-party IDs as partner inputs when available. | Independent profile review checked the evidence against the named service. | The March 2022 vendor explanation says it uses deterministic signals across domains and devices and probabilistic modeling. Its later integration guide identifies hashed email and first-party user IDs as inputs when available. The competing BidPulse test concerns ID5-bridged identifiers and a geographic proxy, and cannot validate the identity of the person or household behind the ID. Retain the two source dates. The competitor study uses fixed-line location history, not independent truth labels for the sold identifier. [55, 56, 57, 58] | supported as stated | ID5 describes using deterministic signals across domains and devices and combining signals in a model that infers identity; its publisher guide names hashed email and first-party IDs as partner inputs when available. [Additional evidence: IDN-VID5-S02, IDN-VID5-S03] | The March 2022 vendor explanation says it uses deterministic signals across domains and devices and probabilistic modeling. Its later integration guide identifies hashed email and first-party user IDs as inputs when available. The competing BidPulse test concerns ID5-bridged identifiers and a geographic proxy, and cannot validate the identity of the person or household behind the ID. Retain the two source dates. The competitor study uses fixed-line location history, not independent truth labels for the sold identifier. | IDN-ID5-F06; data/claims-overlay.json; evidence/verify/id5-id.json |
| X584 | ID5’s 19 October 2026 platform policy says that Twine Data, formerly TrueData Solutions, runs a separate US licensing business and that neither side uses the other’s data for its stated client services or data licensing. | Independent profile review checked the evidence against the named service. | The October 19 v4.4 public policy names Twine Data, formerly TrueData Solutions, as an affiliate with a separately operated US licensing business and expressly describes both data-use boundaries as current policy. It is direct evidence of what ID5 published, not independent observation of its systems. Classify the published policy as a direct record of ID5’s statement; enforcement and technical separation are self-described. This dated disclosure is a vendor statement about data use. It does not establish the lineage of every historical edge. [131] | supported with qualification | In its 19 October 2026 platform privacy policy, ID5 states that its platform does not use data from Twine Data, formerly TrueData Solutions, in client or partner services, and that Twine does not use ID5 platform data in its separate US licensing business. [Evidence class: direct_record] | The October 19 v4.4 public policy names Twine Data, formerly TrueData Solutions, as an affiliate with a separately operated US licensing business and expressly describes both data-use boundaries as current policy. It is direct evidence of what ID5 published, not independent observation of its systems. Classify the published policy as a direct record of ID5’s statement; enforcement and technical separation are self-described. This dated disclosure is a vendor statement about data use. It does not establish the lineage of every historical edge. | IDN-ID5-F07; data/claims-overlay.json; evidence/verify/id5-id.json |
| X585 | The Prebid ID5 module guide uses the id5-sync.com EID source and documents a value of 0 where consent is absent or the user has opted out. | Independent profile review checked the evidence against the named service. | The Prebid module documentation says the id5-sync.com entry should still be present, with the ID in its uids array set to the string “0” for no consent or opt-out. Sender behavior in documentation does not prove recipient execution. A sender-side sentinel does not prove a specific DSP received it or removed an earlier ID. [175] | supported with qualification | Prebid documents that the id5-sync.com EID should remain present, with its ID set to the string “0” when consent is absent or the user opts out. | The Prebid module documentation says the id5-sync.com entry should still be present, with the ID in its uids array set to the string “0” for no consent or opt-out. Sender behavior in documentation does not prove recipient execution. A sender-side sentinel does not prove a specific DSP received it or removed an earlier ID. | IDN-ID5-F08; data/claims-overlay.json; evidence/verify/id5-id.json |
| X586 | ID5 instructs publishers with US-originating traffic to include it in their do-not-sell/share signal flow, such as GPP or GPC, and gives separate TCF or opt-in instructions for EEA and UK traffic. | Independent profile review checked the evidence against the named service. | The US publisher instructions call for ID5 in GPP/GPC or equivalent opt-out flows. EEA and UK instructions require purpose-specific signals for TCF participants or prior valid opt-in for non-TCF participants, and ID5 approval for non-TCF EEA processing. The policy separately identifies purposes using consent versus legitimate interest. Instructions to publishers document collection and signaling responsibilities. They do not certify the legal basis or verify a receiving DSP’s actions. [177, 131] | supported as stated | ID5 instructs publishers with US-originating traffic to include it in their do-not-sell/share signal flow, such as GPP or GPC, and gives separate TCF or opt-in instructions for EEA and UK traffic. [Additional evidence: IDN-ID5-S05] | The US publisher instructions call for ID5 in GPP/GPC or equivalent opt-out flows. EEA and UK instructions require purpose-specific signals for TCF participants or prior valid opt-in for non-TCF participants, and ID5 approval for non-TCF EEA processing. The policy separately identifies purposes using consent versus legitimate interest. Instructions to publishers document collection and signaling responsibilities. They do not certify the legal basis or verify a receiving DSP’s actions. | IDN-ID5-F09; data/claims-overlay.json; evidence/verify/id5-id.json |
| X587 | Adobe documents ID5 ID as a universal-ID targeting option for DSP placements, with one universal-ID type per placement and available choices dependent on the selected geography. | Independent profile review checked the evidence against the named service. | Adobe’s placement instructions allow one universal ID type at a time with choices constrained by geo-targets, and explicitly say ID5 IDs are available at no fee within this Adobe offering. This resolves the route-specific availability of the ID free of charge, not the fees paid by a DSP to ID5. The settings describe a selectable buy-side feature, not its use by every seat, a national coverage estimate or an exported cross-platform identifier. [127] | supported with qualification | Adobe lists ID5 ID as a no-fee universal-ID option in its DSP placement settings. A buyer selects one universal-ID type, and available options depend on the placement’s geographical targets. | Adobe’s placement instructions allow one universal ID type at a time with choices constrained by geo-targets, and explicitly say ID5 IDs are available at no fee within this Adobe offering. This resolves the route-specific availability of the ID free of charge, not the fees paid by a DSP to ID5. The settings describe a selectable buy-side feature, not its use by every seat, a national coverage estimate or an exported cross-platform identifier. | IDN-ID5-F11; data/claims-overlay.json; evidence/verify/id5-id.json |
| X588 | ID5 specifies S3 feeds for opt-outs every 30 minutes and deletion requests hourly and expects its partners to ingest and process the lists. | Independent profile review checked the evidence against the named service. | The regular partner lists update every 30 minutes for opt-outs and hourly for routine deletion. A separate on-demand deletion feed is best-effort, without a fixed schedule; file delivery can lag and actual recipient processing is separate. The feeds may arrive late; publication time is distinct from partner completion, deletion of derivatives and independent proof of suppression. [128] | supported with qualification | ID5 specifies 30-minute opt-out updates, hourly routine deletion updates and a separate best-effort feed for on-demand deletions. Partners are expected to ingest and process these S3 lists; their completion time is unverified. | The regular partner lists update every 30 minutes for opt-outs and hourly for routine deletion. A separate on-demand deletion feed is best-effort, without a fixed schedule; file delivery can lag and actual recipient processing is separate. The feeds may arrive late; publication time is distinct from partner completion, deletion of derivatives and independent proof of suppression. | IDN-ID5-F13; data/claims-overlay.json; evidence/verify/id5-id.json |
| X589 | On 1 October 2026 ID5 described the sale of the former TrueData third-party audience business and brand to Cloud Technologies. | Independent profile review checked the evidence against the named service. | ID5’s 1 October 2026 account explicitly scopes its sale description to the third-party audience business and TrueData brand. Independent of that statement, the buyer filed the conditional agreement and then its 5 October completion. The dated seller statement is separate from the date the buyer reported completion. It does not describe the sale of ID5 ID. [217, 342, 343] | supported as stated | On 1 October 2026 ID5 described the sale of the former TrueData third-party audience business and brand to Cloud Technologies. [Additional evidence: IDN-ID5-S16, IDN-ID5-S17] | ID5’s 1 October 2026 account explicitly scopes its sale description to the third-party audience business and TrueData brand. Independent of that statement, the buyer filed the conditional agreement and then its 5 October completion. The dated seller statement is separate from the date the buyer reported completion. It does not describe the sale of ID5 ID. | IDN-ID5-F17; data/claims-overlay.json; evidence/verify/id5-id.json |
| X590 | The April 2025 and November 2026 ID5 publisher SDK guides both set maxCascades to eight by default when cookie sync is needed and specify minus one to disable syncing. | Independent profile review checked the evidence against the named service. | Both immutable release README files state that maxCascades defaults to 8 when sync is required, and a value of -1 disables it. GitHub commit metadata independently yields v1.0.83 April 18, 2025 and v1.0.100 November 6, 2026. The Prebid integration’s separate canCookieSync switch defaults to false. This demonstrates continuity of one configurable parameter. Actual use, browser settings and assignment decisions may differ. [179, 180, 175] | supported with qualification | The 18 April 2025 and 6 November 2026 ID5 JavaScript SDK guides specify the same default maximum of eight cookie syncs when syncing is required. Setting maxCascades to −1 disables sync on this route. [Additional evidence: IDN-ID5-S06] | Both immutable release README files state that maxCascades defaults to 8 when sync is required, and a value of -1 disables it. GitHub commit metadata independently yields v1.0.83 April 18, 2025 and v1.0.100 November 6, 2026. The Prebid integration’s separate canCookieSync switch defaults to false. This demonstrates continuity of one configurable parameter. Actual use, browser settings and assignment decisions may differ. | IDN-ID5-F21; data/claims-overlay.json; evidence/verify/id5-id.json |
| X591 | The Prebid client example uses 90-day storage and a two-hour refresh. ID5 states a separate 100-day maximum for data used to deliver live services. Neither interval establishes the age at which an individual link ceases to be valid. | Independent profile review checked the evidence against the named service. | Three separate clocks matter for a buyer. The sample is local storage, the cookie clock resets with an update, and a live-service data cap does not define edge validity. [175, 131] | structural | Prebid illustrates 90-day local storage and a two-hour refresh. ID5’s policy separately sets a 90-day duration from the last update for its third-party cookie and a 100-day maximum for live-service data before deletion or aggregation. The records do not give a general graph-link expiry. | Three separate clocks matter for a buyer. The sample is local storage, the cookie clock resets with an update, and a live-service data cap does not define edge validity. | evidence/profiles/id5-id.json, capabilities.refresh_expiry.observed |
| X592 | The US publisher requirements name GPP/GPC signals, with separate EEA/UK TCF or opt-in instructions. The Prebid sender documents a zero EID value if consent is absent or the user opts out. A public end-to-end recipient test was unresolved. | Independent profile review checked the evidence against the named service. | The Prebid sentinel is a string within a specific EID entry, not a zero-valued EID object or proof of enforcement. Operator policy and partner instructions distinguish US opt-outs from EEA/UK signals. [175, 177, 176, 131] | structural | ID5 requires a US sale/sharing opt-out signal path, including GPP or GPC, and separate purpose-specific TCF or prior opt-in handling in EEA/UK traffic. Prebid says the id5-sync.com EID should remain present with the ID string “0” on no consent or opt-out. No named recipient’s enforcement was verified. | The Prebid sentinel is a string within a specific EID entry, not a zero-valued EID object or proof of enforcement. Operator policy and partner instructions distinguish US opt-outs from EEA/UK signals. | evidence/profiles/id5-id.json, capabilities.permission_handling.observed |
| X593 | ["IDN-ID5-S06", "IDN-ID5-S07", "IDN-ID5-S08"] | Independent profile review checked the evidence against the named service. | The Prebid sentinel is a string within a specific EID entry, not a zero-valued EID object or proof of enforcement. Operator policy and partner instructions distinguish US opt-outs from EEA/UK signals. [175, 177, 176, 131] | structural | ["IDN-ID5-S06", "IDN-ID5-S07", "IDN-ID5-S08", "IDN-ID5-S05"] | The Prebid sentinel is a string within a specific EID entry, not a zero-valued EID object or proof of enforcement. Operator policy and partner instructions distinguish US opt-outs from EEA/UK signals. | evidence/profiles/id5-id.json, capabilities.permission_handling.source_ids |
| X594 | The related Data Onboarding service gives the distinct-input denominator and numerator by output type. Its table is illustrative. The US publisher case compares commercial treatments. Neither publishes an observed numerical US bidstream match rate for the registered buying task. | Independent profile review checked the evidence against the named service. | The sample 80% is not a field result. The geography refers to US audiences, not a confirmed US domicile of every participating publisher. The upload is a neighboring service. [129, 130, 32] | structural | The adjacent Data Onboarding guide defines distinct hashed emails matched over distinct inputs for each output ID type. Its sample is illustrative. The Sovrn case compares CPMs on participating publishers’ US audiences. Neither source supplies an observed US bidstream match rate for the ID5 ID. | The sample 80% is not a field result. The geography refers to US audiences, not a confirmed US domicile of every participating publisher. The upload is a neighboring service. | evidence/profiles/id5-id.json, capabilities.match_rate_definition.observed |
| X595 | Adobe documents a DSP targeting option, and Adform describes an ID5-based retargeting workflow. Both sources describe recipient functionality. The Adform announcement gives no market-specific count and its several products belong to one operator. The older Sovrn case does not establish a current destination. | Independent profile review checked the evidence against the named service. | Adobe’s instructions and Adform’s 2025 announcement describe available integration routes. Both lack same-period, US-scoped proof of production consumption by a named seat. The registered dimension counts US-supported production recipients. [127, 178, 129, 130] | structural | Adobe documents an ID5 placement option; Adform reported a single-tag ID5 retargeting integration on 28 May 2025 across products it operates. The records do not establish how many named recipients used this exact ID on US-supported production routes at cutoff. Sovrn’s 2022 test is historical. | Adobe’s instructions and Adform’s 2025 announcement describe available integration routes. Both lack same-period, US-scoped proof of production consumption by a named seat. The registered dimension counts US-supported production recipients. | evidence/profiles/id5-id.json, capabilities.destination_coverage.observed |
| X596 | live | Independent profile review checked the evidence against the named service. | The interfaces exist, but the registered US-supported production-recipient count at cutoff is unconfirmed in these records. [127, 178, 129, 130] | structural | unconfirmed | The interfaces exist, but the registered US-supported production-recipient count at cutoff is unconfirmed in these records. | evidence/profiles/id5-id.json, capabilities.destination_coverage.availability |
| X597 | ID5 issues specific operating requirements for US publisher traffic. Sovrn documented a US publisher deployment in 2022. Adobe documents current placement targeting by selected geography, without enumerating ID5-enabled US seats or supplying a US transaction log. | Independent profile review checked the evidence against the named service. | The historical test documents US audiences of participating publishers, not the publishers’ domicile. The issuer explicitly names its US market business in the 2025 announcement; its current requirements address US traffic. [177, 127, 129, 130, 132] | structural | ID5 publishes requirements for US-originating publisher traffic and identified the US as an existing client market in 2025. Sovrn and ID5 reported a 2022 test on participating publishers’ US audiences. Adobe’s geography-dependent option does not enumerate US-enabled seats or transactions. | The historical test documents US audiences of participating publishers, not the publishers’ domicile. The issuer explicitly names its US market business in the 2025 announcement; its current requirements address US traffic. | evidence/profiles/id5-id.json, capabilities.us_availability.observed |
| X598 | ["IDN-ID5-S07", "IDN-ID5-S10", "IDN-ID5-S12", "IDN-ID5-S22"] | Independent profile review checked the evidence against the named service. | The historical test documents US audiences of participating publishers, not the publishers’ domicile. The issuer explicitly names its US market business in the 2025 announcement; its current requirements address US traffic. [177, 127, 129, 130, 132] | structural | ["IDN-ID5-S07", "IDN-ID5-S10", "IDN-ID5-S12", "IDN-ID5-S22", "IDN-ID5-S14"] | The historical test documents US audiences of participating publishers, not the publishers’ domicile. The issuer explicitly names its US market business in the 2025 announcement; its current requirements address US traffic. | evidence/profiles/id5-id.json, capabilities.us_availability.source_ids |
| X599 | The partner case measures CPM and compares linkage and auction transmission conditions. ID5 describes deterministic and inferred links. The reviewed method, case, onboarding guide and product-specific search did not yield an independent US person- or household-labeled test with error counts. | Independent profile review checked the evidence against the named service. | BidPulse is a competitive measurement of IDs bridged into bid requests. Geographical proximity on selected fixed-line traffic cannot be substituted for independently labeled truth about people or households, or extrapolated to the core token. The Sovrn study measured CPM. [55, 56, 129, 130, 32, 57, 58] | structural | Sovrn and ID5 compared CPM. Competitor Intent IQ separately tested how closely the locations of IDs bridged by ID5 matched bid-request locations. Its method uses fixed-line IP history and distance, without separately labeled person or household membership. Independent accuracy of the core ID5 ID remains unconfirmed in this search. | BidPulse is a competitive measurement of IDs bridged into bid requests. Geographical proximity on selected fixed-line traffic cannot be substituted for independently labeled truth about people or households, or extrapolated to the core token. The Sovrn study measured CPM. | evidence/profiles/id5-id.json, capabilities.output_validation.observed |
| X600 | ["IDN-ID5-S04", "IDN-ID5-S03", "IDN-ID5-S12", "IDN-ID5-S22", "IDN-ID5-S13"] | Independent profile review checked the evidence against the named service. | BidPulse is a competitive measurement of IDs bridged into bid requests. Geographical proximity on selected fixed-line traffic cannot be substituted for independently labeled truth about people or households, or extrapolated to the core token. The Sovrn study measured CPM. [55, 56, 129, 130, 32, 57, 58] | structural | ["IDN-ID5-S04", "IDN-ID5-S03", "IDN-ID5-S12", "IDN-ID5-S22", "IDN-ID5-S13", "IDN-VID5-S02", "IDN-VID5-S03"] | BidPulse is a competitive measurement of IDs bridged into bid requests. Geographical proximity on selected fixed-line traffic cannot be substituted for independently labeled truth about people or households, or extrapolated to the core token. The Sovrn study measured CPM. | evidence/profiles/id5-id.json, capabilities.output_validation.source_ids |
| X601 | ID5 supplies affected IDs by S3 on separate opt-out and deletion schedules. Partners must ingest the files. Observed end-to-end removal latency and suppression of derived segments at an identified recipient were unresolved in the reviewed feed and DSP guide. | Independent profile review checked the evidence against the named service. | Regular deletion files are distinct from best-effort on-demand files; partner configuration, ingestion and removal can lag. ID5 has its own browser/email suppression mechanisms, which are separate from DSP actions. [176, 128, 131] | structural | ID5 documents S3 opt-out files every 30 minutes, routine deletion files hourly and on-demand deletions on a best-effort schedule. Partners configure delivery and are expected to process the files. The public records do not show completion latency or removal from derived segments at a named DSP. | Regular deletion files are distinct from best-effort on-demand files; partner configuration, ingestion and removal can lag. ID5 has its own browser/email suppression mechanisms, which are separate from DSP actions. | evidence/profiles/id5-id.json, capabilities.withdrawal_controls.observed |
| X602 | ["IDN-ID5-S08", "IDN-ID5-S09"] | Independent profile review checked the evidence against the named service. | Regular deletion files are distinct from best-effort on-demand files; partner configuration, ingestion and removal can lag. ID5 has its own browser/email suppression mechanisms, which are separate from DSP actions. [176, 128, 131] | structural | ["IDN-ID5-S08", "IDN-ID5-S09", "IDN-ID5-S05"] | Regular deletion files are distinct from best-effort on-demand files; partner configuration, ingestion and removal can lag. ID5 has its own browser/email suppression mechanisms, which are separate from DSP actions. | evidence/profiles/id5-id.json, capabilities.withdrawal_controls.source_ids |
| X603 | A publisher can retrieve an encrypted ID5 ID through its site integration and pass the identifier in an ad request. ID5 describes a DSP receiving keys and merge instructions outside the auction to recognize a publisher-specific ID. The buyer’s task here is to use that identifier on another seller’s inventory. [IDN-ID5-F05, IDN-ID5-F08, IDN-ID5-F10] | Independent profile review checked the evidence against the named service. | The agreement identifies which item is delivered free and which services the operator offers for purchase. [56, 176, 351] | structural | A publisher can obtain an encrypted ID5 ID and forward it in an ad request. The agreement sells decryption, connections between IDs and use of ID5 data as additional services. ID5’s DSP guide describes out-of-auction keys and merge instructions for linking publisher-specific IDs. [IDN-ID5-F05, IDN-ID5-F08, IDN-ID5-F10] | The agreement identifies which item is delivered free and which services the operator offers for purchase. | evidence/profiles/id5-id.json, what_is_sold |
| X604 | ID5 distinguishes free standard publisher distribution from uses that require a commercial license. Fees for a specific DSP, revenue per ID and US ID5 ID sales were unresolved in the reviewed registration guide, Adobe settings and UK filing. [IDN-ID5-F04, IDN-ID5-F05, IDN-ID5-F11] | Independent profile review checked the evidence against the named service. | Distinguish operator monetization from a DSP’s offer to its advertiser; the buyer feature provides concrete route pricing. [56, 127, 340, 351] | structural | ID5 offers encrypted ID delivery without charge under its agreement. It offers decryption, information linking IDs and use of its data for purchase. Adobe says ID5 IDs carry no fee in its own DSP placement feature. Enterprise licence rates, revenue per ID and US ID5 ID sales remain undisclosed in these records. [IDN-ID5-F04, IDN-ID5-F05, IDN-ID5-F11] | Distinguish operator monetization from a DSP’s offer to its advertiser; the buyer feature provides concrete route pricing. | evidence/profiles/id5-id.json, how_it_earns |
| X605 | Publishers can register to distribute the ID. Adobe describes a placement setting for advertisers using its DSP. Adform describes tagging for retargeting, and Sovrn reports publisher data passing to buyers in a historical US test. [IDN-ID5-F05, IDN-ID5-F11, IDN-ID5-F12, IDN-ID5-F14] | Independent profile review checked the evidence against the named service. | Identify paying versus distributing participants and qualify the geography of the historical case. [56, 127, 178, 129, 130, 351] | structural | Publishers register to distribute the encrypted ID. DSPs may purchase decryption and linkage information. Adobe documents a targeting setting for advertisers. Adform reports a retargeting route, and Sovrn and ID5 describe a historical test with participating publishers’ US audiences. [IDN-ID5-F05, IDN-ID5-F11, IDN-ID5-F12, IDN-ID5-F14] | Identify paying versus distributing participants and qualify the geography of the historical case. | evidence/profiles/id5-id.json, who_buys |
| X606 | ID5 documented distinct opt-out and deletion feeds and partner ingestion. | Independent profile review checked the evidence against the named service. | Technical requirements describe what partners are expected to do, not confirmed receipt. [128] | structural | ID5 documented regular and on-demand rights feeds and its expectations for partner ingestion. | Technical requirements describe what partners are expected to do, not confirmed receipt. | evidence/profiles/id5-id.json, timeline.5.event |
| X607 | [{"claim": "Sovrn and ID5 compare publisher auction conditions with hashed email passed, withheld after linking, or a control condition. They report a pricing effect for the combined treatment.", "evidence_class": "vendor_assertion", "sponsor": "Sovrn and ID5 participated and publicized the work. No separate funder identified.", "design": "Joint partner comparison across three publisher traffic groups. CPM is the outcome. Test duration, publisher count, traffic count, assignment procedure, independent funding and error labels are not specified.", "limits": "The case reports CPM differences for bundled Sovrn and ID5 treatments. There is no count of labeled true or false person links, representative national base, or stated randomization.", "source_ids": ["IDN-ID5-S12", "IDN-ID5-S22"], "finding_ids": ["IDN-ID5-F14"]}, {"claim": "Match Preview reports the share of distinct input hashed emails matched to each output ID type. Its example is illustrative.", "evidence_class": "vendor_assertion", "sponsor": "ID5, provider of Data Onboarding", "design": "Format specification with sample inputs and an example report, configurable countries, lookback and matching mode.", "limits": "A buyer must supply a real population for an observed rate. This adjacent onboarding report does not establish match accuracy or a US bidstream figure.", "source_ids": ["IDN-ID5-S13"], "finding_ids": ["IDN-ID5-F15"]}] | Independent profile review checked the evidence against the named service. | Retain commercial test and worked example, and add the dated, interested competing benchmark that tests a distinct output and proxy. [129, 32, 130, 57, 58] | structural | [{"claim": "Sovrn and ID5 report 72% CPM uplift with hashed-email linkage and 98% with the hashed email passed. The joint case compared three publisher traffic groups and supplies no raw CPMs or counts to reproduce the uplifts.", "evidence_class": "vendor_assertion", "sponsor": "Sovrn and ID5 participated and publicized the work. No separate funder identified.", "design": "Partner comparison of control, hashed email linked and withheld from auction, and hashed email linked and passed. Subjects were participating publishers’ US audiences. The study omits test dates, group counts, assignment procedure, underlying CPMs and uncertainty.", "limits": "The case reports CPM differences for bundled Sovrn and ID5 treatments. There is no count of labeled true or false person links, representative national base, or stated randomization.", "source_ids": ["IDN-ID5-S12", "IDN-ID5-S22"], "finding_ids": ["IDN-ID5-F14"]}, {"claim": "Match Preview reports the share of distinct input hashed emails matched to each output ID type. Its example is illustrative.", "evidence_class": "vendor_assertion", "sponsor": "ID5, provider of Data Onboarding", "design": "Format specification with sample inputs and an example report, configurable countries, lookback and matching mode.", "limits": "A buyer must supply a real population for an observed rate. This adjacent onboarding report does not establish match accuracy or a US bidstream figure.", "source_ids": ["IDN-ID5-S13"], "finding_ids": ["IDN-ID5-F15"]}, {"claim": "In Intent IQ’s dated table, 34.3% of assessed ID5-enriched bid requests passed its strict 100-mile rule between 12 August and 10 October 2026. Its mean within-request proportion was 35.1%.", "evidence_class": "vendor_assertion", "sponsor": "Intent IQ, competing provider; the issuer says an external implementer replicated the method on its data.", "design": "Matches bid-request fixed-line IP location to minimum observed distance of bridged IDs or synced cookies. Requires observations on three separate days within a 30-day history. Strict counts each request passing only if every assessed bridged ID is within threshold; proportional averages each request’s share within threshold.", "limits": "Bridging analysis, not person or household labels or verified accuracy of the native token. No sample geography, sample counts or uncertainty for the dated table. The release’s undated narrative gives different numbers; only the explicitly dated table is used.", "source_ids": ["IDN-VID5-S02", "IDN-VID5-S03"], "finding_ids": ["IDN-ID5-F06"]}] | Retain commercial test and worked example, and add the dated, interested competing benchmark that tests a distinct output and proxy. | evidence/profiles/id5-id.json, performance_evidence |
| X608 | The paired sources do not provide the model weights, acceptance threshold, or validation against an independently labeled US person or household. | Independent profile review checked the evidence against the named service. | The separate comparative test is relevant contrary evidence but cannot settle H2. [55, 56, 57, 58] | structural | Competitor Intent IQ benchmarks IDs that ID5 bridges into bid requests against fixed-line IP location histories. The period-specific result provides a geographic disagreement measure, without independently labeled US person/household truth for ID5 ID. | The separate comparative test is relevant contrary evidence but cannot settle H2. | evidence/profiles/id5-id.json, findings.5.counterevidence |
| X609 | Undated Adform account read 2026-11-11 | Independent profile review checked the evidence against the named service. | The embedded Story data supplies a publication date, distinct from customer deployment dates. [178] | structural | Adform newsroom HTML dates the announcement to 2025-05-28; read 2026-11-11. | The embedded Story data supplies a publication date, distinct from customer deployment dates. | evidence/profiles/id5-id.json, findings.11.scope_time |
| X610 | (empty) | Independent profile review checked the evidence against the named service. | Publisher page source field publishedOn gives 2025-05-28T14:28:00. [178] | structural | 2025-05-28 | Publisher page source field publishedOn gives 2025-05-28T14:28:00. | evidence/profiles/id5-id.json, sources.10.pub_date |
| X611 | Recipient account read 11 November 2026 | Independent profile review checked the evidence against the named service. | The announcement concerns a feature; avoid implying the 2026 access date is its deployment or measurement period. [178] | structural | Integration described in a 28 May 2025 announcement; no measurement interval disclosed. | The announcement concerns a feature; avoid implying the 2026 access date is its deployment or measurement period. | evidence/profiles/id5-id.json, sources.10.measurement_period |
| X612 | registry | Independent profile review checked the evidence against the named service. | The schema provides no registry type. This is a public Companies House register entry; retain direct_record for the finding. [370] | structural | other | The schema provides no registry type. This is a public Companies House register entry; retain direct_record for the finding. | evidence/profiles/id5-id.json, sources.0.source_type |
| X613 | registry | Independent profile review checked the evidence against the named service. | The schema provides no registry type. This is a public Companies House register entry; retain direct_record for the finding. [339] | structural | other | The schema provides no registry type. This is a public Companies House register entry; retain direct_record for the finding. | evidence/profiles/id5-id.json, sources.1.source_type |
| X614 | README, Configuration Parameters, maxCascades row, commit bd6c891f2c63 | Independent profile review checked the evidence against the named service. | The original heading is Available Configuration Options, not Configuration Parameters. [179] | structural | README, Available Configuration Options, maxCascades row, commit bd6c891f2c63 | The original heading is Available Configuration Options, not Configuration Parameters. | evidence/profiles/id5-id.json, findings.20.support.0.locator |
| X615 | README, Configuration Parameters, maxCascades row, commit 576fd16780a9 | Independent profile review checked the evidence against the named service. | The original heading is Available Configuration Options, not Configuration Parameters. [180] | structural | README, Available Configuration Options, maxCascades row, commit 576fd16780a9 | The original heading is Available Configuration Options, not Configuration Parameters. | evidence/profiles/id5-id.json, findings.20.support.1.locator |
| X616 | The feeds may arrive late; publication time is distinct from partner completion, deletion of derivatives and independent proof of suppression. | Independent profile review checked the evidence against the named service. | Retain timing exceptions and separate recipient action from publication. [128] | structural | ID5 also describes best-effort, on-demand deletions with no fixed schedule. Files can arrive late. Publication time does not establish partner deletion, suppression of derived outputs, or measured completion. | Retain timing exceptions and separate recipient action from publication. | evidence/profiles/id5-id.json, findings.12.limitations |
| X617 | The partner case measured auction pricing. No independently labeled US error rate or current national bidstream coverage was established in the reviewed sources. [IDN-ID5-F06, IDN-ID5-F14, IDN-ID5-F15] | Independent profile review checked the evidence against the named service. | Include the strongest contrary product-related measurement in the buyer’s evidence limits. [55, 129, 32, 130, 57, 58] | structural | The partner study measured auction pricing. Intent IQ’s separate comparison uses location history as a proxy for the IDs bridged into bid requests. The reviewed reports establish neither independently labeled US identity-error rates nor national bidstream coverage. [IDN-ID5-F06, IDN-ID5-F14, IDN-ID5-F15] | Include the strongest contrary product-related measurement in the buyer’s evidence limits. | evidence/profiles/id5-id.json, limits.1 |
| X618 | ["ID5’s 2025 announcement anticipated a combination. Its October 2026 policy expressly separates the affiliate’s licensed data from ID5 platform services. [IDN-ID5-F07, IDN-ID5-F16]", "The seller’s 1 October account predates the 5 October completion recorded by the buyer. The 30 September filing was conditional. [IDN-ID5-F17, IDN-ID5-F18, IDN-ID5-F19]", "Cookie-independent identification can coexist with conditional cookie syncing in the publisher SDK. The version comparison establishes continuity for one setting. [IDN-ID5-F06, IDN-ID5-F21]", "The vendor preview defines the denominator for a customer-file test. That does not give the public an observed rate on a US publisher-to-buyer handoff. [IDN-ID5-F14, IDN-ID5-F15]"] | Independent profile review checked the evidence against the named service. | Competing test uses a different output and metric than the labeled identity-accuracy hypothesis. [57, 58] | structural | ["ID5’s 2025 announcement anticipated a combination. Its October 2026 policy expressly separates the affiliate’s licensed data from ID5 platform services. [IDN-ID5-F07, IDN-ID5-F16]", "The seller’s 1 October account predates the 5 October completion recorded by the buyer. The 30 September filing was conditional. [IDN-ID5-F17, IDN-ID5-F18, IDN-ID5-F19]", "Cookie-independent identification can coexist with conditional cookie syncing in the publisher SDK. The version comparison establishes continuity for one setting. [IDN-ID5-F06, IDN-ID5-F21]", "The vendor preview defines the denominator for a customer-file test. That does not give the public an observed rate on a US publisher-to-buyer handoff. [IDN-ID5-F14, IDN-ID5-F15]", "Intent IQ reports geographic mismatches among IDs bridged by ID5. It tested its own data and proxy distance rule. The evidence does not establish natural-person or household error for the encrypted ID itself. [IDN-ID5-F06]"] | Competing test uses a different output and metric than the labeled identity-accuracy hypothesis. | evidence/profiles/id5-id.json, counterevidence_and_tensions |
| X619 | What does an enterprise license cost, and what revenue is attributable to this named US service? | Independent profile review checked the evidence against the named service. | Identify the documented priced service, not the no-charge delivery entitlement or Adobe’s route. [56, 127, 351] | structural | What rate does a DSP pay ID5 for decryption and linkage information, and what revenue is attributable to the named US service? | Identify the documented priced service, not the no-charge delivery entitlement or Adobe’s route. | evidence/profiles/id5-id.json, open_questions.3 |
| X620 | Experian says an unnamed DSP synced its cookies and received a weekly graph file linking them to cookies, mobile ad IDs, CTV IDs and IP addresses. | Independent profile review checked the evidence against the named service. | The anonymous customer’s PDF names its cookies as the sync input. It lists Experian cookies, MAIDs, CTV IDs and IPs returned and weekly file delivery. The issuer’s companion post confirms the description. Neither identifies the cookie setting in the browser, the DSP, the country, the physical file schema or edge lifetime. Apply the route only to this example. The recipient_interface field incorrectly identified the input as first-party. [37, 36] | supported as stated | Experian says an unnamed DSP synced its cookies and received a weekly graph file linking them to cookies, mobile ad IDs, CTV IDs and IP addresses. [Additional evidence: IDN-EXP-S14] | The anonymous customer’s PDF names its cookies as the sync input. It lists Experian cookies, MAIDs, CTV IDs and IPs returned and weekly file delivery. The issuer’s companion post confirms the description. Neither identifies the cookie setting in the browser, the DSP, the country, the physical file schema or edge lifetime. Apply the route only to this example. The recipient_interface field incorrectly identified the input as first-party. | IDN-EXP-F04; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X621 | Experian says MiQ licensed hashed emails, UID2s and partner cookies for its Identity Spine, which its traders and analysts use for targeting and measurement. | Independent profile review checked the evidence against the named service. | The PDF reports that MiQ licensed HEMs, UID2s and third-party partner cookies. MiQ itself confirms putting Experian’s IDs into its multi-source Identity Spine and describes traders’ and analysts’ audience and measurement uses. Neither report locates the tested data in the US or establishes paid transaction counts. The buyer supplies first-hand implementation evidence but is commercially interested. Count this as one customer engagement. [21, 118] | supported as stated | Experian says MiQ licensed hashed emails, UID2s and partner cookies for its Identity Spine, which its traders and analysts use for targeting and measurement. [Additional evidence: IDN-VEXP-S01] | The PDF reports that MiQ licensed HEMs, UID2s and third-party partner cookies. MiQ itself confirms putting Experian’s IDs into its multi-source Identity Spine and describes traders’ and analysts’ audience and measurement uses. Neither report locates the tested data in the US or establishes paid transaction counts. The buyer supplies first-hand implementation evidence but is commercially interested. Count this as one customer engagement. | IDN-EXP-F06; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X622 | Experian reports a 70% match rate from MiQ-supplied IP addresses to cookieless identifiers in the MiQ engagement. | Independent profile review checked the evidence against the named service. | The seller’s MiQ directory entry identifies the direction of matching and 70%. Page 5 of the full case reports the same rate. Neither supplies submitted or eligible record counts, selection rules, test dates or a geography-specific denominator. The named customer independently describes use, not a labeled audit or the 70% rate. This is a seller-reported match yield. The 6.5 devices per matched IP has a different base, and 51% and 64% reach changes have different units. [20, 21] | supported as stated | Experian reports a 70% match rate from MiQ-supplied IP addresses to cookieless identifiers in the MiQ engagement. [Additional evidence: IDN-C-S06] | The seller’s MiQ directory entry identifies the direction of matching and 70%. Page 5 of the full case reports the same rate. Neither supplies submitted or eligible record counts, selection rules, test dates or a geography-specific denominator. The named customer independently describes use, not a labeled audit or the 70% rate. This is a seller-reported match yield. The 6.5 devices per matched IP has a different base, and 51% and 64% reach changes have different units. | IDN-EXP-F07; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X623 | Experian says the unnamed DSP synced 84% of IDs within weeks of using its Digital Graph. | Independent profile review checked the evidence against the named service. | The October 2025 Results section reports within weeks 84% of IDs synced. The same case’s slide deck labels 84% synced IDs and separately reports a 9% match-rate increase. Neither defines the eligible denominator, baseline, relative/percentage-point treatment, ground truth or sampling error. One unnamed DSP case, reported twice. Accuracy language in the headlines is not a measured rate of correct links. [36, 37] | supported as stated | Experian says the unnamed DSP synced 84% of IDs within weeks of using its Digital Graph. [Additional evidence: IDN-EXP-S01] | The October 2025 Results section reports within weeks 84% of IDs synced. The same case’s slide deck labels 84% synced IDs and separately reports a 9% match-rate increase. Neither defines the eligible denominator, baseline, relative/percentage-point treatment, ground truth or sampling error. One unnamed DSP case, reported twice. Accuracy language in the headlines is not a measured rate of correct links. | IDN-EXP-F08; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X624 | Tapad’s own Graph notice says it permanently opts out a submitted identifier and can remove related identifiers for 60 days, subject to reassessment of the link. | Independent profile review checked the evidence against the named service. | Tapad distinguishes a submitted identifier already in its Graph from other identifiers it associates with it. Related IDs can reenter if the link is reassessed. The IP-specific section limits an IPv4 opt-out to 12 months. A cleared browser cookie can require another opt-out. Clients are instructed to discard removed IDs on receipt of refreshed data. No per-recipient timing is reported. These are Tapad’s stated rules. Client obligations, performed deletions and US Digital Graph license terms are distinct. [126] | supported with qualification | Tapad’s Graph notice describes permanent opt-out for an identifier already in its graph and conditional 60-day suppression of related device IDs. It specifies 12 months for a registered IPv4 address. | Tapad distinguishes a submitted identifier already in its Graph from other identifiers it associates with it. Related IDs can reenter if the link is reassessed. The IP-specific section limits an IPv4 opt-out to 12 months. A cleared browser cookie can require another opt-out. Clients are instructed to discard removed IDs on receipt of refreshed data. No per-recipient timing is reported. These are Tapad’s stated rules. Client obligations, performed deletions and US Digital Graph license terms are distinct. | IDN-EXP-F11; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X625 | After Experian announced Snowflake clean-room access in 2024, Snowflake removed its Experian connector from the legacy provider and consumer web interface in September 2026 while preserving existing configurations. | Independent profile review checked the evidence against the named service. | The vendor announced generally available offline/digital collaboration in September 2024. Snowflake’s September 10, 2026 note specifically removes Experian from both legacy web interfaces without deleting configurations in that release. Snowflake’s dated end-of-life schedule separately blocked new legacy rooms in the web app on October 1 and scheduled later retirements. There is no confirmation here of transactions in a saved setup, Experian connector support in the replacement API, or termination of directly licensed graph delivery. [122, 123, 124] | supported with qualification | Experian announced Snowflake clean-room access in 2024. On September 10, 2026, Snowflake removed the Experian connector from its legacy web interface without deleting existing configurations. From October 1, the web app could no longer create new legacy rooms. [Additional evidence: IDN-VEXP-S02] | The vendor announced generally available offline/digital collaboration in September 2024. Snowflake’s September 10, 2026 note specifically removes Experian from both legacy web interfaces without deleting configurations in that release. Snowflake’s dated end-of-life schedule separately blocked new legacy rooms in the web app on October 1 and scheduled later retirements. There is no confirmation here of transactions in a saved setup, Experian connector support in the replacement API, or termination of directly licensed graph delivery. | IDN-EXP-F12; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X626 | Tapad reported that its Device Graph had twice the precision of the next competitor in a 2016 Hotels.com evaluation. | Independent profile review checked the evidence against the named service. | The 2016 issuer report describes Hotels.com’s three-and-a-half-month comparison with two unnamed rivals and a twofold precision result relative to the next-highest offering. There is no absolute precision, holdout construction, sample size or known financial sponsor. Experian acquired Tapad in November 2020. Historical commercial-buyer comparison reported by the winner. It does not validate the later US Digital Graph. [45, 46] | supported as stated | Tapad reported that its Device Graph had twice the precision of the next competitor in a 2016 Hotels.com evaluation. [Additional evidence: IDN-VEXP-S03] | The 2016 issuer report describes Hotels.com’s three-and-a-half-month comparison with two unnamed rivals and a twofold precision result relative to the next-highest offering. There is no absolute precision, holdout construction, sample size or known financial sponsor. Experian acquired Tapad in November 2020. Historical commercial-buyer comparison reported by the winner. It does not validate the later US Digital Graph. | IDN-EXP-F14; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X627 | The Digital Graph sheet credits Experian Marketing Services, Inc., while the US marketing data-practices notice names Experian Marketing Solutions, LLC. | Independent profile review checked the evidence against the named service. | The ©2025 Digital Graph sheet identifies Experian Marketing Services, Inc. The 2026 marketing-data notice names Experian Marketing Solutions, LLC. The plc annual report lists the latter and Tapad as distinct US subsidiaries. A product-sheet copyright, related undertaking and privacy-notice entity are insufficient to identify the signer of a specific US graph license. The source supports the attributed statement at this scope. [119, 336, 337] | supported as stated | The Digital Graph sheet credits Experian Marketing Services, Inc., while the US marketing data-practices notice names Experian Marketing Solutions, LLC. [Additional evidence: IDN-EXP-S08] | The ©2025 Digital Graph sheet identifies Experian Marketing Services, Inc. The 2026 marketing-data notice names Experian Marketing Solutions, LLC. The plc annual report lists the latter and Tapad as distinct US subsidiaries. A product-sheet copyright, related undertaking and privacy-notice entity are insufficient to identify the signer of a specific US graph license. The source supports the attributed statement at this scope. | IDN-EXP-F15; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X628 | Experian plc reported worldwide statutory revenue of US$8,445 million for the year ended March 31, 2026. | Independent profile review checked the evidence against the named service. | The Financial review statutory table reports $8,445m for FY26 and $7,523m for FY25. The FY26 results release confirms a year ended March 31 and distinguishes $8,425m ongoing revenue from $8,445m statutory. Neither table separates US Digital Graph revenue. The source supports the attributed statement at this scope. [337, 338] | supported as stated | Experian plc reported worldwide statutory revenue of US$8,445 million for the year ended March 31, 2026. [Additional evidence: IDN-EXP-S09] | The Financial review statutory table reports $8,445m for FY26 and $7,523m for FY25. The FY26 results release confirms a year ended March 31 and distinguishes $8,425m ongoing revenue from $8,445m statutory. Neither table separates US Digital Graph revenue. The source supports the attributed statement at this scope. | IDN-EXP-F17; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X629 | The UK Upper Tribunal dismissed the Commissioner’s appeal in April 2024, leaving a substituted notice covering certain open-source records used by Experian Limited for direct marketing. | Independent profile review checked the evidence against the named service. | The April 22 Upper Tribunal decision dismissed the ICO appeal. Paragraph 47 reproduces the FTT substitute notice for direct marketing from three public sources, subject to enumerated exceptions. The ICO confirmed on May 24 it would not pursue further appeal. Neither instrument identifies the US Digital Graph or demonstrates completed notice delivery. Legal record outside IQ. Public-register sourcing is distinct from digital open-source software. The order regulates Experian Limited in the UK. [67, 66, 357] | supported as stated | The UK Upper Tribunal dismissed the Commissioner’s appeal in April 2024, leaving a substituted notice covering certain open-source records used by Experian Limited for direct marketing. [Additional evidence: IDN-VEXP-S04] | The April 22 Upper Tribunal decision dismissed the ICO appeal. Paragraph 47 reproduces the FTT substitute notice for direct marketing from three public sources, subject to enumerated exceptions. The ICO confirmed on May 24 it would not pursue further appeal. Neither instrument identifies the US Digital Graph or demonstrates completed notice delivery. Legal record outside IQ. Public-register sourcing is distinct from digital open-source software. The order regulates Experian Limited in the UK. | IDN-EXP-F18; data/claims-overlay.json; evidence/verify/experian-resolution.json |
| X630 | ["IDN-C-S05", "IDN-EXP-S14"] | Independent profile review checked the evidence against the named service. | Seller’s IP-to-ID rate and the separate DSP’s sync proportion both lack eligible raw counts. D is appropriate. A lower or higher percentage is not evidence of lower or higher accuracy. [20, 36, 21] | structural | ["IDN-C-S05", "IDN-EXP-S14", "IDN-C-S06"] | Seller’s IP-to-ID rate and the separate DSP’s sync proportion both lack eligible raw counts. D is appropriate. A lower or higher percentage is not evidence of lower or higher accuracy. | evidence/profiles/experian-resolution.json, capabilities.match_rate_definition.source_ids |
| X631 | An unnamed DSP synced first-party cookies and received linked Experian cookies, mobile ad IDs, CTV IDs and IPs in a weekly file. The source set does not include the physical schema, handling of unmatched rows or duplicate rules. | Independent profile review checked the evidence against the named service. | Both versions say the DSP used its cookies; neither determines the first- or third-party context in which the browser set or read them. Listed returned IDs and a weekly file are seller-reported. Grade D is retained. [37, 36] | structural | An unnamed DSP synced its cookies and received related Experian cookies, mobile ad IDs, CTV IDs and IP addresses in a weekly file. The cases do not specify browser cookie context, file schema, duplicate handling or unmatched rows. | Both versions say the DSP used its cookies; neither determines the first- or third-party context in which the browser set or read them. Listed returned IDs and a weekly file are seller-reported. Grade D is retained. | evidence/profiles/experian-resolution.json, capabilities.recipient_interface.observed |
| X632 | MiQ’s Identity Spine is named and its traders are described using the data. The case leaves its implementation geography unspecified. The DSP case is anonymous. Snowflake’s old web connector was removed in September 2026. A count of named production recipients on a documented US-specific route remains unconfirmed. | Independent profile review checked the evidence against the named service. | The customer itself confirms feed use. Neither the US-facing article, its global counterpart, nor the seller report isolates US traffic. No counted US endpoint or successor Snowflake identity connection is proved. An announced integration cannot fill this gap. [37, 21, 122, 123, 118, 124] | structural | MiQ confirms it used Experian IDs in its Identity Spine. Its case gives no US-specific sample. The DSP is unnamed. Snowflake removed the connector from its legacy UI in September 2026 and blocked new legacy rooms there in October. A US-specific recipient count remains unconfirmed. | The customer itself confirms feed use. Neither the US-facing article, its global counterpart, nor the seller report isolates US traffic. No counted US endpoint or successor Snowflake identity connection is proved. An announced integration cannot fill this gap. | evidence/profiles/experian-resolution.json, capabilities.destination_coverage.observed |
| X633 | ["IDN-EXP-S01", "IDN-C-S06", "IDN-EXP-S07", "IDN-EXP-S06"] | Independent profile review checked the evidence against the named service. | The customer itself confirms feed use. Neither the US-facing article, its global counterpart, nor the seller report isolates US traffic. No counted US endpoint or successor Snowflake identity connection is proved. An announced integration cannot fill this gap. [37, 21, 122, 123, 118, 124] | structural | ["IDN-EXP-S01", "IDN-C-S06", "IDN-EXP-S07", "IDN-EXP-S06", "IDN-VEXP-S01", "IDN-VEXP-S02"] | The customer itself confirms feed use. Neither the US-facing article, its global counterpart, nor the seller report isolates US traffic. No counted US endpoint or successor Snowflake identity connection is proved. An announced integration cannot fill this gap. | evidence/profiles/experian-resolution.json, capabilities.destination_coverage.source_ids |
| X634 | Experian’s US notice gives a browser-level opt-out rule. Tapad’s own Graph notice defines permanent opt-out of the submitted identifier and conditional handling of related IDs. The materials reviewed do not show when an Experian Digital Graph licensee receives a deletion, applies it or prevents reappearance. | Independent profile review checked the evidence against the named service. | The original fails to mention Tapad’s IPv4 exception and its explicit contractual instruction to discard opted-out IDs upon refresh. Those are publisher-described controls for Tapad; no exact-SKU recipient implementation is seen. C grades the written notices, never observed compliance. [37, 125, 126] | structural | The US notice starts GPC opt-outs at the browser ID unless already linked. Tapad describes permanent opt-out for in-graph identifiers, a 60-day conditional rule for related devices, a 12-month IPv4 rule and a recipient discard obligation on refresh. No reviewed record shows a US Digital Graph licensee applying a deletion or preventing reappearance. | The original fails to mention Tapad’s IPv4 exception and its explicit contractual instruction to discard opted-out IDs upon refresh. Those are publisher-described controls for Tapad; no exact-SKU recipient implementation is seen. C grades the written notices, never observed compliance. | evidence/profiles/experian-resolution.json, capabilities.withdrawal_controls.observed |
| X635 | ["IDN-EXP-S01", "IDN-C-S06"] | Independent profile review checked the evidence against the named service. | MiQ’s own account directly documents trade and analyst use. Experian separately reports weekly delivery in the unnamed DSP case. Neither describes the cookie setting as first-party. [37, 21, 118] | structural | ["IDN-EXP-S01", "IDN-C-S06", "IDN-VEXP-S01"] | MiQ’s own account directly documents trade and analyst use. Experian separately reports weekly delivery in the unnamed DSP case. Neither describes the cookie setting as first-party. | evidence/swot/experian-resolution.json, strengths.1.src |
| X636 | ["IDN-C-S05", "IDN-EXP-S14"] | Independent profile review checked the evidence against the named service. | The full MiQ and DSP records omit raw eligibility and count information. Reproduction from the published materials is not possible. [20, 36, 21] | structural | ["IDN-C-S05", "IDN-EXP-S14", "IDN-C-S06"] | The full MiQ and DSP records omit raw eligibility and count information. Reproduction from the published materials is not possible. | evidence/swot/experian-resolution.json, weaknesses.0.src |
| X637 | The removal of Snowflake’s legacy Experian connector can disrupt setup for buyers whose workflows still depend on that interface. | Independent profile review checked the evidence against the named service. | New legacy clean room creation through the web application was already blocked on October 1; the existing connector configurations were not deleted by the September change. Qualify which buyer task faces the issue. [122, 123, 124] | structural | Snowflake blocked new legacy rooms in its web app on October 1. Buyers planning a new Experian setup there need a different route. | New legacy clean room creation through the web application was already blocked on October 1; the existing connector configurations were not deleted by the September change. Qualify which buyer task faces the issue. | evidence/swot/experian-resolution.json, threats.0.t |
| X638 | ["IDN-EXP-S07", "IDN-EXP-S06"] | Independent profile review checked the evidence against the named service. | New legacy clean room creation through the web application was already blocked on October 1; the existing connector configurations were not deleted by the September change. Qualify which buyer task faces the issue. [122, 123, 124] | structural | ["IDN-EXP-S07", "IDN-EXP-S06", "IDN-VEXP-S02"] | New legacy clean room creation through the web application was already blocked on October 1; the existing connector configurations were not deleted by the September change. Qualify which buyer task faces the issue. | evidence/swot/experian-resolution.json, threats.0.src |
| X639 | A sales contact and full-dataset or customer-file options are public. The trial cases show customers using a graph. They do not independently establish the US scale or accuracy of the saleable output. | Independent profile review checked the evidence against the named service. | Distinguish ordinary customer deployments from trials and from independent validation. [117, 21, 37, 118] | structural | The US sales page offers a full dataset or appended links. MiQ describes using the graph, and the seller describes an unnamed DSP. Neither case independently measures US-wide coverage or link accuracy. | Distinguish ordinary customer deployments from trials and from independent validation. | evidence/profiles/experian-resolution.json, entity.status_note |
| X640 | ["IDN-U-S17", "IDN-EXP-S01", "IDN-C-S06", "IDN-C-S05", "IDN-EXP-S02"] | Independent profile review checked the evidence against the named service. | Reference the customer’s direct account of use. [118] | structural | ["IDN-U-S17", "IDN-EXP-S01", "IDN-C-S06", "IDN-C-S05", "IDN-EXP-S02", "IDN-VEXP-S01"] | Reference the customer’s direct account of use. | evidence/profiles/experian-resolution.json, field_evidence.entity.status_note.source_ids |
| X641 | ["IDN-EXP-S01", "IDN-C-S06"] | Independent profile review checked the evidence against the named service. | Include the identified implementing customer’s original. [118] | structural | ["IDN-EXP-S01", "IDN-C-S06", "IDN-VEXP-S01"] | Include the identified implementing customer’s original. | evidence/profiles/experian-resolution.json, field_evidence.who_buys.source_ids |
| X642 | [{"date": "2016-09-14", "event_type": "historical customer evaluation", "event": "Tapad reported results of a Hotels.com comparison for its historical Device Graph. The study is not transferable to the current US SKU.", "source_ids": ["IDN-B-S05"], "finding_ids": ["IDN-EXP-F14"]}, {"date": "2024-04-22", "event_type": "UK decision, outside IQ", "event": "The UK Upper Tribunal dismissed the Commissioner’s appeal about Experian Limited and open-source direct marketing data.", "source_ids": ["IDN-H-S05", "IDN-H-S04"], "finding_ids": ["IDN-EXP-F18"]}, {"date": "2024-09-18", "event_type": "partner availability announcement", "event": "Experian described general availability of collaboration with its offline or digital graph in Snowflake Data Clean Rooms after beta.", "source_ids": ["IDN-EXP-S07", "IDN-EXP-S06"], "finding_ids": ["IDN-EXP-F12"]}, {"date": "2025-10-01", "event_type": "customer case published", "event": "Experian described one DSP’s cookie-sync workflow and weekly graph delivery.", "source_ids": ["IDN-EXP-S01", "IDN-EXP-S14"], "finding_ids": ["IDN-EXP-F04", "IDN-EXP-F08"]}, {"date": "2026-09-10", "event_type": "legacy interface removal", "event": "Snowflake removed its Experian connector from the legacy web interface and preserved existing configurations.", "source_ids": ["IDN-EXP-S07", "IDN-EXP-S06"], "finding_ids": ["IDN-EXP-F12"]}, {"date": "2026-10-20", "event_type": "product description updated", "event": "Experian’s revised article documents its weekly rebuild and the addition of IPv6 and phone-based UID2s. The rollout dates are undisclosed.", "source_ids": ["IDN-EXP-S03"], "finding_ids": ["IDN-EXP-F05"]}] | Independent profile review checked the evidence against the named service. | Add dated ownership, UK litigation and Snowflake setup milestones that delimited the original claims. [46, 357, 124, 123] | structural | [{"date": "2016-09-14", "event_type": "historical customer evaluation", "event": "Tapad reported results of a Hotels.com comparison for its historical Device Graph. The study is not transferable to the current US SKU.", "source_ids": ["IDN-B-S05"], "finding_ids": ["IDN-EXP-F14"]}, {"date": "2020-11-19", "event_type": "completed acquisition", "event": "Experian reported completing its acquisition of Tapad. The 2016 Hotels.com result predates this ownership change.", "source_ids": ["IDN-VEXP-S03"], "finding_ids": ["IDN-EXP-F14"]}, {"date": "2024-04-22", "event_type": "UK decision, outside IQ", "event": "The UK Upper Tribunal dismissed the Commissioner’s appeal about Experian Limited and open-source direct marketing data.", "source_ids": ["IDN-H-S05", "IDN-H-S04"], "finding_ids": ["IDN-EXP-F18"]}, {"date": "2024-05-24", "event_type": "UK regulator update, outside IQ", "event": "The Information Commissioner said it would not seek permission for a further appeal in the UK Experian Limited case.", "source_ids": ["IDN-VEXP-S04"], "finding_ids": ["IDN-EXP-F18"]}, {"date": "2024-09-18", "event_type": "partner availability announcement", "event": "Experian described general availability of collaboration with its offline or digital graph in Snowflake Data Clean Rooms after beta.", "source_ids": ["IDN-EXP-S07", "IDN-EXP-S06"], "finding_ids": ["IDN-EXP-F12"]}, {"date": "2025-10-01", "event_type": "customer case published", "event": "Experian described one DSP’s cookie-sync workflow and weekly graph delivery.", "source_ids": ["IDN-EXP-S01", "IDN-EXP-S14"], "finding_ids": ["IDN-EXP-F04", "IDN-EXP-F08"]}, {"date": "2026-09-10", "event_type": "legacy interface removal", "event": "Snowflake removed its Experian connector from the legacy web interface and preserved existing configurations.", "source_ids": ["IDN-EXP-S07", "IDN-EXP-S06"], "finding_ids": ["IDN-EXP-F12"]}, {"date": "2026-10-01", "event_type": "legacy setup restriction", "event": "Snowflake stopped new legacy clean rooms from being created through its web app. It had already removed the Experian connector from that interface.", "source_ids": ["IDN-VEXP-S02", "IDN-EXP-S06"], "finding_ids": ["IDN-EXP-F12"]}, {"date": "2026-10-20", "event_type": "product description updated", "event": "Experian’s revised article documents its weekly rebuild and the addition of IPv6 and phone-based UID2s. The rollout dates are undisclosed.", "source_ids": ["IDN-EXP-S03"], "finding_ids": ["IDN-EXP-F05"]}] | Add dated ownership, UK litigation and Snowflake setup milestones that delimited the original claims. | evidence/profiles/experian-resolution.json, timeline |
| X643 | ["IDN-C-S05"] | Independent profile review checked the evidence against the named service. | Source the 70% rate to the full seller case and the corresponding directory entry. The buyer blog does not report 70%. [20, 21] | structural | ["IDN-C-S05", "IDN-C-S06"] | Source the 70% rate to the full seller case and the corresponding directory entry. The buyer blog does not report 70%. | evidence/profiles/experian-resolution.json, performance_evidence.0.source_ids |
| X644 | The report does not identify the precise eligible ID base, counts, exclusions or decision threshold for that figure. A synced ID alone does not establish a correct person or household link. | Independent profile review checked the evidence against the named service. | Distinguish within-case change from the 84% share of IDs synced. [37, 36] | structural | The seller also reports a 9% match-rate increase without an absolute baseline or a percentage-point definition. Neither figure gives raw counts, eligibility rules or independently labeled links. | Distinguish within-case change from the 84% share of IDs synced. | evidence/profiles/experian-resolution.json, performance_evidence.1.limits |
| X645 | ["IDN-EXP-S14"] | Independent profile review checked the evidence against the named service. | Both publications describe the same case. [36, 37] | structural | ["IDN-EXP-S14", "IDN-EXP-S01"] | Both publications describe the same case. | evidence/profiles/experian-resolution.json, performance_evidence.1.source_ids |
