Research

Primary research, published here.

Full-length papers with their own method, a scored vendor field and a corrections ledger of the first-pass claims that did not survive review. Published on this site rather than through an outlet, so the data behind every figure stays one click away.

papers
14
vendors scored
363
corrections logged
4,289

Papers

  1. Advertising identity November 2026

    Who Is This?

    The State of Advertising Identity in 2026

    What did the seller match, how was the match checked, and what happens when someone opts out? The Identity Quotient groups products by the job a buyer needs done.

    products scored
    24
    corrections logged
    1,085

    57 products in the register · 12