| X646 | An implementation using the old Snowflake web interface faces a documented connector change. [IDN-EXP-F12] | Independent profile review checked the evidence against the named service. | Make the setup risk specific to the buyer task and date. [123, 124] | structural | Snowflake removed the Experian connector from its legacy web interface in September 2026. Its web app no longer creates new legacy clean rooms. [IDN-EXP-F12] | Make the setup risk specific to the buyer task and date. | evidence/profiles/experian-resolution.json, risks.0 |
| X647 | The US notice sets out a browser-level choice, while Tapad documents device-level effects for its own Graph. The US Digital Graph sales materials do not connect these two policies at the point of export. [IDN-EXP-F01, IDN-EXP-F10, IDN-EXP-F11] | Independent profile review checked the evidence against the named service. | Explicit recipient instruction in the Tapad notice is stronger than a generic privacy policy but provides no measured US Digital Graph execution. [125, 126] | structural | Experian’s US notice describes browser-level GPC handling. Tapad publishes separate device and IPv4 opt-out rules and requires its clients to discard opted-out IDs when refreshed data arrive. The reviewed materials do not show those terms or recipient actions for every US Digital Graph export. [IDN-EXP-F10, IDN-EXP-F11] | Explicit recipient instruction in the Tapad notice is stronger than a generic privacy policy but provides no measured US Digital Graph execution. | evidence/profiles/experian-resolution.json, counterevidence_and_tensions.2 |
| X648 | MiQ Digital Graph case, Results; companion overview, MiQ case entry | Independent profile review checked the evidence against the named service. | The quote is in the directory blurb, not a results section in the standalone case. [20] | structural | Case study resources, “Experian’s Graph increases reach and scale: Case study with MiQ”, descriptive paragraph (line 218). | The quote is in the directory blurb, not a results section in the standalone case. | evidence/profiles/experian-resolution.json, findings.6.support.0.locator |
| X649 | from our products and services for sixty (60) days unless we believe the opted out Identifier | Independent profile review checked the evidence against the named service. | Support the material IPv4 exception with a contiguous quote. The first passage and locator point to the neighboring 60-day device-level paragraph. [126] | structural | Tapad will opt-out the IP address registered for the twelve (12) months | Support the material IPv4 exception with a contiguous quote. The first passage and locator point to the neighboring 60-day device-level paragraph. | evidence/profiles/experian-resolution.json, findings.10.support.1.passage |
| X650 | 2. Offer Meaningful Choice, following paragraph. Read through the condition for later re-entry. | Independent profile review checked the evidence against the named service. | Specify the two separate mechanisms. [126] | structural | IP Address Opt-out, first sentence. For related devices see 2. Offer Meaningful Choice, immediately after the permanent opt-out paragraph. | Specify the two separate mechanisms. | evidence/profiles/experian-resolution.json, findings.10.support.1.locator |
| X651 | The profile re-registered IDN-U-S17, IDN-C-S05, IDN-C-S06, IDN-H-S04, IDN-H-S05, IDN-B-S05 using unchanged Stage 3 source records. | Independent profile review checked the evidence against the named service. | The kit otherwise creates a source alias pointing to itself for each repeated source ID. [117, 20, 21, 66, 67, 45] | structural | Reuse the existing canonical IDs. Retain the independent reopenings in the profile source list and verification receipts. | The kit otherwise creates a source alias pointing to itself for each repeated source ID. | evidence/profiles/experian-resolution.json, sources and reopened_source_ids |
| X652 | At the September 2021 closing, Apollo Funds took over Yahoo and Verizon retained a 10% interest; Verizon later reported $4.1 billion of cash proceeds net of cash transferred for the sale of Verizon Media. | Independent profile review checked the evidence against the named service. | Apollo identified its affiliate-managed funds as acquirers of the former Verizon Media and Verizon’s retained stake at closing. Verizon reports $4.1bn net in liquidity disclosures and $4.3bn at closing before the cash-transferred adjustment in Note 3. The 2025 docket identifies the signatory’s disclosed parent. Dates distinguish group ownership at closing and the later signatory disclosure. Neither supplies a current cap table or ID-specific financials. [329, 347, 348] | supported with qualification | Apollo’s funds acquired Yahoo in September 2021, when Verizon retained 10%. Verizon reported $4.1 billion in cash proceeds net of cash transferred for Verizon Media. In September 2025 Yahoo Ad Tech LLC identified College Parent, L.P. as its corporate parent in a court disclosure. [Additional evidence: IDN-VYAH-S05] | Apollo identified its affiliate-managed funds as acquirers of the former Verizon Media and Verizon’s retained stake at closing. Verizon reports $4.1bn net in liquidity disclosures and $4.3bn at closing before the cash-transferred adjustment in Note 3. The 2025 docket identifies the signatory’s disclosed parent. Dates distinguish group ownership at closing and the later signatory disclosure. Neither supplies a current cap table or ID-specific financials. | IDN-YAH-F02; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X653 | Yahoo’s server integration accepts a publisher ID and a hashed email, then supplies the ConnectID for the ad request; its Prebid workflow can instead start with a domain-specific user identifier. | Independent profile review checked the evidence against the named service. | The authenticated S2S endpoint requires he and pi and documents a ConnectID on an eligible request. Success may instead carry an empty object under opt-out. The Prebid configuration accepts puid, and the Magnite guide describes first-party-cookie fallback. Documented interface behavior only. Neither successful issuance nor first-party state establishes person-link correctness. [165, 189, 163] | supported with qualification | Yahoo documents an authenticated server lookup using a hashed email and publisher ID. An eligible lookup can return a ConnectID for a bid, while an opted-out lookup returns an empty object. Its Prebid route can instead start with a domain-specific publisher ID. | The authenticated S2S endpoint requires he and pi and documents a ConnectID on an eligible request. Success may instead carry an empty object under opt-out. The Prebid configuration accepts puid, and the Magnite guide describes first-party-cookie fallback. Documented interface behavior only. Neither successful issuance nor first-party state establishes person-link correctness. | IDN-YAH-F05; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X654 | Yahoo’s JavaScript module returns a cached ConnectID if available and refreshes it when missing or stale; for its server integration Yahoo recommends a 24-hour cache. | Independent profile review checked the evidence against the named service. | The standalone getIds module is documented to return a local cached value immediately and asynchronously fetch an absent or not-recent value. The S2S recommendation is a 24-hour cache. A distinct pinned Prebid module uses response TTL/default staleness logic and can hand the cached ID back with a refresh callback. Cache freshness is distinct from a guarantee that all recipients reject an old token. [194, 165, 195] | supported with qualification | Yahoo documents immediate use of the standalone JavaScript module’s cached ConnectID and an asynchronous fetch when it is missing or stale. It recommends a 24-hour server cache. A separately versioned Prebid module can return a cached value while refreshing it. [Additional evidence: IDN-VYAH-S04] | The standalone getIds module is documented to return a local cached value immediately and asynchronously fetch an absent or not-recent value. The S2S recommendation is a 24-hour cache. A distinct pinned Prebid module uses response TTL/default staleness logic and can hand the cached ID back with a refresh callback. Cache freshness is distinct from a guarantee that all recipients reject an old token. | IDN-YAH-F06; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X655 | A publisher can also set a local opt-out flag that stops the ConnectID JavaScript module from issuing further IDs on that domain. | Independent profile review checked the evidence against the named service. | The publisher example sets a domain-local storage value to 1 and describes disabling new JS generation. In the 2026 pinned Prebid code getId and decode check the same key and stop when the opt-out is read. Local browser enforcement and a vendor-described server rule do not show retrieval from downstream recipients. [201, 164, 195] | supported as stated | A publisher can also set a local opt-out flag that stops the ConnectID JavaScript module from issuing further IDs on that domain. [Additional evidence: IDN-VYAH-S04] | The publisher example sets a domain-local storage value to 1 and describes disabling new JS generation. In the 2026 pinned Prebid code getId and decode check the same key and stop when the opt-out is read. Local browser enforcement and a vendor-described server rule do not show retrieval from downstream recipients. | IDN-YAH-F08; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X656 | Google lists ConnectID as a secure-signal provider and Yahoo instructs Ad Manager publishers to route it to Yahoo’s App & Display and Video bidders. | Independent profile review checked the evidence against the named service. | Google’s provider row marks the signal supported for web and unsupported for app in this table. In the Authorized Buyers branch Yahoo names two Yahoo bidders. Open Bidding instead uses a signal-sharing toggle. Two bidder configurations in one operator’s DSP. Google’s table describes transport availability, not contract rights for each listed demand channel. [156, 157] | supported with qualification | Google lists ConnectID as a web secure signal. For Authorized Buyers, Yahoo tells Ad Manager publishers to select the Yahoo App & Display and Yahoo Video bidders. Its Open Bidding instructions use a sharing toggle. | Google’s provider row marks the signal supported for web and unsupported for app in this table. In the Authorized Buyers branch Yahoo names two Yahoo bidders. Open Bidding instead uses a signal-sharing toggle. Two bidder configurations in one operator’s DSP. Google’s table describes transport availability, not contract rights for each listed demand channel. | IDN-YAH-F09; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X657 | Yahoo’s Prebid guide identifies version 7.52 as the boundary below which publishers must supply the hashed email parameter. | Independent profile review checked the evidence against the named service. | The help table does print a below-7.52 email requirement. The executable 7.51.0 configuration accepts a string-valued he OR puid; 7.52.0 removed the requirement for either. The release explicitly dates that change to June 1, 2023. The attributed guide wording is accurate; its operational claim is refuted by the tagged code. This is an input-validation change in the integration, not an upstream person-graph change or evidence of a 2025 Chrome sync transition. [164, 189, 191, 192, 193] | contradicted | Yahoo’s guide says an email hash is required below Prebid 7.52. The released 7.51.0 module accepts either an email hash or a publisher user ID. Version 7.52.0, released June 1, 2023, removed the requirement to supply either value. [Additional evidence: IDN-VYAH-S01, IDN-VYAH-S02, IDN-VYAH-S03] | The help table does print a below-7.52 email requirement. The executable 7.51.0 configuration accepts a string-valued he OR puid; 7.52.0 removed the requirement for either. The release explicitly dates that change to June 1, 2023. The attributed guide wording is accurate; its operational claim is refuted by the tagged code. This is an input-validation change in the integration, not an upstream person-graph change or evidence of a 2025 Chrome sync transition. | IDN-YAH-F17; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X658 | On August 26, 2026, the Southern District of New York ordered five plaintiffs’ claims in the private ConnectID suit into arbitration and stayed litigation of those claims. | Independent profile review checked the evidence against the named service. | Signed document 103 grants arbitration and stays the claims of five of six plaintiffs. Separate order 104 instructed parties to confer about possible limited discovery of Baker’s arbitration status; neither decides the merits of privacy allegations. A dated account of the August orders. No verified determination of the sixth plaintiff’s status at the cutoff. [90, 8] | supported as stated | On August 26, 2026, the Southern District of New York ordered five plaintiffs’ claims in the private ConnectID suit into arbitration and stayed litigation of those claims. [Additional evidence: IDN-VYAH-S06] | Signed document 103 grants arbitration and stays the claims of five of six plaintiffs. Separate order 104 instructed parties to confer about possible limited discovery of Baker’s arbitration status; neither decides the merits of privacy allegations. A dated account of the August orders. No verified determination of the sixth plaintiff’s status at the cutoff. | IDN-YAH-F20; data/claims-overlay.json; evidence/verify/yahoo-connectid.json |
| X659 | Form 10-K, liquidity and capital resources, Disposition of Business; also Note 3, Verizon Media Divestiture | Independent profile review checked the evidence against the named service. | Specify the separate locations for net cash and closing cash. [347] | structural | Form 10-K, Cash Flows Used in Investing Activities, Disposition of Business (net cash transferred). Compare Note 3, Verizon Media Divestiture (cash at close). | Specify the separate locations for net cash and closing cash. | evidence/profiles/yahoo-connectid.json, findings.1.support.1.locator |
| X660 | Cache the ConnectID, recommendation is 24 hours; Best Practices, daily cadence to pick up consent changes; contrast the OAuth access-token instruction | Independent profile review checked the evidence against the named service. | The actual section heading is Caching. [165] | structural | Caching, first paragraph (24 hours). Best Practices, instruction to refresh for recent consent changes. OAuth lifetime is covered separately. | The actual section heading is Caching. | evidence/profiles/yahoo-connectid.json, findings.5.support.1.locator |
| X661 | Get the ConnectID, optional GPP and related fields; API Responses, row 200; Opt-Out Handling, success status with empty JSON object | Independent profile review checked the evidence against the named service. | Use the original section names and both parameter and response locations. [165] | structural | GET Yahoo ConnectID, Query Parameters: gpp and gpp_sid. HTTP Status Return Codes, 200. Opt-Out Handling, empty object after opt-out. | Use the original section names and both parameter and response locations. | evidence/profiles/yahoo-connectid.json, findings.6.support.2.locator |
| X662 | Secure signals providers table, Yahoo row, Demand channel column across the wrapped Authorized Buyers and Open Bidding lines | Independent profile review checked the evidence against the named service. | The column is Bidder availability. The row covers web and excludes app. [156] | structural | Secure signals providers, RTB bidders table, Yahoo ConnectID row. Read App and Web, the three Web integration deployment columns, and Bidder availability (Authorized Buyers + Open Bidding). | The column is Bidder availability. The row covers web and excludes app. | evidence/profiles/yahoo-connectid.json, findings.8.support.0.locator |
| X663 | December 1 release, opening paragraph and Grow implementation description | Independent profile review checked the evidence against the named service. | Distinguish the posting day from the dateline. [158] | structural | Newsroom post dated December 2, 2021; December 1 press dateline, first two paragraphs on eCPM and the Grow integration. | Distinguish the posting day from the dateline. | evidence/profiles/yahoo-connectid.json, findings.14.support.0.locator |
| X664 | Configuration table, email requirement and adjacent puid row | Independent profile review checked the evidence against the named service. | Separate the current table from the historical code check. [189] | structural | Prebid project-hosted Configuration table, params.he (optional) and adjacent params.puid; the numbered boundary is in Yahoo’s own guide, not in the pinned April 2025 guide. | Separate the current table from the historical code check. | evidence/profiles/yahoo-connectid.json, findings.16.support.1.locator |
| X665 | 2021-12-01 | Independent profile review checked the evidence against the named service. | The newsroom page labels the post December 2. Its internal release dateline and announcement date are December 1. [158] | structural | 2021-12-02 | The newsroom page labels the post December 2. Its internal release dateline and announcement date are December 1. | evidence/profiles/yahoo-connectid.json, sources.7.pub_date |
| X666 | Yahoo operated as a standalone company under Apollo Funds at the 2021 closing. Verizon then retained a 10% interest. The transaction records screened do not supply a refreshed 2026 cap table for the group or a direct member register for the signatory. [IDN-YAH-F02] | Independent profile review checked the evidence against the named service. | The September 2025 docket supplies a later disclosure identifying the signatory’s corporate parent. [329, 347, 348] | structural | Apollo’s funds acquired Yahoo in September 2021. Verizon retained a 10% interest at closing. In September 2025 Yahoo Ad Tech LLC and Yahoo Inc. each identified College Parent, L.P. as corporate parent and Apollo Global Management as another affiliate in court disclosures. A November 2026 ownership percentage was not established. [IDN-YAH-F02] | The September 2025 docket supplies a later disclosure identifying the signatory’s corporate parent. | evidence/profiles/yahoo-connectid.json, entity.owner_or_parent |
| X667 | Publishers and SSPs integrate. Advertisers and agencies buy through Yahoo DSP. Merkle described an audience-onboarding path, and Truthset described activation of rated segments in that DSP. These are two upstream audience routes to one buyer platform. [IDN-YAH-F09, IDN-YAH-F12, IDN-YAH-F13, IDN-YAH-F21] | Independent profile review checked the evidence against the named service. | The cited release is Yahoo’s, and the evidence explicitly identifies advertisers. A distinct purchase by agencies as principals is unverified. [160, 157, 162, 161, 198] | structural | Participating publishers and approved supply partners integrate the ID. Advertisers use supported audiences in Yahoo DSP. Yahoo’s Merkle announcement and Truthset’s own release identify upstream audiences for that DSP. [IDN-YAH-F04, IDN-YAH-F09, IDN-YAH-F12, IDN-YAH-F13, IDN-YAH-F21] | The cited release is Yahoo’s, and the evidence explicitly identifies advertisers. A distinct purchase by agencies as principals is unverified. | evidence/profiles/yahoo-connectid.json, who_buys |
| X668 | Yahoo links the publisher contract to access to demand in Yahoo DSP. It markets ConnectID inside DSP campaigns. The reviewed contract, onboarding guide and issuer coverage records do not disclose a separate per-ID fee, a take rate, product revenue or margin. [IDN-YAH-F01, IDN-YAH-F02, IDN-YAH-F21] | Independent profile review checked the evidence against the named service. | Separate the channel tied to the service agreement from the financial terms the reviewed documents do not disclose. The acquisition price is a different unit. [160, 199, 198] | structural | Yahoo links the publisher agreement to demand in its DSP and markets ConnectID within DSP campaigns. The reviewed service agreement, integration guide and issuer announcements disclose no separate per-ID fee, take rate, product revenue or margin. [IDN-YAH-F01, IDN-YAH-F04, IDN-YAH-F21] | Separate the channel tied to the service agreement from the financial terms the reviewed documents do not disclose. The acquisition price is a different unit. | evidence/profiles/yahoo-connectid.json, how_it_earns |
| X669 | Net of cash transferred, subject to customary adjustments. Sale of the whole Verizon Media business. Excludes preferred and common units received and is not ConnectID revenue or valuation. | Independent profile review checked the evidence against the named service. | Keep the two cash bases distinct for reconciliation with the same filing. [347] | structural | Verizon reports $4.1 billion net of cash transferred in its 2021 cash-flow discussion. Note 3 states $4.3 billion cash at closing, plus preferred and common units. Both amounts concern Verizon Media. Neither is a ConnectID sale price or operating result. | Keep the two cash bases distinct for reconciliation with the same filing. | evidence/profiles/yahoo-connectid.json, financials.0.definition |
| X670 | Mediavine reported Grow integration and November publisher-yield results. | Independent profile review checked the evidence against the named service. | The dateline supports the event date. The displayed publication date is the next day. [158] | structural | Mediavine’s December 1 release described Grow integration and November publisher yield. The newsroom posted it December 2. | The dateline supports the event date. The displayed publication date is the next day. | evidence/profiles/yahoo-connectid.json, timeline.3.event |
| X671 | SDNY sent five plaintiffs’ claims to arbitration and stayed litigation of those claims. | Independent profile review checked the evidence against the named service. | The sixth plaintiff was the subject of a separate scheduling order, not included in the five-person stay. [90, 8] | structural | SDNY sent five plaintiffs’ claims to arbitration and stayed litigation of those claims. A separate order asked the parties to confer about possible discovery on the sixth plaintiff’s arbitration status. | The sixth plaintiff was the subject of a separate scheduling order, not included in the five-person stay. | evidence/profiles/yahoo-connectid.json, timeline.10.event |
| X672 | ["IDN-YAH-S17"] | Independent profile review checked the evidence against the named service. | Add the separate signed court order on the remaining plaintiff. [90, 8] | structural | ["IDN-YAH-S17", "IDN-VYAH-S06"] | Add the separate signed court order on the remaining plaintiff. | evidence/profiles/yahoo-connectid.json, timeline.10.source_ids |
| X673 | Published interfaces document the handoff into Yahoo demand. They cannot establish a unique reachable person count for an advertiser’s campaign. [IDN-YAH-F09, IDN-YAH-F14] | Independent profile review checked the evidence against the named service. | Separate the reason a campaign count is unavailable from an interface specification. [156, 157, 197, 198] | structural | Published interfaces document the handoff into Yahoo demand. Yahoo’s reach estimates do not establish how many unique people a particular campaign can reach. [IDN-YAH-F09, IDN-YAH-F14] | Separate the reason a campaign count is unavailable from an interface specification. | evidence/profiles/yahoo-connectid.json, limits.0 |
| X674 | A publisher’s cached ID and an already transmitted copy have a different lifecycle from the result of a fresh opted-out lookup. [IDN-YAH-F06, IDN-YAH-F07, IDN-YAH-F08] | Independent profile review checked the evidence against the named service. | The March 2026 implementation distinguishes suppression of locally cached data from recall of delivered values. [165, 201, 195] | structural | A cached Prebid value can accompany a refresh request. The local opt-out can block decoding in the examined module. Neither that control nor an empty new server lookup demonstrates removal of copies already delivered. [IDN-YAH-F06, IDN-YAH-F07, IDN-YAH-F08] | The March 2026 implementation distinguishes suppression of locally cached data from recall of delivered values. | evidence/profiles/yahoo-connectid.json, risks.2 |
| X675 | Which country-specific agreements, current ownership records and product financial disclosures can settle the limits identified in the US contract and the corporate sale filing? | Independent profile review checked the evidence against the named service. | The new disclosure identifies a corporate parent but supplies no current ownership percentage. [160, 347, 348] | structural | Which country-specific agreements, current ownership percentages and product financial disclosures can resolve the limits of the US contract, the dated parent disclosure and the corporate sale filing? | The new disclosure identifies a corporate parent but supplies no current ownership percentage. | evidence/profiles/yahoo-connectid.json, open_questions.4 |
| X676 | 2026-11-11 Web + source repositories: dated Prebid Yahoo module history, version 7.52, Verizon Media December 2020 ConnectID predecessor technical migration and edge lineage. The April 2025 pinned source was opened from the original GitHub raw file. Current guide and integration index examined. A current undated guide supplies no dated comparable post-event edition. No documented default US workflow or corroborating edge migration found in these files. | Independent profile review checked the evidence against the named service. | Extend the search boundary for version history without claiming a Chrome-default result. [164, 202, 191, 192, 193, 195] | structural | 2026-11-11 Web + source repositories: dated Prebid Yahoo module history, version 7.52, Verizon Media December 2020 ConnectID predecessor technical migration and edge lineage. The April 2025 pinned source was opened from the original GitHub raw file. Current guide and integration index examined. A current undated guide supplies no dated comparable post-event edition. No documented default US workflow or corroborating edge migration found in these files. Refutation subsequently compared original Prebid.js releases 7.51.0 and 7.52.0, release #9965 and the March 7, 2026 Yahoo module. Those sources date an input rule and client cache behavior; none designates the main US Chrome workflow. | Extend the search boundary for version history without claiming a Chrome-default result. | evidence/profiles/yahoo-connectid.json, searches.4 |
| X677 | 2026-11-11 Web + publisher originals: Yahoo ConnectID service agreement pricing signatory privacy; Verizon 2021 Form 10-K Verizon Media cash proceeds and Apollo Yahoo closing; current equity and product revenue checks. Checked service agreement, integration index, SEC annual filing and closing release. No later cap table or standalone ConnectID revenue in the opened set. | Independent profile review checked the evidence against the named service. | Disclose the later source used for the entity field. [348] | structural | 2026-11-11 Web + publisher originals: Yahoo ConnectID service agreement pricing signatory privacy; Verizon 2021 Form 10-K Verizon Media cash proceeds and Apollo Yahoo closing; current equity and product revenue checks. Checked service agreement, integration index, SEC annual filing and closing release. No later cap table or standalone ConnectID revenue in the opened set. Refutation read the public September 26, 2025 SDNY docket descriptions of disclosures 53 and 54, which identify the corporate parent. The complete attachments and a refreshed ownership table were not established. | Disclose the later source used for the entity field. | evidence/profiles/yahoo-connectid.json, searches.0 |
| X678 | No contrary product-specific record found in the named sources. | Independent profile review checked the evidence against the named service. | The original claim of no contrary record no longer holds. [191, 192, 193] | structural | Published Prebid.js 7.51.0 accepts either he or puid; the 7.52.0 source and release notes remove the requirement for either. Yahoo’s undated help wording is contradicted for 7.51.0. | The original claim of no contrary record no longer holds. | evidence/profiles/yahoo-connectid.json, findings.16.counterevidence |
| X679 | This version boundary concerns email input. The guide and current publisher documentation do not identify a main US Chrome integration or establish a before/after shift in third-party sync around the April 2025 Chrome decision. | Independent profile review checked the evidence against the named service. | Bound the corrected version claim to the alternate client inputs. [191, 192, 193] | structural | The dated release changes client input validation. It does not show production uptake, person-link accuracy, new graph edges or a shift in the primary US Chrome sync path. | Bound the corrected version claim to the alternate client inputs. | evidence/profiles/yahoo-connectid.json, findings.16.limitations |
| X680 | Attribution, scope and caveats follow the linked sections; issuer, partner and court records establish different propositions. | Independent profile review checked the evidence against the named service. | Replace the generic confidence rationale with the provenance for the repaired claim. [191, 192, 193] | structural | High confidence in the released client rule from a comparison of tagged versions and the June 1, 2023 release note. The help article’s required-email statement conflicts with that rule. | Replace the generic confidence rationale with the provenance for the repaired claim. | evidence/profiles/yahoo-connectid.json, findings.16.confidence_reason |
| X681 | The recommendation concerns the publisher’s cache. The reviewed guides do not establish a universal token lifetime, underlying person-edge age, or maximum lifetime of every historical copy. | Independent profile review checked the evidence against the named service. | Source code adds a specific exception to any assumed hard expiration. [165, 195] | structural | The 24-hour recommendation concerns the publisher’s S2S cache. In the examined Prebid client, a stale ID can accompany a resync; its client TTL does not establish downstream token validity, graph-edge age or maximum lifetime of a previously transmitted copy. | Source code adds a specific exception to any assumed hard expiration. | evidence/profiles/yahoo-connectid.json, findings.5.limitations |
| X682 | The publisher is responsible for controls and the applicable preferences. An empty response establishes suppression at lookup. It does not establish deletion from recipients that already received the token. No live opt-out transaction was performed. | Independent profile review checked the evidence against the named service. | The vendor documentation specifies behavior; no operational test was performed. [160, 165, 200] | structural | The service agreement assigns controls to publishers. The API specifies an empty result on a new opted-out lookup. No live request was tested, and removal of historical copies at recipients was not demonstrated. | The vendor documentation specifies behavior; no operational test was performed. | evidence/profiles/yahoo-connectid.json, findings.6.limitations |
| X683 | On the server route the seller supplies the email hash and the Yahoo-issued publisher ID. On the Prebid route the hashed email field is optional from version 7.52. A publisher user ID, including a first-party cookie, is another documented input. Yahoo’s publisher guide includes an example before registration and a repeat after login. | Independent profile review checked the evidence against the named service. | Versioned project code overturns the guide’s historical requirement. The documented S2S and optional browser inputs remain C technical evidence. [164, 163, 189, 165, 191, 192, 193] | structural | The server route requires the seller’s email hash and Yahoo’s publisher ID. On the Prebid route, 7.51.0 accepted an email hash or a publisher ID. Version 7.52.0 removed the need for either. A publisher can pass a local ID, and the Magnite wrapper documents a first-party-cookie fallback. | Versioned project code overturns the guide’s historical requirement. The documented S2S and optional browser inputs remain C technical evidence. | evidence/profiles/yahoo-connectid.json, capabilities.input_provenance.observed |
| X684 | ["IDN-YAH-S04", "IDN-YAH-S15", "IDN-YAH-S16", "IDN-C-S23"] | Independent profile review checked the evidence against the named service. | Cite the evidence used in the repaired capability observation. [164, 163, 189, 165, 191, 192, 193] | structural | ["IDN-YAH-S04", "IDN-YAH-S15", "IDN-YAH-S16", "IDN-C-S23", "IDN-VYAH-S01", "IDN-VYAH-S02", "IDN-VYAH-S03"] | Cite the evidence used in the repaired capability observation. | evidence/profiles/yahoo-connectid.json, capabilities.input_provenance.source_ids |
| X685 | The JS module uses a cached ID when available. On the server path Yahoo recommends a 24-hour cache and periodic refresh to reflect changed consent. The OAuth credential has its own expiry. The guides do not establish a hard lifetime for the returned ID or its underlying graph edge. | Independent profile review checked the evidence against the named service. | Sources establish three route-specific behaviors. The 14-day limit is a client staleness boundary with exceptions, not a graph-edge or DSP rejection rule. Grade C fits inspected integration code. [194, 165, 195] | structural | Yahoo’s standalone JavaScript may return a cached ID and fetch asynchronously. Yahoo recommends a 24-hour S2S cache. The March 2026 Prebid module checks an external publisher’s stored ConnectID for staleness using the response TTL, up to 14 days, or a 14-day default. A stale ID may accompany the refresh callback. No hard recipient or graph-edge expiry is established. | Sources establish three route-specific behaviors. The 14-day limit is a client staleness boundary with exceptions, not a graph-edge or DSP rejection rule. Grade C fits inspected integration code. | evidence/profiles/yahoo-connectid.json, capabilities.refresh_expiry.observed |
| X686 | ["IDN-YAH-S05", "IDN-C-S23"] | Independent profile review checked the evidence against the named service. | Cite the evidence used in the repaired capability observation. [194, 165, 195] | structural | ["IDN-YAH-S05", "IDN-C-S23", "IDN-VYAH-S04"] | Cite the evidence used in the repaired capability observation. | evidence/profiles/yahoo-connectid.json, capabilities.refresh_expiry.source_ids |
| X687 | The US contract requires preference forwarding. The server guide gives optional GPP and section-ID inputs and an HTTP 200 empty JSON response for a user opted out through Yahoo or a covered framework. A caller with that response has no issued ID to pass. The sources do not give an implementation-wide decision rule for missing signals. | Independent profile review checked the evidence against the named service. | Contract is primary for supplier duties. Yahoo documents the S2S opt-out response. Pinned Prebid code forwards GPP data when available and has distinct local-storage and exception handling. C for technical implementation evidence. [160, 164, 165, 195] | structural | The agreement requires suppliers to forward applicable preferences. Yahoo documents optional GPP inputs and an empty HTTP 200 response on an opted-out S2S lookup. In the March 2026 Prebid module the local flag gates getId and decode. Disabled local storage and a storage exception take different branches. No single rule for all missing signals or recipients is established. | Contract is primary for supplier duties. Yahoo documents the S2S opt-out response. Pinned Prebid code forwards GPP data when available and has distinct local-storage and exception handling. C for technical implementation evidence. | evidence/profiles/yahoo-connectid.json, capabilities.permission_handling.observed |
| X688 | ["IDN-YAH-S01", "IDN-YAH-S04", "IDN-C-S23"] | Independent profile review checked the evidence against the named service. | Cite the evidence used in the repaired capability observation. [160, 164, 165, 195] | structural | ["IDN-YAH-S01", "IDN-YAH-S04", "IDN-C-S23", "IDN-VYAH-S04"] | Cite the evidence used in the repaired capability observation. | evidence/profiles/yahoo-connectid.json, capabilities.permission_handling.source_ids |
| X689 | Yahoo DSP expects the yahoo.com EID in user.ext.eids for OpenRTB 2.5 and user.eids for 2.6, according to the server guide. The publisher Prebid guide supplies a network inspection check. On Google Ad Manager the publisher selects the relevant Yahoo buyers. These are implementation instructions rather than transaction logs. | Independent profile review checked the evidence against the named service. | Versioned OpenRTB examples are a C-grade interface specification. The Authorized Buyers and Open Bidding instructions differ. [164, 156, 157, 165] | structural | Yahoo specifies the yahoo.com EID at user.ext.eids for OpenRTB 2.5 and user.eids for 2.6. Prebid supplies a payload check. Ad Manager’s Authorized Buyers setup names two Yahoo bidders; Open Bidding uses a sharing toggle. The guides provide no transaction log. | Versioned OpenRTB examples are a C-grade interface specification. The Authorized Buyers and Open Bidding instructions differ. | evidence/profiles/yahoo-connectid.json, capabilities.recipient_interface.observed |
| X690 | The consumer-facing email portal provides an opt-out path. A local flag disables new JS issuance on that domain. Yahoo documents an empty opted-out result on a new server lookup. The issuer recommends daily cache refresh. No downstream deletion receipt or bound on existing recipient copies was found in these controls. | Independent profile review checked the evidence against the named service. | Documented S2S responses and scope-limited local state differ from recipient-side deletion. Code allows the effect on a stored value to be described more precisely. Grade C for technical evidence. [164, 200, 201, 165, 195] | structural | Yahoo offers an email opt-out form and specifies an empty opted-out S2S result on a new lookup. A domain-local flag stops new JS generation. In the examined Prebid code it also blocks decoding a stored ID. Yahoo recommends a daily server cache refresh. The reviewed interfaces provide no receipt or deadline for deletion of earlier recipient copies. | Documented S2S responses and scope-limited local state differ from recipient-side deletion. Code allows the effect on a stored value to be described more precisely. Grade C for technical evidence. | evidence/profiles/yahoo-connectid.json, capabilities.withdrawal_controls.observed |
| X691 | ["IDN-YAH-S04", "IDN-YAH-S18", "IDN-A-S13", "IDN-C-S23"] | Independent profile review checked the evidence against the named service. | Cite the evidence used in the repaired capability observation. [164, 200, 201, 165, 195] | structural | ["IDN-YAH-S04", "IDN-YAH-S18", "IDN-A-S13", "IDN-C-S23", "IDN-VYAH-S04"] | Cite the evidence used in the repaired capability observation. | evidence/profiles/yahoo-connectid.json, capabilities.withdrawal_controls.source_ids |
| X692 | The Prebid guide states a version boundary for required hashed email. That lets a buyer distinguish older and newer integration inputs. The screened material does not date a change to graph edges, or a change in US third-party-cookie sync instructions. | Independent profile review checked the evidence against the named service. | The 2023 source pair and release provide a dated before/after change and refute the guide’s historic email requirement. Grade C, a technical record of a client-side release. [164, 189, 202, 191, 192, 193] | structural | Prebid.js 7.51.0 accepted an email hash or a publisher user ID. Version 7.52.0, released June 1, 2023, removed the requirement for either. Yahoo’s current help table overstates the older email requirement. These versions document a publisher configuration change. They do not date a change to Yahoo’s graph edges or the primary US Chrome sync route. | The 2023 source pair and release provide a dated before/after change and refute the guide’s historic email requirement. Grade C, a technical record of a client-side release. | evidence/profiles/yahoo-connectid.json, capabilities.change_traceability.observed |
| X693 | ["IDN-YAH-S04", "IDN-YAH-S16", "IDN-E-S28"] | Independent profile review checked the evidence against the named service. | Cite the evidence used in the repaired capability observation. [164, 189, 202, 191, 192, 193] | structural | ["IDN-YAH-S04", "IDN-YAH-S16", "IDN-E-S28", "IDN-VYAH-S01", "IDN-VYAH-S02", "IDN-VYAH-S03"] | Cite the evidence used in the repaired capability observation. | evidence/profiles/yahoo-connectid.json, capabilities.change_traceability.source_ids |
| X694 | 2026-11-11 | Independent profile review checked the evidence against the named service. | The verifiable effective change precedes the undated guide inspection. [191, 192, 193] | structural | Prebid.js 7.51.0 released May 25, 2023; 7.52.0 released June 1, 2023; Yahoo guide read November 11, 2026. | The verifiable effective change precedes the undated guide inspection. | evidence/profiles/yahoo-connectid.json, findings.16.scope_time |
| X695 | Publishers can configure Google Ad Manager to deliver ConnectID to Yahoo’s named bidders, using a secure-signal provider Google lists. | Independent profile review checked the evidence against the named service. | Google’s product row and Yahoo’s configuration guide agree on the available web transport. Selection by bidder name is specifically the Authorized Buyers branch. [156, 157] | structural | On web inventory, Ad Manager publishers can configure the Google-listed ConnectID signal for Yahoo’s named Authorized Buyers bidders. | Google’s product row and Yahoo’s configuration guide agree on the available web transport. Selection by bidder name is specifically the Authorized Buyers branch. | evidence/swot/yahoo-connectid.json, strengths.0.t |
| X696 | The public contract restricts onward sharing. The reviewed placements document Yahoo DSP as the buyer destination. | Independent profile review checked the evidence against the named service. | The contract directly establishes permitted sharing and directs bids to Yahoo. Use its exact obligations so the record basis covers the whole line. [160] | structural | The US agreement directs publishers to include ConnectID in bids to Yahoo and restricts sharing beyond the company and its approved publishers. | The contract directly establishes permitted sharing and directs bids to Yahoo. Use its exact obligations so the record basis covers the whole line. | evidence/swot/yahoo-connectid.json, weaknesses.0.t |
| X697 | ["IDN-YAH-S01", "IDN-YAH-S07"] | Independent profile review checked the evidence against the named service. | The contract directly establishes permitted sharing and directs bids to Yahoo. Use its exact obligations so the record basis covers the whole line. [160] | structural | ["IDN-YAH-S01"] | The contract directly establishes permitted sharing and directs bids to Yahoo. Use its exact obligations so the record basis covers the whole line. | evidence/swot/yahoo-connectid.json, weaknesses.0.src |
| X698 | Mediavine reports publisher yield and Truthset grades its audiences. Neither record tests the accuracy of Yahoo’s person links. | Independent profile review checked the evidence against the named service. | Truthset describes independently scored audiences; the announcement does not establish who performed each rating or test Yahoo’s person-level matching. The eCPM result concerns yield. [158, 161] | structural | Mediavine reports publisher yield. Truthset describes ratings of its own audiences. Neither account tests whether Yahoo linked the right person. | Truthset describes independently scored audiences; the announcement does not establish who performed each rating or test Yahoo’s person-level matching. The eCPM result concerns yield. | evidence/swot/yahoo-connectid.json, weaknesses.1.t |
| X699 | Publisher consent or a user opt-out can stop new issuance, reducing the identifiable supply available for a planned buy. | Independent profile review checked the evidence against the named service. | The agreement requires notices and permission where applicable. A lack of required consent or an applicable opt-out may suppress supply; obtaining consent does not itself stop issuance. [160, 164, 165] | structural | If required consent is missing or a user opts out, fewer eligible ConnectIDs may be available for a planned buy. | The agreement requires notices and permission where applicable. A lack of required consent or an applicable opt-out may suppress supply; obtaining consent does not itself stop issuance. | evidence/swot/yahoo-connectid.json, threats.0.t |
| X700 | ["IDN-YAH-S01", "IDN-YAH-S04", "IDN-YAH-S18"] | Independent profile review checked the evidence against the named service. | The agreement requires notices and permission where applicable. A lack of required consent or an applicable opt-out may suppress supply; obtaining consent does not itself stop issuance. [160, 164, 165] | structural | ["IDN-YAH-S01", "IDN-YAH-S04", "IDN-C-S23"] | The agreement requires notices and permission where applicable. A lack of required consent or an applicable opt-out may suppress supply; obtaining consent does not itself stop issuance. | evidence/swot/yahoo-connectid.json, threats.0.src |
| X701 | In May 2020, Comcast, Charter and ViacomCBS announced an agreement for equal ownership of Blockgraph, an initiative Charter dates to 2017. | Independent profile review checked the evidence against the named service. | The Charter link now redirects to its newsroom. The same joint release is intact on Blockgraph’s site. Dated May 29, 2020, it records the equal-ownership agreement and dates the initiative’s launch to 2017 with incubation inside FreeWheel. Both URLs are retained so the redirected original and working copy can be audited. [332, 333] | supported with qualification | In a May 29, 2020 joint release, Comcast, Charter and ViacomCBS announced an agreement for equal ownership of Blockgraph and dated the original FreeWheel initiative to 2017. [Additional evidence: IDN-VCTV-S01] | The Charter link now redirects to its newsroom. The same joint release is intact on Blockgraph’s site. Dated May 29, 2020, it records the equal-ownership agreement and dates the initiative’s launch to 2017 with incubation inside FreeWheel. Both URLs are retained so the redirected original and working copy can be audited. | IDN-CTV-F05; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X702 | The v2.29 enterprise guide requires a household record ID and postal address and permits additional signals for that household, including IP and set-top-box IDs. | Independent profile review checked the evidence against the named service. | Version 2.29 specifies a household record ID plus a valid postal address when establishing identity. It lists optional IP, email, device and set-top-box signals. The rule establishes the documented input contract, without testing the resulting links. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. [109] | supported as stated | The v2.29 enterprise guide requires a household record ID and postal address and permits additional signals for that household, including IP and set-top-box IDs. [Evidence class: vendor_assertion] | Version 2.29 specifies a household record ID plus a valid postal address when establishing identity. It lists optional IP, email, device and set-top-box signals. The rule establishes the documented input contract, without testing the resulting links. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. | IDN-CTV-F07; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X703 | Blockgraph’s v2.29 manual distinguishes deleting a subscriber record from opting it out: opt-out makes existing segments and matches stale until refreshed. | Independent profile review checked the evidence against the named service. | The two removal paths differ. Delete leaves existing segments and matches available. Opt Out removes the record from the subscriber file, marks affected existing segments and matches stale, and requires refresh before later use. Recall of previously exported copies is unspecified. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. [115] | supported as stated | Blockgraph’s v2.29 manual distinguishes deleting a subscriber record from opting it out: opt-out makes existing segments and matches stale until refreshed. [Evidence class: vendor_assertion] | The two removal paths differ. Delete leaves existing segments and matches available. Opt Out removes the record from the subscriber file, marks affected existing segments and matches stale, and requires refresh before later use. Recall of previously exported copies is unspecified. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. | IDN-CTV-F08; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X704 | For historical IP lookup, Blockgraph documents optional row timestamps and configurable lookback periods of 30, 60 or 90 days when the provider maintains household signal history. | Independent profile review checked the evidence against the named service. | Version 2.29 permits a per-row optional timestamp for IP. Historical signal resolution depends on the provider keeping a household’s earlier signal values and on its configured 30-, 60- or 90-day retention. No current-link maximum age is supplied. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. Time-indexed historical lookups, post-impression attribution and expiry of the currently accepted link are separate objects. [114] | supported as stated | For historical IP lookup, Blockgraph documents optional row timestamps and configurable lookback periods of 30, 60 or 90 days when the provider maintains household signal history. [Evidence class: vendor_assertion] | Version 2.29 permits a per-row optional timestamp for IP. Historical signal resolution depends on the provider keeping a household’s earlier signal values and on its configured 30-, 60- or 90-day retention. No current-link maximum age is supplied. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. Time-indexed historical lookups, post-impression attribution and expiry of the currently accepted link are separate objects. | IDN-CTV-F09; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X705 | Blockgraph’s v2.29 Universe Match workflow lists partners that granted permission and specifies a downloadable file containing one column of Blockgraph IDs. | Independent profile review checked the evidence against the named service. | The v2.29 workflow lists participants that granted permission. A participant can calculate or refresh the match count, then obtain the universe match result. The output guide specifies a headerless CSV column of BG IDs. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. [110, 111] | supported as stated | Blockgraph’s v2.29 Universe Match workflow lists partners that granted permission and specifies a downloadable file containing one column of Blockgraph IDs. [Evidence class: vendor_assertion] | The v2.29 workflow lists participants that granted permission. A participant can calculate or refresh the match count, then obtain the universe match result. The output guide specifies a headerless CSV column of BG IDs. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. | IDN-CTV-F10; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X706 | Kantar reported a greater-than-tenfold rise in its TV partner match rates after integrating Blockgraph’s IDoS. | Independent profile review checked the evidence against the named service. | Kantar’s November 1 release reports more than ten times the prior rate in its integration with TV partners and notes direct matching of panel and client data. Neither absolute rate, test period, denominator nor correctness is provided. Blockgraph’s CEO later rounds the same result to tenfold. The Kantar release says over tenfold. The CEO’s rounded wording and his later November 7 article are not separate study results. [112, 113] | supported with qualification | Kantar reported that its match rates with TV partners rose to more than ten times their prior level after integrating Blockgraph’s IDoS. The partners disclosed neither absolute rate nor the eligible record counts. [Additional evidence: IDN-CTV-S17] | Kantar’s November 1 release reports more than ten times the prior rate in its integration with TV partners and notes direct matching of panel and client data. Neither absolute rate, test period, denominator nor correctness is provided. Blockgraph’s CEO later rounds the same result to tenfold. The Kantar release says over tenfold. The CEO’s rounded wording and his later November 7 article are not separate study results. | IDN-CTV-F11; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X707 | Spectrum Reach describes a production workflow in which it supplies household TV exposure, Innovid supplies household website visits and Blockgraph performs the match. | Independent profile review checked the evidence against the named service. | Spectrum says its set-top-box and streaming-device data supply household exposure, Innovid collects visits, and Blockgraph performs the household join. It sells reporting based on this workflow. This is evidence from a participating seller, not independently observed production performance. [38, 167] | supported as stated | Spectrum Reach describes a production workflow in which it supplies household TV exposure, Innovid supplies household website visits and Blockgraph performs the match. [Additional evidence: IDN-CTV-S09] | Spectrum says its set-top-box and streaming-device data supply household exposure, Innovid collects visits, and Blockgraph performs the household join. It sells reporting based on this workflow. This is evidence from a participating seller, not independently observed production performance. | IDN-CTV-F12; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X708 | Spectrum Reach’s showcase describes visits counted within seven days of a TV exposure in its multiscreen attribution workflow. | Independent profile review checked the evidence against the named service. | The Spectrum contribution on printed page 240 says reporting counted site visits within seven days of a traditional or streaming TV exposure. On Spectrum’s separate FAQ, seven concurrent days of input data before reporting is a different clock. Printed PDF page 240 was rendered and read because normal text extraction misdecodes its typeface. [167, 38] | supported as stated | Spectrum Reach’s showcase describes visits counted within seven days of a TV exposure in its multiscreen attribution workflow. [Additional evidence: IDN-CTV-S08] | The Spectrum contribution on printed page 240 says reporting counted site visits within seven days of a traditional or streaming TV exposure. On Spectrum’s separate FAQ, seven concurrent days of input data before reporting is a different clock. Printed PDF page 240 was rendered and read because normal text extraction misdecodes its typeface. | IDN-CTV-F13; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X709 | The August 2026 release added SHA256-hashed email as a preset signal for identity upload, batch resolution and other workflows. | Independent profile review checked the evidence against the named service. | Release 2.29 is marked August 2026. It introduces pre-hashed SHA256 email as a preset option for identity upload, segment creation, batch identity resolution and measurement data processing. Later release 3.0.0 does not change this historical introduction date. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. [116, 172] | supported as stated | The August 2026 release added SHA256-hashed email as a preset signal for identity upload, batch resolution and other workflows. [Evidence class: vendor_assertion] [Additional evidence: IDN-VCTV-S02] | Release 2.29 is marked August 2026. It introduces pre-hashed SHA256 email as a preset option for identity upload, segment creation, batch identity resolution and measurement data processing. Later release 3.0.0 does not change this historical introduction date. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. | IDN-CTV-F15; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X710 | An April 2025 CIMM report describes an IDoS footprint above 130 million households without specifying the active-period count, deduplication procedure or US eligible base for that figure. | Independent profile review checked the evidence against the named service. | Printed page 26 of CIMM’s April 2025 report describes use across 130M+ households. It gives no study, cited census, deduplication rule, national eligible universe or counting period for the number. Neither current reachable US households nor accuracy follows. CIMM’s descriptive report supplies a figure with no identified underlying count or named source in this bullet. Use synthesis for the report plus the disclosure check; retain a D-grade footprint claim. Single-column raw PDF text and the page layout place 130M+ in the same Blockgraph bullet. [6] | supported with qualification | An April 2025 CIMM report describes IDoS as operating across 130 million-plus households. It gives no geography, active period, deduplication rule or eligible household base for that figure. [Evidence class: synthesis] | Printed page 26 of CIMM’s April 2025 report describes use across 130M+ households. It gives no study, cited census, deduplication rule, national eligible universe or counting period for the number. Neither current reachable US households nor accuracy follows. CIMM’s descriptive report supplies a figure with no identified underlying count or named source in this bullet. Use synthesis for the report plus the disclosure check; retain a D-grade footprint claim. Single-column raw PDF text and the page layout place 130M+ in the same Blockgraph bullet. | IDN-CTV-F16; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X711 | Blockgraph’s 2021 notice directs clients’ customers to the client and describes assistance with removal or amendment at the client’s direction. | Independent profile review checked the evidence against the named service. | Section 6 explicitly concerns data received from clients. It directs a customer seeking to stop contact back to the client. If that client requests help removing or amending its data, Blockgraph will cooperate under its contract. No deadline or automatic signal propagation is specified. [170] | supported with qualification | Blockgraph’s 2021 notice directs customers seeking to stop contact to the client they deal with and says Blockgraph will assist with removing or amending client data at that client’s direction under its contract. | Section 6 explicitly concerns data received from clients. It directs a customer seeking to stop contact back to the client. If that client requests help removing or amending its data, Blockgraph will cooperate under its contract. No deadline or automatic signal propagation is specified. | IDN-CTV-F17; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X712 | Blockgraph’s CEO gave a broad match-rate definition in the commentary about Kantar, without specifying how that case counted eligible or matched records. | Independent profile review checked the evidence against the named service. | The CEO defines match rate generically as the share of a marketer’s audience data linkable to media purchases. Neither his account nor the original Kantar release says how Kantar counted eligible records or matched households. The accounts describe one relationship. [113, 112] | supported as stated | Blockgraph’s CEO gave a broad match-rate definition in the commentary about Kantar, without specifying how that case counted eligible or matched records. [Additional evidence: IDN-CTV-S07] | The CEO defines match rate generically as the share of a marketer’s audience data linkable to media purchases. Neither his account nor the original Kantar release says how Kantar counted eligible records or matched households. The accounts describe one relationship. | IDN-CTV-F18; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X713 | A complete manual refresh in version 2.29 replaces the node’s identity layer and omits records absent from the replacement file. | Independent profile review checked the evidence against the named service. | A complete v2.29 manual refresh replaces the whole identity layer. Records absent from the new file are removed; users can run the refresh at any time. It is distinct from the separate Delete and Opt Out actions. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. [115] | supported as stated | A complete manual refresh in version 2.29 replaces the node’s identity layer and omits records absent from the replacement file. [Evidence class: vendor_assertion] | A complete v2.29 manual refresh replaces the whole identity layer. Records absent from the new file are removed; users can run the refresh at any time. It is distinct from the separate Delete and Opt Out actions. The deciding source is the seller’s technical manual or release note (grade C). It establishes the documented behavior of the specified version; execution or accuracy has not been independently observed. | IDN-CTV-F21; data/claims-overlay.json; evidence/verify/blockgraph.json |
| X714 | full | Independent profile review checked the evidence against the named service. | The redirected original no longer delivers the article. [332, 333] | structural | Original Charter URL redirects to its general newsroom. The complete joint announcement was read in the issuer newsroom under IDN-VCTV-S01. | The redirected original no longer delivers the article. | evidence/profiles/blockgraph.json, sources.4.access_limits |
| X715 | full | Independent profile review checked the evidence against the named service. | Document content was recoverable through the public web renderer. [423] | structural | Full two-page issuer PDF via indexed web PDF extraction. The direct Mozilla request to the CDN returned HTTP 503. | Document content was recoverable through the public web renderer. | evidence/profiles/blockgraph.json, sources.3.access_limits |
| X716 | 2022-01-21 | Independent profile review checked the evidence against the named service. | The ©2022 footer identifies the year. The day appears within a sheet identifier, not a labelled publication date. [423] | structural | 2022 | The ©2022 footer identifies the year. The day appears within a sheet identifier, not a labelled publication date. | evidence/profiles/blockgraph.json, sources.3.pub_date |
| X717 | 2022-01-21 | Independent profile review checked the evidence against the named service. | Keep document identifier separate from the publication date. [423] | structural | 2022 solution sheet; document identifier ends in 01.21.2022 | Keep document identifier separate from the publication date. | evidence/profiles/blockgraph.json, findings.3.scope_time |
| X718 | (field absent) | Independent profile review checked the evidence against the named service. | The page path is versioned; the release note dates the release. [109, 116] | structural | 2.29 (release dated August 2026) | The page path is versioned; the release note dates the release. | evidence/profiles/blockgraph.json, sources.10.version |
| X719 | 2026-08 | Independent profile review checked the evidence against the named service. | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. [109, 116] | structural | (empty) | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. | evidence/profiles/blockgraph.json, sources.10.pub_date |
| X720 | (field absent) | Independent profile review checked the evidence against the named service. | The page path is versioned; the release note dates the release. [115, 116] | structural | 2.29 (release dated August 2026) | The page path is versioned; the release note dates the release. | evidence/profiles/blockgraph.json, sources.11.version |
| X721 | 2026-08 | Independent profile review checked the evidence against the named service. | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. [115, 116] | structural | (empty) | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. | evidence/profiles/blockgraph.json, sources.11.pub_date |
| X722 | (field absent) | Independent profile review checked the evidence against the named service. | The page path is versioned; the release note dates the release. [110, 116] | structural | 2.29 (release dated August 2026) | The page path is versioned; the release note dates the release. | evidence/profiles/blockgraph.json, sources.12.version |
| X723 | 2026-08 | Independent profile review checked the evidence against the named service. | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. [110, 116] | structural | (empty) | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. | evidence/profiles/blockgraph.json, sources.12.pub_date |
| X724 | (field absent) | Independent profile review checked the evidence against the named service. | The page path is versioned; the release note dates the release. [111, 116] | structural | 2.29 (release dated August 2026) | The page path is versioned; the release note dates the release. | evidence/profiles/blockgraph.json, sources.13.version |
| X725 | 2026-08 | Independent profile review checked the evidence against the named service. | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. [111, 116] | structural | (empty) | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. | evidence/profiles/blockgraph.json, sources.13.pub_date |
| X726 | (field absent) | Independent profile review checked the evidence against the named service. | The page path is versioned; the release note dates the release. [114, 116] | structural | 2.29 (release dated August 2026) | The page path is versioned; the release note dates the release. | evidence/profiles/blockgraph.json, sources.15.version |
| X727 | 2026-08 | Independent profile review checked the evidence against the named service. | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. [114, 116] | structural | (empty) | No individual page publication date appears on the versioned manual. Preserve the release month in version metadata. | evidence/profiles/blockgraph.json, sources.15.pub_date |
| X728 | United States | Independent profile review checked the evidence against the named service. | The source gives a North American publication context, not the test population’s country. [112] | structural | North American release; case sample geography undisclosed | The source gives a North American publication context, not the test population’s country. | evidence/profiles/blockgraph.json, sources.6.geography |
| X729 | United States | Independent profile review checked the evidence against the named service. | The cited bulletin does not identify a country for the clients using the integration. [169] | structural | Commercial bulletin; geography of participating campaigns unspecified | The cited bulletin does not identify a country for the clients using the integration. | evidence/profiles/blockgraph.json, sources.14.geography |
| X730 | United States | Independent profile review checked the evidence against the named service. | The Kochava bulletin does not locate campaigns. [169] | structural | US market context; geography of the cited implementation unspecified | The Kochava bulletin does not locate campaigns. | evidence/profiles/blockgraph.json, findings.13.scope_geo |
| X731 | United States | Independent profile review checked the evidence against the named service. | The report does not give a country for the figure itself. [6] | structural | US market report; geography of the reported 130M+ households unspecified | The report does not give a country for the figure itself. | evidence/profiles/blockgraph.json, findings.15.scope_geo |
| X732 | The cited source identifies the named actor and its own reported rule or result. The scope and limits are recorded alongside the claim. | Independent profile review checked the evidence against the named service. | The generic vendor attribution overstates the available provenance. [6] | structural | The number is present on page 26; its original provenance, period and population are not reported. | The generic vendor attribution overstates the available provenance. | evidence/profiles/blockgraph.json, findings.15.confidence_reason |
| X733 | Descriptive report cites an IDoS household figure, without a product-specific count window, US eligible universe or deduplication method. | Independent profile review checked the evidence against the named service. | The source is a market description, not a published footprint census. [6] | structural | Page 26 describes IDoS as serving 130M+ households. The bullet gives no geographical base, measurement period, deduplication method or provenance for that quantity. | The source is a market description, not a published footprint census. | evidence/profiles/blockgraph.json, sources.18.reliability_notes |
| X734 | Opening paragraph | Independent profile review checked the evidence against the named service. | Supply the working copy for both excerpts. [332, 333] | structural | Joint release, first paragraph in the Blockgraph issuer copy under IDN-VCTV-S01; original Charter URL redirects. | Supply the working copy for both excerpts. | evidence/profiles/blockgraph.json, findings.4.support.0.locator |
| X735 | Paragraph starting with this phrase | Independent profile review checked the evidence against the named service. | The issuer copy contains the original release. [332, 333] | structural | Joint release, history paragraph beginning with the 2017 launch, Blockgraph issuer copy under IDN-VCTV-S01. | The issuer copy contains the original release. | evidence/profiles/blockgraph.json, findings.4.support.1.locator |
| X736 | Spectrum Reach Multiscreen Attribution, Measuring Multiscreen Attribution slide, final step | Independent profile review checked the evidence against the named service. | The first Spectrum summary on p. 34 does not contain the quoted time window. [167] | structural | Printed p. 240, Spectrum Reach, Measuring Multiscreen Attribution, fifth process column (Analytics show). | The first Spectrum summary on p. 34 does not contain the quoted time window. | evidence/profiles/blockgraph.json, findings.12.support.0.locator |
| X737 | Section 3, Addressability, Blockgraph bullet preceding clean-room discussion | Independent profile review checked the evidence against the named service. | Column order matters in the PDF. [6] | structural | Printed p. 26, section 3 Addressability, Blockgraph bullet at left. | Column order matters in the PDF. | evidence/profiles/blockgraph.json, findings.15.support.0.locator |
| X738 | First-party workflow account. The seven-day conversion window concerns ad attribution. | Independent profile review checked the evidence against the named service. | This locates the operative slide and distinguishes separate case origins. [167] | structural | Spectrum’s contribution on printed p. 240 gives the seven-day visit window. The same PDF carries separate contributor-authored cases, including Effectv/Mancini’s, at pp. 19–20. | This locates the operative slide and distinguishes separate case origins. | evidence/profiles/blockgraph.json, sources.8.reliability_notes |
| X739 | marks existing segments and matches as stale and must be refreshed before further use | Independent profile review checked the evidence against the named service. | The short excerpt captures the distinction. The adjacent words, independently checked, require refresh before reuse. [115] | structural | marks existing segments and matches as stale | The short excerpt captures the distinction. The adjacent words, independently checked, require refresh before reuse. | evidence/profiles/blockgraph.json, findings.7.support.0.passage |
| X740 | ["IDN-CTV-S01", "IDN-CTV-S05", "IDN-CTV-S06", "IDN-CTV-S18", "IDN-CTV-S10"] | Independent profile review checked the evidence against the named service. | Link the intact JV release and latest checked release to historical and current status. [333, 172] | structural | ["IDN-CTV-S01", "IDN-CTV-S05", "IDN-CTV-S06", "IDN-CTV-S18", "IDN-CTV-S10", "IDN-VCTV-S01", "IDN-VCTV-S02"] | Link the intact JV release and latest checked release to historical and current status. | evidence/profiles/blockgraph.json, entity.source_ids |
| X741 | The US service is documented in the company notice, customer accounts and the dated August 2026 enterprise release. Individual buyers’ current contracts are not public. | Independent profile review checked the evidence against the named service. | A version 3.0.0 release predates the cutoff. A release is not a deployment census. [38, 116, 172] | structural | US partner workflows, version 2.29 documentation and the October 2026 release support active service status. Each participant’s current contract and deployed version remain unverified. | A version 3.0.0 release predates the cutoff. A release is not a deployment census. | evidence/profiles/blockgraph.json, entity.status_note |
| X742 | Initiative launched in 2017 within FreeWheel according to Charter. Joint venture announced in May 2020. The legal incorporation date was not verified. | Independent profile review checked the evidence against the named service. | Use the same original joint release through its working issuer copy. [332, 333] | structural | The parties date the original FreeWheel initiative to 2017. Their joint venture agreement was announced in May 2020. The reviewed sources do not establish the legal incorporation date. | Use the same original joint release through its working issuer copy. | evidence/profiles/blockgraph.json, entity.founded |
| X743 | [{"date": "2017", "event_type": "initiative launch", "event": "Charter dates the original initiative to 2017 and identifies FreeWheel as its incubator.", "source_ids": ["IDN-CTV-S05"]}, {"date": "2020-05-29", "event_type": "ownership agreement", "event": "Three parties announced an agreement for equal ownership of Blockgraph.", "source_ids": ["IDN-CTV-S05"]}, {"date": "2022-02", "event_type": "implemented route", "event": "TVSquared described use of IDoS in its response to NBCUniversal’s measurement framework.", "source_ids": ["IDN-CTV-S02"]}, {"date": "2022-11-01", "event_type": "partner claim", "event": "Kantar reported a relative change in panel matching after integrating IDoS.", "source_ids": ["IDN-CTV-S07"]}, {"date": "2023-06-08", "event_type": "product extension", "event": "Blockgraph announced GraphPort to map third-party identity graphs to IDoS.", "source_ids": ["IDN-G-S07"]}, {"date": "2023-10-25", "event_type": "implemented route", "event": "Kochava documented an update to its existing Blockgraph implementation.", "source_ids": ["IDN-CTV-S15"]}, {"date": "2026-08", "event_type": "release", "event": "Version 2.29 added pre-hashed email as a preset input across identity workflows.", "source_ids": ["IDN-CTV-S10"]}] | Independent profile review checked the evidence against the named service. | The joint release has a working issuer copy. Release 3.0.0 adds a later dated change directly affecting the identity-node interface. [332, 333, 172] | structural | [{"date": "2017", "event_type": "initiative launch", "event": "The joint release dates the original initiative to 2017 and names FreeWheel as incubator.", "source_ids": ["IDN-CTV-S05", "IDN-VCTV-S01"]}, {"date": "2020-05-29", "event_type": "ownership agreement", "event": "Three parties announced an agreement for equal ownership of Blockgraph.", "source_ids": ["IDN-CTV-S05", "IDN-VCTV-S01"]}, {"date": "2022-02", "event_type": "implemented route", "event": "TVSquared described use of IDoS in its response to NBCUniversal’s measurement framework.", "source_ids": ["IDN-CTV-S02"]}, {"date": "2022-11-01", "event_type": "partner claim", "event": "Kantar reported a relative change in panel matching after integrating IDoS.", "source_ids": ["IDN-CTV-S07"]}, {"date": "2023-06-08", "event_type": "product extension", "event": "Blockgraph announced GraphPort to map third-party identity graphs to IDoS.", "source_ids": ["IDN-G-S07"]}, {"date": "2023-10-25", "event_type": "implemented route", "event": "Kochava documented an update to its existing Blockgraph implementation.", "source_ids": ["IDN-CTV-S15"]}, {"date": "2026-08", "event_type": "release", "event": "Version 2.29 added pre-hashed email as a preset input across identity workflows.", "source_ids": ["IDN-CTV-S10"]}, {"date": "2026-10", "event_type": "release", "event": "Version 3.0.0 placed identity-node automations on the Identity Data page and showed additional signal information for the latest upload.", "source_ids": ["IDN-VCTV-S02"]}] | The joint release has a working issuer copy. Release 3.0.0 adds a later dated change directly affecting the identity-node interface. | evidence/profiles/blockgraph.json, timeline |
| X744 | The study compared supplier IP linkages with reference data from other providers. Participant results were masked. | Independent profile review checked the evidence against the named service. | Specify the comparison, modeling approach, time window and reference providers to explain why the results cannot be assigned to IDoS. [40, 41] | structural | Truthset evaluated IPv4-to-postal and IPv4-to-hashed-email links from six masked suppliers, using assignments sampled from two ISPs and one virtual MVPD. Input files covered December 1, 2024 through February 28, 2025. Its model estimated linkage accuracy against that reference data. | Specify the comparison, modeling approach, time window and reference providers to explain why the results cannot be assigned to IDoS. | evidence/profiles/blockgraph.json, performance_evidence.2.design |
| X745 | ["IDN-B-S01"] | Independent profile review checked the evidence against the named service. | The report gives methods; the commissioning body explicitly names the sponsors and researcher. Both concern the same study. [40, 41] | structural | ["IDN-B-S01", "IDN-B-S02"] | The report gives methods; the commissioning body explicitly names the sponsors and researcher. Both concern the same study. | evidence/profiles/blockgraph.json, performance_evidence.2.source_ids |
| X746 | US software for TV distributors and measurement partners to match signals to household records across participant nodes. The enterprise operator can generate a household BG ID from its subscriber data. With permission, a partner can obtain a count and identifiers for the households the nodes share, or resolve impression and outcome signals for measurement. External graphs can enter through GraphPort. Self-service OnDemand onboarding and a clean room are distinct workflows. | Independent profile review checked the evidence against the named service. | Scope the sale to the documented identity route. Both the customer-facing output and the distinct GraphPort path have named sources. [109, 110, 111, 114, 420] | structural | US TV identity software lets a participant establish a postal-address household record, match its permitted partner’s records, retrieve overlap counts or IDs, and resolve impression or outcome signals. GraphPort maps external graphs into IDoS. The enterprise identity node is the delivery route covered here. | Scope the sale to the documented identity route. Both the customer-facing output and the distinct GraphPort path have named sources. | evidence/profiles/blockgraph.json, what_is_sold |
| X747 | Published documentation distinguishes partner authorization, consumer withdrawal and report access. Completed deletion from exported files or another participant’s node was not demonstrated. | Independent profile review checked the evidence against the named service. | Authorization, a node operation and downstream receipt are different. [170, 115, 110, 111] | structural | Partner authorization and record removal are documented at the participant node. The public sources reviewed do not show a consumer refusal delivered to a downstream destination or erasure of files previously exported. | Authorization, a node operation and downstream receipt are different. | evidence/profiles/blockgraph.json, limits.2 |
| X748 | Legacy partner announcements and active US measurement accounts cover different dates and scopes. Summing them would overstate current destination availability. | Independent profile review checked the evidence against the named service. | The partner sources do not establish continuing client contracts. [168, 112, 38, 169] | structural | Dated partner descriptions establish activity at different times. Counting all of them as active at the cutoff could overstate the number of current destinations. | The partner sources do not establish continuing client contracts. | evidence/profiles/blockgraph.json, risks.1 |
| X749 | {"observed": "A participant builds a household layer from its own record ID and a valid postal address. The node generates a BG ID and maps supplied IPs, set-top-box IDs, mobile ad IDs or emails to it. A measurement partner can resolve an input signal against the chosen provider. One home can supply several signals. The guide leaves duplicate addresses, multi-unit buildings and conflicting IP records unresolved.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S11", "IDN-CTV-S16", "IDN-G-S05"], "date": "2026-08"} | Independent profile review checked the evidence against the named service. | The manual explicitly accommodates unit numbers and illustrates two record IDs sharing an IP. Narrow the unresolved question to collision handling. [109, 115, 114, 166] | structural | {"observed": "In v2.29 a node maps IP, email and device signals to a household BG ID generated using a valid postal address and participant record ID. The upload supports an apartment/unit field and multiple signals. The guides do not specify how identity resolution breaks ties when different households share a signal or an address is incomplete.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S11", "IDN-CTV-S12", "IDN-CTV-S16", "IDN-G-S05"], "date": "2026-08"} | The manual explicitly accommodates unit numbers and illustrates two record IDs sharing an IP. Narrow the unresolved question to collision handling. | evidence/profiles/blockgraph.json, capabilities.assignment_method |
| X750 | {"observed": "CIMM lists a reported footprint above 130 million households without an eligible US base or count window for that total. Spectrum attributes its own exposure input to over 30 million homes. Its source footprint, the national graph count and a buyer’s matched or reachable households are different populations.", "availability": "live", "evidence_grade": "D", "source_ids": ["IDN-CTV-S19", "IDN-CTV-S08"], "date": "2026-11-11"} | Independent profile review checked the evidence against the named service. | The larger figure lacks an express national population; the smaller describes an input source. [6, 38] | structural | {"observed": "CIMM’s April 2025 report describes 130M+ households without giving the figure’s geography, eligible base or counting window. Spectrum’s FAQ separately describes collecting exposure inputs from more than 30 million households. Neither count establishes unique US households currently reachable through IDoS for one buyer.", "availability": "live", "evidence_grade": "D", "source_ids": ["IDN-CTV-S19", "IDN-CTV-S08"], "date": "2026-11-11"} | The larger figure lacks an express national population; the smaller describes an input source. | evidence/profiles/blockgraph.json, capabilities.us_coverage_measure |
| X751 | {"observed": "The customer contacts the data-supplying client. The client can delete a subscriber record and leave existing segments and matches available, or use Opt Out to mark them stale and require refresh before further use. No deadline for processing or mechanism to recall earlier exports was established in the reviewed node guide and policy.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S01", "IDN-CTV-S12", "IDN-CTV-S14"], "date": "2026-08"} | Independent profile review checked the evidence against the named service. | Describe the contractual contact route and explicit node controls without treating every consumer request as a platform signal. [170, 115, 111] | structural | {"observed": "Section 6 of the 2021 notice directs a customer who wants contact to stop to the client. In v2.29, Delete removes a subscriber record but leaves existing segments and matches usable. Opt Out removes the record and requires refresh of affected segments and matches. No processing deadline or mechanism to recall earlier exports was established in these sources.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S01", "IDN-CTV-S12", "IDN-CTV-S14"], "date": "2026-08"} | Describe the contractual contact route and explicit node controls without treating every consumer request as a platform signal. | evidence/profiles/blockgraph.json, capabilities.withdrawal_controls |
| X752 | {"observed": "The August 2026 release added a preset for SHA256-hashed email and specifies the identity upload, segment, batch resolution and measurement workflows affected. It also describes changes to the measurement-processing interface. A dated implementation change establishes the supported input, without quantifying its effect on reach or accuracy.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S10"], "date": "2026-08"} | Independent profile review checked the evidence against the named service. | A newer identity-interface change was publicly documented before the cutoff. [116, 172] | structural | {"observed": "Release 2.29 introduced pre-hashed email across several identity workflows in August 2026. Version 3.0.0 in October placed node-specific automations on the Identity Data page and displayed more information on signals in the most recent upload. Neither note reports changes in household accuracy or reachable scale.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S10", "IDN-VCTV-S02"], "date": "2026-10"} | A newer identity-interface change was publicly documented before the cutoff. | evidence/profiles/blockgraph.json, capabilities.change_traceability |
| X753 | {"observed": "The Kantar case reports changes in match yield; Spectrum reports its own MVPD match rate. Neither tests assigned BG IDs against independently labeled household truth. The screened CIMM/Truthset IP benchmark masks the participating data suppliers. A named, product-specific test with denominators for correct and false links was not located in the screened sources.", "availability": "live", "evidence_grade": "D", "source_ids": ["IDN-CTV-S07", "IDN-CTV-S08", "IDN-B-S01"], "date": "2026-11-11"} | Independent profile review checked the evidence against the named service. | The service can be live even though a validation capability for its actual output is unconfirmed. A commercial yield claim does not test household correctness. [112, 38, 40] | structural | {"observed": "Kantar reports a relative match-yield change. Spectrum describes an MVPD-level match rate. Neither source reports the correctness of BG household assignments against reference labels. The commissioned IP study masks suppliers. A product-specific, externally referenced validation of IDoS household links was not established in the sources screened by the cutoff.", "availability": "unconfirmed", "evidence_grade": "D", "source_ids": ["IDN-CTV-S07", "IDN-CTV-S08", "IDN-B-S01"], "date": "2026-11-11"} | The service can be live even though a validation capability for its actual output is unconfirmed. A commercial yield claim does not test household correctness. | evidence/profiles/blockgraph.json, capabilities.output_validation |
| X754 | The native route depends on the participant’s own household records and valid postal addresses. Optional IP, email, device and set-top-box signals link back to that household unit. The measurement workflow needs an enabled identity partner and recipient permissions. GraphPort can map an external provider’s IDs into the platform. Coverage depends on which distributor, signal and purpose the parties authorize. | Independent profile review checked the evidence against the named service. | The guide documents participant permissions and delivery scopes. It does not demonstrate per-purpose consumer approval for every flow. [109, 110, 114, 420] | structural | Native node identity requires the participant’s household records and valid postal addresses. Optional IP, email, mobile advertising IDs and set-top-box IDs can be mapped to these records. Batch resolution needs Client Services and a partner to enable permissions and can restrict results by a delivery scope. GraphPort can connect external graph IDs. Actual coverage depends on the chosen participant and the records and signals available to that integration. | The guide documents participant permissions and delivery scopes. It does not demonstrate per-purpose consumer approval for every flow. | evidence/profiles/blockgraph.json, supply_and_data_dependencies |
| X755 | {"observed": "Universe Match requires partner permission; only granting partners appear in the UI. A batch report needs the partner and Client Services to enable permissions and can be filtered by delivery scope. The guide describes node-to-node authorization. The input and output contracts reviewed do not trace a household’s consent or refusal to a destination.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S13", "IDN-CTV-S16", "IDN-CTV-S01"], "date": "2026-08"} | Independent profile review checked the evidence against the named service. | Bilateral partner authorization and a household’s own consent have distinct evidence. The manual actually shows an empty data-rights slot. [115, 110, 111, 114] | structural | {"observed": "Universe Match lists authorized partners. Batch resolution requires permissions enabled by the partner and Client Services, with an optional delivery-scope filter. In the v2.29 manual upload template the Data Rights column is left empty. The reviewed input and output contracts do not demonstrate transfer of a household’s consent or refusal to a destination.", "availability": "live", "evidence_grade": "C", "source_ids": ["IDN-CTV-S12", "IDN-CTV-S13", "IDN-CTV-S14", "IDN-CTV-S16"], "date": "2026-08"} | Bilateral partner authorization and a household’s own consent have distinct evidence. The manual actually shows an empty data-rights slot. | evidence/profiles/blockgraph.json, capabilities.permission_handling |
| X756 | The US service notice and the vendor’s UK and European website language describe different activities. Historical equal ownership is not evidence of present equity percentages. | Independent profile review checked the evidence against the named service. | Anchor the two comparisons to the correct dates and scope. [170, 334, 333] | structural | The 2021 notice describes US service availability while its UK and European wording covers website visitors. The 2020 agreement records equal ownership at announcement. The cutoff ownership page lists names without percentages. | Anchor the two comparisons to the correct dates and scope. | evidence/profiles/blockgraph.json, counterevidence_and_tensions.2 |
| X757 | Second page, ordering paragraph, August 26, 2026 | Independent profile review checked the evidence against the named service. | The quoted words appear at the foot of the first physical page of order 104. The sentence continues onto the second page. [8] | structural | Order 104, ordering paragraph at foot of p. 1, continued on p. 2; signed August 26, 2026. | The quoted words appear at the foot of the first physical page of order 104. The sentence continues onto the second page. | evidence/verify/yahoo-connectid.json, additional_sources.IDN-VYAH-S06.locator |
| X758 | LiveRamp RampID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — ATS is an integration path; Known Identity Resolution is a different deliverable. US terms cover SSP Sidecar and DSP/DMP roles. [133, 140, 183, 345, 185, 141, 318, 234, 181] | structural | LiveRamp RampID — Eligible for scored-cell review | ATS is an integration path; Known Identity Resolution is a different deliverable. US terms cover SSP Sidecar and DSP/DMP roles. | data/product-register.json, IDN-P001 |
| X759 | Unified ID 2.0 as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Private operator instances, the EUID namespace, The Trade Desk’s DSP and Identity Alliance have separate roles. Ohio and Oregon production endpoints and the Roku Exchange buying route are documented. [142, 143, 365, 366, 153, 154, 331, 203] | structural | Unified ID 2.0 — Eligible for scored-cell review | Private operator instances, the EUID namespace, The Trade Desk’s DSP and Identity Alliance have separate roles. Ohio and Oregon production endpoints and the Roku Exchange buying route are documented. | data/product-register.json, IDN-P002 |
| X760 | European Unified ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — EU/EEA and named additional territories; United Kingdom explicitly listed. [211, 367, 368] | structural | European Unified ID — Listed | European territory comparator. EUID lists the UK and European territories; a US offer for this distinct namespace was not documented. | data/product-register.json, IDN-P003 |
| X761 | ID5 ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — The SDK and Prebid modules implement the same service. Twine Data’s US licensing and the sold TrueData audience business do not constitute ID5 ID. The partner case documents participating US audiences. [370, 339, 56, 131, 177, 176, 129, 351, 280, 369, 132, 177] | structural | ID5 ID — Eligible for scored-cell review | The SDK and Prebid modules implement the same service. Twine Data’s US licensing and the sold TrueData audience business do not constitute ID5 ID. The partner case documents participating US audiences. | data/product-register.json, IDN-P004 |
| X762 | Yahoo ConnectID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Google documents the web secure-signal path and Yahoo names its App & Display and Video bidders. Publisher software integration does not confer rights to share an ID with any DSP. [160, 347, 199, 156, 157, 196, 329, 348, 165, 371, 372] | structural | Yahoo ConnectID — Eligible for scored-cell review | Google documents the web secure-signal path and Yahoo names its App & Display and Video bidders. Publisher software integration does not confer rights to share an ID with any DSP. | data/product-register.json, IDN-P005 |
| X763 | Lotame Panorama ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Lotame documents its publisher and SSP EID. TripleLift names a USA route and a Michigan case names a publisher and SSP. Lotame is an author-declared former employer. [374, 373, 375, 279, 215, 263, 264] | structural | Lotame Panorama ID — Eligible for scored-cell review | Lotame documents its publisher and SSP EID. TripleLift names a USA route and a Michigan case names a publisher and SSP. Lotame is an author-declared former employer. | data/product-register.json, IDN-P006 |
| X764 | Criteo OneTag identity service as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Prebid identifies the criteo.com EID and exchange adapter path to Criteo demand. Equativ maps that namespace separately from Criteo OB. TripleLift lists a US Criteo ID passthrough. The OneTag provisional name remains an unverified alias. [250, 376, 263, 377, 250] | structural | Criteo ID for Exchanges — Eligible for scored-cell review | Prebid identifies the criteo.com EID and exchange adapter path to Criteo demand. Equativ maps that namespace separately from Criteo OB. TripleLift lists a US Criteo ID passthrough. The OneTag provisional name remains an unverified alias. | data/product-register.json, IDN-P007 |
| X765 | Amazon Publisher Services publisher ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Amazon cites US publisher data for its measurement tool; proprietary interoperable ID unconfirmed. [378] | structural | Amazon Publisher Services publisher ID (unconfirmed) — Listed | Signal IQ measures publisher bidstream signals. An APS-issued independent interoperable ID with a separate US route was not established. | data/product-register.json, IDN-P008 |
| X766 | SharedID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Prebid provides the publisher-scoped identifier, available to its adapters. TripleLift names US passthrough. The recipient can receive a publisher’s identifier without establishing any persistent identity across unrelated publishers. [258, 286, 278, 263] | structural | SharedID — Eligible for scored-cell review | Prebid provides the publisher-scoped identifier, available to its adapters. TripleLift names US passthrough. The recipient can receive a publisher’s identifier without establishing any persistent identity across unrelated publishers. | data/product-register.json, IDN-P009 |
| X767 | Utiq identity service as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Named European markets; US eligibility unconfirmed. [379, 380, 263] | structural | Utiq identity service — Listed | Utiq names six European markets. TripleLift lists US/European passthrough. US forwarding does not establish a supported US domestic connection from a participating telecom for issuance. Keep this territorial conflict visible. | data/product-register.json, IDN-P010 |
| X768 | Parrable ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US eligibility unconfirmed. [249, 381, 382, 377, 249] | structural | Parrable ID — Listed | A device identifier, a partner module and exchange namespace are documented. US timezones in sample filters are configuration examples. The checked documents do not establish that a US publisher can obtain the named service on a US inventory route. | data/product-register.json, IDN-P011 |
| X769 | LiveRamp identity resolution as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — LiveRamp documents maintained US AbiliTec links and an available REST API for known PII. Microsoft documents US-only preview enrichment requiring active LiveRamp and AbiliTec subscriptions; this preview does not mean the standalone API is beta. [235, 141, 318, 212, 243, 383] | structural | LiveRamp Known Identity Resolution (AbiliTec API) — Eligible for scored-cell review | Direct AbiliTec API for known US records. Microsoft Dynamics 365 is a separately named US preview integration and does not establish a completed production transaction. | data/product-register.json, IDN-P012 |
| X770 | Experian identity resolution as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Sales page labels US Digital Graph, advertises license or first-party file append, and provides sales access. The customer cases do not identify a US sample. [117, 119, 336, 337, 338] | structural | Experian Digital Graph — Eligible for scored-cell review | Direct Digital Graph licensing, full dataset or buyer-file append. The MiQ IP and unnamed DSP case rates remain separate geographic-unknown cases. | data/product-register.json, IDN-P013 |
| X771 | TransUnion TruAudience Identity Resolution as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — TransUnion offers TruAudience Identity Resolution & Enrichment as a Snowflake native marketplace workflow that links customer records against historical and current attributes for over 250 million US consumers. The Data Health Assessment insurer case is a different service. [384, 246, 17, 18, 19, 316] | structural | TransUnion TruAudience Identity Resolution — Eligible for scored-cell review | TruAudience Identity Resolution & Enrichment for Snowflake (customer records to persistent identities); excludes Data Health Assessment, Graph Extracts, streaming media activation. | data/product-register.json, IDN-P014 |
| X772 | Epsilon CORE ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — US solution page advertises Core ID recognition for US consumers; processing description identifies the USA for identity resolution and describes client file onboarding. US processing location alone would be insufficient. [385, 386, 22, 319, 320, 216] | structural | Epsilon CORE ID — Eligible for scored-cell review | Core ID Services customer-file onboarding (not Intent Messaging or a combined platform). | data/product-register.json, IDN-P015 |
| X773 | Acxiom Real Identity as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — A US retail deployment exists for Digital Recognition powered by Real ID. The public records do not establish standalone US access to the specifically registered brand-cloud Real ID application. Do not conflate the two offers. [387, 388, 327, 328] | structural | Acxiom Real ID in the cloud — Listed | Brand-cloud Real ID app; no qualifying US route frozen. Distinct Digital Recognition noted. | data/product-register.json, IDN-P016 |
| X774 | Merkle Merkury as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — US private-graph offer at launch. Dentsu reports a later US Hy-Vee/RedMedia test of the Merkury integration, and the ensuing agency appointment identifies a continuing planned use. These do not measure delivery in every commercial configuration. [389, 359, 360] | structural | Merkle Merkury — Eligible for scored-cell review | Merkury private identity platform; US Merkury-to-The Trade Desk test with Hy-Vee’s RedMedia. | data/product-register.json, IDN-P017 |
| X775 | Zeta identity resolution as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Zeta announced US-graph enrichment inside the Snowflake native app; the current regional guide specifies US-only Streamlit availability, a named CGI first-party data-share request, and the account-based views containing enriched data. [244, 391, 390] | structural | Zeta Customer Growth Intelligence — Eligible for scored-cell review | Customer Growth Intelligence Snowflake native app, buyer first-party records against Zeta Data Cloud. Do not transfer Customer Matcher rules for two client tables. | data/product-register.json, IDN-P018 |
| X776 | Permutive publisher cohorts as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Permutive standard cohorts are documented on TelevisaUnivision US inventory through Google Ad Manager. OpenX separately advertised a US buyer route in 2022. [205, 206, 207, 208] | structural | Permutive publisher cohorts — Eligible for scored-cell review | US deployed cookieless publisher cohort route confirmed. Contract price unconfirmed. Google Ad Manager is the documented receiving implementation; OpenX is a separate alternative. | data/product-register.json, IDN-P019 |
| X777 | Audigent SmartPMP as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US service terms unresolved. [236, 237, 238, 239, 321, 322] | structural | Audigent SmartPMP — Listed | SmartPMP is a curated media deal. Hadron ID is a separately documented identity module. Do not transfer its features to SmartPMP or count the two names as aliases. | data/product-register.json, IDN-P020 |
| X778 | Ogury Personified Advertising as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Ogury’s identifiable group-targeting offer and US market activation inputs are documented. The preceding Japan campaign remains Japan-only. [248, 210, 356] | structural | Ogury Personified Advertising — Eligible for scored-cell review | Ogury’s identifiable group-targeting offer and US market activation inputs are documented. The preceding Japan campaign remains Japan-only. | data/product-register.json, IDN-P021 |
| X779 | Verve Moments.AI as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — US general offering is explicit in original 2022 issuer release; Q1 2026 report again describes this product for ID-deprecation environments. Provider markets high-quality contextual segments for brands to replace personalised targeting. [392, 393] | structural | Verve Moments.AI — Eligible for scored-cell review | US general offering is explicit in original 2022 issuer release; Q1 2026 report again describes this product for ID-deprecation environments. Provider markets high-quality contextual segments for brands to replace personalised targeting. | data/product-register.json, IDN-P022 |
| X780 | Seedtag contextual audiences as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Documented Michigan targeting option. Distinguish availability from actual US deployment. Eurail’s separate campaign documents US activity but does not establish which formal segment type it used. [394, 395, 349, 355] | structural | Seedtag contextual audiences — Eligible for scored-cell review | Seedtag sells separate named Signature, Seasonal and Custom audience groups as replacement for traditional tracking. The 2026 managed route offers custom contextual targeting including named US-relevant segments. The package with Aristotle voter profiles remains separate. | data/product-register.json, IDN-P023 |
| X781 | TransUnion TruAudience streaming household identity as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US household graph and CTV audience route explicitly described. [245, 246, 247] | structural | TransUnion TruAudience streaming household identity — Listed | Epsilon via the TruAudience Data Marketplace and Comcast addressable campaigns establish TV activation routes. Tru Optik is the former marketplace name. Neither release confirms a separately purchasable household resolution service distinct from TruAudience Identity. | data/product-register.json, IDN-P024 |
| X782 | Blockgraph Identity Operating System as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — US commercial scope in 2021 notice; Spectrum Reach and TVSquared describe US TV matching use. Active dated v2.29/2026 product documents. [396, 170, 332, 334, 335, 116, 333, 172, 168] | structural | Blockgraph Identity Operating System — Eligible for scored-cell review | Stage 4 records distinguish enterprise household matching from purchasing advertiser measurement reports and supersede the Stage 3 uncertainty. | data/product-register.json, IDN-P025 |
| X783 | Viant Household ID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US adult scope claimed in a Viant/TransUnion partner release. [312, 313] | structural | Viant Household ID — Listed | Viant says its DSP recognizes a Household ID profile from bid request inputs, rather than receiving the Household ID in the bidstream. The US adult figure is a vendor reach assertion. Verify a separate household deliverable. | data/product-register.json, IDN-P026 |
| X784 | OpenAP OpenID as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US publisher adoption claimed in 2021. [397, 398] | structural | OpenAP OpenID — Listed | OpenAP called Open Identity a 2025 evolution of its 2021 person-level OpenID. The later product is federated infrastructure, not evidence that OpenID assigns a household. Keep a separate household purchase unconfirmed. | data/product-register.json, IDN-P027 |
| X785 | FreeWheel Identity Network as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US service geography unresolved for the specific implementation. [305, 399] | structural | FreeWheel Identity Network — Listed | A network of third-party and operator signals is not sufficient to establish delivery of a distinct household graph product. | data/product-register.json, IDN-P028 |
| X786 | Cadent Aperture Viewer Graph as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Cadent markets US CTV buying using its Viewer Graph within Total Video. Separate graph access or licensing is unconfirmed. [306, 314, 315] | structural | Cadent Viewer Graph — Listed | Viewer Graph is the named graph. Aperture’s 2024 targeting sheet and the current Total Video offer are platforms or routes using it. Premion’s 2021 announcement described planned integration. | data/product-register.json, IDN-P029 |
| X787 | Samba TV household graph as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US eligibility immaterial to exclusion. [400] | structural | Samba TV household graph — Listed | Permanent exclusion: author’s current employer. | data/product-register.json, IDN-P030 |
| X788 | AWS Clean Rooms as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Product territory outside IQ scope. [240] | structural | AWS Clean Rooms — Listed | Keep the collaboration service outside the IQ. Separately sold identity-resolution services would need their own screening record. | data/product-register.json, IDN-P031 |
| X789 | Microsoft Curate as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Product territory outside IQ scope. [241] | structural | Microsoft Curate — Listed | Keep the deal curation service outside IQ. Check platform availability separately before referring to a live offer. | data/product-register.json, IDN-P032 |
| X790 | OneTrust Consent Management Platform as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Product territory outside IQ scope. [242] | structural | OneTrust Consent Management Platform — Listed | Record this as an upstream signal dependency only. | data/product-register.json, IDN-P033 |
| X791 | IAB Tech Lab Seller Defined Audiences as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Standards document; no US product assumption. [251] | structural | IAB Tech Lab Curated Audiences (formerly Seller Defined Audiences) — Listed | Do not count the renamed specification as a retired vendor or as a separate cohort service. Its own description identifies its Project Rearc origin. | data/product-register.json, IDN-P034 |
| X792 | Chrome Topics API as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | historical — Chrome and Android named separately in the announcement. [282] | structural | Chrome Topics API — Historical | The announcement establishes a decision. Use stream E for separately qualified release or removal evidence. | data/product-register.json, IDN-P035 |
| X793 | Chrome Protected Audience API as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | historical — Chrome 153 release note; distinct from Android. [283] | structural | Chrome Protected Audience API — Historical | Record this precise scope and refer to stream E for other endpoint or version milestones. A headline alone cannot establish removal from every installation. | data/product-register.json, IDN-P036 |
| X794 | Chrome Federated Learning of Cohorts as a provisional seed in listed | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | historical — Chrome proposal; no vendor admission. [284] | structural | Chrome Federated Learning of Cohorts — Historical | Classify the replaced proposal as historical and keep trial, replacement and later Topics events separate. | data/product-register.json, IDN-P037 |
| X795 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Prebid states US support and a partner-ID authorization path. TripleLift identifies US bridging. The 33Across User ID module implements the Lexicon service. [401, 263] | structural | 33Across Lexicon — Eligible for scored-cell review | Prebid states US support and a partner-ID authorization path. TripleLift identifies US bridging. The 33Across User ID module implements the Lexicon service. | data/product-register.json, IDN-P038 |
| X796 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Historical issuer document; current US offer unknown. [402, 377, 403] | structural | Zeotap ID+ — Listed | The issuer’s North America statement refers to the underlying graph. PubMatic names five European countries for the ID+ implementation; Equativ lists its namespace without a country. US availability of the distinct ID+ offer remains unconfirmed. | data/product-register.json, IDN-P039 |
| X797 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — US political TV exposure segments explicitly offered with the ID-free route, independently of the sold ID-based variants. The current product page offers ID-free audience creation; 2023 Trade Desk evidence is historical access, not a current seat count. [404, 209, 350] | structural | Proximic by Comscore Predictive Audiences — Eligible for scored-cell review | US political TV exposure segments explicitly offered with the ID-free route, independently of the sold ID-based variants. The current product page offers ID-free audience creation; 2023 Trade Desk evidence is historical access, not a current seat count. | data/product-register.json, IDN-P040 |
| X798 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US-specific eligibility not tested for distinct identifier. [405] | structural | The Trade Desk Galileo — Listed | Do not double-count a UID2 integration as a second interoperable ID. Reopen if a distinct service meets a registered task. | data/product-register.json, IDN-P041 |
| X799 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Explicit US outdoor/RV retailer submitted its CRM for matching and appending links under this service. A separate call to purchase offers a match test. [27, 27, 28] | structural | Deep Sync Profile Hygiene & Enrichment — Eligible for scored-cell review | Profile Hygiene & Enrichment, customer CRM file against Deep Sync graph. | data/product-register.json, IDN-P042 |
| X800 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Dedicated Transaction Matching case, buyer call to send distinct POS data and a monthly production pipeline establish a separate US identity deliverable: POS transactions resolved to individuals/households, with attribution an optional use. Different from CRM record resolution. [12, 12] | structural | Deep Sync Transaction Matching — Eligible for scored-cell review | Transaction Matching via the monthly POS matching waterfall. | data/product-register.json, IDN-P043 |
| X801 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Kochava describes a US QSR CTV campaign using household mapping. The source does not establish a Kochava Collective product or standalone graph licence. [35] | structural | Kochava advanced measurement household graph (QSR case) — Listed | Trace the relation of that deployment to Kochava Collective and to independently purchasable household mapping; a CTV measurement case cannot establish a current graph licence. Track legal scope separately in H. | data/product-register.json, IDN-P044 |
| X802 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — LtvAdx markets a US TV campaign route with HouseholdID targeting in a paid-impression plan. Separate licensing of household resolution is unconfirmed. [227, 229, 230, 231] | structural | LtvAdx Identity API / HouseholdID — Listed | Identity API documents household creation and lookup. The US page and pricing evidence establish paid media access to HouseholdID targeting, without a separately priced graph licence. Distinguish device-link and household retention from VAST serving fallback. | data/product-register.json, IDN-P045 |
| X803 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US purchase or deployment unconfirmed for this programmer workflow. [307, 141, 318] | structural | LiveRamp Advanced TV programmer CTV Audience Recognition — Listed | Check separate TV purchase and deliverable to avoid duplicating P001 or P012; the guide establishes a named workflow. | data/product-register.json, IDN-P046 |
| X804 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Factori markets the CTV/cross-device dataset globally through AWS Marketplace. The listing does not identify US dataset coverage or a US deployment. [308, 309] | structural | Factori Cross Device / CTV Graph — Listed | A contract subscription for the cross-device/CTV data is documented. Confirm US-specific records or a US use case, and a rule for household-link aging. The daily feed cadence does not document that rule. | data/product-register.json, IDN-P047 |
| X805 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US eligibility unconfirmed. [406] | structural | Datonics Cookieless Audience Targeting — Listed | Datonics describes 1:1 targeting through an ID graph and a limited release for existing clients. The record does not identify a buyable cohort, modelled audience or seller-defined output meeting the registered C task. | data/product-register.json, IDN-P048 |
| X806 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Standards proposal; no US availability assigned. [407] | structural | Project Rearc Accountability Platform draft — Listed | No dated withdrawal was located in the registered IAB searches. Keep publication of the draft and possible retirement as distinct questions. | data/product-register.json, IDN-P049 |
| X807 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Dedicated Prebid module and namespace. TripleLift lists the product by name for US passthrough. Google identifies a beta curation path. The ID is distinct from Audigent SmartPMP. [237, 238, 263] | structural | Audigent Hadron ID — Eligible for scored-cell review | Dedicated Prebid module and namespace. TripleLift lists the product by name for US passthrough. Google identifies a beta curation path. The ID is distinct from Audigent SmartPMP. | data/product-register.json, IDN-P050 |
| X808 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — The documented Prebid module requires a registration key. Equativ maps neustar.biz to Fabrick. TripleLift describes US passthrough with the same issuer and spells the identifier “Fabrik.” No equivalence to TransUnion’s general person graph is inferred. [408, 263, 377] | structural | Neustar Fabrick ID — Eligible for scored-cell review | The documented Prebid module requires a registration key. Equativ maps neustar.biz to Fabrick. TripleLift describes US passthrough with the same issuer and spells the identifier “Fabrik.” No equivalence to TransUnion’s general person graph is inferred. | data/product-register.json, IDN-P051 |
| X809 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — Publisher Link requires its own enrollment for publisher first-party authenticated inputs. It issues a distinct publisher ID linked to Epsilon CORE ID, taken by TripleLift in US passthrough. The US exchange path is selectable independently of a buyer-side CORE ID resolution engagement. [263, 362, 385] | structural | Epsilon PubLink — Eligible for scored-cell review | Publisher Link requires its own enrollment for publisher first-party authenticated inputs. It issues a distinct publisher ID linked to Epsilon CORE ID, taken by TripleLift in US passthrough. The US exchange path is selectable independently of a buyer-side CORE ID resolution engagement. | data/product-register.json, IDN-P052 |
| X810 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — The provider module offers a persistent probabilistic device identifier for publishers’ bidding routes. TripleLift names FTrack US passthrough. Optional household output is outside this A row. [263, 409] | structural | Flashtalking FTrack — Eligible for scored-cell review | The provider module offers a persistent probabilistic device identifier for publishers’ bidding routes. TripleLift names FTrack US passthrough. Optional household output is outside this A row. | data/product-register.json, IDN-P053 |
| X811 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | yes — A publisher uses the LiveIntent module and product-registration route to obtain the email-linked nonID. TripleLift lists LiveIntent for US bridging. The ID route is distinct from general graph or media services. [263, 377, 260] | structural | LiveIntent nonID — Eligible for scored-cell review | A publisher uses the LiveIntent module and product-registration route to obtain the email-linked nonID. TripleLift lists LiveIntent for US bridging. The ID route is distinct from general graph or media services. | data/product-register.json, IDN-P054 |
| X812 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — Prebid attributes the identifier to Merkury’s private identity graph. TripleLift uses Merkle ID for its own audiences, and Merkle describes US Merkury services. Distinct direct A purchase in addition to P017 B remains unconfirmed. [263, 377, 363, 410, 389] | structural | Merkle ID — Listed | Prebid attributes the identifier to Merkury’s private identity graph. TripleLift uses Merkle ID for its own audiences, and Merkle describes US Merkury services. Distinct direct A purchase in addition to P017 B remains unconfirmed. | data/product-register.json, IDN-P055 |
| X813 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — TripleLift lists a US passthrough identifier but provides no unit or relationship to the separately documented Viant Household ID route, on which Viant says the household ID is not passed in the bidstream. A separate selectable A offer is unconfirmed. [263, 312, 313] | structural | Viant ID (exchange label) — Listed | TripleLift lists a US passthrough identifier but provides no unit or relationship to the separately documented Viant Household ID route, on which Viant says the household ID is not passed in the bidstream. A separate selectable A offer is unconfirmed. | data/product-register.json, IDN-P056 |
| X814 | Unregistered discovery lead | Verify buying task, US availability, boundaries, aliases and commercial unit before scoring. | unconfirmed — US-only Streamlit application; matches records between two customer-supplied tables. The documented processing does not establish a sold first-party-to-Zeta-graph link for this operation, or equivalence to CGI. [244] | structural | Zeta Customer Matcher — Listed | Two input tables, threshold-based matched pairs, separate from CGI. | data/product-register.json, IDN-P057 |
| X815 | One or more broad promotional statements were initially treated as a result. | Reopen the source and test the exact anchor on the frozen service and route. | An aim to provide accurate targeting gives no result of an assessment. For anchor 1 identify an actual reported performance or accuracy result for the sold output. | supported with qualification | R11 applied to every assessed output-validation cell. | An aim to provide accurate targeting gives no result of an assessment. For anchor 1 identify an actual reported performance or accuracy result for the sold output. | data/scoring-rulings.json R11 |
| X816 | TripleLift named provider LiveIntent in the source table. | Reopen the source and test the exact anchor on the frozen service and route. | The table uses the provider label. Joining two records requires a separately evidenced product mapping. [263, 260] | supported with qualification | Final claim attributes bridging to LiveIntent; the separate module guide names nonID. | The table uses the provider label. Joining two records requires a separately evidenced product mapping. | IDN-A5-F01 |
| X817 | Lotame says its browser opt-out stops its use of Panorama ID and that a previously stored ID may take up to seven days to expire. This describes Lotame’s browser handling; the notice does not demonstrate a bidder acting on the same choice. | Reopen the source and test the exact anchor on the frozen service and route. | Lotame provides a browser-specific opt-out. A stored Panorama ID can take up to seven days to expire, and the platform may take several hours to process the request. If third-party cookies are unavailable, probabilistic browser recognition can fail and a new ID can be assigned. The platform policy describes a service rule without supplying a bidder receipt. [261] | supported with qualification | Lotame says an opt-out applies to the requesting browser and a stored Panorama ID can take up to seven days to expire. If third-party cookies are blocked, Lotame says its probabilistic recognition may fail and it may assign a new ID. The notice does not demonstrate a bidder acting on the request. | Lotame provides a browser-specific opt-out. A stored Panorama ID can take up to seven days to expire, and the platform may take several hours to process the request. If third-party cookies are unavailable, probabilistic browser recognition can fail and a new ID can be assigned. The platform policy describes a service rule without supplying a bidder receipt. | IDN-HX-F03; data/claims-extra.json |
| X818 | LiveIntent documents sending the GDPR flag and consent string, US Privacy string, GPP string and applicable sections to its identity-resolution endpoint. The guide does not prescribe how a downstream bidder must act on a particular nonID. | Reopen the source and test the exact anchor on the frozen service and route. | The source lists all five parameters and shows an example ID-resolution URL. Where its blocked-call handler applies, it says identity resolution and other network calls are blocked entirely. It names no downstream bidder treatment of a particular nonID. [219] | supported with qualification | LiveIntent’s guide lists a GDPR flag and consent string, US Privacy string, GPP string and applicable sections on eligible identity-resolution requests; in privacy mode, the guide says the hub blocks identity-resolution requests. It gives no downstream bidder rule for a particular nonID. | The source lists all five parameters and shows an example ID-resolution URL. Where its blocked-call handler applies, it says identity resolution and other network calls are blocked entirely. It names no downstream bidder treatment of a particular nonID. | IDN-HX-F04; data/claims-extra.json |
| X819 | TripleLift’s California FAQ says it forwards consent status with the bid request and leaves the targeting response to the DSP. Its exchange listing of an identifier in a US bid path does not alone establish how a recipient must use that identifier after refusal. | Reopen the source and test the exact anchor on the frozen service and route. | Page 1, lines 14–20 of the publisher PDF assign interpretation of the 2020 US privacy signal to the recipient. That statement is historical and cannot establish GPP rules at the 2026 cutoff. [262] | supported with qualification | In a 2020 California publisher FAQ, TripleLift said it forwarded the bid request and the consent status to the DSP and left targeting to the DSP. This historical exchange policy does not identify how a particular bidder handles a particular advertising ID. | Page 1, lines 14–20 of the publisher PDF assign interpretation of the 2020 US privacy signal to the recipient. That statement is historical and cannot establish GPP rules at the 2026 cutoff. | IDN-HX-F05; data/claims-extra.json |
| X820 | Neustar launched the particular Fabrick ID in December, distinguished from its July Fabrick platform launch. The release markets a consumer- and household-graph identifier to publishers as an alternative to third-party cookies. | Reopen the source and test the exact anchor on the frozen service and route. | The December 16, 2020 body says the Fabrick ID accepts phone/email and links a consumer/household graph to publisher data; SSPs can substitute an ID via API. The December release is specifically for the ID, not the earlier suite. Its Virginia dateline alone does not establish the population of buyers at launch; the later TripleLift table records US passthrough. It does not establish continuity of an identified set of edges. [426] | supported with qualification | On December 16, 2020, Neustar announced Fabrick ID for graph-based linkage of publisher and advertiser data without third-party cookies. The later TripleLift table documents a US passthrough route. The announcement does not show a US deployment at launch or continued predecessor edges. | The December 16, 2020 body says the Fabrick ID accepts phone/email and links a consumer/household graph to publisher data; SSPs can substitute an ID via API. The December release is specifically for the ID, not the earlier suite. Its Virginia dateline alone does not establish the population of buyers at launch; the later TripleLift table records US passthrough. It does not establish continuity of an identified set of edges. | IDN-HY-F02; data/claims-extra.json |
| X821 | The SDK integration offered custom cohort creation from Topics and optional Topics retrieval for bid requests. It does not establish use in the US standard-cohort PPS route, a US default, or third-party-cookie syncing. | Reopen the source and test the exact anchor on the frozen service and route. | Permutive’s own July 26, 2024 changelog lists Topics-based custom cohorts and an optional Prebid bid-request path. After workspace enablement opted-in Topics are automatically synced via the SDK; making them available in bid requests additionally depends on Prebid integration and the option selected. These are Topics, not third-party cookie ID sync. The record supplies no US-standard-cohort implementation, general US default, before/after pair or TV household accuracy test. [427] | supported with qualification | Permutive’s July 26, 2024 product changelog describes optional Topics-based custom cohorts and conditional sending of opted-in Topics through Prebid. It supplies H4 implementation context, not an H8 US graph launch, a US Chrome cookie-sync default, or H5 link validation. | Permutive’s own July 26, 2024 changelog lists Topics-based custom cohorts and an optional Prebid bid-request path. After workspace enablement opted-in Topics are automatically synced via the SDK; making them available in bid requests additionally depends on Prebid integration and the option selected. These are Topics, not third-party cookie ID sync. The record supplies no US-standard-cohort implementation, general US default, before/after pair or TV household accuracy test. | IDN-HY-F04; data/claims-extra.json |
| X822 | On January 4, 2024, TransUnion said its enhanced marketing identity graph represented 250 million adults and 98% of US adults. Its Snowflake Resolution & Enrichment offer uses that graph to match client records. | Reopen the source and test the exact anchor on the frozen service and route. | qualified [316, 19] | supported with qualification | On January 4, 2024, TransUnion said the enhanced graph used by TruAudience Identity Resolution represented 250 million adults and 98% of US adults. In its January 2026 Snowflake guide, it said the Resolution & Enrichment offer matched client records using its identity graph. | qualified | IDN-B5-F01; data/claims-extra.json |
| X823 | Lotame’s launch statement claimed higher accuracy for Panorama ID and a persistent consumer opt-out across touchpoints. The statement supplies no reference accuracy test or enforcement sequence. | Reopen the source and test the exact anchor on the frozen service and route. | Lotame’s October 28, 2020 launch statement promoted high targeting accuracy and a one-time, persistent opt-out. It reports no test of Panorama’s right-person link accuracy and gives no implementation sequence establishing enforcement. [375] | supported with qualification | Lotame’s October 28, 2020 launch statement promoted high targeting accuracy and a one-time, persistent opt-out. It reports no test of Panorama’s right-person link accuracy and gives no implementation sequence establishing enforcement. | Lotame’s October 28, 2020 launch statement promoted high targeting accuracy and a one-time, persistent opt-out. It reports no test of Panorama’s right-person link accuracy and gives no implementation sequence establishing enforcement. | IDN-AS5-F02; data/claims-extra.json |
| X824 | SharedID reads or sets a random first-party ID. Permitted publisher page loads can set the cookie, while missing Purpose 1 consent with the TCF module, an opt-out marker, or a COPPA flag inhibits specified activity. | Reopen the source and test the exact anchor on the frozen service and route. | On the first-party SharedID path the module can read or set a random ID. Its TCF check can skip reading and setting cookies when Purpose 1 consent is absent, the publisher opt-out cookie prevents further cookie setting, and the COPPA flag suppresses the ID. Prebid’s versioned walkthrough distinguishes the older sharedid.org EID from the pubcid.org EID after the 5.0 merge. [286, 285] | supported with qualification | On the first-party SharedID path the module can read or set a random ID. Its TCF check can skip reading and setting cookies when Purpose 1 consent is absent, the publisher opt-out cookie prevents further cookie setting, and the COPPA flag suppresses the ID. Prebid’s versioned walkthrough distinguishes the older sharedid.org EID from the pubcid.org EID after the 5.0 merge. | On the first-party SharedID path the module can read or set a random ID. Its TCF check can skip reading and setting cookies when Purpose 1 consent is absent, the publisher opt-out cookie prevents further cookie setting, and the COPPA flag suppresses the ID. Prebid’s versioned walkthrough distinguishes the older sharedid.org EID from the pubcid.org EID after the 5.0 merge. | IDN-AS5-F03; data/claims-extra.json |
| X825 | The SharedID configuration defines per-page-load extension, configurable storage lifetime and a sample bidRequest.userId.sharedid object. Unrestricted publishers can install the module. | Reopen the source and test the exact anchor on the frozen service and route. | The module is available without registration. A publisher can configure the storage period and choose per-load expiry extension. The sample bidRequest.userId.sharedid is an adapter-facing field. Prebid’s versioned guide uses pubcid.org as the later outgoing EID source. [258, 286, 285] | supported with qualification | The module is available without registration. A publisher can configure the storage period and choose per-load expiry extension. The sample bidRequest.userId.sharedid is an adapter-facing field. Prebid’s versioned guide uses pubcid.org as the later outgoing EID source. | The module is available without registration. A publisher can configure the storage period and choose per-load expiry extension. The sample bidRequest.userId.sharedid is an adapter-facing field. Prebid’s versioned guide uses pubcid.org as the later outgoing EID source. | IDN-AS5-F04; data/claims-extra.json |
| X826 | 33Across calls its Lexicon connection probabilistic, accepts optional SHA256 hashed email, and lists prioritized publisher settings, a global variable and browser storage as sources. | Reopen the source and test the exact anchor on the frozen service and route. | 33Across describes Lexicon’s publisher-to-demand connection as probabilistic. The optional SHA-256 email lookup checks configured hem, global _33across.hem.sha256 and then browser storage. The external source is 33across.com and the adapter field is 33acrossId. [401] | supported with qualification | 33Across describes Lexicon’s publisher-to-demand connection as probabilistic. The optional SHA-256 email lookup checks configured hem, global _33across.hem.sha256 and then browser storage. The external source is 33across.com and the adapter field is 33acrossId. | 33Across describes Lexicon’s publisher-to-demand connection as probabilistic. The optional SHA-256 email lookup checks configured hem, global _33across.hem.sha256 and then browser storage. The external source is 33across.com and the adapter field is 33acrossId. | IDN-AS5-F05; data/claims-extra.json |
| X827 | The SharedID guide describes publisher-side opt-out input and suppression at forthcoming module runs. Its script example allows the cookie to expire and then be created again. | Reopen the source and test the exact anchor on the frozen service and route. | The publisher can set the opt-out marker to prevent further cookie setting on module runs. Separately, an example normal script configuration lets an existing cookie expire before creating another. It describes ordinary expiry with that configuration, not recreation after opt-out. [286] | supported with qualification | The publisher can set the opt-out marker to prevent further cookie setting on module runs. Separately, an example normal script configuration lets an existing cookie expire before creating another. It describes ordinary expiry with that configuration, not recreation after opt-out. | The publisher can set the opt-out marker to prevent further cookie setting on module runs. Separately, an example normal script configuration lets an existing cookie expire before creating another. It describes ordinary expiry with that configuration, not recreation after opt-out. | IDN-AS5-F13; data/claims-extra.json |
| X828 | The v1.0 Prebid example supplies the identityLink configuration and read call for idl_env. An enabled SSP Sidecar decodes a valid envelope and translates it to the DSP-specific ID; an expired envelope cannot be decrypted. [4] | Reopen the source and test the exact anchor on the frozen service and route. | The Prebid snippet configures and reads idl_env; it does not actually send a worked envelope to Sidecar. The overview defines the handoff and the output unit, including the downstream DSP-specific RampID. Preserve documented expiry separately. [184, 133, 137] | supported with qualification | The Prebid snippet configures and reads idl_env; it does not actually send a worked envelope to Sidecar. The overview defines the handoff and the output unit, including the downstream DSP-specific RampID. Preserve documented expiry separately. [2, verified] | The Prebid snippet configures and reads idl_env; it does not actually send a worked envelope to Sidecar. The overview defines the handoff and the output unit, including the downstream DSP-specific RampID. Preserve documented expiry separately. | IDN-P001 liveramp-rampid / recipient_interface; data/scores.json or data/matrix-extra.json |
| X829 | The dated IdentityLink-to-RampID rename identifies the label change; documented legacy Prebid integration names and the idl_env storage key explain how implementers recognize the old representation. No versioned before/after output test is supplied. [3] | Reopen the source and test the exact anchor on the frozen service and route. | The May 24, 2021 notice describes a name and branding rollout, explicitly saying the change does not affect the product. The legacy code names confirm compatibility but show no dated change to the sold output, field or rule. [185, 184] | withdrawn | The May 24, 2021 notice describes a name and branding rollout, explicitly saying the change does not affect the product. The legacy code names confirm compatibility but show no dated change to the sold output, field or rule. [ne, unverified] | The May 24, 2021 notice describes a name and branding rollout, explicitly saying the change does not affect the product. The legacy code names confirm compatibility but show no dated change to the sold output, field or rule. | IDN-P001 liveramp-rampid / change_traceability; data/scores.json or data/matrix-extra.json |
| X830 | The operator may issue a usable token or report an ecosystem opt-out. Its DSP webhook distributes the affected raw ID. The documented DSP bidder must stop using it in real-time bids, even if a previous token still decrypts. The rules were read at cutoff. [4] | Reopen the source and test the exact anchor on the frozen service and route. | The operator and DSP guide specify generation and opt-out behavior and a model webhook. The sample names a generic DSP role and an illustrative endpoint. It does not trace a denied case to a named receiving implementation. [149, 186, 144] | supported with qualification | The operator and DSP guide specify generation and opt-out behavior and a model webhook. The sample names a generic DSP role and an illustrative endpoint. It does not trace a denied case to a named receiving implementation. [3, verified] | The operator and DSP guide specify generation and opt-out behavior and a model webhook. The sample names a generic DSP role and an illustrative endpoint. It does not trace a denied case to a named receiving implementation. | IDN-P002 ttd-uid2 / permission_handling; data/scores.json or data/matrix-extra.json |
| X831 | Deep Sync reports 51% before and 66% after enrichment for a US retailer’s nominal 141.7 million-record CRM file. The two matched record counts are published. The observation window and exact destination-eligible base are undisclosed; cutoff is the reading date. [2] | Reopen the source and test the exact anchor on the frozen service and route. | Deep Sync supplies before/after counts and percentages for matching a retailer CRM file to a named activation destination. It does not identify token generation or credentialed bidder decryption on the frozen public-operator route. [31] | withdrawn | Deep Sync supplies before/after counts and percentages for matching a retailer CRM file to a named activation destination. It does not identify token generation or credentialed bidder decryption on the frozen public-operator route. [ne, unverified] | Deep Sync supplies before/after counts and percentages for matching a retailer CRM file to a named activation destination. It does not identify token generation or credentialed bidder decryption on the frozen public-operator route. | IDN-P002 ttd-uid2 / match_rate_definition; data/scores.json or data/matrix-extra.json |
| X832 | The Prebid module guide, read at cutoff, identifies 8.33.0 as the boundary for externalModuleUrl interoperability with the ID5 API. No paired older/updated handoff example or migration treatment is supplied for that change. The SDK cookie-sync defaults did not change across the inspected releases. [2] | Reopen the source and test the exact anchor on the frozen service and route. | The module guide mentions a version boundary for externalModuleUrl interoperability. It provides no release date for that behavior. The dated SDK README snapshots concern separate versions and do not establish the date of this Prebid change. [175, 179, 180] | withdrawn | The module guide mentions a version boundary for externalModuleUrl interoperability. It provides no release date for that behavior. The dated SDK README snapshots concern separate versions and do not establish the date of this Prebid change. [ne, unverified] | The module guide mentions a version boundary for externalModuleUrl interoperability. It provides no release date for that behavior. The dated SDK README snapshots concern separate versions and do not establish the date of this Prebid change. | IDN-P004 id5-id / change_traceability; data/scores.json or data/matrix-extra.json |
| X833 | Calling the module at every opportunity and its private cache do not specify when Lotame refreshes or expires the Panorama ID. [ne] | Reopen the source and test the exact anchor on the frozen service and route. | The separately reopened service notice describes a time-bounded post-opt-out expiry on this browser. That narrow clock is not the lifespan of an active graph edge. [261] | supported with qualification | The cutoff privacy notice gives a specific browser-output clock: a previously stored Panorama ID may take up to seven days to expire after the opt-out. The active graph link and remote bidder copy have no documented time-to-live here. [3, verified] | The separately reopened service notice describes a time-bounded post-opt-out expiry on this browser. That narrow clock is not the lifespan of an active graph edge. | IDN-P006 lotame-panorama / refresh_expiry; data/scores.json or data/matrix-extra.json |
| X834 | Lotame identifies a consumer opt-out for Panorama ID. Its launch statement does not specify a request-time state evaluator or demonstrate recipient behavior. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The current privacy notice describes the browser input and distinct outcomes depending on third-party-cookie support. The original launch summary alone understated documented handling. [261, 374] | supported with qualification | The module issues on eligible publisher requests. The cutoff privacy notice describes browser-specific opt-out, a cookie marker on supporting browsers and a conditional probabilistic recognition attempt where third-party cookies are blocked. No bidder’s restricted-state handling is established. [3, verified] | The current privacy notice describes the browser input and distinct outcomes depending on third-party-cookie support. The original launch summary alone understated documented handling. | IDN-P006 lotame-panorama / permission_handling; data/scores.json or data/matrix-extra.json |
| X835 | Lotame claimed targeting accuracy for Panorama ID’s device-to-person triangulation. It published no reference labels, denominator or error estimate with that claim. [1] | Reopen the source and test the exact anchor on the frozen service and route. | Lotame links attributes to a general assertion of high precision; the release reports no result from testing the ID’s correctness. Attribute counts and customer yield do not satisfy R11. [375] | withdrawn | Lotame describes triangulation and claims the highest targeting accuracy. The cited release reports no evaluation of actual Panorama ID assignments or measured accuracy result. Attribute counts do not substantiate correctness. [ne, unverified] | Lotame links attributes to a general assertion of high precision; the release reports no result from testing the ID’s correctness. Attribute counts and customer yield do not satisfy R11. | IDN-P006 lotame-panorama / output_validation; data/scores.json or data/matrix-extra.json |
| X836 | Lotame’s product announcement offers consumers a persistent opt-out across touchpoints. The evidence does not identify the key or the authentication required. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The published policy identifies the browser, the individual initiating the request and what Lotame does to the corresponding IDs. Re-identification in a blocked-cookie browser is fallible. [261] | supported with qualification | A user requests opt-out for a particular browser. Lotame sets a marker or attempts probabilistic recognition to honor the choice; prior browser storage may take seven days to expire. A new ID may be assigned when recognition fails. The source does not demonstrate treatment of earlier Panorama IDs at recipients. [3, verified] | The published policy identifies the browser, the individual initiating the request and what Lotame does to the corresponding IDs. Re-identification in a blocked-cookie browser is fallible. | IDN-P006 lotame-panorama / withdrawal_controls; data/scores.json or data/matrix-extra.json |
| X837 | The Prebid module can extend the stored ID on each page load; its example sets a 365-day cookie life. Each setting applies to the publisher-scoped ID, and browser limits may shorten it. Evidence as of November 11, 2026. [3] | Reopen the source and test the exact anchor on the frozen service and route. | The dedicated guide’s script route provides expInterval and extend, explains expiry from creation or renewal on execution and illustrates allowing expiry before a new ID is created. [286] | supported with qualification | On Prebid’s published SharedID script route, extend can renew expiry at each execution or leave it fixed from creation. expInterval sets the clock, and the example allows expiry before a fresh ID is created. This is the publisher-scoped ID, not a downstream identity graph. Guide read at cutoff. [4, verified] | The dedicated guide’s script route provides expInterval and extend, explains expiry from creation or renewal on execution and illustrates allowing expiry before a new ID is created. | IDN-P009 sharedid / refresh_expiry; data/scores.json or data/matrix-extra.json |
| X838 | The module defines bidRequest.userId.sharedid and sharedid.org for bidder adapters. Its header example abbreviates the UUID. It does not demonstrate a complete delivery to a named receiving implementation. Evidence as of November 11, 2026. [2] | Reopen the source and test the exact anchor on the frozen service and route. | Prebid’s long, version-specific guide includes a complete UUID in both EIDs and dates a namespace boundary; the module header’s abbreviated example alone misses that distinction. [286, 285] | supported with qualification | The version-specific Prebid guide shows a complete sample EID with a UUID. It distinguishes the pre-5.0 sharedid.org EID from the pubcid.org EID adopted at the merge. Bid adapters forward the eligible value; handling of an invalid incoming ID at the buyer is unspecified. [3, verified] | Prebid’s long, version-specific guide includes a complete UUID in both EIDs and dates a namespace boundary; the module header’s abbreviated example alone misses that distinction. | IDN-P009 sharedid / recipient_interface; data/scores.json or data/matrix-extra.json |
| X839 | The guide identifies the 4.x-to-5.0 removal of third-party synchronization. The checked material does not date that technical release. The pinned module snapshot is not its change date. [ne] | Reopen the source and test the exact anchor on the frozen service and route. | GitHub’s release record dates the module merge to June 10, 2021. The Prebid guide names version 5, documents both EIDs and explains which namespace survived for buyers. [285, 286] | supported with qualification | Prebid.js 5.0.0 merged SharedID and PubCommon. The guide documents the prior third-party sync, explains why pubcid.org was retained for buyers and gives complete before-and-after EID payloads. The June 10, 2021 release dates the merge. [4, verified] | GitHub’s release record dates the module merge to June 10, 2021. The Prebid guide names version 5, documents both EIDs and explains which namespace survived for buyers. | IDN-P009 sharedid / change_traceability; data/scores.json or data/matrix-extra.json |
| X840 | The US sales page and Digital Graph seller claim more accurate cross-device targeting and attribution, including in the DSP case. It provides no labeled references or denominator for accuracy. Its 84% is a sync yield and receives no accuracy credit. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The DSP case reports synced IDs and a match-rate increase. Its account relates improved accuracy to media attribution. No test result for correct person or household links on the sold US Graph appears. The US page’s promotional accuracy language supplies no such result. [117, 36, 37] | withdrawn | The DSP case reports synced IDs and a match-rate increase. Its account relates improved accuracy to media attribution. No test result for correct person or household links on the sold US Graph appears. The US page’s promotional accuracy language supplies no such result. [1, unverified] | The DSP case reports synced IDs and a match-rate increase. Its account relates improved accuracy to media attribution. No test result for correct person or household links on the sold US Graph appears. The US page’s promotional accuracy language supplies no such result. | IDN-P013 experian-resolution / output_validation; data/scores.json or data/matrix-extra.json |
| X841 | TruAudience offers a named Snowflake native application that returns an augmented customer view. The separately advertised Graph Extracts deliverable is excluded. The integrated app record does not specify a transmitted field name or schema for the buyer’s enriched view. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The native offer says customers can use the same individual identifier to unite customer data with digital engagement data. That specifies the delivered key, unit and user role. The publication gives no column-level schema or worked row for this app. [19] | supported with qualification | The native offer says customers can use the same individual identifier to unite customer data with digital engagement data. That specifies the delivered key, unit and user role. The publication gives no column-level schema or worked row for this app. [2, qualified] | The native offer says customers can use the same individual identifier to unite customer data with digital engagement data. That specifies the delivered key, unit and user role. The publication gives no column-level schema or worked row for this app. | IDN-P014 truaudience-identity-resolution / recipient_interface; data/scores.json or data/matrix-extra.json |
| X842 | The provider cites attributes for over 250 million US consumers used in the match but states no period or actual inventory date. [ne] | Reopen the source and test the exact anchor on the frozen service and route. | TransUnion’s January 4, 2024 release describes its enhanced reference graph at that time as representing 250 million adults and 98% of the US adult population. The named Snowflake app uses this graph to match buyer records. The same-period adult numerator and base are attributed to the issuer. No deduplication or model/observation split is given. [316, 19] | supported with qualification | TransUnion’s January 4, 2024 release describes its enhanced reference graph at that time as representing 250 million adults and 98% of the US adult population. The named Snowflake app uses this graph to match buyer records. The same-period adult numerator and base are attributed to the issuer. No deduplication or model/observation split is given. [3, qualified] | TransUnion’s January 4, 2024 release describes its enhanced reference graph at that time as representing 250 million adults and 98% of the US adult population. The named Snowflake app uses this graph to match buyer records. The same-period adult numerator and base are attributed to the issuer. No deduplication or model/observation split is given. | IDN-P014 truaudience-identity-resolution / us_coverage_measure; data/scores.json or data/matrix-extra.json |
| X843 | The seller claims more consistent resolution and improved identity accuracy. The US native-app description does not give reference labels, error counts or an accuracy denominator. No health-assessment yield is counted as accuracy. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The cloud offer describes accuracy as a benefit. The separate issuer report describes a Truthset ranking on linking hashed emails to physical addresses. This is a different link direction and unit from accepted customer-record-to-persistent-identity output on the Snowflake service. The product materials were available at the November 11, 2026 cutoff. [19, 316] | withdrawn | The cloud offer describes accuracy as a benefit. The separate issuer report describes a Truthset ranking on linking hashed emails to physical addresses. This is a different link direction and unit from accepted customer-record-to-persistent-identity output on the Snowflake service. The product materials were available at the November 11, 2026 cutoff. [1, unverified] | The cloud offer describes accuracy as a benefit. The separate issuer report describes a Truthset ranking on linking hashed emails to physical addresses. This is a different link direction and unit from accepted customer-record-to-persistent-identity output on the Snowflake service. The product materials were available at the November 11, 2026 cutoff. | IDN-P014 truaudience-identity-resolution / output_validation; data/scores.json or data/matrix-extra.json |
| X844 | The processing specification identifies tagged customer transactions on the client’s own website or app, the client property supplying the signal, and the client profile where that event is placed. The offline file is sent by the client. The complete mandatory-versus-optional input set is not documented. [3] | Reopen the source and test the exact anchor on the frozen service and route. | The client supplies an offline file of names, addresses, emails, customer IDs and transactions. The contract names the supplier and fields without stating how the client collected those fields. The explicitly observed website and app events belong to a separate tag flow. The product materials were available at the November 11, 2026 cutoff. [385] | supported with qualification | The client supplies an offline file of names, addresses, emails, customer IDs and transactions. The contract names the supplier and fields without stating how the client collected those fields. The explicitly observed website and app events belong to a separate tag flow. The product materials were available at the November 11, 2026 cutoff. [2, qualified] | The client supplies an offline file of names, addresses, emails, customer IDs and transactions. The contract names the supplier and fields without stating how the client collected those fields. The explicitly observed website and app events belong to a separate tag flow. The product materials were available at the November 11, 2026 cutoff. | IDN-P015 epsilon-core-id / input_provenance; data/scores.json or data/matrix-extra.json |
| X845 | The specification says first-party cookies incorporated for US data subjects are restricted to the contributing client. This is a scoped client-use control. It does not identify a per-record consent or evaluation event on the onboarding file. [1] | Reopen the source and test the exact anchor on the frozen service and route. | Client-exclusive first-party cookies are documented only in the digital tagging clause. The offline-file clause describes dropping contact information from the resulting client profile. It specifies neither a use-restriction input nor a use-permission decision at the file handoff. The product materials were available at the November 11, 2026 cutoff. [385] | withdrawn | Client-exclusive first-party cookies are documented only in the digital tagging clause. The offline-file clause describes dropping contact information from the resulting client profile. It specifies neither a use-restriction input nor a use-permission decision at the file handoff. The product materials were available at the November 11, 2026 cutoff. [1, unverified] | Client-exclusive first-party cookies are documented only in the digital tagging clause. The offline-file clause describes dropping contact information from the resulting client profile. It specifies neither a use-restriction input nor a use-permission decision at the file handoff. The product materials were available at the November 11, 2026 cutoff. | IDN-P015 epsilon-core-id / permission_handling; data/scores.json or data/matrix-extra.json |
| X846 | The named COREid sales page claims accuracy for its resolution product. It provides no reference source, denominator or independently measured error. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The US product page calls the service the industry’s most accurate. It gives no returned-link correctness result or reported test of its claims. The processing specification supplies a workflow rather than a validation study. The product materials were available at the November 11, 2026 cutoff. [386] | withdrawn | The US product page calls the service the industry’s most accurate. It gives no returned-link correctness result or reported test of its claims. The processing specification supplies a workflow rather than a validation study. The product materials were available at the November 11, 2026 cutoff. [1, unverified] | The US product page calls the service the industry’s most accurate. It gives no returned-link correctness result or reported test of its claims. The processing specification supplies a workflow rather than a validation study. The product materials were available at the November 11, 2026 cutoff. | IDN-P015 epsilon-core-id / output_validation; data/scores.json or data/matrix-extra.json |
| X847 | Dentsu identifies a US test with the Hy-Vee RedMedia operation through its Trade Desk integration. The October appointment still describes broader use prospectively. Freeze status is beta; this does not establish general US deployment. [2] | Reopen the source and test the exact anchor on the frozen service and route. | The January account documents a US test with Hy-Vee’s RedMedia. The later agency announcement describes plans to extend usage. Neither source establishes a generally accessible US beta or production access through the named integration for a US buyer beyond the test. [359, 360, 389] | supported with qualification | The January account documents a US test with Hy-Vee’s RedMedia. The later agency announcement describes plans to extend usage. Neither source establishes a generally accessible US beta or production access through the named integration for a US buyer beyond the test. [1, qualified] | The January account documents a US test with Hy-Vee’s RedMedia. The later agency announcement describes plans to extend usage. Neither source establishes a generally accessible US beta or production access through the named integration for a US buyer beyond the test. | IDN-P017 merkle-merkury / us_availability; data/scores.json or data/matrix-extra.json |
| X848 | The dated announcement describes a tested change in integration: UID2 connections let Merkury audiences reach The Trade Desk and support closed-loop measurement. It supplies no release version, effective date or migration record. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The dated January announcement documents new Trade Desk connectivity and closed-loop measurement. It does not identify a change in the rule, definition or delivered private identity link sold under this row. [359] | withdrawn | The dated January announcement documents new Trade Desk connectivity and closed-loop measurement. It does not identify a change in the rule, definition or delivered private identity link sold under this row. [1, unverified] | The dated January announcement documents new Trade Desk connectivity and closed-loop measurement. It does not identify a change in the rule, definition or delivered private identity link sold under this row. | IDN-P017 merkle-merkury / change_traceability; data/scores.json or data/matrix-extra.json |
| X849 | The issuer describes updating the eligible inventory pool by activating content as it is published. It supplies this event for adding content but no rule for reclassifying an existing URL or removing stale inventory. [2] | Reopen the source and test the exact anchor on the frozen service and route. | The cited trigger admits newly published pages. It says nothing about re-evaluating or removing an existing membership, or that the eligible pool is rebuilt each time. [392] | withdrawn | The cited trigger admits newly published pages. It says nothing about re-evaluating or removing an existing membership, or that the eligible pool is rebuilt each time. [2, unverified] | The cited trigger admits newly published pages. It says nothing about re-evaluating or removing an existing membership, or that the eligible pool is rebuilt each time. | IDN-P022 verve-moments-ai / refresh_expiry; data/scores.json or data/matrix-extra.json |
| X850 | Verve explicitly claims accuracy for its segmentation of content for targeting. The US launch gives no labeled URL benchmark, sample size or denominator for that claim. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. [392] | withdrawn | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. [1, unverified] | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. | IDN-P022 verve-moments-ai / output_validation; data/scores.json or data/matrix-extra.json |
| X851 | Seedtag claims accurate targeting of specific interests by the audience groups but supplies no independent classification sample, label source or denominator for correct custom assignment on the managed route. [1] | Reopen the source and test the exact anchor on the frozen service and route. | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. [355] | withdrawn | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. [1, unverified] | The release describes desired targeting precision. It reports no result from checking the correctness of assignments of the sold segment. R11 requires an actual reported result; this wording cannot meet anchor 1. | IDN-P023 seedtag-contextual-audiences / output_validation; data/scores.json or data/matrix-extra.json |
| X852 | The guide calls linking probabilistic but gives no observed match rate. [ne] | Reopen the source and test the exact anchor on the frozen service and route. | A separate November 2024 release claims 90% average bidstream matches for the joint MediaWallah–Lexicon onboarding offer. It does not isolate a yield for the native 33Across ID on its publisher route. [401, 34] | supported with qualification | The joint MediaWallah–Lexicon offering claims a 90% average bidstream match rate. Its source does not isolate the native Lexicon service’s contribution or identify its eligible input population. The publisher module itself gives no observed numeric match rate. [ne, verified] | A separate November 2024 release claims 90% average bidstream matches for the joint MediaWallah–Lexicon onboarding offer. It does not isolate a yield for the native 33Across ID on its publisher route. | IDN-P038 33across-lexicon / match_rate_definition; data/scores.json or data/matrix-extra.json |
| X853 | The report describes one month of the US grocer’s full POS volume, and a 73.0% cumulative rate linking transactions to inferred or known buyers. It specifies direction, input owner, unit and period length, but not the calendar month. The approximate pass counts do not reconcile to the headline and there is no exact deduplication or eligibility rule. [3] | Reopen the source and test the exact anchor on the frozen service and route. | The seller gives 73.0% of the grocery chain’s POS volume in a single month, measured as transaction rows assigned to a buyer across three passes. The month is unnamed and the publication date does not date it. The rounded stage counts do not exactly reproduce the headline. [12] | supported with qualification | The seller gives 73.0% of the grocery chain’s POS volume in a single month, measured as transaction rows assigned to a buyer across three passes. The month is unnamed and the publication date does not date it. The rounded stage counts do not exactly reproduce the headline. [2, qualified] | The seller gives 73.0% of the grocery chain’s POS volume in a single month, measured as transaction rows assigned to a buyer across three passes. The month is unnamed and the publication date does not date it. The rounded stage counts do not exactly reproduce the headline. | IDN-P043 deep-sync-transaction-matching / match_rate_definition; data/scores.json or data/matrix-extra.json |
| X854 | On module initialization LiveConnect receives the us_privacy string from Prebid’s consent module and says it respects opt-outs. It does not state the distinct bid payload produced in each state. Evidence as of November 11, 2026. [2] | Reopen the source and test the exact anchor on the frozen service and route. | The separate privacy and consent guide identifies the allowed URL fields and explicitly says its blocked handler stops identity requests. It establishes two sender outcomes, but no bidder disposition. [260, 219] | supported with qualification | For eligible identity-resolution calls Live Connect Hub sends the documented consent fields. A named blocked-call handler prevents the identity request in privacy mode. Both sender outcomes are specified; downstream bidder handling of an absent or denied nonID remains unverified. Guide read at cutoff. [3, verified] | The separate privacy and consent guide identifies the allowed URL fields and explicitly says its blocked handler stops identity requests. It establishes two sender outcomes, but no bidder disposition. | IDN-P054 liveintent-nonid / permission_handling; data/scores.json or data/matrix-extra.json |
| X855 | IDN-B5-F01 used 19 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-C-S12 [316] | supported with qualification | Retained the exact 6-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-B5-F01 |
| X856 | IDN-B5-F01 used 14 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-CV-S02 [19] | supported with qualification | Retained the exact 3-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-B5-F01 |
| X857 | IDN-B5-F02 used 15 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-CV-S02 [19] | supported with qualification | Retained the exact 3-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-B5-F02 |
| X858 | IDN-VA5-F01 used 14 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-U-S13 [286] | supported with qualification | Retained the exact 5-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-VA5-F01 |
| X859 | IDN-AS5-F01 used 15 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-A-S15 [374] | supported with qualification | Retained the exact 11-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F01 |
| X860 | IDN-AS5-F03 used 13 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-U-S13 [286] | supported with qualification | Retained the exact 6-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F03 |
| X861 | IDN-AS5-F04 used 16 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-A-S18 [258] | supported with qualification | Retained the exact 5-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F04 |
| X862 | IDN-AS5-F08 used 15 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-UV-S08 [408] | supported with qualification | Retained the exact 3-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F08 |
| X863 | IDN-AS5-F09 used 11 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-UV-S08 [408] | supported with qualification | Retained the exact 7-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F09 |
| X864 | IDN-AS5-F13 used 11 quoted words in the cited extract. | Apply cumulative source quotation budget without changing the factual finding. | IDN-U-S13 [286] | supported with qualification | Retained the exact 3-word excerpt and its original locator. | Apply cumulative source quotation budget without changing the factual finding. | IDN-AS5-F13 |
| X865 | H3 counted Yahoo and Criteo inside the EU and UK scopes using a generic Europe route. | A European region label cannot place an exact service in each legal territory. | IDN-YAH-S01, IDN-A5-S01 [160, 263] | supported with qualification | Moved each product to geographic scope unresolved in those subgroups; confirmed EU and UK each move from 1/5–3/5 to 1/3. US and primary results retain their prior scope. | A European region label cannot place an exact service in each legal territory. | H3 regional scope |
| X866 | H6 case matrix omitted the LiveRamp competition docket and UK and French complaints although its evidence file discussed them. | An omission could obscure why nearby cases fail the named-service implementation test. | IDN-LR-S17, IDN-LR-S18, IDN-LR-S19 [87, 88, 89] | supported with qualification | Added the interim competition order, proposed remedy and two reported complaints; the private Yahoo proceeding is set apart. The implementation test remains unresolved. | An omission could obscure why nearby cases fail the named-service implementation test. | H6 legal search |
| X867 | Deep Sync says it found a buyer for 73.0% of the sales in one US grocer’s monthly file [claim IDN-C-F16]. | The issuer counts transaction rows across three passes. The last infers a buyer and does not verify who paid. | IDN-C-S20 [12] | supported with qualification | Deep Sync says it linked 73.0% of the transactions in one US grocer’s monthly file to an inferred buyer [claim IDN-C-F16]. | The issuer counts transaction rows across three passes. The last infers a buyer and does not verify who paid. | 00-abstract.md |
| X868 | The case gives no test of who paid for each shop [claim IDN-C-F16]. | State the untested unit as a purchase and leave the published yield intact. | IDN-C-S20 [12] | supported with qualification | The case gives no separate test of who paid for each purchase [claim IDN-C-F16]. | State the untested unit as a purchase and leave the published yield intact. | 00-abstract.md |
| X869 | The search left a different claim open. | A separate sentence reports the H2 result and needs its own marker. | IDN-METHOD-S06 [14] | supported with qualification | The search left the test of link accuracy open [claim IDN-HYP-H2]. | A separate sentence reports the H2 result and needs its own marker. | 00-abstract.md |
| X870 | In the group of US IDs known to be in scope, too few firms showed how a user’s choice reached the next party and changed what it could do [claim IDN-HYP-H3]. | H3 counts products offered in the US and gives credit for a documented path in any eligible jurisdiction. The US-only calculation is distinct. | IDN-METHOD-S04, IDN-METHOD-S06 [13, 14] | supported with qualification | Among the IDs known to be sold in the US, too few showed how a choice reached the next party and changed what it could do in at least one market where the ID was offered [claim IDN-HYP-H3]. | H3 counts products offered in the US and gives credit for a documented path in any eligible jurisdiction. The US-only calculation is distinct. | 00-abstract.md |
| X871 | A score can change when the date of the proof is wrong or when the source describes a different service. | A statement about scoring is traceable to the frozen cell and date rules. | IDN-METHOD-S03 [11] | supported with qualification | A score can change when the date of the proof is wrong or when the source describes a different service [claim IDN-METHOD-F02]. | A statement about scoring is traceable to the frozen cell and date rules. | 14-appendix.md |
| X872 | A firm can look back at old web addresses without saying how long it keeps the current link. | Distinguish network addresses from web URLs; the manual gives a configurable historical IP lookup. | IDN-CTV-S16 [114] | supported with qualification | A firm can look up a past IP address without saying how long it keeps the link to a home. | Distinguish network addresses from web URLs; the manual gives a configurable historical IP lookup. | plain.md |
| X873 | Some products had too few facts to get a score. | Three assessed products fell below the four-cell floor. Others entered the matrix without a band. | IDN-METHOD-S05, IDN-METHOD-S07 [10, 218] | supported with qualification | Some products had too few checked facts to enter the score table. | Three assessed products fell below the four-cell floor. Others entered the matrix without a band. | plain.md |
| X874 | A publisher can ask for a token, a mapper can send in client data and a bidder can open a token [claim IDN-UID-F06]. | The permission matrix specifies email/phone inputs and accepted hashes, not arbitrary client data. | IDN-UID-S02, IDN-UID-S03 [143, 144] | supported with qualification | A publisher can ask for a token, a mapper can submit an email, phone number or its hash, and a bidder can open a token [claim IDN-UID-F06]. | The permission matrix specifies email/phone inputs and accepted hashes, not arbitrary client data. | 01-defining.md |
| X875 | The Trade Desk describes a CRM route that lets a client target ads in its own seat but keeps the IDs and the map from client rows to IDs [claim IDN-UID-F09]. | The license table withholds mapping from the client. The old sentence leaves it unclear which party keeps the map. | IDN-UID-S11, IDN-U-S03 [188, 203] | supported with qualification | The Trade Desk describes a CRM route that lets a client target ads in its own seat without receiving the IDs or the map from client rows to IDs [claim IDN-UID-F09]. | The license table withholds mapping from the client. The old sentence leaves it unclear which party keeps the map. | 01-defining.md |
| X876 | A Google Ad Manager case names TelevisaUnivision as a user of its standard groups on US ad space [claim IDN-U-F19]. | The case attributes the cohorts to Permutive. Name the owner rather than using an ambiguous pronoun after Google Ad Manager. | IDN-U-S25, IDN-U-S26, IDN-UV-S02, IDN-UV-S03 [205, 206, 207, 208] | supported with qualification | A Google Ad Manager case names TelevisaUnivision as a user of Permutive’s standard groups on US ad space [claim IDN-U-F19]. | The case attributes the cohorts to Permutive. Name the owner rather than using an ambiguous pronoun after Google Ad Manager. | 01-defining.md |
| X877 | OpenRTB 2.6 has fields for the party that inserted an ID, the technology that matched it and the match method [claim IDN-A-F01]. | OpenRTB 2.6 §3.2.27 describes matcher as the technology providing the method. The field is a provenance declaration, not observation of a correct match. | IDN-A-S02, IDN-A-S01 [252, 1] | supported with qualification | OpenRTB 2.6 has fields for who inserted an ID, whose technology supplied the match method, and which method was used [claim IDN-A-F01]. | OpenRTB 2.6 §3.2.27 describes matcher as the technology providing the method. The field is a provenance declaration, not observation of a correct match. | 02-standards.md |
| X878 | From Prebid.js 11.14, ID modules skip retrieval if a check with a configured consent platform fails [claim IDN-A-F05]. | Prebid User ID Module limits this failure behavior to the configured TCF/GPP module, not all consent mechanisms. | IDN-A-S06, IDN-AV-S01 [239, 255] | supported with qualification | From Prebid.js 11.14, ID modules skip retrieval if a configured TCF or GPP consent-platform lookup fails [claim IDN-A-F05]. | Prebid User ID Module limits this failure behavior to the configured TCF/GPP module, not all consent mechanisms. | 02-standards.md |
| X879 | One names the pilot, one shows a join and one says where the group goes. | The short recap states what three actual Instacart records show and needs its own source markers. | IDN-F-S14, IDN-F-S15, IDN-F-S16 [296, 297, 298] | supported with qualification | One names the pilot, one shows a join and one says where the group goes [claim IDN-F-F14] [claim IDN-F-F15] [claim IDN-F-F16]. | The short recap states what three actual Instacart records show and needs its own source markers. | 06-platforms.md |
| X880 | A US grocer gave Deep Sync a month of sales [claim IDN-C-F16]. | The issuer describes a monthly POS file and production workflow; it does not identify a particular month or certify a single handover. | IDN-C-S20 [12] | supported with qualification | Deep Sync describes a US grocer’s monthly sales file [claim IDN-C-F16]. | The issuer describes a monthly POS file and production workflow; it does not identify a particular month or certify a single handover. | 03-match-rates.md |
| X881 | You can see how the share grew. | Added the relevant case citation to the unmarked cumulative comparison. | IDN-C-S20 [12] | supported with qualification | You can see how the share grew [claim IDN-C-F16]. | Added the relevant case citation to the unmarked cumulative comparison. | 03-match-rates.md |
| X882 | You still cannot tell who paid for a given bag of food [claim IDN-C-F16]. | Bound the absence to the published probabilistic third pass. A statement about what any reader could know was broader. | IDN-C-S20 [12] | supported with qualification | The case gives no check against known buyers for that last step [claim IDN-C-F16]. | Bound the absence to the published probabilistic third pass. A statement about what any reader could know was broader. | 03-match-rates.md |
| X883 | TransUnion says it found a person for 99.5% of a US insurer's records [claim IDN-C-F04]. | Preserves the asserted graph link as the output of the Data Health Assessment, distinct from identifying a verified person. | IDN-C-S04, IDN-CV-S01 [17, 18] | supported with qualification | TransUnion says it linked 99.5% of a US insurer’s records to a person in its offline graph [claim IDN-C-F04]. | Preserves the asserted graph link as the output of the Data Health Assessment, distinct from identifying a verified person. | 03-match-rates.md |
| X884 | A sale found on the first pass is still in the count at the end. | The first pass stays in the explicitly cumulative monthly POS waterfall. Added source to the unmarked statement. | IDN-C-S20 [12] | supported with qualification | A sale found on the first pass is still in the count at the end [claim IDN-C-F16]. | The first pass stays in the explicitly cumulative monthly POS waterfall. Added source to the unmarked statement. | 03-match-rates.md |
| X885 | The source gives no full rule for the 99.5%, count of rows sent or time spent gathering the file [claim IDN-C-F04]. | The check concerns undisclosed study period. The research did not seek time spent gathering records. | IDN-CV-S01 [18] | supported with qualification | The source gives no full match rule for the 99.5%, count of rows sent or dates for the test [claim IDN-C-F04]. | The check concerns undisclosed study period. The research did not seek time spent gathering records. | 03-match-rates.md |
| X886 | MiQ also bought access to hashed emails, UID2s and partner cookies for its own Identity Spine [claim IDN-EXP-F06]. | The issuer confirms licensing on PDF p. 4. The commercial fee is not disclosed. | IDN-C-S06 [21] | supported with qualification | Experian says MiQ licensed hashed emails, UID2s and partner cookies for its own Identity Spine [claim IDN-EXP-F06]. | The issuer confirms licensing on PDF p. 4. The commercial fee is not disclosed. | 03-match-rates.md |
| X887 | The case labels those rows 51% and 66% [claim IDN-C-F26]. | These are before/after entries in a single UID2 output row, not two output rows. | IDN-C-S30 [31] | supported with qualification | The case labels those two results 51% and 66% [claim IDN-C-F26]. | These are before/after entries in a single UID2 output row, not two output rows. | 03-match-rates.md |
| X888 | Google put an 81% revenue gain on restricted-cookie browsers in a New York Times Advertising case that used both UID2 and RampID [claim IDN-LR-F10]. | The page describes Ad Manager impact estimates and both IDs. This is not an observed controlled effect of one ID. | IDN-LR-S06 [59] | supported with qualification | Google reported an estimated 81% revenue gain on restricted-cookie browsers in a New York Times Advertising case that used both UID2 and RampID [claim IDN-LR-F10]. | The page describes Ad Manager impact estimates and both IDs. This is not an observed controlled effect of one ID. | 03-match-rates.md |
| X889 | In a CIMM TV study the group left after matching skewed older, richer and more keen on TV [claim IDN-B-F03]. | CIMM reports income and heavier viewing, not wealth or a preference for television. | IDN-B-S03 [42] | supported with qualification | In a CIMM TV study the matched group skewed older, had higher incomes and watched more TV [claim IDN-B-F03]. | CIMM reports income and heavier viewing, not wealth or a preference for television. | 03-match-rates.md |
| X890 | The grocery case overturns the claim that no scored product gives a US rate, a clear base, the matched group and the method [claim IDN-HYP-H1]. | The preregistered counterexample is a product admitted to the IQ; applying H1 to any scored-cell review would change the frame. | IDN-METHOD-S04, IDN-METHOD-S06, IDN-C-S20 [13, 14, 12] | supported with qualification | The grocery case overturns the claim that no admitted product gives a US rate, a clear base, the matched group and the method [claim IDN-HYP-H1]. | The preregistered counterexample is a product admitted to the IQ; applying H1 to any scored-cell review would change the frame. | 03-match-rates.md |
| X891 | The paper could not settle whether a public, independent test verified a scored graph's person or household links in the US against known truth [claim IDN-HYP-H2]. | Use the admitted, named graph buying-task frame, with people or households and US reference population. | IDN-METHOD-S04, IDN-METHOD-S06, IDN-B-S01, IDN-B-S04, IDN-B-S05, IDN-HX-S06 [13, 14, 40, 43, 45, 62] | supported with qualification | The paper could not settle whether a public, independent test verified an admitted graph’s US person or household links against known truth [claim IDN-HYP-H2]. | Use the admitted, named graph buying-task frame, with people or households and US reference population. | 03-match-rates.md |
| X892 | Experian sells its Digital Graph as a dataset or as links it adds to a buyer's records [claim IDN-EXP-F01]. | The official US product sheet offers both forms; the page supplies no public price and does not establish terms for other territories. | IDN-U-S17 [117] | supported with qualification | Experian offers its US Digital Graph as a full dataset or as links it adds to a buyer’s records [claim IDN-EXP-F01]. | The official US product sheet offers both forms; the page supplies no public price and does not establish terms for other territories. | 12-buyer-guide.md |
| X893 | An October agency announcement called for Merkury to link enterprise media, RedMedia and loyalty programs [claim IDN-I-F07]. | The source dates the iProspect appointment and names one retailer; it is a plan. | IDN-I-S05 [360] | supported with qualification | An October 2025 agency announcement called for Merkury to link Hy-Vee’s enterprise media, RedMedia and loyalty programs [claim IDN-I-F07]. | The source dates the iProspect appointment and names one retailer; it is a plan. | 12-buyer-guide.md |
| X894 | To claim the ad drove the visit, plan a test with a group that could have seen the ad and did not. | The original advice suggested comparison by observed exposure. The revised reader instruction specifies assignment to a holdout where feasible. | 12-buyer-guide.md | supported with qualification | To test if the ad drove a visit, randomly hold it back from some eligible homes where feasible, then compare visits across the groups. | The original advice suggested comparison by observed exposure. The revised reader instruction specifies assignment to a holdout where feasible. | 12-buyer-guide.md |
| X895 | The DSP gets a separate message and must stop using that UID2 to bid, but the guide lets it use another ID or pass on the bid [claim IDN-UID-F13]. | The technical guide states the intended webhook and bidder behavior; actual receipt in a campaign was not observed. | IDN-A-S09 [186] | supported with qualification | The UID2 guide calls for a separate message to the DSP and tells it to stop using that UID2 to bid, while allowing another ID or no bid [claim IDN-UID-F13]. | The technical guide states the intended webhook and bidder behavior; actual receipt in a campaign was not observed. | 12-buyer-guide.md |
| X896 | A full refresh can replace the node's identity layer and leave out records missing from the new file [claim IDN-CTV-F21]. | The operator explicitly specifies a complete manual refresh in v2.29. It is not a universal scheduled expiry rule. | IDN-CTV-S12 [115] | supported with qualification | In version 2.29, a full manual refresh replaces the node’s identity layer and leaves out rows missing from the new file [claim IDN-CTV-F21]. | The operator explicitly specifies a complete manual refresh in v2.29. It is not a universal scheduled expiry rule. | 12-buyer-guide.md |
| X897 | UID2 describes a free public operator for a contracted participant, while a private operator pays for its own systems [claim IDN-UID-F09]. | Scope the no-charge route to contracted use. The source describes host and engineering resources rather than a quantified cost. | IDN-U-S03, IDN-UID-S11 [203, 188] | supported with qualification | UID2 says a firm under contract can use a public operator at no charge, while a private operator provides its own hosting and engineering [claim IDN-UID-F09]. | Scope the no-charge route to contracted use. The source describes host and engineering resources rather than a quantified cost. | 12-buyer-guide.md |
| X898 | One tells her how the send went. | Clarify an instruction about which denominator to use. | 03-match-rates.md | supported with qualification | Use the first to see how the send went. | Clarify an instruction about which denominator to use. | 03-match-rates.md |
| X899 | You can still value the effect the case describes. | Avoid implying an independently established causal effect. | 03-match-rates.md | supported with qualification | Use the reach and revenue figures for their stated purposes. | Avoid implying an independently established causal effect. | 03-match-rates.md |
| X900 | You can prove that 60 of the 80 are right. | Explicitly label all purported pairs in the arithmetic example so the 20 false positives are established rather than inferred from missing proof. | 03-match-rates.md | supported with qualification | Of the 80 claimed links, 60 are known to be true and 20 are known to be false. | Explicitly label all purported pairs in the arithmetic example so the 20 false positives are established rather than inferred from missing proof. | 03-match-rates.md |
| X901 | A site can stop issuing an ID [claim IDN-A-F11]. | LiveRamp’s original cited excerpt concerns removal of a stored envelope. The UID2 participant rule directly supports stop-generation at a particular site. | IDN-A-S12, IDN-A-S08, IDN-UID-S18 [257, 149, 150] | supported with qualification | A site can stop making new UID2s after a local opt-out [claim IDN-UID-F12]. | LiveRamp’s original cited excerpt concerns removal of a stored envelope. The UID2 participant rule directly supports stop-generation at a particular site. | 04-consent.md |
| X902 | The test covered [stat h3_t] IDs sold for use across firms [claim IDN-HYP-H3]. | US buying-task availability includes an open integration and gives no implication that every ID carries a price. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-A-S08, IDN-UID-S18, IDN-A-S09, IDN-ID5-S06, IDN-ID5-S09, IDN-LR-S03, IDN-HX-S04 [14, 13, 149, 150, 186, 175, 128, 182, 219] | supported with qualification | The test covered [stat h3_t] IDs available for use across firms [claim IDN-HYP-H3]. | US buying-task availability includes an open integration and gives no implication that every ID carries a price. | 04-consent.md |
| X903 | In Prebid.js 11.14 and later, the type of lookup failure can change the bid [claim IDN-A-F05]. | The versioned guide governs user-ID retrieval and inclusion, not bidder pricing, bid submission or other IDs. | IDN-A-S06, IDN-AV-S01 [239, 255] | supported with qualification | From Prebid.js 11.14, lookup failures can change which IDs its modules add to a bid request [claim IDN-A-F05]. | The versioned guide governs user-ID retrieval and inclusion, not bidder pricing, bid submission or other IDs. | 04-consent.md |
| X904 | It carries the raw ID and the opt-out time and tends to arrive within a few minutes [claim IDN-UID-F13]. | The issuer describes its dispatch time. The guide does not give a guarantee of receipt at every DSP in that interval. | IDN-A-S09 [186] | supported with qualification | The notice carries the raw ID and the opt-out time, and UID2 says it typically sends it within a few minutes [claim IDN-UID-F13]. | The issuer describes its dispatch time. The guide does not give a guarantee of receipt at every DSP in that interval. | 04-consent.md |
| X905 | A token issued before the notice could still decrypt [claim IDN-UID-F13]. | The directly documented order of operations establishes that decryption is not permission to bid, without a claim about every earlier-issued token. | IDN-A-S09, IDN-A-S08 [186, 149] | supported with qualification | The DSP guide tells the bidder to check opt-out status after it decrypts the token [claim IDN-A-F08]. | The directly documented order of operations establishes that decryption is not permission to bid, without a claim about every earlier-issued token. | 04-consent.md |
| X906 | The documentation gives the buyer a key, a notice and an action to implement [claim IDN-A-F07] [claim IDN-A-F08]. | A raw UID2 and a decryption key are different objects. The DSP guide identifies the raw UID2 in the notice. | IDN-A-S08, IDN-A-S09 [149, 186] | supported with qualification | The DSP guide ties the raw UID2 to a withdrawal notice and a rule for bidding [claim IDN-UID-F13]. | A raw UID2 and a decryption key are different objects. The DSP guide identifies the raw UID2 in the notice. | 04-consent.md |
| X907 | If a reader opts out on one site, they should ask which ID the site sent to the wider system [claim IDN-UID-F12]. | The previous question implied the participant might send an ID onward as part of the local request; the written rule says it is not relayed. | IDN-A-S08, IDN-UID-S18 [149, 150] | supported with qualification | A site-level opt-out stays with that participant and is not sent to the wider UID2 system [claim IDN-UID-F12]. | The previous question implied the participant might send an ID onward as part of the local request; the written rule says it is not relayed. | 04-consent.md |
| X908 | If consent is absent, or the user opts out, the id5-sync.com field stays in the bid and its ID takes the string value zero [claim IDN-ID5-F08]. | Restore the issuer’s normative language and identify the actual EID entry and string. The guide does not observe real bids. | IDN-ID5-S06 [175] | supported with qualification | When consent is absent or the user opts out, the guide says the id5-sync.com EID should stay in the bid with its ID set to the string “0” [claim IDN-ID5-F08]. | Restore the issuer’s normative language and identify the actual EID entry and string. The guide does not observe real bids. | 04-consent.md |
| X909 | LiveRamp also sends lists of opt-outs and deletions to firms that get audience files [claim IDN-LR-F07]. | The source lists APIs and segment updates for destinations. Not every workflow pushes a list to every receiver. | IDN-LR-S05 [139] | supported with qualification | LiveRamp makes opt-out and deletion requests available to destinations that might receive the consumer’s data [claim IDN-LR-F07]. | The source lists APIs and segment updates for destinations. Not every workflow pushes a list to every receiver. | 04-consent.md |
| X910 | The firm that gets the file must process it [claim IDN-LR-F07]. | The source explicitly leaves processing to the destination’s discretion. The original sentence incorrectly imposed a duty to process. | IDN-LR-S05 [139] | supported with qualification | LiveRamp expects destinations to respect the requests and leaves processing to their discretion [claim IDN-LR-F07]. | The source explicitly leaves processing to the destination’s discretion. The original sentence incorrectly imposed a duty to process. | 04-consent.md |
| X911 | A Michigan case puts names to the firms on one bid path [claim IDN-A5-F02]. | Distinguish targeting at the SSP and delivery through two DSPs from a traced ID through each bidder. | IDN-A5-S02 [264] | supported with qualification | Lotame’s case names the firms involved in a Michigan media buy [claim IDN-A5-F02]. | Distinguish targeting at the SSP and delivery through two DSPs from a traced ID through each bidder. | 04-consent.md |
| X912 | In California, TripleLift says it sends consent status with the bid and leaves the targeting choice to the DSP [claim IDN-HX-F05]. | Add the historical date and attribution rather than representing a 2020 FAQ as a current promise. | IDN-HX-S05 [262] | supported with qualification | In a 2020 California FAQ, TripleLift said it sent consent status with the bid and left targeting to the DSP [claim IDN-HX-F05]. | Add the historical date and attribution rather than representing a 2020 FAQ as a current promise. | 04-consent.md |
| X913 | The paper set a strict bar. | A statement about the registered test should carry the hypothesis marker. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-A-S08, IDN-UID-S18, IDN-A-S09, IDN-ID5-S06, IDN-ID5-S09, IDN-LR-S03, IDN-HX-S04 [14, 13, 149, 150, 186, 175, 128, 182, 219] | supported with qualification | The paper set a strict majority test [claim IDN-HYP-H3]. | A statement about the registered test should carry the hypothesis marker. | 04-consent.md |
| X914 | Then ask what the firm that gets the ID must do when the reader says no. One more privacy policy would settle neither point. | A future detailed product policy might settle either question. Change the categorical assertion to an evidence request. | 04-consent.md | supported with qualification | Then ask what the firm that gets the ID must do when the reader says no. Ask the firm to show where its policy covers that exact route. | A future detailed product policy might settle either question. Change the categorical assertion to an evidence request. | 04-consent.md |
| X915 | A publisher can use the AdCP Go client to send a typed ID token to a buyer’s service [claim IDN-J-F03]. | The source client posts to routerURL/tmp/identity. That post alone does not establish receipt by an independent buyer’s service. | IDN-J-S03, IDN-JV-S01, IDN-JV-S02 [92, 93, 94] | supported with qualification | A publisher can use the AdCP Go client to send a typed ID token to the Trusted Match router [claim IDN-J-F03]. | The source client posts to routerURL/tmp/identity. That post alone does not establish receipt by an independent buyer’s service. | 10-agents.md |
| X916 | The code at the other end can check who is in an audience and return IDs for the packages that qualify [claim IDN-J-F04] [claim IDN-J-F05]. | Specify the tested input and the reference implementation. The inspected membership call uses a token and audience ID, not a natural-person truth test. | IDN-J-S05, IDN-JV-S01, IDN-J-S04 [95, 93, 108] | supported with qualification | The reference code can check whether a token belongs to an audience and return IDs for packages that qualify [claim IDN-J-F04] [claim IDN-J-F05]. | Specify the tested input and the reference implementation. The inspected membership call uses a token and audience ID, not a natural-person truth test. | 10-agents.md |
| X917 | Evgeny Popov co-leads the Signals & Measurement working group at AgenticAdvertising.org, which develops AdCP [claim IDN-METHOD-F01]. | Keep the declaration exactly at the scope of the supplied author record; development of the protocol is established independently of his role. | IDN-METHOD-S01, IDN-METHOD-S02 [232, 233] | supported with qualification | Evgeny Popov co-leads the Signals & Measurement working group at AgenticAdvertising.org [claim IDN-METHOD-F01]. | Keep the declaration exactly at the scope of the supplied author record; development of the protocol is established independently of his role. | 10-agents.md |
| X918 | The same test applies to its work and to the IAB records here. | Attach the predeclared test to this factual statement about equal treatment. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-J-S03, IDN-J-S05, IDN-J-S04, IDN-J-S16 [14, 13, 92, 95, 108, 101] | supported with qualification | The study applies the same rule to AdCP and the IAB protocols it screened [claim IDN-HYP-H7]. | Attach the predeclared test to this factual statement about equal treatment. | 10-agents.md |
| X919 | The study calls an input an identity signal when it helps a system recognize a unit or check a link that matters to the buy. | Make the analytic definition explicit and traceable to the registered hypothesis. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-J-S03, IDN-J-S05, IDN-J-S04, IDN-J-S16 [14, 13, 92, 95, 108, 101] | supported with qualification | For this test, an identity signal is an input used to recognize a unit or check a link that matters to an ad buy [claim IDN-HYP-H7]. | Make the analytic definition explicit and traceable to the registered hypothesis. | 10-agents.md |
| X920 | It lets the buyer send segments and a list of people or records to leave out [claim IDN-J-F08]. | The draft assigns fields by role; it does not constrain the suppression list to a verified natural-person identity or require one specific record type. | IDN-J-S08 [97] | supported with qualification | The draft lets the buyer supply segments and a suppression list [claim IDN-J-F08]. | The draft assigns fields by role; it does not constrain the suppression list to a verified natural-person identity or require one specific record type. | 10-agents.md |
| X921 | PubMatic says Clubtails ran a CTV campaign on its Activate platform with AdCP in December 2025 [claim IDN-J-F14]. | Use the issuer’s present-running account at the announcement date. The record does not document completion or an identity-bearing message. | IDN-J-S13 [106] | supported with qualification | In December 2025, PubMatic said a Clubtails CTV campaign was running on Activate with AdCP [claim IDN-J-F14]. | Use the issuer’s present-running account at the announcement date. The record does not document completion or an identity-bearing message. | 10-agents.md |
| X922 | Did that bring back cookie sync as the first choice for US web IDs? | H4 tests an increase on the four-point scale for most products. A rise can occur without sync becoming the first choice. | IDN-METHOD-S04, IDN-METHOD-S06 [13, 14] | supported with qualification | Did most US web ID products give cookie sync a bigger role in their guides after that choice [claim IDN-HYP-H4]? | H4 tests an increase on the four-point scale for most products. A rise can occur without sync becoming the first choice. | 05-chrome.md |
| X923 | Chrome 144 said removal was planned [claim IDN-E-F03]. | Chrome 144 names Private Aggregation. The immediately preceding paragraph also discusses Topics. | IDN-E-S03, IDN-E-S04 [265, 266] | supported with qualification | Chrome 144 said it planned to remove the Private Aggregation API [claim IDN-E-F03]. | Chrome 144 names Private Aggregation. The immediately preceding paragraph also discusses Topics. | 05-chrome.md |
| X924 | Chrome 152 said the feature had been removed [claim IDN-E-F03]. | Chrome 152 describes the completed Private Aggregation removal for that release. | IDN-E-S03, IDN-E-S04 [265, 266] | supported with qualification | Chrome 152 said it had removed that API [claim IDN-E-F03]. | Chrome 152 describes the completed Private Aggregation removal for that release. | 05-chrome.md |
| X925 | A July Chromium change says the Topics interfaces that sites could call had already become stubs [claim IDN-E-F04]. | The pinned d37e2f6 source change is from 2026. The preceding platform timeline includes 2025. | IDN-E-S05 [267] | supported with qualification | A July 2026 Chromium change says the Topics interfaces that sites could call had already become stubs [claim IDN-E-F04]. | The pinned d37e2f6 source change is from 2026. The preceding platform timeline includes 2025. | 05-chrome.md |
| X926 | In September the team said a trial to take out the stubs had reached 1% on milestone 153 and later in the Stable channel [claim IDN-E-F07]. | The blink-dev progress note of September 24, 2026 identifies milestone 153 and later on Stable. | IDN-E-S19, IDN-E-S05 [270, 267] | supported with qualification | In September 2026 the team said a trial to take out the stubs had reached 1% on milestone 153 and later in the Stable channel [claim IDN-E-F07]. | The blink-dev progress note of September 24, 2026 identifies milestone 153 and later on Stable. | 05-chrome.md |
| X927 | That was a share in the trial, which gave no count of devices or US share [claim IDN-E-F07]. | The Google engineering update reports an experiment rollout percentage on eligible builds, with no device count or US fraction. | IDN-E-S19, IDN-E-S05 [270, 267] | supported with qualification | The 1% was the rollout on those Stable builds [claim IDN-E-F07]. The team gave no device count or US share [claim IDN-E-F07]. | The Google engineering update reports an experiment rollout percentage on eligible builds, with no device count or US fraction. | 05-chrome.md |
| X928 | On July 14, Chromium changed its Protected Audience renderer calls, including `runAdAuction` [claim IDN-E-F05]. | Commit 4bcc11c identifies the year as well as day for the renderer change. | IDN-E-S06, IDN-EV-S01 [268, 5] | supported with qualification | On July 14, 2026, Chromium changed its Protected Audience renderer calls, including `runAdAuction` [claim IDN-E-F05]. | Commit 4bcc11c identifies the year as well as day for the renderer change. | 05-chrome.md |
| X929 | Chrome 153 still listed deprecation and removal as planned [claim IDN-U-F36]. | The Chrome 153 release note describes intent at that stage. Commit 2ae93f9 on October 30 removed the remaining stubbed WebIDL bindings. The source change gives no rollout share. | IDN-U-S48, IDN-EV-S01 [283, 5] | supported with qualification | Chrome 153 still listed deprecation and removal as planned [claim IDN-U-F36]. On October 30, 2026, Chromium removed the last web bindings for Protected Audience from its source tree [source IDN-EV-S01]. | The Chrome 153 release note describes intent at that stage. Commit 2ae93f9 on October 30 removed the remaining stubbed WebIDL bindings. The source change gives no rollout share. | 05-chrome.md |
| X930 | A July 14 change to Attribution Reporting kept the setter used with `XMLHttpRequest` but made it do nothing [claim IDN-E-F06]. | Chromium commit d7a74fd dates the removal of renderer implementation and preservation of a no-op setter. | IDN-E-S07 [269] | supported with qualification | A July 14, 2026 change to Attribution Reporting kept the setter used with `XMLHttpRequest` but made it do nothing [claim IDN-E-F06]. | Chromium commit d7a74fd dates the removal of renderer implementation and preservation of a no-op setter. | 05-chrome.md |
| X931 | It planned a staged move for the rest of the API, with removal at 100% of Stable after October 24 [claim IDN-E-F08]. | The October 8 update gives a prospective rollout schedule. Completion and the exact last date are unconfirmed. | IDN-E-S20, IDN-E-S07 [271, 269] | supported with qualification | It planned to roll out the rest of the change in steps, with full removal on Stable set for after October 24 [claim IDN-E-F08]. | The October 8 update gives a prospective rollout schedule. Completion and the exact last date are unconfirmed. | 05-chrome.md |
| X932 | Yahoo’s April 2025 Prebid guide says ConnectID can work without third-party cookies [claim IDN-YAH-F18]. | The pinned April 4 module guide is a Prebid project document. Yahoo provides the separate choices of integration. | IDN-E-S28, IDN-YAH-S03 [202, 199] | supported with qualification | Prebid’s April 2025 guide says Yahoo ConnectID can work without third-party cookies [claim IDN-YAH-F18]. | The pinned April 4 module guide is a Prebid project document. Yahoo provides the separate choices of integration. | 05-chrome.md |
| X933 | Lotame’s Panorama guide describes an ID for a device that a publisher can use to enrich data or activate an audience [claim IDN-E-F25]. | The dated page is the project-maintained Prebid module guide. | IDN-E-S31 [279] | supported with qualification | Prebid’s guide to Lotame Panorama describes an ID for a device that a publisher can use to enrich data or activate an audience [claim IDN-E-F25]. | The dated page is the project-maintained Prebid module guide. | 05-chrome.md |
| X934 | Lotame also has a sync pixel for Spherical, a separate route [claim IDN-E-F25]. | The attached January Panorama revision describes the Prebid module. It does not document a Spherical pixel. A conditional example preserves the workflow comparison. | IDN-E-S31 [279] | supported with qualification | Suppose a seller also gives a buyer a pixel for a different product. | The attached January Panorama revision describes the Prebid module. It does not document a Spherical pixel. A conditional example preserves the workflow comparison. | 05-chrome.md |
| X935 | It cannot fill the gap for Panorama [claim IDN-E-F25]. | The January Panorama manual independently documents its module but leaves ranked same-route before/after guidance unresolved. | IDN-E-S31 [279] | supported with qualification | That pixel would leave the gap in Panorama’s own guide open [claim IDN-E-F25]. | The January Panorama manual independently documents its module but leaves ranked same-route before/after guidance unresolved. | 05-chrome.md |
| X936 | A TV report can start with a home address and end with a visit to a site. | Spectrum identifies household exposures and visits. Its FAQ does not establish postal address as the input in that customer attribution route. | IDN-CTV-S08 [38] | supported with qualification | What can a TV report tell a brand about site visits from homes that saw its ads? | Spectrum identifies household exposures and visits. Its FAQ does not establish postal address as the input in that customer attribution route. | 07-ctv.md |
| X937 | The study found no cap on the age of a live link [claim IDN-HYP-H5]. | Blockgraph documents history lookback and manual data-layer refresh. Its public H5 screen found no elapsed-time refresh or expiry rule for the same live edge. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-CTV-S11, IDN-CTV-S16, IDN-CTV-S12, IDN-G-S15 [14, 13, 109, 114, 115, 204] | supported with qualification | For Blockgraph, the study found no time limit on a live link in the guides it checked [claim IDN-HYP-H5]. | Blockgraph documents history lookback and manual data-layer refresh. Its public H5 screen found no elapsed-time refresh or expiry rule for the same live edge. | 07-ctv.md |
| X938 | FreeWheel describes a built-in network with signals from Comcast and other publishers [claim IDN-G-F08]. | The product description distinguishes Comcast deterministic signals from proprietary partner IDs supplied by publishers. | IDN-G-S08 [305] | supported with qualification | FreeWheel describes a built-in network with identity signals from Comcast and IDs from its publisher partners [claim IDN-G-F08]. | The product description distinguishes Comcast deterministic signals from proprietary partner IDs supplied by publishers. | 07-ctv.md |
| X939 | In 2020, Skydeo struck a deal to map mobile data to the Tru Optik Household Graph in the US [claim IDN-G-F15]. | The January 23 announcement says the data will be mapped. It does not certify mapping had finished. | IDN-G-S16 [310] | supported with qualification | In January 2020, Skydeo announced a deal to map mobile data to the Tru Optik Household Graph in the US [claim IDN-G-F15]. | The January 23 announcement says the data will be mapped. It does not certify mapping had finished. | 07-ctv.md |
| X940 | The patent leaves open use of that number in the live product and says nothing about when a link expires after its last sighting [claim IDN-G-F17]. | Patent Figure 5 step 508 applies the threshold to IP assignment age before proceeding. This example does not specify the lifetime of a deployed edge. | IDN-G-S18 [311] | supported with qualification | The 45-day example gives no expiry date for a link after its last sighting [claim IDN-G-F17]. Its use in Viant’s live product remains unconfirmed [claim IDN-G-F17]. | Patent Figure 5 step 508 applies the threshold to IP assignment age before proceeding. This example does not specify the lifetime of a deployed edge. | 07-ctv.md |
| X941 | Blockgraph’s version 2.29 guide says the new file takes the place of the old identity layer at that node [claim IDN-CTV-F21]. | The manual separately supports add/remove operations. Full replacement is specifically the Refresh of Identity Data operation. | IDN-CTV-S12 [115] | supported with qualification | Blockgraph’s version 2.29 guide says a full manual refresh replaces the identity layer at that node with a new file [claim IDN-CTV-F21]. | The manual separately supports add/remove operations. Full replacement is specifically the Refresh of Identity Data operation. | 07-ctv.md |
| X942 | LiveRamp’s guide to IP-based CTV events calls for new data each day or week once the programmer is set up [claim IDN-G-F11]. | LiveRamp’s Guidelines for Sending IP Address-Based Data recommends daily or weekly delivery after the initial file. | IDN-G-S11 [307] | supported with qualification | LiveRamp’s guide to IP-based CTV events recommends new data each day or week once the programmer is set up [claim IDN-G-F11]. | LiveRamp’s Guidelines for Sending IP Address-Based Data recommends daily or weekly delivery after the initial file. | 07-ctv.md |
| X943 | It gives a six-month lookback for past events [claim IDN-G-F11]. | The programmer guide gives a maximum historical ingestion/lookback capability, not a fixed window or edge age. | IDN-G-S11 [307] | supported with qualification | It can look back up to six months for past events [claim IDN-G-F11]. | The programmer guide gives a maximum historical ingestion/lookback capability, not a fixed window or edge age. | 07-ctv.md |
| X944 | It claims a 95% addressable match rate for its own multichannel TV service [claim IDN-CTV-F19]. | Spectrum expressly attributes its 95% claim to its MVPD role. It is not a Blockgraph accuracy statistic. | IDN-CTV-S08 [38] | supported with qualification | It claims a 95% addressable match rate for its own pay-TV distribution service [claim IDN-CTV-F19]. | Spectrum expressly attributes its 95% claim to its MVPD role. It is not a Blockgraph accuracy statistic. | 07-ctv.md |
| X945 | Kantar says its match rates with TV partners rose to more than ten times their old level once it linked to Blockgraph [claim IDN-CTV-F11]. | The joint release reports the over-tenfold case and gives neither absolute rate. This is a participant claim. | IDN-CTV-S07, IDN-CTV-S17 [112, 113] | supported with qualification | Kantar and Blockgraph say Kantar’s match rates with TV partners rose to more than ten times their old level after the firms linked their systems [claim IDN-CTV-F11]. | The joint release reports the over-tenfold case and gives neither absolute rate. This is a participant claim. | 07-ctv.md |
| X946 | The study found no clock for a live link [claim IDN-HYP-H5] [claim IDN-CTV-F09]. | The Blockgraph historical IP query and manual refresh are available, without an elapsed-time rule for the same active edge. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-CTV-S11, IDN-CTV-S16, IDN-CTV-S12, IDN-G-S15 [14, 13, 109, 114, 115, 204] | supported with qualification | Its guides gave no time rule for when that live link should be renewed or dropped [claim IDN-HYP-H5] [claim IDN-CTV-F09]. | The Blockgraph historical IP query and manual refresh are available, without an elapsed-time rule for the same active edge. | 07-ctv.md |
| X947 | In that narrow frame, let all the firms with open scope count as unknowns and the result could fall on either side of half [claim IDN-HYP-H5]. | The confirmed frame was 0/1. Adding eight uncertain-scope candidates gives the separate sensitivity interval 0/9–8/9. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-CTV-S11, IDN-CTV-S16, IDN-CTV-S12, IDN-G-S15 [14, 13, 109, 114, 115, 204] | supported with qualification | When the study also counted the products of uncertain scope as unknowns, the share could fall on either side of half [claim IDN-HYP-H5]. | The confirmed frame was 0/1. Adding eight uncertain-scope candidates gives the separate sensitivity interval 0/9–8/9. | 07-ctv.md |
| X948 | A media team can now ask one group for both. | The November 2 buyer release describes both identity inputs in its new offer. It does not establish that either can be acquired separately through one purchase. | IDN-D-S08 [216] | supported with qualification | Publicis markets both in that offer [claim IDN-D-F08]. | The November 2 buyer release describes both identity inputs in its new offer. It does not establish that either can be acquired separately through one purchase. | 08-market.md |
| X949 | The name on the merger filing was LiveRamp Holdings, Inc. [claim IDN-LR-F22]. | The country-specific service terms identify the operating company. The closing 8-K identifies the acquired parent. The revised sentence attaches the filing to the latter claim and handles its corporate suffix. | IDN-LR-S13, IDN-D-S02, IDN-D-S03 [345, 141, 318] | supported with qualification | The merger filing names the parent, LiveRamp Holdings, Inc., as the acquired firm [claim IDN-LR-F22] [claim IDN-D-F02]. | The country-specific service terms identify the operating company. The closing 8-K identifies the acquired parent. The revised sentence attaches the filing to the latter claim and handles its corporate suffix. | 08-market.md |
| X950 | It broke out subscription and marketplace sales within that segment [claim IDN-D-F05]. | The SEC 10-K, Note 3 revenue disaggregation, calls the second category Marketplace and Other. It includes transactions outside the marketplace. | IDN-D-S04 [7] | supported with qualification | It split sales into Subscription and Marketplace and Other [claim IDN-D-F05] [source IDN-D-S04]. | The SEC 10-K, Note 3 revenue disaggregation, calls the second category Marketplace and Other. It includes transactions outside the marketplace. | 08-market.md |
| X951 | The rights also vary by route. | The UID2 licence matrix separates API access from the CRM onboarding route and its non-return of identifiers. | IDN-UID-S11, IDN-U-S03 [188, 203] | supported with qualification | The rights also vary by route [claim IDN-UID-F09]. | The UID2 licence matrix separates API access from the CRM onboarding route and its non-return of identifiers. | 08-market.md |
| X952 | There is a fee for the platform inside that bill. | This is the premise of the buyer example. No actual buyer invoice or universal billing rule is implied. | 08-market.md | supported with qualification | Suppose the platform fee is part of that bill. | This is the premise of the buyer example. No actual buyer invoice or universal billing rule is implied. | 08-market.md |
| X953 | One shows who has control. | Companies House records the French SAS as a person with significant control of the UK company. | IDN-ID5-S02, IDN-ID5-S19 [339, 340] | supported with qualification | The UK register shows who has control [claim IDN-ID5-F02]. | Companies House records the French SAS as a person with significant control of the UK company. | 08-market.md |
| X954 | One names firms that put up funds. | ID5’s funding release names TransUnion and S4S Ventures. | IDN-ID5-S18 [341] | supported with qualification | The funding release names two investors [claim IDN-ID5-F03]. | ID5’s funding release names TransUnion and S4S Ventures. | 08-market.md |
| X955 | The US Digital Graph sheet credits Experian Marketing Services, Inc. [claim IDN-EXP-F15]. | The footer is a copyright credit. It does not identify the signatory of a buyer agreement. | IDN-EXP-S02, IDN-EXP-S05, IDN-EXP-S08 [119, 336, 337] | supported with qualification | The US Digital Graph sheet gives its copyright to Experian Marketing Services, Inc., in its footer [claim IDN-EXP-F15]. | The footer is a copyright credit. It does not identify the signatory of a buyer agreement. | 08-market.md |
| X956 | MiQ licensed hashed emails, UID2s and partner cookies for use in its Identity Spine [claim IDN-EXP-F06]. | The attribution matches a sales case published by Experian. Page 4 identifies the licensed inputs. | IDN-C-S06, IDN-VEXP-S01 [21, 118] | supported with qualification | Experian says MiQ licensed hashed emails, UID2s and partner cookies for use in its Identity Spine [claim IDN-EXP-F06]. | The attribution matches a sales case published by Experian. Page 4 identifies the licensed inputs. | 08-market.md |
| X957 | Its traders and analysts use the data [claim IDN-EXP-F06]. | The page 4 user and uses are reported by the commercial supplier. | IDN-C-S06, IDN-VEXP-S01 [21, 118] | supported with qualification | The case says MiQ traders and analysts use the data [claim IDN-EXP-F06]. | The page 4 user and uses are reported by the commercial supplier. | 08-market.md |
| X958 | That later account gives a buyer a claim about how the graph was built. | The January 2024 supplier account describes clustering and scoring. This is an issuer description. | IDN-C-S12 [316] | supported with qualification | That later account gives a buyer a claim about how the graph was built [claim IDN-C-F11]. | The January 2024 supplier account describes clustering and scoring. This is an issuer description. | 08-market.md |
| X959 | Its US ConnectID agreement names Yahoo Ad Tech LLC as the signer, based in New York [claim IDN-YAH-F01]. | The US form names the counterparty and its principal place of business. It is a template, not evidence of a particular signed order. | IDN-YAH-S01 [160] | supported with qualification | Its US ConnectID terms name Yahoo Ad Tech LLC as the contracting party, with its main office in New York [claim IDN-YAH-F01]. | The US form names the counterparty and its principal place of business. It is a template, not evidence of a particular signed order. | 08-market.md |
| X960 | The records give names at the dates they cite. | The two ownership records are dated differently. This is advice on how to use them. | 08-market.md | supported with qualification | Keep each set of owner names with the date of its source. | The two ownership records are dated differently. This is advice on how to use them. | 08-market.md |
| X961 | The buyer paid it once. | The prior sentence supposes a package quote. No real transaction or paid invoice is documented. | 08-market.md | supported with qualification | In that example, the buyer would pay the package price once. | The prior sentence supposes a package quote. No real transaction or paid invoice is documented. | 08-market.md |
| X962 | In the US, Mobilewalla said it changed its location-data filters after an FTC order [claim IDN-H-F09] [claim IDN-H-F10]. | The filing reports implementation of a third-party filter, not a measured change to multiple filters. | IDN-H-S09, IDN-H-S10, IDN-HV-S02 [71, 72, 73] | supported with qualification | Mobilewalla told investors it had applied a third-party location filter in its US business after an FTC order [claim IDN-H-F09] [claim IDN-H-F10]. | The filing reports implementation of a third-party filter, not a measured change to multiple filters. | 09-regulation.md |
| X963 | Users could do so with a device or system control of the kind set out in the order [claim IDN-H-F08]. | Provision VIII permits the specified mechanism. It does not establish that the firm deployed it or that consumers actually used it. | IDN-H-S08 [70] | supported with qualification | The order lets the firm use a qualifying device or operating-system control for that purpose [claim IDN-H-F08]. | Provision VIII permits the specified mechanism. It does not establish that the firm deployed it or that consumers actually used it. | 09-regulation.md |
| X964 | In January 2025 the FTC placed a limit on Mobilewalla’s use of data from an online ad auction [claim IDN-H-F09]. | Provision II restricts collection or retention for other purposes. It is not an unrestricted general rule on use of all auction data. | IDN-H-S09 [71] | supported with qualification | In January 2025 the FTC limited Mobilewalla’s collection or retention of covered data from online ad auctions [claim IDN-H-F09]. | Provision II restricts collection or retention for other purposes. It is not an unrestricted general rule on use of all auction data. | 09-regulation.md |
| X965 | It set a clock of 180 days to put a sensitive-location program in place [claim IDN-H-F19]. | The order defines issuance as the start event, distinct from the effective-date rule for Gravy. | IDN-H-S19 [84] | supported with qualification | It gave them 180 days from issuance to put a sensitive-location program in place [claim IDN-H-F19]. | The order defines issuance as the start event, distinct from the effective-date rule for Gravy. | 09-regulation.md |
| X966 | An April order bars InMarket from the sale or licence of the location data it defines in return for something of value [claim IDN-H-F20]. | Provision I defines the data and the consideration. The FTC April 2024 order is distinct from the next provision on sensitive-location products. | IDN-H-S20 [85] | supported with qualification | An April 2024 order bars InMarket from selling or licensing defined location data for something of value [claim IDN-H-F20]. | Provision I defines the data and the consideration. The FTC April 2024 order is distinct from the next provision on sensitive-location products. | 09-regulation.md |
| X967 | A different rule applies when the person logs out [claim IDN-H-F11]. | California judgment paragraph 26(b) covers no account and not logged in. It does not require a previous logout. | IDN-H-S11, IDN-H-S12 [74, 75] | supported with qualification | A different rule applies when the person is not logged in or has no account [claim IDN-H-F11]. | California judgment paragraph 26(b) covers no account and not logged in. It does not require a previous logout. | 09-regulation.md |
| X968 | The author works at Samba TV [claim IDN-H-F13]. | The author’s biography, disclosure and register establish the employment. The Texas pleading does not. | IDN-METHOD-S01, IDN-METHOD-S02 [232, 233] | supported with qualification | The author works at Samba TV [claim IDN-METHOD-F01]. | The author’s biography, disclosure and register establish the employment. The Texas pleading does not. | 09-regulation.md |
| X969 | The study keeps Samba out of its scores and includes it in this legal screen [claim IDN-H-F13] [claim IDN-HYP-H6]. | The author declaration and registered screen support these two study decisions. | IDN-METHOD-S01, IDN-METHOD-S02, IDN-METHOD-S06, IDN-METHOD-S04, IDN-H-S03, IDN-H-S06, IDN-H-S08, IDN-H-S15 [232, 233, 14, 13, 65, 68, 70, 79] | supported with qualification | The study keeps Samba out of its scores and includes it in this legal screen [claim IDN-METHOD-F01] [claim IDN-HYP-H6]. | The author declaration and registered screen support these two study decisions. | 09-regulation.md |
| X970 | The public copy has a blank space for the judge’s name and date [claim IDN-H-F14]. | The posted original, printed page 9, has blank judicial signature and date lines. A missing printed name is a different claim. | IDN-H-S14, IDN-HV-S01 [78, 77] | supported with qualification | The posted copy leaves the judge’s signature and date blank [claim IDN-H-F14]. | The posted original, printed page 9, has blank judicial signature and date lines. A missing printed name is a different claim. | 09-regulation.md |
| X971 | In May 2025 the Belgian authority gave its account of the Market Court’s ruling [claim IDN-H-F17]. | The reopened news item identifies the speaker, which matters for the characterization of the court’s decision. | IDN-H-S17 [81] | supported with qualification | In May 2025 the Belgian data protection authority gave its account of the Market Court’s ruling [claim IDN-H-F17]. | The reopened news item identifies the speaker, which matters for the characterization of the court’s decision. | 09-regulation.md |
| X972 | It said the court rejected joint control for work done wholly within OpenRTB [claim IDN-H-F17]. | The May DPA account attributes this limited finding to the court. It is not an independent finding about a particular bidder. | IDN-H-S17 [81] | supported with qualification | The authority said the court rejected joint control for processing that takes place wholly within OpenRTB [claim IDN-H-F17]. | The May DPA account attributes this limited finding to the court. It is not an independent finding about a particular bidder. | 09-regulation.md |
| X973 | It dealt with the sixth plaintiff in a separate order [claim IDN-YAH-F20]. | Docket 104 asks about discovery on whether Baker agreed to arbitrate. It does not refer him to arbitration with the other five. | IDN-YAH-S17, IDN-VYAH-S06 [90, 8] | supported with qualification | A separate August 26 order asked the parties whether limited discovery was needed for the sixth plaintiff [claim IDN-YAH-F20] [source IDN-VYAH-S06]. | Docket 104 asks about discovery on whether Baker agreed to arbitrate. It does not refer him to arbitration with the other five. | 09-regulation.md |
| X974 | In an interim order on October 19, 2026, the FTC said standalone services had to stay open [claim IDN-LR-F20]. | The Order to Maintain Assets requires continued, nondiscriminatory access to LiveRamp standalone offerings. It is a separate operative order. | IDN-LR-S18, IDN-LR-S19, IDN-LR-S17, IDN-VLR-S02 [88, 89, 87, 9] | supported with qualification | The FTC’s October 19, 2026 interim order required continued access to LiveRamp’s standalone services [claim IDN-LR-F20]. | The Order to Maintain Assets requires continued, nondiscriminatory access to LiveRamp standalone offerings. It is a separate operative order. | 09-regulation.md |
| X975 | A proposed five-year order was still open for public comment at the cutoff [claim IDN-LR-F20]. | The October 23 Federal Register notice gives the November 23 comment deadline. The recorded November 11 cutoff precedes it. | IDN-LR-S18, IDN-LR-S19, IDN-LR-S17, IDN-VLR-S02 [88, 89, 87, 9] | supported with qualification | The proposed five-year order was open for public comment through November 23, 2026 [claim IDN-LR-F20] [source IDN-VLR-S02]. | The October 23 Federal Register notice gives the November 23 comment deadline. The recorded November 11 cutoff precedes it. | 09-regulation.md |
| X976 | An engineer could count each as one new connection. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | A team might log these as two new links. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X977 | Get the draft agreement before assigning the purchase to a master contract already in use. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | Get the draft deal. Does the contract the buyer has on file cover this job too? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X978 | On a long-running deal, check the terms for any transfer of rights. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | For an old deal, check who could pass the rights to a new firm. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X979 | Ask for a fresh order or assignment record before attributing that fee to one of the current groups. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | Ask for the new order or the terms that moved it. Which firm now earns the fee? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X980 | Then ask for a test in which the second partner gets an output it can use. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | Run a test. Can the other firm use what it gets? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X981 | A researcher could use both to map the links between firms without counting a new sale each time a source is named. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | Use the two lists to map the firms. When a source shows up on both, has a second sale taken place? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X982 | A price alone cannot tell the buyer whether it may send the file on. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md | supported with qualification | Read the terms. Does the price include the right to send the file on? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 08-market.md |
| X983 | Could we trace a change that clearly for the US ad identity products in this paper? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Could we trace such a change for the firms that sell US ad IDs? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X984 | Did an enforcement action cause a change in a named identity service? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Did a case cause a change in a named ID service? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X985 | A team may have good evidence of a legal duty and little evidence on how two devices were matched. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | A team might know what the law requires. It may still lack proof of how two devices were linked. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X986 | The tester has learned how the site chose that ad. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | The test would show how the site chose that ad. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X987 | Where the firm relies on an exception, what reason does it keep on file? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | If the firm says a row is exempt, what grounds does it keep on file? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X988 | Use an authorized test and a known record. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Get leave to run the test. Use a row the team knows. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X989 | Could a routine refresh or a different source have changed the link that day? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Could a fresh file or a new source have changed the link that day? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X990 | A buyer with access to the first feed may need help from the party that holds the graph to answer those questions. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Who holds the graph? The buyer may see the first feed and still need that firm to check the link. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X991 | Each choice would leave a different trace in what the client receives. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Which of those paths fits the file the buyer got? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X992 | To find out what reached the client, seek a record from someone who could see the output. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Who could see the file that reached the buyer? Get their log. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X993 | A graph buyer who knows a supplier uses location as one input should ask what kind of data the buyer would get. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | Say a firm uses location data to build a graph. What would the buyer get? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X994 | If a seller cites a deadline, ask what sets that clock in motion. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | If a rule has a due date, what starts the clock? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X995 | To work out where the viewing signal goes, it must trace what happens after the viewer uses the screen. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | What left the TV once the user made that choice? Which firm got it? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X996 | For a claim about one of the APIs, follow that API into its own records. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | For a claim about an API, check the records for that code. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X997 | A regulator who confirmed a change in how ads look in Gmail has answered a question about those ads [claim IDN-H-F16]. | Plain-language precision in the buyer test; the source scope and factual result are preserved. | IDN-H-S16, IDN-H-S15 [80, 79] | supported with qualification | The CNIL confirmed the change to the Gmail ads it had named [claim IDN-H-F16]. | Plain-language precision in the buyer test; the source scope and factual result are preserved. | 09-regulation.md |
| X998 | Neither the European court ruling nor the Belgian case verified how a named US ad ID handles each of those states at a bidder [claim IDN-H-F17] [claim IDN-H-F22]. | Plain-language precision in the buyer test; the source scope and factual result are preserved. | IDN-H-S17, IDN-H-S22 [81, 82] | supported with qualification | Those cases did not test what a US bidder did with a named ad ID under each choice [claim IDN-H-F17] [claim IDN-H-F22]. | Plain-language precision in the buyer test; the source scope and factual result are preserved. | 09-regulation.md |
| X999 | It also helps a buyer decide what it can check on its own, and what it must ask a partner to check. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | What can the buyer check on its own? What does a second firm need to check? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X1000 | Nor does it tell us that enforcement never changed the way a seller links records. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | It also leaves open the effect of past cases on how firms link rows. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X1001 | A new record that confirms the change for the same product could settle the claim. | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md | supported with qualification | A new check could settle the claim. Did the same product change? | Plain-language buyer guidance or a hypothetical test; no new claim about the seller or case. | 09-regulation.md |
| X1002 | The matrix has [stat vendors_scored] scored rows and [stat cells_scored] cells with scores [claim IDN-IQ-F01]. | Summing k in the 24 admitted rows gives 142. The stock cells_scored statistic also includes six proposals marked unverified and yields 148. | IDN-METHOD-S05, IDN-METHOD-S07, IDN-METHOD-S03 [10, 218, 11] | supported with qualification | The matrix has [stat vendors_scored] scored rows and [stat counted_checked_cells] cells that count toward the IQ [claim IDN-IQ-F01]. | Summing k in the 24 admitted rows gives 142. The stock cells_scored statistic also includes six proposals marked unverified and yields 148. | 11-iq.md |
| X1003 | The guide names US sites where the operator accepts calls [claim IDN-UID-F10]. | The guide lists servers in Ohio and Oregon. Calling them sites risks implying publisher properties or a user reach measure. | IDN-U-S04, IDN-I-S08 [153, 154] | supported with qualification | The guide names US production endpoints where the operator accepts calls [claim IDN-UID-F10]. | The guide lists servers in Ohio and Oregon. Calling them sites risks implying publisher properties or a user reach measure. | 11-iq.md |
| X1004 | It says the device link drops out thirty days since the device was last seen [claim IDN-G-F13]. | The seller’s clear event boundary is 30 days after the device was last seen. | IDN-G-S13, IDN-G-S14, IDN-GV-S01, IDN-GV-S02, IDN-GV-S03 [227, 228, 229, 230, 231] | supported with qualification | It says the device link drops out thirty days after the device was last seen [claim IDN-G-F13]. | The seller’s clear event boundary is 30 days after the device was last seen. | 13-method.md |
| X1005 | It replaces the node's identity layer with the supplied file [claim IDN-CTV-F21]. | The source describes a user-initiated replacement file. Name the distinct operation so readers do not attach this predicate to the historical IP resolver. | IDN-CTV-S12 [115] | supported with qualification | A manual refresh replaces the node's identity layer with the supplied file [claim IDN-CTV-F21]. | The source describes a user-initiated replacement file. Name the distinct operation so readers do not attach this predicate to the historical IP resolver. | 13-method.md |
| X1006 | A gap has no value in that sum. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A gap has no value in that sum [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1007 | Where a row has a score for a few strong cells and gaps in the rest, those strong cells can still give it a high raw score. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | Where a row has a score for a few strong cells and gaps in the rest, those strong cells can still give it a high raw score [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1008 | Two rows can thus share a printed whole number and sit on either side of a line. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | Two rows can thus share a printed whole number and sit on either side of a line [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1009 | A row can meet its own test and have too few peers for a band. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A row can meet its own test and have too few peers for a band [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1010 | A thin row can have a score near one with a band. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A thin row can have a score near one with a band [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1011 | A score from a different peer group gives no basis for choosing between the two. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A score from a different peer group gives no basis for choosing between the two [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1012 | Nothing about the source records changes in these calculations. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | Nothing about the source records changes in these calculations [claim IDN-METHOD-F02]. | The sentence states a consequence of the registered quotient, band or weighting rules. Attach the explicit method anchor. | 11-iq.md |
| X1013 | A case that is in scope but has an unknown result stays in the main count. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A case that is in scope but has an unknown result stays in the main count [claim IDN-METHOD-F02]. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | 13-method.md |
| X1014 | A case whose scope is in doubt goes into a wider check. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | A case whose scope is in doubt goes into a wider check [claim IDN-METHOD-F02]. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | 13-method.md |
| X1015 | If the high end is exactly half, the claim of more than half fails. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | If the high end is exactly half, the claim of more than half fails [claim IDN-METHOD-F02]. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | 13-method.md |
| X1016 | If the low end is half and the high end above it, leave the result open. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | IDN-METHOD-S03, IDN-METHOD-S04 [11, 13] | supported with qualification | If the low end is half and the high end above it, leave the result open [claim IDN-METHOD-F02]. | The sentence states how the registered majority test treats unknown records or an exact half. Add an explicit method anchor. | 13-method.md |
| X1017 | The case gives a reader a reported route to the DSP and a claim about how many IDs synced. | The sentence makes an unmarked factual synthesis of the named vendor case, including its delivery route and credited figure. | IDN-EXP-S01, IDN-EXP-S14 [37, 36] | supported with qualification | The case gives a reader a reported route to the DSP and a claim about how many IDs synced [claim IDN-EXP-F04][claim IDN-EXP-F08]. | The sentence makes an unmarked factual synthesis of the named vendor case, including its delivery route and credited figure. | 13-method.md |
| X1018 | It is a clear example of what a TV buyer might ask for, even as the right to buy that service on its own stays open. | Qualified territorial and purchased-service conclusion should carry the operator guides that bound it. | IDN-G-S13, IDN-G-S14, IDN-GV-S01, IDN-GV-S02, IDN-GV-S03 [227, 228, 229, 230, 231] | supported with qualification | It is a clear example of what a TV buyer might ask for, even as the right to buy that service on its own stays open [claim IDN-G-F13][claim IDN-U-F45]. | Qualified territorial and purchased-service conclusion should carry the operator guides that bound it. | 13-method.md |
| X1019 | This preserves the scope of the legal review while keeping the author's employer out of scored comparisons. | Attach the registered author-interest declaration to the factual statement about the scored exclusion. | IDN-METHOD-S01, IDN-METHOD-S02 [232, 233] | supported with qualification | This preserves the scope of the legal review while keeping the author's employer out of scored comparisons [claim IDN-METHOD-F01]. | Attach the registered author-interest declaration to the factual statement about the scored exclusion. | 13-method.md |
| X1020 | Figure accompanying the method chapter labelled 148 numerical entries as scored cells. | Use the project’s paper hook for an unambiguous label; the stock numeric-cell tally includes unverified cells. | IDN-METHOD-S05, IDN-METHOD-S07 [10, 218] | supported with qualification | The method chapter now renders separate panels for proposed cell positions and 142 counted cells across admitted rows. | Use the project’s paper hook for an unambiguous label; the stock numeric-cell tally includes unverified cells. | 13-method.md / fig-evidence-review |
| X1021 | Buyer-tool consent, household and lineage questions had references to adjacent product features. | Each question should point to the evidence needed to test the purchased service. | IDN-UID-S18, IDN-CTV-S12, IDN-CTV-S16, IDN-METHOD-S06 [150, 115, 114, 14] | supported with qualification | Use the recipient opt-out path, the manual-refresh and historical lookup rules, and the product-level lineage test as references for the three questions. | Each question should point to the evidence needed to test the purchased service. | data/decision-tool.json |
| X1022 | 33Across’s additional DSP-rate case was listed as inaccessible after the original URL could not be opened. | Two readers located the same original on the UK hostname, and a fresh fetch verified the passage. | IDN-SW-S01 [39] | supported with qualification | Added the publisher-hosted copy and H1-M22. Reporting nearly 90% of publisher traffic still lacks US case scope and a suitable assignment rule; P038 does not qualify. | Two readers located the same original on the UK hostname, and a fresh fetch verified the passage. | H1 / estimates |
| X1023 | The historical 33Across launch story was unavailable on its first URL. | Separate a dated namespace description from evidence that the same links survived. | IDN-SW-S02 [428] | supported with qualification | Added the recovered June 2021 issuer-hosted media interview. Its Pub Token was scoped to 33Across’s exchange. Current graph reuse remains unresolved. | Separate a dated namespace description from evidence that the same links survived. | H8 / source access |
| X1024 | The research search log lacked the recovery routes used after rate limiting on native scholarly indexes. | An index result is a search receipt, not original study evidence. | IDN-METHOD-S04, IDN-METHOD-S06 [13, 14] | supported with qualification | Synchronized all 186 native scholarly-index requests and their individual outcomes, including successful public index responses for the nine confirmed graph tasks. | An index result is a search receipt, not original study evidence. | checks/hypothesis-search-records.json |
| X1025 | CNIL SAN-2023-009 and Conseil d’État 482872 — Criteo; partner-site cookie and retargeting ID activity. Seed IDN-P007 is a possible functional match; confirm exact SKU. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S01, IDN-H-S02, IDN-H-S03 [63, 64, 65] | supported with qualification | CNIL SAN-2023-009 and Conseil d’État 482872 — CNIL examined Criteo’s partner-site cookies and retargeting. It recorded partner audits during the investigation. The decision and appeal do not establish a later change to ID for Exchanges. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1026 | Information Commissioner v Experian, [2024] UKUT 105 (AAC) — UK direct marketing and the ConsumerView database discussed in tribunal paragraphs 9–12; no automatic attribution to IDN-P013. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S04, IDN-H-S05, IDN-H-S06 [66, 67, 68] | supported with qualification | Information Commissioner v Experian, [2024] UKUT 105 (AAC) — The UK ConsumerView direct-marketing database is named. Experian projected compliance by July 24, 2025. Separate verification of implementation and a change to the US Digital Graph remain open. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1027 | FTC v Kochava, 2:22-cv-00377-BLW — Kochava and Collective Data Solutions; complaint describes household mapping. Separate purchased service and graph scope unresolved. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S07, IDN-H-S08 [69, 70] | supported with qualification | FTC v Kochava, 2:22-cv-00377-BLW — The complaint describes household mapping. The order imposes duties on Collective Data Solutions for precise device location. The record does not show a separate household-graph change. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1028 | FTC v Mobilewalla, docket C-4811 — Mobilewalla US data operations; place a specific identity deliverable in the universe before treating it as an H6 product. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S09, IDN-H-S10, IDN-HV-S02 [71, 72, 73] | supported with qualification | FTC v Mobilewalla, docket C-4811 — Mobilewalla’s SEC filings describe its own US data-screening program. The filings do not independently verify compliance or establish a change to a separately identified identity product. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1029 | Texas v Sony, original petition — law / Sony Corporation of America and Sony Electronics, Inc., were named defendants in the Nueces County petition. The pleading mentions Samba Interactive TV. The Samba household graph, IDN-P030, is an author interest. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S13, IDN-HV-S01 [76, 77] | supported with qualification | Texas v Sony, original petition — complaint / Texas named the Sony entities. The petition depicts a Samba Interactive TV screen. Texas reported the case ongoing on May 11, 2026. The filing establishes no change to the Samba household graph. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1030 | Texas v LG Electronics U.S.A., Inc., Bell County, 169th District Court, 25DCV358507 — law / LG viewing data; affirmative express consent. | Distinguish the affected service and procedural status from a completed change to a registered product. | IDN-H-S14, IDN-HV-S01 [78, 77] | supported with qualification | Texas v LG Electronics U.S.A., Inc., Bell County, 169th District Court, 25DCV358507 — settlement / Texas announced a settlement concerning LG viewing data. The posted agreed terms lack a verified judicial signature or entry date. Neither the trigger date nor a product rollout is established. | Distinguish the affected service and procedural status from a completed change to a registered product. | data/rules.json |
| X1031 | No entry for FTC, Publicis Groupe / LiveRamp, C-4841, file 261 0115 | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | IDN-LR-S17, IDN-LR-S18, IDN-LR-S19, IDN-VLR-S02 [87, 88, 89, 9] | supported with qualification | FTC, Publicis Groupe / LiveRamp, C-4841, file 261 0115 — Operative interim asset-maintenance order and separate proposed consent remedy. 2026-10-19 interim order issued. A separate proposal was posted for comment through 2026-11-23. Publicis and LiveRamp, competition and supply access. The interim order protects standalone services and confidential information. The record does not independently verify a completed RampID data-policy change. | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | data/rules.json |
| X1032 | No entry for Open Rights Group complaint to UK Information Commissioner’s Office concerning LiveRamp | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | IDN-LR-S20, IDN-LR-S07 [60, 61] | supported with qualification | Open Rights Group complaint to UK Information Commissioner’s Office concerning LiveRamp — Reported complaint, no verified operative disposition. Complaint reported 2024-02-28. No operative outcome verified. Open Rights Group reported the submission about European RampID processing after commissioning the Cracked Labs review. No linked regulator ruling or separately verified product change was found. | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | data/rules.json |
| X1033 | No entry for Open Rights Group complaint to French CNIL concerning LiveRamp | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | IDN-LR-S20, IDN-LR-S07 [60, 61] | supported with qualification | Open Rights Group complaint to French CNIL concerning LiveRamp — Reported complaint, no verified operative disposition. Complaint reported 2024-02-28. No operative outcome verified. Open Rights Group reported the submission about European RampID processing after commissioning the Cracked Labs review. No linked regulator ruling or separately verified product change was found. | The final H6 public-authority register includes this distinct record. Its procedural and implementation limits belong in the reader-facing instrument. | data/rules.json |
| X1034 | No private-case status row for Baker v. Yahoo. | The final H6 register keeps this private case as context outside its public-enforcement population. | IDN-YAH-S17, IDN-VYAH-S06 [90, 8] | supported with qualification | Baker v. Yahoo Inc., 25-cv-2797, orders 103 and 104 — The court sent five plaintiffs’ claims to arbitration and stayed those claims. It separately directed consultation about limited discovery for the sixth plaintiff. Private procedural case. The orders resolve no merits issue and establish no change to ConnectID. Outside the H6 public-enforcement population. | The final H6 register keeps this private case as context outside its public-enforcement population. | data/rules.json |
| X1035 | AdCP Trusted Match — 2026-11-11 / Specified request from publisher through router to buyer agents / Maturity shown as schema; public specification alone does not establish production status. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S01 [91] | supported with qualification | AdCP Trusted Match — October 29, 2026 (version record) / Experimental 3.2.8 schema: publisher via router to buyer agents; opaque tokens, page context excluded. / Schema evidence. The request definition supplies no live identity exchange. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1036 | AdCP Go reference client — Implemented HTTP connector / Publisher-side POST; no production status inferred. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S03, IDN-JV-S01, IDN-JV-S02 [92, 93, 94] | supported with qualification | AdCP Go reference client — Source code constructs a typed token request and posts it to the router’s identity route. / Reference connector and publisher-client tests at the September 11 commit. External buyer receipt, production identity traffic and an identity-bearing trade remain unverified. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1037 | AdCP Go reference identity engine — Implemented token-to-audience lookup / Requires a configured audience service. No campaign or third-party interoperability trace. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S05, IDN-JV-S01 [95, 93] | supported with qualification | AdCP Go reference identity engine — Code submits request tokens and target IDs to a batch membership lookup. / Reference connector. The lookup needs a configured audience service, identities and target audiences. No independent two-party trace. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1038 | AdCP Go UID2/EUID decoder — Implemented provider-specific decoder / Configuration-dependent and function-level evidence. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S06, IDN-JV-S01 [96, 93] | supported with qualification | AdCP Go UID2/EUID decoder — UID2 and EUID decoder code calls the configured client to decrypt an advertising token. / Reference code for a provider-specific operation. Key access, counterpart receipt and paid use are unverified. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1039 | AAMP OpenProposal — Draft object model / Proposal expectation, not measured performance. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S08, IDN-J-S07 [97, 98] | supported with qualification | AAMP OpenProposal — Draft seller field for expected match rate and buyer fields for segments and suppression. / Draft object model. The expectation is supplied by the seller. No measured match or executed suppression is reported. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1040 | AAMP buyer and seller SDKs — 2026-11-11 / Issuer-described demo / Identity payload and live transaction status are unconfirmed. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S15 [99] | supported with qualification | AAMP buyer and seller SDKs — October 27, 2026 (overview revision) / IAB describes buyer–seller negotiation and a handoff to Google Ad Manager. / Issuer demonstration. The exact demo date and identity payload are unverified. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1041 | IAB Agentic Audiences — 2026-11-11 / OpenRTB community-extension representation / Adjacent interface; confirm buyer-agent scope separately. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S09, IDN-J-S16, IDN-J-S17 [100, 101, 102] | supported with qualification | IAB Agentic Audiences — August 2026 (v1.1 release listing) / The OpenRTB community extension defines integer type 1 for an identity embedding. / Ordinary programmatic auctions can also use the extension. A type code does not identify a buyer agent or a transaction. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1042 | Prebid Agentic Audiences RTD — Merged module / Prebid records a merge into master on April 14, 2026. The module constructs the auction fragment. A buyer receipt and agent mediation remain unverified. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S10, IDN-J-S11, IDN-J-S12, IDN-J-S09, IDN-J-S16 [103, 104, 105, 100, 101] | supported with qualification | Prebid Agentic Audiences RTD — Merged code copies a stored type and vector into user.data[].segment[].ext.aa. / Prebid handoff inside the auction system. A stored entry can represent identity; the code supplies no receipt by a buyer agent. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1043 | PubMatic Activate — Seller reports live AdCP-enabled buy / Identity-specific production use or transaction is unconfirmed. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S13 [106] | supported with qualification | PubMatic Activate — PubMatic reports an AdCP-enabled CTV campaign for Clubtails. / Seller account of a campaign. No token, identity-bearing field or transaction trace links its delivery to the protocol’s identity route. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1044 | Optable and PubMatic — Commercial partners describe integration / No field-level trace in cited record. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | IDN-J-S14 [107] | supported with qualification | Optable and PubMatic — Optable reports publisher first-party audience planning via PubMatic. / Partner account describes access without transfer of underlying data. No identity field or recipient trace establishes an identity-bearing trade. | Show the observed interface, its evidence level and the remaining gap to an identity-bearing transaction. | data/agentic-map.json |
| X1045 | No HTTP identity-handler row. | The reference handler is a separate implemented operation recorded by IDN-J-F05. | IDN-J-S04, IDN-JV-S01 [108, 93] | supported with qualification | AdCP Go reference HTTP identity handler — The handler decodes the request, calls its service and returns eligible package IDs. Reference connector. This code supplies no two-party exchange record, named production use or identity-bearing trade. | The reference handler is a separate implemented operation recorded by IDN-J-F05. | data/agentic-map.json |
| X1046 | Nielsen historical Tapad verification — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-B-S06, IDN-B-S12, IDN-BV-S01 [52, 53, 54] | supported with qualification | Nielsen historical Tapad verification — 2015 evaluation; 2017 program overview. Accepted-link denominator and raw counts undisclosed. An investor reported 91% accuracy for Tapad. The original test protocol and current product mapping were unavailable. Funding filter: commissioned. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1047 | Intent IQ BidPulse release (October 2026) — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-ID5-S04, IDN-ID5-S03, IDN-VID5-S02, IDN-VID5-S03 [55, 56, 57, 58] | supported with qualification | Intent IQ BidPulse release (October 2026) — 2026. Person- or household-labelled panel size unreported. The competitor reports geographic consistency. The release does not give an independent person- or household-link error rate. Funding filter: vendor. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1048 | Google/New York Times Advertising Secure Signals — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-LR-S06 [59] | supported with qualification | Google/New York Times Advertising Secure Signals — 2025 publisher results. US-only sample and single-ID control undisclosed. Google reports an 81% gain with RampID and UID2 both in use. The contribution of either ID alone is unknown. Funding filter: vendor. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1049 | ORG/Cracked Labs technical review: LiveRamp — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-LR-S20, IDN-LR-S07 [60, 61] | supported with qualification | ORG/Cracked Labs technical review: LiveRamp — 2023 review; published 2024. Documentation review; no test panel. ORG reported complaints to the UK ICO and French CNIL. The review supplies no independently labelled link accuracy or regulatory finding. Funding filter: commissioned. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1050 | Deep Sync grocery and RV retailer studies — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-C-S19, IDN-C-S18, IDN-C-S20 [27, 28, 12] | supported with qualification | Deep Sync grocery and RV retailer studies — Cases linked September 2026. Separate bases: full monthly POS file and full retailer CRM file. Seller reports 73.0% of monthly grocery POS after three passes and 83.5% of a separate RV retailer’s CRM file. Funding filter: vendor. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1051 | Conversant/comScore cross-device announcement, April 2016 as reposted in May — blank study year, sample, setting, format, outcome and result; absent from funding filters. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | IDN-HX-S06 [62] | supported with qualification | Conversant/comScore cross-device announcement, April 2016 as reposted in May — 2016. Subset of cross-device links; accepted-link test denominator unavailable. The issuer describes a range for the top four of six tiers. The source does not connect these results to current Epsilon CORE ID. Funding filter: unknown. | Make the recorded study design, result and funding provenance visible in the stock table and its filters. | data/study-evidence.json |
| X1052 | Activation match rate definition — metric displays the value twice (No observed percentage). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S01 [15] | supported with qualification | Activation match rate definition — Activation match (definition): No observed percentage. Distinct from recognition and true-match accuracy. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1053 | Worked illustration — metric displays the value twice (25% recognition and 100% activation). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S02 [16] | supported with qualification | Worked illustration — Device recognition and person activation (worked example): 25% recognition and 100% activation. A worked example rather than an observed counterexample. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1054 | US insurer assessment — metric displays the value twice (99.5%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S04, IDN-CV-S01, IDN-CV-S02 [17, 18, 19] | supported with qualification | US insurer assessment — Assessment file records resolved to a person: 99.5%. TransUnion reports a file-to-person result from its assessment. The diagnostic assessment and commercial resolution/enrichment service have separate catalog entries. The remaining acceptance paths and direct scored-service attribution are not shown. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1055 | Experian / MiQ Digital Graph — metric displays the value twice (70%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S05, IDN-C-S06 [20, 21] | supported with qualification | Experian / MiQ Digital Graph — Submitted IPs matched to cookieless identifiers: 70%. Global operations do not identify the US test stratum. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1056 | IRI transaction data / CORE ID — metric displays the value twice (95%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S07 [22] | supported with qualification | IRI transaction data / CORE ID — Transaction data linked to CORE IDs: 95%. Missing denominator unit and sample scope. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1057 | Intent Messaging, known email subscribers — metric displays the value twice (77%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S08 [23] | supported with qualification | Intent Messaging, known email subscribers — Match for messaging among known email subscribers: 77%. CORE ID is described for the distinct unauthenticated-reach outcome. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1058 | Cadent audience segments — metric displays the value twice (Above 90% average). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S09 [24] | supported with qualification | Cadent audience segments — Mean reported segment match: Above 90% average. No file counts, uncertainty or segment weights. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1059 | Viant/Acxiom direct match — metric displays the value twice (More than 275 million unique IDs). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S10 [25] | supported with qualification | Viant/Acxiom direct match — Count of matched unique IDs: More than 275 million unique IDs. Count only; no percent and no licensed-SKU mapping. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1060 | Forrester composite identity match gain — metric displays the value twice (20 percentage points). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S11 [26] | supported with qualification | Forrester composite identity match gain — Modeled change in enterprise identity-match rate: 20 percentage points. Modeled change, not a product-level observed percentage. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1061 | US RV retailer CRM file — metric displays the value twice (83.5% resolved). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S19, IDN-C-S18 [27, 28] | supported with qualification | US RV retailer CRM file — Full CRM records resolved: 83.5% resolved. Case also displays a 700K incoming audience; no raw counts to derive exact global matched numerator. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1062 | US grocery chain monthly POS — metric displays the value twice (73.0% cumulative, after pass 3). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S20 [12] | supported with qualification | US grocery chain monthly POS — Cumulative monthly POS match after three passes: 73.0% cumulative, after pass 3. Vendor also reports about 70% of anonymous transactions and 64–73% of monthly volume. Approximate pass counts total 31.0m, against a 32m headline. The final pass infers buyers from geography and households; no labeled accuracy. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1063 | Kantar / Blockgraph panel overlap with TV partners — metric displays the value twice (Above 10× before/after). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S25 [29] | supported with qualification | Kantar / Blockgraph panel overlap with TV partners — Relative increase in panel-to-partner matching: Above 10× before/after. Relative change, no absolute levels. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1064 | Yahoo DataX sample response — metric displays the value twice (80 of 100 (80%)). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S27 [30] | supported with qualification | Yahoo DataX sample response — Partner-ID match in a sample upload response: 80 of 100 (80%). No measured result or automatic transfer to ConnectID. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1065 | Mass retailer, UID2 activation — metric displays the value twice (73.2M / 51% before; 94.4M / 66% after). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S30 [31] | supported with qualification | Mass retailer, UID2 activation — Matched CRM records at UID2 before and after enrichment: 73.2M / 51% before; 94.4M / 66% after. Deep Sync’s labeled UID2 row gives 51% and 66%. Dividing by the 141.7m nominal CRM size gives about 51.7% and 66.6%; rounding or destination eligibility is unspecified. Meta’s upper estimate exceeds 100%. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1066 | ID5 onboarding preview — metric displays the value twice (Definition only). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S31 [32] | supported with qualification | ID5 onboarding preview — Distinct input hashed emails with at least one match, by output ID type (definition): Definition only. Preview gives a definition; a US measured numerical case for this particular metric is absent from the inspected evidence. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1067 | Merkury Data Enhancement / DataSource to Data Cloud — metric displays the value twice (At least 95%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S32 [33] | supported with qualification | Merkury Data Enhancement / DataSource to Data Cloud — DataSource matches to Data Cloud profiles: At least 95%. Published by Merkle; no defined US client file or rule. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1068 | Joint onboarding and Lexicon — metric displays the value twice (90% average). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-HX-S01 [34] | supported with qualification | Joint onboarding and Lexicon — Average onboarding match, joint Lexicon claim: 90% average. Joint production claim; no denominator or averaging weights. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1069 | US QSR exposure/control campaign — metric displays the value twice (66.1%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-U-S56 [35] | supported with qualification | US QSR exposure/control campaign — Household match of exposure and holdout records: 66.1%. Reported by Kochava for a measurement engagement; original screened unit is not a separately verified licensed US graph. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1070 | US insurer field-agreement slice — metric displays the value twice (97.4%). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-C-S04, IDN-CV-S01, IDN-CV-S02 [17, 18, 19] | supported with qualification | US insurer field-agreement slice — Agreement on names and at least one other personal field: 97.4%. The 97.4% is a field-agreement slice. The 99.5% is a wider file-to-person assessment result. Exact slice denominator and attribution to a licensed TruAudience service are unverified. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1071 | Unnamed DSP cookie-sync case — metric displays the value twice (84% synced). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-EXP-S14, IDN-EXP-S01 [36, 37] | supported with qualification | Unnamed DSP cookie-sync case — IDs synced by a DSP: 84% synced. Secondary issuer report on Digital Graph with a different input and output from MiQ. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1072 | Spectrum own MVPD service — metric displays the value twice (95% addressable match rate). The product and denominator qualification is outside the rendered columns. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | IDN-CTV-S08 [38] | supported with qualification | Spectrum own MVPD service — Addressable match for Spectrum’s own service: 95% addressable match rate. Adjacent Spectrum service. No attribution to a Blockgraph match rate. | Identify the estimand independently of its numeric value and expose the existing case restriction in the rendered table. | data/estimates.json |
| X1073 | 33Across publisher traffic resolved to DSP ID — Unspecified exclusions in rendered estimate. | The new independent sweep row records geography, unit and service-attribution limits in a field the stock table hides. | IDN-SW-S01 [39] | supported with qualification | 33Across publisher traffic resolved to DSP ID — No identified US sample, exact eligible traffic unit or assignment rule. Distinct from the MediaWallah joint offering. | The new independent sweep row records geography, unit and service-attribution limits in a field the stock table hides. | data/estimates.json |
| X1074 | Abstract, method and plain edition did not identify who made the paper. | Disclose research, writing, scoring and checks. | IDN-PROC-S01 [364] | supported with qualification | Identify the seven OpenAI stages, internal model checks, the subsequent 36-claim Claude Code review, its four dispositions and the scope of human review. | Disclose research, writing, scoring and checks. | Abstract, chapter 13, plain edition, process record and claim IDN-PROC-F01. |
| X1075 | The abstract and method could imply a full day of coverage. | Distinguish the registered date from the observed source snapshot. | IDN-PROC-S01, IDN-METHOD-S04 [364, 13] | supported with qualification | State the actual 8:37 a.m. EST close on November 11, 2026. | Distinguish the registered date from the observed source snapshot. | Abstract, chapter 13 and handoff report. |
| X1076 | H5 headline generalized from one confirmed buyer service. | Apply the frozen decision rule to the confirmed population and show eligibility uncertainty. | IDN-METHOD-S04, IDN-METHOD-S06, IDN-CTV-S11, IDN-CTV-S16, IDN-CTV-S12 [13, 14, 109, 114, 115] | supported with qualification | Give the confirmed result as 0/1 on Blockgraph and the sensitivity range as 0/9 to 8/9. The wider market finding is unresolved. | Apply the frozen decision rule to the confirmed population and show eligibility uncertainty. | Abstract, chapter 7, method, plain edition, H5 ledger and handoff. |
| X1077 | The H5 headline named LtvAdx without citing the relevant records. | Separate a US ad purchase from a household-resolution service purchase. | IDN-G-S13, IDN-G-S14, IDN-GV-S01, IDN-GV-S02, IDN-GV-S03 [227, 228, 229, 230, 231] | supported with qualification | Attach five LtvAdx sources, identify the 30-day last-seen rule and US ad plans, and explain the eligibility boundary. | Separate a US ad purchase from a household-resolution service purchase. | IDN-HYP-H5 and chapter 7. |
| X1078 | The H7 ledger described the workflow as buying and omitted the author interest at the stand-alone verdict. | Trace the publisher client and separate implementation from a paid event. | IDN-METHOD-S06, IDN-METHOD-S04, IDN-J-S03, IDN-J-S05, IDN-J-S04, IDN-J-S16, IDN-METHOD-S01 [14, 13, 92, 95, 108, 101, 232] | supported with qualification | Identify the publisher client, Trusted Match router and conditional membership lookup. State the limit of transaction evidence and the author’s working-group role next to the verdict. | Trace the publisher client and separate implementation from a paid event. | IDN-HYP-H7, hypothesis results and handoff. |
| X1079 | The relationship between H1 and Deep Sync’s 2/5 grade D cell was implicit. | Compare H1’s three required disclosures with the registered match-rate documentation ladder. | IDN-C-S20, IDN-METHOD-S03, IDN-METHOD-S05 [12, 11, 10] | supported with qualification | Explain the missing calendar month, rounded steps and different rate summaries on the case page. | Compare H1’s three required disclosures with the registered match-rate documentation ladder. | Chapter 11. |
| X1080 | The scoring table labelled unverified cells individually but readers had to count the exclusions. | Count all numeric cells and those used in the IQ. | IDN-METHOD-S05 [10] | supported with qualification | Add 148 numeric, 142 counted and six unverified, treated as n/e. | Count all numeric cells and those used in the IQ. | Chapter 11 matrix note. |
| X1081 | A LiveIntent sample request was described as a worked bidder result. | Does the Prebid example show actual downstream receipt? | IDN-A5-S07 [260] | supported with qualification | Describe an illustrative response field and the documented adapter path. Mark the claim supported with qualification. | Does the Prebid example show actual downstream receipt? | IDN-AS5-F12, chapter and matrix rationale. |
| X1082 | The abbreviated support fields for H4 and H8 joined passages from several different records. | Can a reviewer assign each excerpt to its issuer? | IDN-E-S17, IDN-E-S18, IDN-YAH-S01, IDN-D-S05, IDN-D-S06 [221, 222, 160, 214, 215] | supported with qualification | Label the twelve H4 and seven H8 excerpts separately by source and locator. | Can a reviewer assign each excerpt to its issuer? | IDN-HYP-H4 and IDN-HYP-H8. |
| X1083 | The Permutive case wording risked applying the US figure to an entire integration. | Identify the measured inventory, unit and separate buying route. | IDN-U-S25, IDN-UV-S02, IDN-UV-S03 [205, 207, 208] | supported with qualification | Tie 1.24 audience categories per impression to US inventory in the Google case. Attribute activation steps and the OpenX route to their own sources. | Identify the measured inventory, unit and separate buying route. | IDN-U-F19, chapter and matrix rationale. |
| X1084 | The Yahoo order locator did not lead directly to the signed stay. | Read the signed conclusion and the separate order for Baker. | IDN-YAH-S17, IDN-VYAH-S06 [90, 8] | supported with qualification | Retain the five-plaintiff stay and give the physical page of the signed August 26 order. | Read the signed conclusion and the separate order for Baker. | IDN-YAH-F20. |
| X1085 | The Markdown appendix could be mistaken for the complete claim register. | Compare the instrument selection with the downloadable claim ledger. | IDN-PROC-S01 [364] | supported with qualification | Label the table as the claims used in the paper or marked as headlines and link the complete claim-level data. | Compare the instrument selection with the downloadable claim ledger. | Chapter 14 and claim-ledger download. |

### Take the records with you

The downloads include the source and claim registers, the score rules, the product rows and the data behind the figures [11, 13]. Each stage of the full handoff has a counted self-check [11, 13]. The final report names the sources that could be read only in part or could not be opened [11, 13].

- `advertising-identity-2026.pdf` — the print edition: A4, with page numbers and chapter bookmarks
- `advertising-identity-2026.md` — the reading edition in Markdown, with numbered references
- `data/source-register.csv` — every source with publication, event and access dates, type, geography, funding, access limits and reliability notes
- `data/claim-ledger.csv` — every material claim with sources, passage locator, scope, evidence class, review outcome, confidence and final wording
- `data/capability-matrix-wide.csv` — vendors × dimensions, scores only (0–5, n/e, n/a)
- `data/capability-matrix-long.csv` — one row per cell with availability, evidence grade, date, rationale and sources
- `data/capability-dimensions.csv` — the dimension definitions and anchors
- `data/iq.csv` — Identity Quotient per vendor: group means, coverage, band, the reweightings and every dimension score
- `data/iq-method.csv` — IQ groups, weights, shrinkage, field mean, band thresholds, and the reweighting, shrinkage and missing-evidence tests
- `data/iq-missing-evidence.csv` — every scored cell turned into n/e one at a time, with the IQ before and after
- `data/swot.csv` — every SWOT line for the profiled vendors with its basis (record, vendor documentation, vendor claim, inference) and sources
- `data/deep-profiles.csv` — the deep-profile summaries: what is sold, buyers, earnings, dependencies, fit, limits
- `data/deep-profile-records.csv` — every profiled capability, performance claim, financial figure and timeline event with sources
- `data/study-evidence.csv` — studies with design, sample, sponsor and limits
- `data/corrections-ledger.csv` — every change made after verification and challenge passes, with implementation and verification
- `data/vendor-universe.csv` — the reasoned vendor sample with scope group, owner, formats, routes and status
- `data/agentic-map.csv` — AI and agentic capabilities with status labels and observability
- `data/rules.csv` — rules and governance instruments by jurisdiction
- `data/estimates.csv` — market-size estimates with inclusions, exclusions and combinability
- `data/decision-tool.csv` — the decision tool rows
- `data/product-register.csv` — the product register: every scored or listed product with its purchased service, operating entity, owner, cohort, inclusion reason, aliases and US availability
- `data/hypotheses.csv` — the hypotheses registered before extraction, with the rules for supported, contradicted and unresolved
- `data/figure-data.json` — the canonical dataset behind every data figure
- `data/derived-checks.json` — every derived number with its formula and inputs
- `data/stats.json` — every count the prose and figures quote about the research itself, computed from the ledgers
- `data/scoring-rules.json` — dimension definitions, anchors, cell rules and the IQ definition, fixed before scoring



## References

The 430 sources cited in this paper, numbered in order of first citation. The full register of 442 source records, with access dates, access limits and reliability notes, is in data/source-register.csv.

[1] OpenRTB 2.6, EID and UID objects. IAB Tech Lab. https://github.com/InteractiveAdvertisingBureau/openrtb2.x/blob/main/2.6.md Accessed 2026-11-11.

[2] Prebid.js TCF Control Module. Prebid.org. https://docs.prebid.org/dev-docs/modules/tcfControl Accessed 2026-11-11.

[3] Global Privacy Control, W3C Working Draft of 24 September 2026. W3C Privacy Working Group, 2026-09-24. https://www.w3.org/TR/gpc/ Accessed 2026-11-11.

[4] Next steps for Privacy Sandbox and tracking protections in Chrome. Google, 2025-04-22. https://privacysandbox.google.com/blog/privacy-sandbox-next-steps Accessed 2026-11-11.

[5] Remove Protected Audience Web platform bindings and features. Chromium project, 2026-10-30. https://chromium.googlesource.com/chromium/src/%2B/2ae93f9b43b9bb8d54e382a9257ef950feaa308c?format=JSON Accessed 2026-11-11.

[6] Passing the Baton: Programmatic TV Advertising as a Relay Race. Coalition for Innovative Media Measurement, 2025-04. https://cimm-us.org/wp-content/uploads/2025/04/Passing-the-Baton-Programmatic-TV-Advertising-as-a-Relay-Race_April-2025-v2.pdf Accessed 2026-11-11.

[7] Annual report on Form 10-K for the fiscal year ended March 31, 2026. LiveRamp Holdings, Inc., 2026-05-21. https://www.sec.gov/Archives/edgar/data/733269/000073326926000025/ramp-20260331.htm Accessed 2026-11-11.

[8] Baker v. Yahoo Inc., SDNY, order 104. US District Court for the Southern District of New York, 2026-08-26. https://cases.justia.com/federal/district-courts/new-york/nysdce/1:2025cv02797/639983/104/0.pdf Accessed 2026-11-11.

[9] Publicis and LiveRamp; Analysis of Proposed Agreement Containing Consent Order To Aid Public Comment. Federal Trade Commission, Federal Register via U.S. Government Publishing Office, 2026-10-23. https://www.govinfo.gov/content/pkg/FR-2026-10-23/html/2026-21659.htm Accessed 2026-11-11.

[10] (Author's work) Computed capability matrix and sensitivity scenarios. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/capability-matrix.json Accessed 2026-11-11.

[11] (Author's work) Preregistered dimension anchors, cohort masks and quotient. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/scoring-rules.json Accessed 2026-11-11.

[12] Turning 43.8 million anonymous transactions into real people. Deep Sync. https://deepsync.com/case-grocery-transaction-matching Accessed 2026-11-11.

[13] (Author's work) Registered hypotheses, populations and decision rules. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/hypotheses.json Accessed 2026-11-11.

[14] (Author's work) Study-level test results and public-record case registers. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/hypothesis-results.json Accessed 2026-11-11.

[15] Match Rates. LiveRamp, 2026-08-06. https://docs.liveramp.com/connect/en/match-rates.html Accessed 2026-11-11.

[16] Interpreting RampID, LiveRamp’s People-Based Identifier. LiveRamp, 2025-07-10. https://docs.liveramp.com/connect/en/interpreting-rampid,-liveramp-s-people-based-identifier.html Accessed 2026-11-11.

[17] Filling the Blanks: How a National Insurance Provider Enriched Its Customer Profiles With TransUnion. TransUnion. https://www.transunion.com/case-study/dha-insurance-identity Accessed 2026-11-11. Access: partial extract.

[18] Filling the Blanks: How a National Insurance Provider Enriched its Customer Profiles with TransUnion (full case PDF). TransUnion, 2025. https://www.transunion.com/content/dam/transunion/us/business/collateral/use-case/us-gms-25-3444829-insurance-health-check-case-study.pdf Accessed 2026-11-11.

[19] TransUnion Identity and Snowflake: Solving Foundational Marketing Use Cases at Scale. TransUnion, 2026-01-27. https://www.transunion.com/lp/identity-snowflake Accessed 2026-11-11.

[20] Case study resources: Experian’s Graph increases reach and scale with MiQ. Experian. https://www.experian.com/marketing/resources/case-study Accessed 2026-11-11.

[21] MiQ increases reach and scale for their advertisers using Experian’s Graph. Experian and MiQ. https://www.experian.com/content/dam/marketing/na/assets/ems/marketing-services/documents/case-studies/Graph-MIQ-CaseStudy.pdf Accessed 2026-11-11.

[22] As third-party cookies crumble, CPGs look to smarter clean rooms. Epsilon / Kate Jacobson, 2022-09-09. https://www.epsilon.com/us/insights/blog/cpg-clean-rooms-iri Accessed 2026-11-11.

[23] Driving revenue and better customer experience with Intent Messaging. Epsilon. https://www.epsilon.com/us/client-success/case-studies/specialty-retailer-intent-messaging Accessed 2026-11-11.

[24] Cadent Aperture Viewer Graph Continues to Scale, Delivering Unprecedented TV Audience Insights at Market-Leading Speed. Cadent via PR Newswire, 2022-03-30. https://www.prnewswire.com/news-releases/cadent-aperture-viewer-graph-continues-to-scale-delivering-unprecedented-tv-audience-insights-at-market-leading-speed-301513209.html Accessed 2026-11-11.

[25] Viant and Acxiom Advance Deterministic Partnership With High Match Results. Viant, 2019-09-11. https://www.viantinc.com/company/news/press-releases/viant-and-acxiom-advance-deterministic-partnership-with-high-match-results/ Accessed 2026-11-11.

[26] The Total Economic Impact of TransUnion Audience and Identity Solutions. Forrester Consulting, 2026-07. https://tei.forrester.com/go/transunion/audienceandidentitysolutions/?lang=en-us Accessed 2026-11-11.

[27] 83.5% match rate, and 4.9× device IDs per customer. Deep Sync. https://deepsync.com/case-retail-device-reach Accessed 2026-11-11.

[28] The Retail Identity Gap is costing you money every day. Here’s how we closed it. Deep Sync / Julia Andrade, 2026-09-01. https://deepsync.com/blog/retail-identity-gap/ Accessed 2026-11-11.

[29] Blockgraph Kantar Partnership Elevates Convergent TV Measurement Capabilities. Kantar, 2022-11-01. https://www.kantar.com/north-america/Company-news/Blockgraph-Kantar-Partnership-Elevates-Convergent-TV-Measurement-Capabilities Accessed 2026-11-11.

[30] Match Rates (Yahoo DataX). Yahoo. https://help.yahooinc.com/datax/docs/match-rates Accessed 2026-11-11.

[31] A top retail loyalty program still had a 26% blind spot. Deep Sync. https://deepsync.com/case-retail-enrichment Accessed 2026-11-11.

[32] Data Onboarding – Match Preview. ID5. https://wiki.id5.io/docs/id-data-onboarding-match-preview Accessed 2026-11-11.

[33] Merkle at Adobe Summit 2024, Merkury Data Enhancement slide. Merkle, 2024. https://go.merkle.com/rs/442-SZV-721/images/2024_Merkle_Adobe_Summit_Booth_Talks.pdf?version=1 Accessed 2026-11-11.

[34] 33Across and MediaWallah Partner to Innovate Data Onboarding and Identity Resolution for Programmatic Campaigns. 33Across and MediaWallah, issuer release via GlobeNewswire, 2024-11-19. https://rss.globenewswire.com/news-release/2024/11/19/2983661/0/en/33across-and-mediawallah-partner-to-innovate-data-onboarding-and-identity-resolution-for-programmatic-campaigns.html Accessed 2026-11-11.

[35] QSR Proves Incremental Lift from Connected TV Ad Campaign. Kochava. https://www.kochava.com/case-studies/qsr-connectedtv/ Accessed 2026-11-11.

[36] How Experian’s Digital Graph improved attribution accuracy for a DSP. Experian Marketing Services, 2025-10-01. https://www.experian.com/blogs/marketing-forward/how-experians-digital-graph-improved-attribution-accuracy-for-a-dsp-with-84-of-ids-resolved/ Accessed 2026-11-11.

[37] How Graph enabled accurate attribution for a demand-side platform (DSP). Experian Marketing Services. https://www.experian.com/content/dam/marketing/na/assets/ems/marketing-services/documents/case-studies/Graph_Case_Study.pdf Accessed 2026-11-11.

[38] Leverage the Power of Multiscreen. Spectrum Reach, 2024-03-07. https://www.spectrumreach.com/articles/multiscreen-attribution Accessed 2026-11-11.

[39] How Lexicon Increases Buying Power for DSPs. 33Across. https://www.33across.co.uk/insights/case-study/how-lexicon-increases-buying-power-for-dsps Accessed 2026-11-11.

[40] The Challenges of Using IP Address for Audience Identification and Measurement. Truthset for CIMM and Go Addressable, 2025-11-05. https://cimm-us.org/wp-content/uploads/2025/11/CIMM-_-GoAddressable-_-Truthset-Challenges-of-Using-IP-Address-for-Audience-Identification-and-Measurement-November-2025.pdf Accessed 2026-11-11.

[41] The Challenges of Using IP Address for Audience Identification. Go Addressable, 2025-11. https://goaddressable.com/report/challenges-using-ip-address-for-audience-identification-measurement/ Accessed 2026-11-11.

[42] Television Data Match Rate & Match Bias Study. CIMM; Janus Strategy & Insights, Pre-Meditated Media and Sequent Partners, 2023-05. https://cimm-us.org/wp-content/uploads/2024/04/Television-Data-Match-Rate-and-Match-Bias-Study_Janus-Pre-Meditated-Sequent_May-2023.pdf Accessed 2026-11-11.

[43] Deterministic Matching. LiveRamp. https://docs.liveramp.com/connect/en/deterministic-matching.html Accessed 2026-11-11.

[44] Cross-Device Methodologies: Deterministic vs. Probabilistic. LiveRamp. https://storage.googleapis.com/lr-tech-docs-resources/PDFs/Device-Graph%20Methodologies-%20Deterministic%20vs.%20Probabilistic%20%281%29.pdf Accessed 2026-11-11.

[45] Tapad, part of Experian, scores highest in every category of measurement in Hotels.com partner evaluation. Tapad / Experian, 2016-09-14. https://www.experian.com/blogs/marketing-forward/tapad-part-of-experian-scores-highest-in-every-category-of-measurement-in-hotels-com-partner-evaluation/ Accessed 2026-11-11.

[46] Experian acquires Tapad, a leading digital identity resolution provider. Experian plc, 2020-11-19. https://www.experianplc.com/newsroom/press-releases/2020/experian-acquires-tapad Accessed 2026-11-11.

[47] A Privacy Analysis of Cross-device Tracking. Sebastian Zimmeck, Jie S. Li, Hyungtae Kim, Steven M. Bellovin and Tony Jebara, 2017-08. https://www.usenix.org/system/files/conference/usenixsecurity17/sec17-zimmeck.pdf Accessed 2026-11-11.

[48] Reasons Dynamic Addresses Change. Ramakrishna Padmanabhan, Amogh Dhamdhere, Emile Aben, kc claffy and Neil Spring, 2016-11. https://www.caida.org/~amogh/papers/dynamic-IMC16.pdf Accessed 2026-11-11.

[49] Economic Consequences of Online Tracking Restrictions: Evidence from Cookies. Klaus M. Miller and Bernd Skiera, 2023-03. https://arxiv.org/pdf/2303.09147 Accessed 2026-11-11.

[50] Cookie-Based Counting Overstates Size of Web Site Audiences. comScore, 2007-04-16. https://ir.comscore.com/news-releases/news-release-details/cookie-based-counting-overstates-size-web-site-audiences Accessed 2026-11-11.

[51] Bridging The Gap: Advanced Identity Solutions For People-Based Marketers. Forrester Consulting, 2019-07. https://tools.marketimpacttools.com/go/neustar/identityresolution/docs/Neustar_Identity_Resolution_TLP.pdf Accessed 2026-11-11.

[52] Nielsen Launches Cross-Device Provider Verification Powered by Digital Ad Ratings. Nielsen, 2017-01. https://www.nielsen.com/news-center/2017/nielsen-launches-cross-device-provider-verification-powered-by-digital-ad-ratings/ Accessed 2026-11-11.

[53] Portfolio Company News: Tapad Cross-Device Connections 91% Accurate. Firsthand Technology Value Fund, 2015. https://firsthandtvf.com/docs/newsletter/q414.pdf Accessed 2026-11-11.

[54] 2016 Edition: A Marketer’s Guide To Cross-Device Identity. Allison Schiff / AdExchanger, 2016-02-29. https://www.adexchanger.com/data-exchanges/2016-edition-marketers-guide-cross-device-identity/ Accessed 2026-11-11.

[55] First-party IDs and identity resolution methods explained. ID5, 2022-03-24. https://id5.io/news/first-party-ids-and-identity-resolution-methods-explained/ Accessed 2026-11-11.

[56] ID5 Javascript Library. ID5, 2026-02-05. https://wiki.id5.io/docs/id5-javascript-library Accessed 2026-11-11.

[57] Intent IQ Launches BidPulse, Bringing New Transparency to Identity Accuracy in Programmatic Advertising. Intent IQ, original release via GlobeNewswire, 2026-10-12. https://www.globenewswire.com/news-release/2026/10/12/3379132/0/en/intent-iq-launches-bidpulsesm-bringing-new-transparency-to-identity-accuracy-in-programmatic-advertising.html Accessed 2026-11-11.

[58] BidPulse: Online Identity Provider Accuracy Measurement System. Intent IQ, 2026-10. https://www.intentiq.com/wp-content/uploads/2026/10/bidpulse-white-paper.pdf Accessed 2026-11-11.

[59] Secure Signals delivers 81% revenue uplift for New York Times Advertising in low-cookie environments. Google Ad Manager / New York Times Advertising. https://admanager.google.com/home/success-stories/secure-signals-delivers-81-percent-revenue-uplift-for-new-york-times-advertising-in-low-cookie-environments/ Accessed 2026-11-11.

[60] ORG submits complaints about intrusive LiveRamp adtech system. Open Rights Group, 2024-02-28. https://www.openrightsgroup.org/press-releases/org-complaint-liveramp-adtech/ Accessed 2026-11-11.

[61] Pervasive identity surveillance for marketing purposes. Wolfie Christl and Alan Toner, Cracked Labs / Open Rights Group, 2024-02-28. https://www.openrightsgroup.org/publications/report-pervasive-identity-surveillance-for-marketing-purposes/ Accessed 2026-11-11.

[62] Conversant Unveils Findings of Consumer Cross-Device Matching Accuracy Study. Chris Backschies, repost of Conversant announcement, 2016-05-18. https://www.linkedin.com/pulse/conversant-unveils-findings-consumer-cross-device-study-backschies Accessed 2026-11-11.

[63] CNIL restricted committee decision SAN-2023-009. CNIL via Légifrance, 2023-06-22. https://www.legifrance.gouv.fr/cnil/id/CNILTEXT000047707063 Accessed 2026-11-11.

[64] Conseil d’État, decision 482872, Criteo. Conseil d’État, 2026-03-04. https://www.conseil-etat.fr/fr/arianeweb/CE/decision/2026-03-04/482872 Accessed 2026-11-11.

[65] One-stop-shop summary of the French retargeting decision. French lead supervisory authority via European Data Protection Board, 2023-06-22. https://www.edpb.europa.eu/system/files/2023-08/fr_external_summary_final_decision_art_60.pdf Accessed 2026-11-11.

[66] Information Commissioner v Experian Limited [2024] UKUT 105 (AAC), full judgment. Upper Tribunal (Administrative Appeals Chamber), 2024-04-29. https://assets.publishing.service.gov.uk/media/662fa61624347c67e8e3cba0/UA_2023_000512_GIA.pdf Accessed 2026-11-11.

[67] Information Commissioner v Experian Ltd, official judicial summary. HM Courts & Tribunals Service, 2024-04-29. https://www.gov.uk/administrative-appeals-tribunal-decisions/information-commissioner-v-experian-ltd-2024-ukut-105-aac Accessed 2026-11-11.

[68] Listing particulars, 12 June 2025, section on use of personal data for direct marketing. Experian plc, 2025-06-12. https://www.experianplc.com/content/dam/marketing/global/plc/en/assets/documents/debt-information/emtns-and-1441as/12-june-2025-us-notes-listing-particulars.pdf Accessed 2026-11-11.

[69] FTC v Kochava, second amended complaint, document 86. Federal Trade Commission, filed in District of Idaho, 2024-07-15. https://www.ftc.gov/system/files/ftc_gov/pdf/86-SecondAmendedComplaint.pdf Accessed 2026-11-11.

[70] FTC v Kochava and Collective Data Solutions, stipulated order, document 138. US District Court for the District of Idaho, 2026-06-25. https://www.ftc.gov/system/files/ftc_gov/pdf/Kochava-Order.pdf Accessed 2026-11-11.

[71] Mobilewalla decision and order, docket C-4811. Federal Trade Commission, 2025-01-13. https://www.ftc.gov/system/files/ftc_gov/pdf/Mobilewalla-D&O.pdf Accessed 2026-11-11.

[72] Form S-4, SEC accession 0001193125-26-347101, Mobilewalla business and global data governance. Mobilewalla Holdco / SPACSphere via SEC EDGAR, 2026-08-12. https://www.sec.gov/Archives/edgar/data/2149218/000119312526347101/d144744ds4.htm Accessed 2026-11-11.

[73] Amendment No. 1 to Form S-4, Mobilewalla business, Global Data Governance. SPACSphere Acquisition Corp. / Mobilewalla business disclosure, 2026-10-16. https://www.sec.gov/Archives/edgar/data/2081300/000119312526424301/ck0002081300-20261016.htm Accessed 2026-11-11.

[74] People v Disney DTC et al., final judgment and permanent injunction, 26STCV04425. Superior Court of California, County of Los Angeles, 2026-02. https://oag.ca.gov/system/files/attachments/press-docs/CA_SUP_LAX_26STCV04425__Final_Judgment_and_Permanent_Injunction.pdf Accessed 2026-11-11.

[75] Attorney General announces streaming opt-out settlement with Disney. California Department of Justice, 2026-02-11. https://oag.ca.gov/news/press-releases/california-wont-let-it-go-attorney-general-bonta-announces-275-million Accessed 2026-11-11.

[76] State of Texas v Sony, original petition and request for injunctive relief. Texas Attorney General, 2025-12-15. https://www.oag.state.tx.us/sites/default/files/images/press/Sony%20TV%20Petition%20Filed.pdf Accessed 2026-11-11.

[77] Texas announces settlement with LG Electronics U.S.A., Inc. and publishes agreed terms. Office of the Attorney General of Texas, 2026-05-11. https://www.texasattorneygeneral.gov/news/releases/attorney-general-ken-paxton-secures-major-agreement-lg-protect-texans-privacy-and-stop-data-being Accessed 2026-11-11.

[78] State of Texas v LG Electronics U.S.A., Inc., agreed final judgment and permanent injunction, posted copy without judicial signature, 25DCV358507. State of Texas and LG Electronics U.S.A., Inc.; agreed filing for the 169th District Court, Bell County, Texas, 2026. https://www.oag.state.tx.us/sites/default/files/images/press/LG.pdf Accessed 2026-11-11.

[79] CNIL sanction and injunction concerning Google LLC and Google Ireland, English translation. Commission nationale de l’informatique et des libertés, 2025-09-01. https://www.cnil.fr/sites/default/files/2025-09/cnil_sanction_google_en.pdf Accessed 2026-11-11.

[80] Closure of the injunction against Google. Commission nationale de l’informatique et des libertés, 2026-10-14. https://www.cnil.fr/en/closure-injunction-google Accessed 2026-11-11.

[81] The Market Court rules in the IAB Europe case. Belgian Data Protection Authority, 2025-05-14. https://www.dataprotectionauthority.be/the-market-court-rules-in-the-iab-europe-case Accessed 2026-11-11.

[82] IAB Europe v Gegevensbeschermingsautoriteit, C-604/22, ECLI:EU:C:2024:214. Court of Justice of the European Union, 2024-03-07. https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=ecli%3AECLI%3AEU%3AC%3A2024%3A214 Accessed 2026-11-11.

[83] Conseil d’État, Yahoo EMEA Limited, decision 494300. Conseil d’État, 2025-10-07. https://www.conseil-etat.fr/fr/arianeweb/CE/decision/2025-10-07/494300 Accessed 2026-11-11.

[84] X-Mode Social and Outlogic, decision and order. Federal Trade Commission, 2024-04-11. https://www.ftc.gov/system/files/ftc_gov/pdf/X-ModeSocialDecisionandOrder.pdf Accessed 2026-11-11.

[85] InMarket Media, decision and order. Federal Trade Commission, 2024-04-29. https://www.ftc.gov/system/files/ftc_gov/pdf/InMarketMedia-DecisionandOrder.pdf Accessed 2026-11-11.

[86] Gravy Analytics and Venntel, final consent package, docket C-4810. Federal Trade Commission, 2025-01-13. https://www.ftc.gov/system/files/ftc_gov/pdf/212_3035_-_gravy_analytics_final_consent_package_without_signatures.pdf Accessed 2026-11-11.

[87] Publicis-LiveRamp, In the Matter of, C-4841 / file 261 0115. Federal Trade Commission, 2026-10-19. https://www.ftc.gov/legal-library/browse/cases-proceedings/261-0115-publicis-liveramp-matter Accessed 2026-11-11.

[88] Order to Maintain Assets, Publicis Groupe S.A. and LiveRamp Holdings, Inc.. Federal Trade Commission, 2026-10-19. https://www.ftc.gov/system/files/ftc_gov/pdf/2610115publicis-liverampoma.pdf Accessed 2026-11-11.

[89] Analysis of Agreement Containing Consent Orders to Aid Public Comment, Publicis / LiveRamp. Federal Trade Commission, 2026-10-19. https://www.ftc.gov/system/files/ftc_gov/pdf/2610115publicis-liverampaapc.pdf Accessed 2026-11-11.

[90] Baker v. Yahoo Inc., No. 25cv2797, Opinion and Order, docket 103. US District Court for the Southern District of New York, 2026-08-26. https://law.justia.com/cases/federal/district-courts/new-york/nysdce/1:2025cv02797/639983/103/ Accessed 2026-11-11.

[91] Trusted Match Protocol specification, AdCP 3.2.8 documentation. AgenticAdvertising.org. https://docs.adcontextprotocol.org/dist/docs/3.2.8/trusted-match/specification Accessed 2026-11-11.

[92] adcp-go, tmpclient/client.go at a6c741b. AgenticAdvertising.org / AdCP project. https://github.com/adcontextprotocol/adcp-go/blob/a6c741b2b7387412bf72e2992c31b3402acf8986/tmpclient/client.go Accessed 2026-11-11.

[93] adcp-go commit metadata, a6c741b. adcontextprotocol/adcp-go project, 2026-09-11. https://github.com/adcontextprotocol/adcp-go/commit/a6c741b2b7387412bf72e2992c31b3402acf8986 Accessed 2026-11-11.

[94] adcp-go publisher-client identity and activation tests at a6c741b. adcontextprotocol/adcp-go project. https://raw.githubusercontent.com/adcontextprotocol/adcp-go/a6c741b2b7387412bf72e2992c31b3402acf8986/tmpclient/client_test.go Accessed 2026-11-11.

[95] adcp-go, targeting/identity_engine.go at a6c741b. AgenticAdvertising.org / AdCP project. https://github.com/adcontextprotocol/adcp-go/blob/a6c741b2b7387412bf72e2992c31b3402acf8986/targeting/identity_engine.go Accessed 2026-11-11.

[96] adcp-go, targeting/tmpxdecoders/uid2.go at a6c741b. AgenticAdvertising.org / AdCP project. https://github.com/adcontextprotocol/adcp-go/blob/a6c741b2b7387412bf72e2992c31b3402acf8986/targeting/tmpxdecoders/uid2.go Accessed 2026-11-11.

[97] AAMP 3.0, OpenProposal v3.0-draft-1, object model specification. IAB Tech Lab. https://github.com/IABTechLab/OpenProposal/blob/v3.0-draft-1/spec/openproposal-3.0.md Accessed 2026-11-11.

[98] OpenProposal repository and draft status. IAB Tech Lab. https://github.com/IABTechLab/OpenProposal Accessed 2026-11-11.

[99] AAMP Agentic Advertising Management Protocols and reference SDK demonstration. IAB Tech Lab, 2026-02-26. https://iabtechlab.com/standards/aamp-agentic-advertising-management-protocols/ Accessed 2026-11-11.

[100] Agentic Audiences in OpenRTB, community extension. IAB Tech Lab. https://github.com/InteractiveAdvertisingBureau/openrtb/blob/main/extensions/community_extensions/agentic-audiences.md Accessed 2026-11-11.

[101] Operationalizing Agentic Audiences and AI-driven targeting. IAB Tech Lab, 2026-08-04. https://iabtechlab.com/operationalizing-agentic-audiences-and-ai-driven-targeting/ Accessed 2026-11-11.

[102] IAB Tech Lab agentic advertising and AI initiatives. IAB Tech Lab, 2026-01-07. https://iabtechlab.com/standards/agentic-advertising-and-ai/ Accessed 2026-11-11.

[103] Prebid.js PR 14626, Agentic Audiences RTD module, source and tests. Prebid.org. https://github.com/prebid/Prebid.js/pull/14626/files Accessed 2026-11-11.

[104] Prebid.js PR 14626, initial release of the Agentic Audiences RTD module. Prebid.org, 2026-03-22. https://github.com/prebid/Prebid.js/pull/14626 Accessed 2026-11-11.

[105] Agentic Audiences Real-Time Data Module. Prebid.org. https://docs.prebid.org/dev-docs/modules/agenticAudienceAdapter.html Accessed 2026-11-11.

[106] PubMatic and Butler/Till Launch AdCP-Enabled Agentic AI Campaign Across CTV. PubMatic, 2025-12-15. https://investors.pubmatic.com/news-releases/news-release-details/pubmatic-and-butlertill-launch-adcp-enabled-agentic-ai-campaign Accessed 2026-11-11.

[107] Optable and PubMatic Partner to Advance Agentic Audience Discovery and Real-Time Activation Across PubMatic AgenticOS. Optable, 2026-03-12. https://www.optable.co/newsroom/optable-and-pubmatic-partner Accessed 2026-11-11.

[108] adcp-go, targeting/identityagent/handler.go at a6c741b. AgenticAdvertising.org / AdCP project. https://github.com/adcontextprotocol/adcp-go/blob/a6c741b2b7387412bf72e2992c31b3402acf8986/targeting/identityagent/handler.go Accessed 2026-11-11.

[109] Enterprise 2.29 – Setup Household Identity. Blockgraph. https://docs.blockgraph.co/user-guides/2.29/core-functionality/identity/setup-household-identity.md Accessed 2026-11-11.

[110] Enterprise 2.29 – Identity Universe Match. Blockgraph. https://docs.blockgraph.co/user-guides/2.29/core-functionality/identity/identity-universe-match.md Accessed 2026-11-11.

[111] Enterprise 2.29 – Outputs. Blockgraph. https://docs.blockgraph.co/user-guides/2.29/data-format-guidelines/outputs.md Accessed 2026-11-11.

[112] Blockgraph Kantar Partnership Elevates Convergent TV Measurement Capabilities. Kantar and Blockgraph, 2022-11-01. https://upg-cd-ncus.kantar.com/north-america/Company-news/Blockgraph-Kantar-Partnership-Elevates-Convergent-TV-Measurement-Capabilities Accessed 2026-11-11.

[113] Let’s Talk About Match Rates. Jason Manningham, Blockgraph, 2022-11-07. https://www.blockgraph.co/newsroom/lets-talk-about-match-rates Accessed 2026-11-11.

[114] Enterprise 2.29 – ID Resolution. Blockgraph. https://docs.blockgraph.co/user-guides/2.29/core-functionality/measurement-data-and-reports/id-resolution.md Accessed 2026-11-11.

[115] Enterprise 2.29 – Manual Identity Upload. Blockgraph. https://docs.blockgraph.co/user-guides/2.29/core-functionality/identity/setup-household-identity/manual-identity-upload.md Accessed 2026-11-11.

[116] Release 2.29. Blockgraph, 2026-08. https://docs.blockgraph.co/release-notes/release-2.29.md Accessed 2026-11-11.

[117] Digital Graph | Experian Marketing. Experian. https://www.experian.com/marketing/consumer-sync/identity-resolution/digital-graph Accessed 2026-11-11.

[118] Identity Spine Boosts Cookieless Reach. MiQ. https://www.wearemiq.com/resources/blog/how-miq-identity-spine-boosted-cookieless-reach-by-64-with-experian-graph Accessed 2026-11-11.

[119] Digital Graph – Connect confidently. Experian Marketing Services, 2025. https://www.experian.com/content/dam/marketing/na/assets/ems/marketing-services/documents/product-sheets/Graph_product_sheet.pdf Accessed 2026-11-11.

[120] Offline Graph. Experian Marketing Services. https://www.experian.com/marketing/consumer-sync/identity-resolution/offline-graph Accessed 2026-11-11.

[121] Digital Graph and Marketing Attributes – Powering your data-driven business. Experian Marketing Services, 2024. https://www.experian.com/content/dam/marketing/na/assets/ems/marketing-services/documents/product-sheets/DigitalGraph-and-MarketingAttributes_product_sheet.pdf Accessed 2026-11-11.

[122] How Experian and Snowflake are creating more connectivity in programmatic. Experian Marketing Services, 2024-09-18. https://www.experian.com/blogs/marketing-forward/experian-and-snowflake-creating-more-connectivity-in-programmatic-advertising/ Accessed 2026-11-11.

[123] Sep 10, 2026: Snowflake Data Clean Rooms updates. Snowflake, 2026-09-10. https://docs.snowflake.com/en/release-notes/2026/other/2026-09-10-dcr Accessed 2026-11-11.

[124] Jul 23, 2026: Snowflake Data Clean Rooms updates. Snowflake, 2026-07-23. https://docs.snowflake.com/en/release-notes/2026/other/2026-07-23-dcr Accessed 2026-11-11.

[125] U.S. Consumer Data Privacy Notice. Experian, 2026-06-24. https://www.experian.com/privacy/us-consumer-data-privacy-policy Accessed 2026-11-11.

[126] Tapad Global Privacy Notice. Tapad, Inc., 2026-06-24. https://www.tapad.com/global-privacy-notice Accessed 2026-11-11.

[127] Placement settings — Adobe Advertising DSP. Adobe, 2026-10-02. https://experienceleague.adobe.com/en/docs/advertising/dsp/campaign-management/placements/placement-settings Accessed 2026-11-11.

[128] User Rights Propagation. ID5, 2026-03-31. https://wiki.id5.io/docs/user-rights-propagation Accessed 2026-11-11.

[129] Addressability Benefits of a Data Collective. Sovrn and ID5. https://www.sovrn.com/wp-content/uploads/ID5_Sovrn_Case_Study_Final.pdf Accessed 2026-11-11.

[130] Sovrn and ID5: Addressability at Scale. Sovrn, 2022-09-21. https://www.sovrn.com/blog/case-study-addressability-at-scale/ Accessed 2026-11-11.

[131] ID5 Platform Privacy Policy, v4.4. ID5 Technology Limited, 2026-10-19. https://id5.io/legal/agreements/v4.4%20ID5%20Platform%20Privacy%20Policy%2020261019.pdf Accessed 2026-11-11.

[132] ID5 Acquires TrueData. ID5, 2025-11-04. https://id5.io/news/id5-acquires-truedata Accessed 2026-11-11.

[133] Authenticated Traffic Solution. LiveRamp, 2026-09-22. https://docs.liveramp.com/identity/en/authenticated-traffic-solution.html Accessed 2026-11-11.

[134] ATS Security and Pseudonymization. LiveRamp, 2025-08-04. https://docs.liveramp.com/privacy-manager/en/ats-security-and-pseudonymization.html Accessed 2026-11-11.

[135] Customize ad workflows with AWS Elemental MediaTailor Monetization Functions. Kevin Nguyen, Amazon Web Services, 2026-05-28. https://aws.amazon.com/blogs/media/customize-ad-workflows-with-aws-elemental-mediatailor-monetization-functions/ Accessed 2026-11-11.

[136] Form 8-K, furnished Exhibit 99.1, fourth quarter and fiscal 2023 release. LiveRamp Holdings, Inc., 2023-05-24. https://investors.liveramp.com/node/18881/html Accessed 2026-11-11.

[137] 5. Store the Identity Envelope With a Timestamp. LiveRamp, 2026-05-04. https://developers.liveramp.com/authenticatedtraffic-api/docs/5-store-the-identity-envelope Accessed 2026-11-11.

[138] Opt-Out Requirements and Methodologies. LiveRamp, 2026-08-31. https://docs.liveramp.com/privacy-manager/en/opt-out-requirements-and-methodologies.html Accessed 2026-11-11.

[139] Retrieving Opt-Outs and Deletions from LiveRamp. LiveRamp, 2026-02-25. https://docs.liveramp.com/connect/en/retrieving-opt-outs-and-deletions-from-liveramp.html Accessed 2026-11-11.

[140] DSPs, and DMPs RampID Services Country-Specific Terms. LiveRamp. https://liveramp.com/privacy/dsp-dmp-country-specific-terms Accessed 2026-11-11.

[141] Form 8-K, Item 2.01, completion of merger. LiveRamp Holdings, Inc., 2026-11-02. https://www.sec.gov/Archives/edgar/data/733269/000110465926121353/tm2628592d27_8k.htm Accessed 2026-11-11.

[142] Overview of UID2 participants. UID2 documentation, 2026-11-11. https://unifiedid.com/docs/overviews/participants-overview Accessed 2026-11-11.

[143] API permissions. UID2 documentation, 2026-11-11. https://unifiedid.com/docs/getting-started/gs-permissions Accessed 2026-11-11.

[144] FAQs. UID2 documentation, 2026-11-11. https://unifiedid.com/docs/getting-started/gs-faqs Accessed 2026-11-11.

[145] Documentation updates. UID2 documentation. https://unifiedid.com/docs/ref-info/updates-doc Accessed 2026-11-11.

[146] POST /v3/identity/map. UID2 documentation, 2026-11-11. https://unifiedid.com/docs/endpoints/post-identity-map Accessed 2026-11-11.

[147] Identity Report 2024, USA. The Trade Desk Intelligence, 2024. https://assets.thetradedesk.com/assets/documents/TTDI_IdentityReport_2024_V6.pdf Accessed 2026-11-11.

[148] ‘Cookie-less’ identification for/against privacy?. Ido Sivan-Sevilla, Patrick Parham and Lee McGuigan, 2025-08-06. https://policyreview.info/pdf/policyreview-2025-3-2025.pdf Accessed 2026-11-11.

[149] UID2 user opt-out. Unified ID 2.0. https://unifiedid.com/docs/getting-started/gs-opt-out Accessed 2026-11-11.

[150] UID2 Administrator Privacy Policy. The Trade Desk, 2025-04-01. https://transparentadvertising.com/privacy Accessed 2026-11-11.

[151] Blocking Identity Resolution Infrastructure. WebKit, 2026-10-19. https://webkit.org/blog/18438/blocking-identity-resolution-infrastructure/ Accessed 2026-11-11.

[152] Client-side integration guide for Prebid.js | Unified ID 2.0. Unified ID 2.0 documentation. https://unifiedid.com/docs/guides/integration-prebid-client-side Accessed 2026-11-11.

[153] Environments | Unified ID 2.0. Unified ID 2.0. https://unifiedid.com/docs/getting-started/gs-environments Accessed 2026-11-11.

[154] Roku + The Trade Desk UID2: Key takeaways. Roku Advertising, 2024-08-01. https://advertising.roku.com/learn/resources/roku-the-trade-desk-uid-2-key-takeaways Accessed 2026-11-11.

[155] A Publisher Didn’t Get Its UID2 Setup Right. The Trade Desk Didn’t Notice. What Went Wrong?. Anthony Vargas, AdExchanger, 2026-05-07. https://www.adexchanger.com/identity/a-publisher-didnt-get-its-uid2-setup-right-the-trade-desk-didnt-notice-what-went-wrong/ Accessed 2026-11-11.

[156] Secure signals providers. Google Ad Manager. https://support.google.com/admanager/answer/14750072?hl=en Accessed 2026-11-11.

[157] Integrate with Yahoo ConnectID: Google Secure Signals. Yahoo. https://help.yahooinc.com/connect-id/docs/google-secure-signals Accessed 2026-11-11.

[158] Mediavine’s First-Party Data Toolkit, Grow, Integrates with Yahoo ConnectID. Mediavine, 2021-12-02. https://www.mediavine.com/newsroom/mediavines-first-party-data-toolkit-grow-integrates-with-yahoo-connectid/ Accessed 2026-11-11.

[159] The Yahoo DSP and LiveRamp Announce Full Integration of Yahoo ConnectID Across Major Publishers. Yahoo and LiveRamp, 2024-07-17. https://www.yahooinc.com/press/the-yahoo-dsp-and-liveramp-announce-full-integration-of-yahoo-connectid-across-major-publishers Accessed 2026-11-11.

[160] ConnectID Service Agreement. Yahoo Ad Tech LLC. https://www.yahooinc.com/connectid-service-agreement Accessed 2026-11-11.

[161] Truthset And Yahoo DSP Partner To Help Advertisers Improve Profitability With Data Rated Audiences. Truthset, 2026-07-23. https://truthset.com/news/truthset-yahoo-dsp-integration/ Accessed 2026-11-11.

[162] Yahoo & Merkle Strengthen Identity-Based Targeting, Performance & Measurement for Advertisers. Yahoo, with Merkle / Merkury representative, 2021-10-28. https://unified.www.yahooinc.com/press/yahoo-merkle-strengthen-identity-based-targeting-performance-measurement-for-advertisers Accessed 2026-11-11.

[163] Integrate with Yahoo ConnectID: Magnite’s Control Center. Yahoo. https://help.yahooinc.com/connect-id/docs/integrate-with-yahoo-connectid-magnites-control-center Accessed 2026-11-11.

[164] Integrate with Yahoo ConnectID: Prebid Method. Yahoo. https://help.yahooinc.com/connect-id/docs/integrate-with-connectid-prebid-method Accessed 2026-11-11.

[165] Integrate with Yahoo ConnectID: Server-to-Server Method. Yahoo. https://help.yahooinc.com/connect-id/docs/integrate-with-connectid-server-to-server-method Accessed 2026-11-11.

[166] Household Identity. Blockgraph. https://www.blockgraph.co/capabilities/household-identity Accessed 2026-11-11.

[167] Local TV and Video Innovation Showcase, Spectrum Reach Multiscreen Attribution. Spectrum Reach contribution, CIMM and TVB showcase, 2025-02. https://www.tvb.org/wp-content/uploads/2025/02/CIMM-TVB-Local-TV-and-Video-Innovation-Showcase.pdf Accessed 2026-11-11.

[168] NBCUniversal Measurement Framework Look Book, TVSquared response. TVSquared, published by NBCUniversal, 2022-02. https://together.nbcuni.com/wp-content/uploads/sites/3/2022/02/NBCU-Measurement-Framework-Look-Book-V1-Feb-2022-1.pdf Accessed 2026-11-11.

[169] Kochava Product & Partnership Updates Bulletin – October 2023. Jeff Richardson, Kochava, 2023-10-25. https://www.kochava.com/blog/kochava-product-partnership-updates-bulletin-october-2023/ Accessed 2026-11-11.

[170] Website Privacy Policy. Blockgraph, LLC, 2021-01-21. https://www.blockgraph.co/privacy Accessed 2026-11-11.

[171] Can Comcast’s Blockgraph Bring Data Matching And Crypto Tech To TV?. James Hercher interviews Blockgraph GM Jason Manningham, AdExchanger, 2020-02-27. https://www.adexchanger.com/on-tv-and-video/can-comcasts-blockgraph-bring-data-matching-and-crypto-tech-to-tv/ Accessed 2026-11-11.

[172] Release 3.0.0. Blockgraph, 2026-10. https://docs.blockgraph.co/release-notes/release-3.0.0.md Accessed 2026-11-11.

[173] How Experian’s digital graph tackles next-generation signal-agnostic fragmentation. Experian Marketing Services, 2026-10-20. https://www.experian.com/blogs/marketing-forward/experian-digital-graph-next-generation-signal-fragmentation-advertising/ Accessed 2026-11-11.

[174] How accurate third-party data leads the way for advertisers. Experian Marketing Services, 2026-10-20. https://www.experian.com/blogs/marketing-forward/how-accurate-third-party-data-leads-the-way-for-advertisers/ Accessed 2026-11-11.

[175] ID5 ID for Header Bidding. Prebid.org, ID5 module documentation. https://docs.prebid.org/dev-docs/modules/userid-submodules/id5.html Accessed 2026-11-11.

[176] DSP: ID5 ID decryption. ID5. https://wiki.id5.io/docs/dsp Accessed 2026-11-11.

[177] Read First: Privacy Policy Requirements. ID5, 2026-02-01. https://wiki.id5.io/docs/privacy-policy-requirements Accessed 2026-11-11.

[178] Adform supercharges site tracking with ID5 partnership. Adform, 2025-05-28. https://site.adform.com/resources/newsroom/adform-supercharges-site-tracking-with-id5-partnership/ Accessed 2026-11-11.

[179] id5io/id5-api.js / README.md at bd6c891f2c63. ID5, 2025-04-18. https://raw.githubusercontent.com/id5io/id5-api.js/bd6c891f2c63416ae6dbca162dbd887c6f7bda07/README.md Accessed 2026-11-11.

[180] id5io/id5-api.js / README.md at 576fd16780a9. ID5, 2026-11-06. https://raw.githubusercontent.com/id5io/id5-api.js/576fd16780a990df3025e39c43d89a052db15f19/README.md Accessed 2026-11-11.

[181] RampID Methodology. LiveRamp. https://docs.liveramp.com/identity/en/rampid-methodology.html Accessed 2026-11-11.

[182] ATS and Privacy Regulations. LiveRamp, 2026-11-03. https://docs.liveramp.com/privacy-manager/en/gdpr%2C-ccpa%2C-and-ats.html Accessed 2026-11-11.

[183] SSPs – Sidecar and RampID Services Country-Specific Terms. LiveRamp. https://liveramp.com/privacy/authenticated-traffic-solution-country-specific-terms-joint-controller Accessed 2026-11-11.

[184] Configuring Prebid for ATS. LiveRamp. https://developers.liveramp.com/authenticatedtraffic-api/v1.0/docs/configuring-prebidjs-for-ats Accessed 2026-11-11.

[185] Announcement: “IdentityLink” is Now “RampID”. LiveRamp, 2021-05-24. https://docs.liveramp.com/connect/en/announcement---identitylink--is-now--rampid---5-24-21-.html Accessed 2026-11-11.

[186] DSP integration guide. Unified ID 2.0. https://unifiedid.com/docs/guides/dsp-guide Accessed 2026-11-11.

[187] TD Data Processing Addendum for UID2 Agreement. The Trade Desk, 2025-04. https://assets.thetradedesk.com/assets/documents/UID2-TTD-Global-Client-Data-Processing-Addendum-As-of-April-2025.pdf Accessed 2026-11-11.

[188] Licenses to generate and maintain UID2 identifiers. The Trade Desk, 2024-10-01. https://assets.thetradedesk.com/assets/documents/UID2-LICENSES-10-01-24.pdf Accessed 2026-11-11.

[189] Yahoo ConnectID for Header Bidding. Prebid project. https://docs.prebid.org/dev-docs/modules/userid-submodules/yahoo.html Accessed 2026-11-11.

[190] Verizon Media Launches Unified ID Solution “Verizon Media ConnectID”. Verizon Communications via GlobeNewswire, 2020-12-01. https://www.globenewswire.com/news-release/2020/12/01/2137657/0/en/Verizon-Media-Launches-Unified-ID-Solution-Verizon-Media-ConnectID.html Accessed 2026-11-11.

[191] Prebid.js 7.51.0 — Yahoo ConnectID module. Prebid.js project, 2023-05-25. https://raw.githubusercontent.com/prebid/Prebid.js/7.51.0/modules/connectIdSystem.js Accessed 2026-11-11.

[192] Prebid.js 7.52.0 release. Prebid.js project, 2023-06-01. https://api.github.com/repos/prebid/Prebid.js/releases/tags/7.52.0 Accessed 2026-11-11.

[193] Prebid.js 7.52.0 — Yahoo ConnectID module. Prebid.js project, 2023-06-01. https://raw.githubusercontent.com/prebid/Prebid.js/7.52.0/modules/connectIdSystem.js Accessed 2026-11-11.

[194] Integrate with Yahoo ConnectID: JavaScript Method. Yahoo. https://help.yahooinc.com/connect-id/docs/integrate-with-connectid-javascript-method Accessed 2026-11-11.

[195] Prebid.js Yahoo ConnectID module, commit c306d57d. Prebid.js project, 2026-03-07. https://raw.githubusercontent.com/prebid/Prebid.js/c306d57d135984c81c5f2e640e37be91f469b793/modules/connectIdSystem.js Accessed 2026-11-11.

[196] Yahoo Identity Solutions Continues Momentum, Extends Publisher Availability to Provide Publishers Better ROI on their Recognized Audiences. Yahoo, with statements from APS, InMobi, Magnite and PubMatic, 2024-10-17. https://unified.www.yahooinc.com/press/yahoo-identity-solutions-continues-momentum-extends-publisher-availability-to-provide-publishers-better-roi-on-their-recognized-audiences Accessed 2026-11-11.

[197] Yahoo DSP Launches Conversion API to Enhance Campaign Measurement and Optimization. Yahoo, 2025-04-28. https://unified.www.yahooinc.com/press/yahoo-dsp-launches-conversion-api-to-enhance-campaign-measurement-and-optimization Accessed 2026-11-11.

[198] What Strong Data and Identity Actually Do For Your DSP Campaigns. Yahoo, 2026-10-01. https://www.yahooinc.com/blog/what-strong-data-and-identity-actually-do-for-your-dsp-campaigns Accessed 2026-11-11.

[199] Integrating with Yahoo ConnectID. Yahoo. https://help.yahooinc.com/connect-id/docs/integrating-with-yahoo-connectid Accessed 2026-11-11.

[200] ConnectID Opt-out Portal. Yahoo. https://connectid.yahoo.com/ Accessed 2026-11-11.

[201] Create an Easy Opt-In Opt-out Toggle. Yahoo. https://help.yahooinc.com/connect-id/docs/create-an-easy-opt-in-opt-out-toggle Accessed 2026-11-11.

[202] prebid/prebid.github.io / dev-docs/modules/userid-submodules/yahoo.md at 6439acd8604b. Prebid project, 2025-04-04. https://raw.githubusercontent.com/prebid/prebid.github.io/6439acd8604bbbc6d52f69b318d5c830a6a2222e/dev-docs/modules/userid-submodules/yahoo.md Accessed 2026-11-11.

[203] The UID2 Operator | Unified ID 2.0. Unified ID 2.0. https://unifiedid.com/docs/ref-info/ref-operators-public-private Accessed 2026-11-11.

[204] Measure the impact of your CTV campaigns. Criteo. https://help.criteo.com/kb/guide/en/measure-the-impact-of-your-ctv-campaigns-VGx2VMXDMD/Steps/6070682 Accessed 2026-11-11.

[205] Activation Guide for Advertisers – Advertiser Support. Permutive, 2023-01-13. https://advertiser.permutive.com/hc/en-us/articles/7319252685724--Activation-Guide-for-Advertisers Accessed 2026-11-11.

[206] Explainer: What Are Publisher Cohorts? - Permutive. Permutive (excerpt credited to AdMonsters). https://permutive.com/explainer-what-are-publisher-cohorts Accessed 2026-11-11.

[207] TelevisaUnivision and Permutive use Google's publisher provided signals to scale first-party data. Google Ad Manager. https://admanager.google.com/home/success-stories/televisaunivision-and-permutive-use-googles-publisher-provided-signals-to-scale-first-party-data/ Accessed 2026-11-11.

[208] First-Party Data Media Buying with Permutive - OpenX. OpenX, 2022-11-03. https://www.openx.com/press-releases/permutive-first-party-data-media-buying/ Accessed 2026-11-11.

[209] ID-Free Predictive Audiences. Proximic by Comscore. https://www.proximic.com/home/Programmatic-Targeting-Solutions/Predictive-Audiences Accessed 2026-11-11.

[210] Ogury Launches Ogury One, A New Advertising Platform To Address Signal Loss. Ogury, 2025-04-08. https://www.ogury.com/ogury-launches-ogury-one-a-new-advertising-platform-to-address-signal-loss/ Accessed 2026-11-11.

[211] EUID areas of coverage | European Unified ID. European Unified ID, 2024-07-18. https://euid.eu/docs/ref-info/ref-areas-of-coverage Accessed 2026-11-11.

[212] Enrich customer profiles with identity data from LiveRamp (preview). Microsoft, 2026-07-14. https://learn.microsoft.com/en-us/dynamics365/customer-insights/data/enrichment-liveramp Accessed 2026-11-11.

[213] About the customer matching process - Google Ads Help. Google. https://support.google.com/google-ads/answer/7474263?hl=en Accessed 2026-11-11.

[214] Publicis to acquire Lotame. Publicis Groupe, 2025-03-06. https://www.publicisgroupe.com/en/news/press-releases/publicis-to-acquire-lotame-the-world-s-leading-independent-end-to-end-data-solution Accessed 2026-11-11.

[215] Consolidated Financial Statements 2025, Note 3.1. Publicis Groupe, 2026. https://documents.publicisgroupe.com/resultats2025/comptes-consolides-2025-12-ENvdef.pdf Accessed 2026-11-11.

[216] Publicis Groupe completes LiveRamp acquisition and describes its packaged offering. Publicis Groupe, 2026-11-02. https://www.publicisgroupe.com/en/news/press-releases/publicis-groupe-completes-liveramp-acquisition-launches-the-industry-s-first-plug-and-play-ai-growth-engine-that-clients-fully-own-and-control Accessed 2026-11-11.

[217] Staying Focused on Identity. ID5, 2026-10-01. https://id5.io/news/staying-focused-on-identity Accessed 2026-11-11.

[218] (Author's work) Evidence after the independent cell review. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/product-admission.json Accessed 2026-11-11.

[219] Privacy and Consent Handling in Live Connect Hub. LiveIntent / Zeta Global. https://knowledgebase.zetaglobal.com/liveintent/privacy-and-consent-handling-in-live-connect-hub Accessed 2026-11-11.

[220] Update on Plans for Privacy Sandbox Technologies. Google, 2025-10-17. https://privacysandbox.google.com/blog/update-on-plans-for-privacy-sandbox-technologies Accessed 2026-11-11.

[221] Initiate Cookie Sync to ID5 | ID5 Knowledge Base. ID5. https://wiki.id5.io/docs/initiate-cookie-sync-to-id5 Accessed 2026-11-11.

[222] RampID Mapping Files. LiveRamp. https://docs.liveramp.com/identity/en/rampid-mapping-files.html Accessed 2026-11-11.

[223] id5io/id5-api.js / README.md at 45f4bf0d6866. ID5, 2025-10-15. https://raw.githubusercontent.com/id5io/id5-api.js/45f4bf0d686623e743645b10c1b9dbf348d5b7d4/README.md Accessed 2026-11-11.

[224] IABTechLab/uid2docs / docs/guides/integration-javascript-client-side.md at d2169d3f3098. Unified ID 2.0 documentation, 2025-03-10. https://raw.githubusercontent.com/IABTechLab/uid2docs/d2169d3f3098ca19acc504fa757d082ecd1efed5/docs/guides/integration-javascript-client-side.md Accessed 2026-11-11.

[225] IABTechLab/uid2docs / docs/guides/integration-javascript-client-side.md at d133f5faf05d. Unified ID 2.0 documentation, 2025-07-08. https://raw.githubusercontent.com/IABTechLab/uid2docs/d133f5faf05d460e08ae7997017e25decff15920/docs/guides/integration-javascript-client-side.md Accessed 2026-11-11.

[226] IABTechLab/uid2docs / docs/guides/integration-javascript-client-side.md at 79552e551fc8. Unified ID 2.0 documentation, 2026-10-13. https://raw.githubusercontent.com/IABTechLab/uid2docs/79552e551fc8aa2dcbebc5a1b40ef3ebc89413b6/docs/guides/integration-javascript-client-side.md Accessed 2026-11-11.

[227] Identity API & HouseholdID. LtvAdx. https://www.ltvadx.com/docs/identity-api Accessed 2026-11-11.

[228] Viewer Privacy and Consent on LtvAdx. LtvAdx. https://www.ltvadx.com/help/privacy-and-consent Accessed 2026-11-11.

[229] CTV & Linear TV Ad Exchange for the US. LtvAdx. https://www.ltvadx.com/us Accessed 2026-11-11.

[230] Pricing & Fees. LtvAdx. https://www.ltvadx.com/pricing Accessed 2026-11-11.

[231] CTV Household Identity & Targeting | HouseholdID. LtvAdx. https://www.ltvadx.com/product/identity Accessed 2026-11-11.

[232] (Author's work) Evgeny Popov • biography and standards affiliations. Evgeny Popov / No Fluff Advisory. https://nofluffadvisory.com/about/ Accessed 2026-11-11.

[233] (Author's work) Frozen product register and declared interests. Evgeny Popov / No Fluff Advisory, 2026-11-11. ./data/product-register.json Accessed 2026-11-11.

[234] About RampIDs. LiveRamp. https://developers.liveramp.com/rampid-api/reference/rampids Accessed 2026-11-11.

[235] Known Identity Resolution. LiveRamp. https://docs.liveramp.com/identity/en/known-identity-resolution.html Accessed 2026-11-11.

[236] Over 300 SmartPMPs Powered by Syndicated TransUnion Audiences Now Available | Audigent | The Leading Data Activation, Curation and Identity Platform. Audigent. https://audigent.com/blog/transunion-audigent-smartpmps/ Accessed 2026-11-11.

[237] Hadron ID from Audigent for Header Bidding. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/hadron.html Accessed 2026-11-11.

[238] About secure signals for curation partners (Beta) - Google Ad Manager Help. Google Ad Manager. https://support.google.com/admanager/answer/15559440?hl=en Accessed 2026-11-11.

[239] Prebid.js User ID module, GDPR, permissions and opt-out. Prebid.org. https://docs.prebid.org/dev-docs/modules/userId Accessed 2026-11-11.

[240] Analyze data in a collaboration - AWS Clean Rooms. Amazon Web Services. https://docs.aws.amazon.com/clean-rooms/latest/userguide/analyze-data.html Accessed 2026-11-11.

[241] Microsoft Curate - Curated Deals | Microsoft Learn. Microsoft. https://learn.microsoft.com/en-us/xandr/curate/curated-deals Accessed 2026-11-11.

[242] Consent Management Platform | Products | OneTrust. OneTrust. https://www.onetrust.com/products/consent-management/ Accessed 2026-11-11.

[243] AbiliTec API: The Basics. LiveRamp. https://developers.liveramp.com/abilitec-api/reference/the-basics Accessed 2026-11-11.

[244] Streamlit Applications. Zeta Global. https://knowledgebase.zetaglobal.com/kb/zeta-customer-intelligence-streamlit-application Accessed 2026-11-11.

[245] TransUnion and Epsilon Collaborate to Bring ID-Agnostic Audiences to Connected TV and Streaming Audio Campaigns. TransUnion and Epsilon, 2022-06-07. https://newsroom.transunion.com/transunion-and-epsilon-collaborate-to-bring-id-agnostic-audiences--to-connected-tv-and-streaming-audio-campaigns/ Accessed 2026-11-11.

[246] TransUnion Introduces TruAudience Marketing Solutions. TransUnion, 2021-08-24. https://newsroom.transunion.com/transunion-introduces-truaudience-marketing-solutions--to-power-privacy-centric-identity-and-data-capabilities-for-omnichannel-advertising/ Accessed 2026-11-11.

[247] Comcast Advertising and TransUnion Partner to Enable Advertisers to Create and Activate Audiences for Addressable TV Campaigns Across Linear and Streaming. TransUnion and Comcast Advertising, 2024-06-06. https://newsroom.transunion.com/comcast-advertising-and-transunion-partner-to-enable-advertisers-to-create-and-activate-audiences-for-addressable-tv-campaigns-across-linear-and-streaming/ Accessed 2026-11-11.

[248] Soaring Above The Rest: ANA Lands Big Wins With Ogury’s Personas – Ogury. Ogury. https://www.ogury.com/soaring-above-the-rest-ana-lands-big-wins-with-ogurys-personas/ Accessed 2026-11-11.

[249] Parrable ID for Header Bidding. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/parrable.html Accessed 2026-11-11.

[250] Criteo ID for Exchanges. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/criteo.html Accessed 2026-11-11.

[251] Curated Audiences. IAB Tech Lab. https://iabtechlab.com/sda/ Accessed 2026-11-11.

[252] ID Provenance Added to OpenRTB. IAB Tech Lab, 2024-09-23. https://iabtechlab.com/id-provenance-added-to-openrtb/ Accessed 2026-11-11.

[253] Global Privacy Protocol. IAB Tech Lab. https://iabtechlab.com/gpp/ Accessed 2026-11-11.

[254] All You Need to Know About the Transition to TCF v2.3. IAB Europe, 2025-12-15. https://iabeurope.eu/all-you-need-to-know-about-the-transition-to-tcf-v2-3/ Accessed 2026-11-11.

[255] Prebid.js 11.14.0 release record. Prebid.org, 2026-05-27. https://api.github.com/repos/prebid/Prebid.js/releases/tags/11.14.0 Accessed 2026-11-11.

[256] ID5 User ID Module for Prebid Server Java. ID5. https://wiki.id5.io/docs/id5-user-id-module-for-prebid-server Accessed 2026-11-11.

[257] Basic requirements for ATS. LiveRamp. https://docs.liveramp.com/privacy-manager/en/basic-requirements.html Accessed 2026-11-11.

[258] SharedID for Header Bidding. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/sharedid.html Accessed 2026-11-11.

[259] Between Expansion and Collapse: An Identity Buyer’s Guide. Sable Mi / Coalition for Innovative Media Measurement, 2026-04. https://cimm-us.org/wp-content/uploads/2026/04/CIMM-IDR-Buyers-Guide-Between-Expansion-and-Collapse-April-2026.pdf Accessed 2026-11-11.

[260] LiveIntent nonID for Header Bidding. Prebid.org with LiveIntent. https://docs.prebid.org/dev-docs/modules/userid-submodules/liveintent.html Accessed 2026-11-11.

[261] Services Privacy Notice. Lotame. https://www.lotame.com/privacy/services-privacy-notice/ Accessed 2026-11-11.

[262] CCPA FAQs for Publishers. TripleLift. https://triplelift.com/wp-content/uploads/2020/01/CCPA_FAQs-Publisher.pdf Accessed 2026-11-11.

[263] List of Processors, Partners and Other Third Parties. TripleLift, 2026-04. https://triplelift.com/triplelift-processors-and-other-third-parties/ Accessed 2026-11-11.

[264] Gaming case study: cookieless identity solution drives better performance. Lotame with Advance Local and Rush Street Interactive. https://my.lotame.com/media/download/x1rs9c/Gaming_Rush-Advance_Identity-LDX_Lotame-Case-Study.pdf Accessed 2026-11-11.

[265] Chrome 144  |  Release notes  |  Chrome for Developers. Google, 2026-01-13. https://developer.chrome.com/release-notes/144 Accessed 2026-11-11.

[266] Chrome 152  |  Release notes  |  Chrome for Developers. Google, 2026-08-25. https://developer.chrome.com/release-notes/152 Accessed 2026-11-11.

[267] Chromium commit d37e2f6: Topics feature flag cleanup. Chromium project, 2026-07-31. https://chromium.googlesource.com/chromium/src/%2B/d37e2f6c7d5e31cb140dee0d2c6f54dfa338c54e Accessed 2026-11-11.

[268] Chromium commit 4bcc11c: Stub out Protected Audience renderer APIs. Chromium project, 2026-07-14. https://chromium.googlesource.com/chromium/src/%2B/4bcc11c1788374c99b303132c9d5448fafbc681b Accessed 2026-11-11.

[269] Chromium commit d7a74fd: Attribution Reporting renderer cleanup. Chromium project, 2026-07-14. https://chromium.googlesource.com/chromium/src/%2B/d7a74fd8c6906f9a067a3a26c7a21cc58e0c61db Accessed 2026-11-11.

[270] Intent to Deprecate and Remove: Topics API. Chromium project. https://groups.google.com/a/chromium.org/g/blink-dev/c/_R85yctz4Rs/m/0Mf9ObIdCQAJ Accessed 2026-11-11.

[271] [blink-dev] Intent to Deprecate and Remove: Attribution Reporting API. Chromium project. https://groups.google.com/a/chromium.org/g/blink-dev/c/4K2RRt6VYCQ/m/rnT3F5qHBgAJ Accessed 2026-11-11.

[272] Privacy Sandbox on Android  |  Google for Developers. Google. https://developers.google.com/ad-manager/mobile-ads-sdk/android/next-gen/privacy/sandbox Accessed 2026-11-11.

[273] Android Topics package, API level 37 deprecation notice. Google. https://developer.android.com/reference/android/adservices/topics/package-summary Accessed 2026-11-11.

[274] Advertising ID - Play Console Help. Google. https://support.google.com/googleplay/android-developer/answer/6048248?hl=en-EN Accessed 2026-11-11.

[275] Upcoming AppTrackingTransparency requirements - Latest News - Apple Developer. Apple, 2021-04-20. https://developer.apple.com/news/?id=ecvrtzt2 Accessed 2026-11-11.

[276] Full Third-Party Cookie Blocking and More | WebKit. WebKit, 2020-03-24. https://webkit.org/blog/10218/full-third-party-cookie-blocking-and-more/ Accessed 2026-11-11.

[277] Firefox Rolls Out Total Cookie Protection By Default. Mozilla, 2022-06-14. https://blog.mozilla.org/en/mozilla/firefox-rolls-out-total-cookie-protection-by-default-to-all-users-worldwide/ Accessed 2026-11-11.

[278] prebid/prebid.github.io / dev-docs/modules/userid-submodules/sharedid.md at 44c1b8e99908. Prebid project, 2024-11-28. https://raw.githubusercontent.com/prebid/prebid.github.io/44c1b8e999080f74e6077df22eb0520be82e2ade/dev-docs/modules/userid-submodules/sharedid.md Accessed 2026-11-11.

[279] prebid/prebid.github.io / dev-docs/modules/userid-submodules/lotame.md at 9aaf096a3746. Prebid project, 2025-01-06. https://raw.githubusercontent.com/prebid/prebid.github.io/9aaf096a3746136c37c3f366bbd8a621b2cf8fcd/dev-docs/modules/userid-submodules/lotame.md Accessed 2026-11-11.

[280] Mediavine’s Grow.me First-Party Data Toolkit Integrates with ID5’s Universal ID. Mediavine, 2021-04-22. https://www.mediavine.com/newsroom/mediavines-grow-me-first-party-data-toolkit-integrates-with-id5s-universal-id/ Accessed 2026-11-11.

[281] Trusted Media Brands Privacy Policy. Trusted Media Brands. https://www.trustedmediabrands.com/privacy-policy/ Accessed 2026-11-11.

[282] Update on Plans for Privacy Sandbox Technologies. Google Privacy Sandbox, 2025-10-17. https://privacysandbox.google.com/blog/update-on-plans-for-privacy-sandbox-technologies?rd=1&visit_id=639270403938925233-2028017831 Accessed 2026-11-11.

[283] Chrome 153  |  Release notes  |  Chrome for Developers. Google Chrome. https://developer.chrome.com/release-notes/153?authuser=4 Accessed 2026-11-11.

[284] Get to know the new Topics API for Privacy Sandbox. Google Chrome, 2022-01-25. https://blog.google/products-and-platforms/products/chrome/get-know-new-topics-api-privacy-sandbox/ Accessed 2026-11-11.

[285] Prebid.js 5.0.0 release. Prebid.js project on GitHub, 2021-06-10. https://api.github.com/repos/prebid/Prebid.js/releases/tags/5.0.0 Accessed 2026-11-11.

[286] SharedID for Header Bidding. Prebid. https://docs.prebid.org/identity/sharedid.html Accessed 2026-11-11.

[287] Customer Match Best Practices - Google Ads Help. Google. https://support.google.com/google-ads/answer/10010286?hl=en Accessed 2026-11-11.

[288] Manage customer lists  |  Google Ads API  |  Google for Developers. Google. https://developers.google.com/google-ads/api/docs/remarketing/audience-segments/customer-match/manage?hl=en Accessed 2026-11-11.

[289] Share secure signals with bidders - Authorized Buyers Help. Google. https://support.google.com/authorizedbuyers/answer/10728857?hl=en Accessed 2026-11-11.

[290] Advertiser Audiences and Events Manager Agreement. Amazon Ads. https://advertising.amazon.com/dsp/agreement/advertiserAudience/en Accessed 2026-11-11.

[291] Increase campaign relevance with advertiser hashed audiences | Amazon Ads. Amazon Ads, 2017-06-12. https://advertising.amazon.com/blog/advertiser-audiences Accessed 2026-11-11.

[292] Customer File Custom Audiences | Meta for Developers. Meta. https://developers.facebook.com/docs/marketing-api/audiences/guides/custom-audiences/ Accessed 2026-11-11.

[293] Expanding access to Walmart first-party data across leading platforms. Walmart Connect, 2026-05-28. https://www.walmartconnect.com/resources/articles/2026/expanding-access-to-walmart-first-party-data-across-leading-platforms Accessed 2026-11-11.

[294] Search Ads - Roundel™ Media Designed by Target. Target / Roundel, 2023-05-03. https://roundel.com/what-we-offer/search-ads/ Accessed 2026-11-11.

[295] Kroger Precision Marketing Announces Programmatic Solution with Flexible Controls and Decoupled Pricing. Kroger Precision Marketing, 2023-11-15. https://www.krogerprecisionmarketing.com/knowledge-hub/news/kroger-precision-marketing-announces-programmatic-solution-with-flexible/ Accessed 2026-11-11.

[296] Instacart Debuts Data Hub, a Clean Room Offering for Enhanced Media Performance | Instacart Newsroom. Instacart, 2026-01-06. https://company.instacart.com/pressreleases/instacart-debuts-clean-room-offering Accessed 2026-11-11.

[297] Configure and view clean rooms | Instacart Docs. Instacart. https://docs.instacart.com/data_hub/cleanroom/ Accessed 2026-11-11.

[298] Activation templates | Instacart Docs. Instacart. https://docs.instacart.com/data_hub/templates/activation_templates/overview/ Accessed 2026-11-11.

[299] Roku announced as a launch partner for Google’s new Confidential Publisher Match solution. Roku Advertising, 2026-03-23. https://advertising.roku.com/learn/resources/roku-is-a-launch-partner-in-googles-new-confidential-publisher-match-identity-solution-in-display-video-360 Accessed 2026-11-11.

[300] Create a Customer Match list by uploading a data file - Google Ads Help. Google. https://support.google.com/google-ads/answer/10589050?hl=en Accessed 2026-11-11.

[301] Amazon Marketing Cloud: Discover advertising insights | Amazon Ads. Amazon Ads. https://advertising.amazon.com/solutions/products/amazon-marketing-cloud Accessed 2026-11-11.

[302] Customer List Custom Audiences Terms. Meta. https://www.facebook.com/legal/terms/customaudience Accessed 2026-11-11.

[303] LiveRamp – Advertiser Data Destination Terms. Disney Advertising, 2023-07-26. https://www.disneyadvertising.com/advertiser-data-destination-terms/ Accessed 2026-11-11.

[304] Deterministic Householding for Household Targeting. Viant, 2026-07-10. https://www.viantinc.com/insights/blog/deterministic-householding/ Accessed 2026-11-11.

[305] FreeWheel Identity Network: Smarter Identity for Better Outcomes. FreeWheel. https://www.freewheel.com/insights/videos/identity-network Accessed 2026-11-11.

[306] Aperture Platform: Next-Gen Contextual Targeting. Cadent. https://cadent.tv/wp-content/uploads/2024/05/Cadent_NextGen_Contextual_Targeting.pdf Accessed 2026-11-11.

[307] Programmer’s Guide to Connected TV Audience Recognition. LiveRamp. https://docs.liveramp.com/advanced-tv/en/programmer-s-guide-to-connected-tv-audience-recognition.html Accessed 2026-11-11.

[308] Cross Device / CTV Graph. Factori. https://docs.factori.ai/docs/cross-device Accessed 2026-11-11.

[309] Factori CTV/Cross Device Data-2 Billion+ across the Globe. Factori Technologies Pte. Ltd. (seller on AWS Marketplace). https://aws.amazon.com/marketplace/pp/prodview-ofuh7mf7leose Accessed 2026-11-11.

[310] Tru Optik & Skydeo Partner on Connected TV (CTV) Household Graph. Skydeo and Tru Optik, 2020-01-23. https://skydeo.com/2020/01/tru-optik-skydeo-partner-on-connected-tv-ctv-household-graph/ Accessed 2026-11-11.

[311] US Patent 11,310,195: Internet connected household identification for online measurement and dynamic content delivery. US patent specification credited to Viant, reproduced by Google Patents, 2022-04-19. https://patents.google.com/patent/US11310195B2/en Accessed 2026-11-11.

[312] Viant Household ID™: Precise Household Advertising | Viant. Viant. https://www.viantinc.com/solutions/planning-and-buying/household-id/ Accessed 2026-11-11.

[313] Viant and TransUnion Partner to Enrich Viant Household ID | Viant. Viant, 2025-01-28. https://investors.viantinc.com/news-releases/news-release-details/viant-and-transunion-partner-enrich-viant-household-id Accessed 2026-11-11.

[314] Cadent Announces Expansion of TV. Cadent, 2021-02-02. https://cadent.tv/wp-content/uploads/2021/02/Cadent-Announces-Expansion-of-TV-Identity-Graph-Partnership-with-Premion.pdf Accessed 2026-11-11.

[315] Total Video | Linear, CTV, Digital & YouTube | Cadent. Cadent. https://www.cadent.com/total-video Accessed 2026-11-11.

[316] TransUnion Announces Enhanced Identity Graph for Marketing Solutions. TransUnion, 2024-01-04. https://newsroom.transunion.com/transunion-announces-enhanced-identity-graph-for-marketing-solutions/ Accessed 2026-11-11.

[317] Form 8-K, Item 1.01, entry into the merger agreement. LiveRamp Holdings, Inc., 2026-05-18. https://www.sec.gov/Archives/edgar/data/733269/000110465926062908/tm2614904d1_8k.htm Accessed 2026-11-11.

[318] Form 12b-25, notification of late filing for September quarter. LiveRamp Holdings, Inc., 2026-11-10. https://www.sec.gov/Archives/edgar/data/733269/000110465926125180/tm2629306d1_nt10q.htm Accessed 2026-11-11.

[319] Publicis Groupe completes acquisition of Epsilon. Publicis Groupe, 2019-07-02. https://www.publicisgroupe.com/sites/default/files/press-releases/2019-07/07.02.2019%20PR%20Closing%20Epsilon%20EN.pdf Accessed 2026-11-11.

[320] Form 8-K, item 8.01, completed Epsilon disposition. Alliance Data Systems Corporation, 2019-07-17. https://www.sec.gov/Archives/edgar/data/1101215/000155837019006119/f8-k.htm Accessed 2026-11-11.

[321] Experian accelerates identity and activation capabilities with acquisition of Audigent. Experian, 2024-12-04. https://www.experianplc.com/newsroom/press-releases/2024/experian-accelerates-identity-and-activation-capabilities-with-a?sf_campaign_id=_7016S000002W62RQAS Accessed 2026-11-11.

[322] Experian annual report 2025, Board activities and acquisition of Audigent. Experian plc, 2025. https://www.experianplc.com/content/dam/marketing/global/plc/en/assets/documents/reports/2025/experian-annual-report-2025.pdf Accessed 2026-11-11.

[323] TransUnion and Neustar Announce Transaction Close. TransUnion and Neustar, 2021-12-01. https://newsroom.transunion.com/transunion-and-neustar-announce-transaction-close/ Accessed 2026-11-11.

[324] Exhibit 99.3, unaudited pro forma combined financial data, acquisition introduction. TransUnion, 2022-02-16. https://www.sec.gov/Archives/edgar/data/1552033/000119312522043629/d264925dex993.htm Accessed 2026-11-11.

[325] IPG completes acquisition of Acxiom LLC, issuer exhibit on EDGAR. Interpublic Group, 2018-10-01. https://www.sec.gov/Archives/edgar/data/51644/000005164418000051/pressrelease1012018.htm Accessed 2026-11-11.

[326] Form 8-K/A, item 2.01, Acxiom acquisition and pre-closing distribution of LiveRamp. The Interpublic Group of Companies, Inc., 2018-10-02. https://www.sec.gov/Archives/edgar/data/51644/000005164418000057/form8-ka1012018.htm Accessed 2026-11-11.

[327] Omnicom completes acquisition of Interpublic. Omnicom, 2025-11-26. https://www.omc.com/newsroom/omnicom-completes-acquisition-of-interpublic-forming-the-worlds-leading-marketing-and-sales-company-built-for-intelligent-growth-in-the-next-era/?trk=article-ssr-frontend-pulse_little-text-block Accessed 2026-11-11.

[328] Omnicom announces strategy and executive leadership following acquisition of Interpublic. Omnicom, 2025-12-01. https://www.omc.com/newsroom/omnicom-announces-strategy-and-executive-leadership-following-acquisition-of-interpublic/ Accessed 2026-11-11.

[329] Apollo Funds Complete Acquisition of Yahoo. Apollo Global Management, 2021-09-01. https://ir.apollo.com/_assets/_354abe828742146affc9210d18fd72a7/apollo/news/2021-09-01_Apollo_Funds_Complete_Acquisition_of_67.pdf Accessed 2026-11-11.

[330] Quarterly financial statements note, Verizon Media Divestiture. Verizon Communications Inc., 2022-07-28. https://www.sec.gov/Archives/edgar/data/732712/000073271222000043/R10.htm Accessed 2026-11-11.

[331] Annual report on Form 10-K for the year ended December 31, 2025. The Trade Desk, Inc., 2026-02-27. https://www.sec.gov/Archives/edgar/data/1671933/000167193326000014/ttd-20251231.htm Accessed 2026-11-11.

[332] Comcast, Charter and ViacomCBS Announce Blockgraph Partnership. Charter Communications, 2020-05-29. https://corporate.charter.com/newsroom/comcast-charter-and-viacomcbs-announce-blockgraph-partnership Accessed 2026-11-11. Access: partial extract.

[333] Comcast, Charter and ViacomCBS Announce Blockgraph Partnership. Comcast, Charter, ViacomCBS, joint release on Blockgraph’s newsroom, 2020-05-29. https://www.blockgraph.co/newsroom/comcast-charter-and-viacomcbs-announce-blockgraph-partnership Accessed 2026-11-11.

[334] We Are Blockgraph. Blockgraph. https://www.blockgraph.co/about-us Accessed 2026-11-11.

[335] Form 10-K for the year ended December 31, 2025. Comscore, Inc., SEC EDGAR, 2026-03-25. https://www.sec.gov/Archives/edgar/data/1158172/000115817226000009/scor-20251231.htm Accessed 2026-11-11.

[336] Experian Marketing Solutions – Data Practices Notice. Experian Marketing Solutions, LLC, 2026-04-14. https://www.experian.com/privacy/marketing-data-practices-and-policies Accessed 2026-11-11.

[337] Experian plc Annual Report 2026. Experian plc, 2026-06-11. https://www.experianplc.com/content/dam/marketing/global/plc/en/assets/documents/reports/2026/experian-annual-report-2026.pdf Accessed 2026-11-11.

[338] Full-year results FY26. Experian plc, 2026-05-20. https://www.experianplc.com/newsroom/press-releases/2026/full-year-results-fy26 Accessed 2026-11-11.

[339] ID5 Technology Limited — persons with significant control. Companies House. https://find-and-update.company-information.service.gov.uk/company/10735438/persons-with-significant-control Accessed 2026-11-11.

[340] ID5 Technology Limited, accounts made up to 30 June 2025. ID5 Technology Limited via Companies House, 2026-03-15. https://find-and-update.company-information.service.gov.uk/company/10735438/filing-history/MzUxMDA1ODQyMWFkaXF6a2N4/document?format=pdf&download=0 Accessed 2026-11-11.

[341] ID5 Announce $20 Million in Series B Funding. ID5, 2024-04-09. https://id5.io/news/id5-announce-20-million-in-series-b-funding Accessed 2026-11-11.

[342] Cloud Technologies S.A., ESPI 23/2026: Nabycie biznesu i marki TrueData. Cloud Technologies S.A. via Warsaw Stock Exchange, 2026-09-30. https://www.gpw.pl/komunikat?geru_id=498060 Accessed 2026-11-11.

[343] Cloud Technologies S.A., ESPI 24/2026: Realizacja nabycia biznesu i marki TrueData. Cloud Technologies S.A. via Warsaw Stock Exchange, 2026-10-05. https://www.gpw.pl/komunikat?geru_id=498374 Accessed 2026-11-11.

[344] Form 10-Q for the three months ended June 30, 2026. LiveRamp Holdings, Inc. / SEC EDGAR, 2026-08-05. https://www.sec.gov/Archives/edgar/data/733269/000073326926000047/ramp-20260630.htm Accessed 2026-11-11.

[345] LiveRamp Terms of Service Agreement. LiveRamp. https://liveramp.com/legal/liveramp-terms-of-service Accessed 2026-11-11.

[346] Quarterly report on Form 10-Q for the quarter ended September 30, 2026. The Trade Desk, Inc., 2026-11-09. https://www.sec.gov/Archives/edgar/data/1671933/000167193326000116/ttd-20260930.htm Accessed 2026-11-11.

[347] 2021 Form 10-K, Verizon Communications Inc.. Verizon Communications Inc. / SEC EDGAR, 2022-02-11. https://www.sec.gov/Archives/edgar/data/732712/000073271222000008/vz-20211231.htm Accessed 2026-11-11.

[348] Baker v. Yahoo Inc., SDNY docket, entries 53 and 54. US District Court for the Southern District of New York, docket text reproduced by Justia, 2025-09-26. https://dockets.justia.com/docket/new-york/nysdce/1:2025cv02797/639983 Accessed 2026-11-11.

[349] Seedtag and Aristotle Combine Voter Data and Neuro-Contextual AI to Help Political Campaigns Reach Voters in Relevant Moments. Seedtag, Aristotle and Multilocal partner release, 2026-10-21. https://www.seedtag.com/press/seedtag-and-aristotle-combine-voter-data-and-neuro-contextual-ai Accessed 2026-11-11.

[350] Comscore Launches Political Targeting Solutions for 2024 Election Season, exclusively on The Trade Desk. Comscore, with named recipient The Trade Desk, 2023-10-24. https://www.comscore.com/Insights/Press-Releases/2023/10/Political-Targeting-Solutions-for-2024-Election-Season Accessed 2026-11-11.

[351] ID5 ID Agreement v1.2. ID5 Technology Limited, 2025-12-17. https://id5.io/legal/agreements/v1.2%20ID5%20ID%20Agreement%20-%20Embedded%20Platform.pdf Accessed 2026-11-11.

[352] LiveRamp Authenticated Traffic Solution Terms and Conditions. LiveRamp, 2020-09-03. https://liveramp.com/legal/liveramp-authenticated-traffic-solution-terms-and-conditions Accessed 2026-11-11.

[353] LiveRamp Authenticated Traffic Solution: Publisher Terms and Conditions. LiveRamp. https://d7umqicpi7263.cloudfront.net/eula/product/9a89b342-79ba-47b2-8024-33898eff66ba/e81cfef2-6c2c-4e64-901c-ea55cdc5928a.pdf Accessed 2026-11-11.

[354] Q4 2021 Earnings Call, corrected transcript. LiveRamp Holdings, Inc.; Scott E. Howe, CEO, as transcribed by FactSet CallStreet, 2021-05-25. https://investors.liveramp.com/static-files/53777f2b-af22-458d-b8fa-d814b6294589 Accessed 2026-11-11.

[355] Seedtag’s new Contextual Audiences connect brands with unique consumer interests without personal data. Seedtag, original issuer release via PR Newswire, 2023-10-18. https://www.prnewswire.com/news-releases/seedtags-new-contextual-audiences-connect-brands-with-unique-consumer-interests-without-personal-data-301959423.html Accessed 2026-11-11.

[356] NIQ and Ogury Collaborate to Enhance Media Planning and Audience Strategy with GeoPurchase Insights. NIQ, 2026-06-18. https://nielseniq.com/global/en/news-center/2026/niq-and-ogury-collaborate-to-enhance-media-planning-and-audience-strategy-with-geopurchase-insights/ Accessed 2026-11-11.

[357] ICO statement on Upper Tribunal ruling. UK Information Commissioner’s Office, 2024-04-24. https://ico.org.uk/about-the-ico/media-centre/news-and-blogs/2024/04/ico-statement-on-upper-tribunal-ruling/ Accessed 2026-11-11.

[358] October Town Hall: protocol advances, new ways to build, and introducing our Board. AgenticAdvertising.org, 2026-10-22. https://adcontextprotocol.org/perspectives/october-town-hall-protocol-advances-new-ways-to-build-and-introducing-our-board-mvis1aq7 Accessed 2026-11-11.

[359] Dentsu Leverages Merkury to Establish End-to-End Measurement across The Trade Desk. dentsu, 2025-01-08. https://www.dentsu.com/us/en/media-and-investors/dentsu-leverages-merkury-to-establish-end-to-end-measurement-across-the-trade-desk Accessed 2026-11-11.

[360] iProspect Named Media Agency of Record for Hy-Vee. iProspect, 2025-10-16. https://www.iprospect.com/en-us/insights/iprospect-named-media-agency-of-record-for-hy-vee Accessed 2026-11-11.

[361] RedMedia Solutions. Hy-Vee RedMedia. https://www.hredmedia.com/solutions Accessed 2026-11-11.

[362] Publisher Link for Header Bidding. Prebid.org with Epsilon. https://docs.prebid.org/dev-docs/modules/userid-submodules/publisherlink.html Accessed 2026-11-11.

[363] Merkle ID for Header Bidding. Prebid.org with Merkle. https://docs.prebid.org/dev-docs/modules/userid-submodules/merkle.html Accessed 2026-11-11.

[364] (Author's work) How this edition was researched and reviewed. No Fluff Advisory research project, 2026-11-11. ./data/process-record.json Accessed 2026-11-11.

[365] The Impact of The Trade Desk Contributing UID 2.0 for Industry Development. IAB Tech Lab, 2021-05-25. https://iabtechlab.com/the-impact-of-the-trade-desk-contributing-uid-2-0-for-industry-development/ Accessed 2026-11-11.

[366] Tech Lab Update on UID2.0. Anthony Katsur, IAB Tech Lab, 2022-02-28. https://iabtechlab.com/tech-lab-update-on-uid2-0/ Accessed 2026-11-11.

[367] Unified IDs. The Trade Desk. https://open.thetradedesk.com/advertiser/docsApp/Glossary/data/doc/UnifiedIDs Accessed 2026-11-11.

[368] Publisher information. EUID documentation / The Trade Desk, 2026-11-02. https://euid.eu/docs/legal/euid-participant-information-publishers Accessed 2026-11-11.

[369] Publisher | ID5 Knowledge Base. ID5, 2025-12-04. https://wiki.id5.io/docs/publisher Accessed 2026-11-11.

[370] ID5 Technology Limited — company overview. Companies House. https://find-and-update.company-information.service.gov.uk/company/10735438 Accessed 2026-11-11.

[371] LiveRamp and Yahoo Partner to Scale Addressability Across the Ad Ecosystem. LiveRamp and Yahoo (issuer release on Business Wire), 2023-10-16. https://www.businesswire.com/news/home/20231016463270/en/LiveRamp-and-Yahoo-Partner-to-Scale-Addressability-Across-the-Ad-Ecosystem Accessed 2026-11-11.

[372] LiveRamp and Yahoo Partner to Scale Addressability Across the Ad Ecosystem. LiveRamp and Yahoo, 2023-10-16. https://s29.q4cdn.com/969268860/files/doc_news/LiveRamp-and-Yahoo-Partner-to-Scale-Addressability-Across-the-Ad-Ecosystem-2023.pdf Accessed 2026-11-11.

[373] Instructions for SSPs Adopting the Panorama ID in the Bidstream - Lotame Panorama ID—Prebid, Google ESP & Amazon CxM - Lotame Knowledge Base. Lotame. https://my.lotame.com/t/y4hjvdv/instructions-for-ssps-adopting-the-panorama-id-in-the-bidstream Accessed 2026-11-11.

[374] Lotame Panorama ID for Header Bidding. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/lotame.html Accessed 2026-11-11.

[375] Lotame Unveils Panorama ID, the First Global Cookieless Identity Solution for a Privacy-First Open Web. Lotame, 2023-09-12. https://www.lotame.com/resources/lotame-unveils-panorama-id-the-first-global-cookieless-identity-solution-for-a-privacy-first-open-web/ Accessed 2026-11-11.

[376] prebid/prebid.github.io / dev-docs/modules/userid-submodules/criteo.md at 44c1b8e99908. Prebid project, 2024-11-28. https://raw.githubusercontent.com/prebid/prebid.github.io/44c1b8e999080f74e6077df22eb0520be82e2ade/dev-docs/modules/userid-submodules/criteo.md Accessed 2026-11-11.

[377] Extended ID integration (alternative/universal IDs). Equativ. https://help.equativ.com/supply-inventory-with-web-ad-tags/extended-id-integration-alternativeuniversal-ids Accessed 2026-11-11.

[378] Signal IQ - Amazon Publisher Services. Amazon Publisher Services. https://aps.amazon.com/aps/solutions/signal-iq/ Accessed 2026-11-11.

[379] 70 million users: Advertising. Reconnected. - Utiq. Utiq, 2025-11-15. https://utiq.com/70-million-users-advertising-reconnected/ Accessed 2026-11-11.

[380] Web Consent Management. Utiq. https://docs.utiq.com/docs/utiq-consent-management Accessed 2026-11-11.

[381] prebid/prebid.github.io / dev-docs/modules/userid-submodules/parrable.md at 44c1b8e99908. Prebid project, 2024-11-28. https://raw.githubusercontent.com/prebid/prebid.github.io/44c1b8e999080f74e6077df22eb0520be82e2ade/dev-docs/modules/userid-submodules/parrable.md Accessed 2026-11-11.

[382] prebid/prebid.github.io / dev-docs/modules/userid-submodules/parrable.md at c3125df5a719. Prebid project, 2026-10-26. https://raw.githubusercontent.com/prebid/prebid.github.io/c3125df5a719219b50af510b9bf75340972fec14/dev-docs/modules/userid-submodules/parrable.md Accessed 2026-11-11.

[383] AbiliTec Identifiers. LiveRamp. https://developers.liveramp.com/abilitec-api/reference/abilitec-identifiers-1 Accessed 2026-11-11.

[384] Marketing Identity | TruAudience | TransUnion. TransUnion. https://www.transunion.com/solution/truaudience/identity Accessed 2026-11-11.

[385] Epsilon Legal. Epsilon. https://legal.epsilon.com/us/description-of-processing Accessed 2026-11-11.

[386] Identity Resolution | Digital Identification Platform. Epsilon. https://www.epsilon.com/us/products-and-services/identity-core-id Accessed 2026-11-11.

[387] IPG unveils cloud-based identity resolution platform. Acxiom / Interpublic Group, 2023-10-23. https://www.acxiom.co.uk/news/ipg-modernizes-identity-resolution-with-launch-of-real-id-cloud-solution/ Accessed 2026-11-11.

[388] Acxiom launches digital recognition, connecting online and offline identity for true omnichannel personalization - Omnicom. Omnicom / Acxiom, 2026-07-14. https://www.omc.com/newsroom/acxiom-launches-digital-recognition-connecting-online-and-offline-identity-for-true-omnichannel-personalization/ Accessed 2026-11-11.

[389] dentsu. dentsu / Merkle, 2020-02-11. https://www.dentsu.com/us/en/news-releases/merkle-strengthens-position-as-identity-technology-leader-with-launch-of-merkury-identity-resolution-platform Accessed 2026-11-11.

[390] Data and Identity Resolution - Zeta Global. Zeta Global. https://www.zetaglobal.com/products/zeta-data-identity/ Accessed 2026-11-11.

[391] Zeta Launches Customer Growth Intelligence, a Snowflake Native App in the Data Cloud. Zeta Global, 2023-08-28. https://investors.zetaglobal.com/news/news-details/2023/Zeta-Launches-Customer-Growth-Intelligence-a-Snowflake-Native-App-in-the-Data-Cloud/default.aspx Accessed 2026-11-11.

[392] Verve Group launches Moments.AI™ (formerly Beemray), expanding contextual targeting capabilities to North America | Verve. Verve Group, 2022-03-08. https://verve.com/press/verve-group-launches-moments-ai-formerly-beemray/ Accessed 2026-11-11.

[393] Interim Report Q1 2026. Verve Group SE, 2026-05-27. https://investors.verve.com/wp-content/uploads/2026/05/20260527_VER_Report_Q12026_eng_fin.pdf Accessed 2026-11-11.

[394] Seedtag. Seedtag. https://www.seedtag.com/success-stories/henkel Accessed 2026-11-11.

[395] Eurail’s contextual strategy strengthened awareness in the USA. Seedtag; includes Eurail buyer statement. https://www.seedtag.com/success-stories/eurail Accessed 2026-11-11.

[396] Blockgraph and FullThrottle Announce Integration to Unlock New Levels of Addressable Targeting for TV Advertisements | Blockgraph Newsroom. Blockgraph, 2022-06-09. https://www.blockgraph.co/newsroom/blockgraph-and-fullthrottle-addressable-targeting-for-tv-advertisements Accessed 2026-11-11.

[397] OpenAP | Press - OpenAP Unveils OpenID, a Unique Identifier that Powers Resolution of Linear and Digital Audiences into a Common Identity Framework for TV | OpenAP. OpenAP, 2021-04-28. https://www.openap.tv/press/OpenAP-Unveils-OpenID Accessed 2026-11-11.

[398] OpenAP unveils Open Identity, a converged ID solution for streaming. OpenAP, 2025-04-15. https://www.openap.tv/press/open-identity-converged-id-solution-audience-identity-streaming Accessed 2026-11-11.

[399] Identity Network. FreeWheel. https://api-docs.freewheel.tv/advertiser/reference/identity-network Accessed 2026-11-11.

[400] Samba TV Identity. Samba TV. https://info.samba.tv/hubfs/Sales%20Sheets/2021_Identity%202-sheeter.pdf Accessed 2026-11-11.

[401] 33Across ID for Header Bidding. Prebid. https://docs.prebid.org/dev-docs/modules/userid-submodules/33across.html Accessed 2026-11-11.

[402] 3. ZEOTAP’S UNIVERSAL ID: ID+. Zeotap. https://zeotap.com/wp-content/uploads/2025/06/1.-Whitepaper-Digital-identity-is-at-a-turning-point-What-now_.pdf Accessed 2026-11-11.

[403] Zeotap and PubMatic announce expanded partnership on identity. PubMatic, 2021-05-18. https://pubmatic.com/news/zeotap-pubmatic-announce-expanded-partnership-on-identity/ Accessed 2026-11-11.

[404] Scalable, ID-less Targeting Solutions for Media Buyers. PubMatic and Proximic by Comscore, 2023-03-23. https://pubmatic.com/news/pubmatic-partners-with-proximic-by-comscore-giving-media-buyers-scalable-id-less-targeting-soutions/ Accessed 2026-11-11.

[405] The Trade Desk Announces Galileo, a New Approach for Activating Advertiser Data | The Trade Desk. The Trade Desk, 2023-01-05. https://www.thetradedesk.com/press-room/the-trade-desk-announces-galileo-a-new-approach-for-activating-advertiser-data Accessed 2026-11-11.

[406] Cookieless Audience Targeting | Datonics. Datonics. https://www.datonics.com/explore-cookieless Accessed 2026-11-11.

[407] Accountability Platform. IAB Tech Lab, 2021-03. https://iabtechlab.com/wp-content/uploads/2021/03/iabtechlab_accountability_platform_rfc_2021_march.pdf Accessed 2026-11-11.

[408] Fabrick ID by Neustar for Header Bidding. Prebid.org. https://docs.prebid.org/dev-docs/modules/userid-submodules/fabrick.html Accessed 2026-11-11.

[409] FTrack ID from Flashtalking By Mediaocean for Header Bidding. Prebid.org with Flashtalking by Mediaocean. https://docs.prebid.org/dev-docs/modules/userid-submodules/ftrack.html Accessed 2026-11-11.

[410] Data Product Privacy Notice. Merkle, a dentsu company, 2026-07-07. https://www.merkle.com/en/privacy-policy/data-product-privacy-notice.html Accessed 2026-11-11.

[411] User opt-out | European Unified ID. European Unified ID. https://euid.eu/docs/getting-started/gs-opt-out Accessed 2026-11-11.

[412] Pubfinity Boosts Global Revenue per User with ID5 Identity Solutions. ID5. https://id5.io/resources/pubfinity-boosts-global-revenue-per-user-with-id5-identity-solutions Accessed 2026-11-11.

[413] FAQs (Pixels) | Zeta DSP. Zeta Global. https://knowledgebase.zetaglobal.com/pug/faqs-pixels Accessed 2026-11-11.

[414] Power Personalized Experiences with Unified Customer Views. Deep Sync. https://deepsync.com/wp-content/uploads/2025/05/Product-Sheet_Power-Personalized-Experiences-with-Identity-Resolution.pdf Accessed 2026-11-11.

[415] IP Protection Overview  |  Privacy Sandbox. Google. https://privacysandbox.google.com/protections/ip-protection Accessed 2026-11-11.

[416] prebid/prebid.github.io / dev-docs/modules/userid-submodules/utiq.md at d767b07ea9e5. Prebid project, 2025-03-17. https://raw.githubusercontent.com/prebid/prebid.github.io/d767b07ea9e5af85940216c70448bc2d91c68106/dev-docs/modules/userid-submodules/utiq.md Accessed 2026-11-11.

[417] prebid/prebid.github.io / dev-docs/modules/userid-submodules/utiq.md at cdcaee60a7d7. Prebid project, 2026-06-23. https://raw.githubusercontent.com/prebid/prebid.github.io/cdcaee60a7d74c3366398790d04cb6bfd4a7322f/dev-docs/modules/userid-submodules/utiq.md Accessed 2026-11-11.

[418] Utiq ID user ID submodule integration guide, July revision. Prebid project, 2025-07-07. https://raw.githubusercontent.com/prebid/prebid.github.io/eebce06bf4cc54848c49b389f4669583ffa30e1d/dev-docs/modules/userid-submodules/utiq.md Accessed 2026-11-11.

[419] Cadent Integrates Blockgraph to Enable More Effective Targeting of Addressable TV Audiences for its Aperture Clients. Blockgraph and Cadent, 2022-04-27. https://www.blockgraph.co/newsroom/cadent-integrates-blockgraph-addressable-tv-audiences Accessed 2026-11-11.

[420] Blockgraph Announces Launch of GraphPort. Blockgraph, 2023-06-08. https://www.blockgraph.co/newsroom/blockgraph-announces-launch-of-graphport Accessed 2026-11-11.

[421] Nasdaq Website Third Party Participant List. Nasdaq. https://www.nasdaq.com/website-third-party-list Accessed 2026-11-11.

[422] TruAudience Identity Extracts. TransUnion via AWS Marketplace. https://aws.amazon.com/marketplace/pp/prodview-fggqyyuyijri4 Accessed 2026-11-11.

[423] TV Analytics – Powered by Neustar, solution sheet. Neustar, a TransUnion company, 2022. https://www.cdn.neustar/resources/product-literature/marketing/neustar-tv-analytics-solution-sheet.pdf Accessed 2026-11-11.

[424] Investor FAQs. LiveRamp investor relations. https://investors.liveramp.com/investor-resources/investor-faqs Accessed 2026-11-11.

[425] Investor relations presentation, Q4 2024. The Trade Desk, 2025. https://s29.q4cdn.com/168520777/files/doc_earnings/2024/q4/presentation/TheTradeDesk_Q424_Investor_Presentation.pdf Accessed 2026-11-11.

[426] Neustar Launches the Fabrick ID to Future Proof Media Buying and Selling in a Post-Cookie World. Neustar via Business Wire, 2020-12-16. https://www.businesswire.com/news/home/20201216005093/en/Neustar-Launches-the-Fabrick-ID-to-Future-Proof-Media-Buying-and-Selling-in-a-Post-Cookie-World Accessed 2026-11-11.

[427] Privacy Sandbox: Topics integration. Permutive, 2024-07-26. https://changelog.permutive.com/privacy-sandbox-topics-integration-297702 Accessed 2026-11-11.

[428] 33Across Throws Itself Into Cookieless Targeting. AdExchanger interview, republished by 33Across, 2021-06-02. https://www.33across.co.uk/news/2021/06/02/33across-throws-itself-into-cookieless-targeting/ Accessed 2026-11-11.

[429] Publicis Groupe and The Trade Desk Join Forces to Champion New Approach to Privacy-First Identity and Personalization at Scale in a Cookie-less World. Publicis Groupe, 2021. https://www.publicisgroupe.com/en/news/press-releases/publicis-groupe-and-the-trade-desk-join-forces-to-champion-new-approach Accessed 2026-11-11.

[430] Privacy & Cybersecurity: In re The Trade Desk, Inc. Data Privacy Litigation. Lieff Cabraser Heimann & Bernstein. https://www.lieffcabraser.com/privacy/ Accessed 2026-11-11.
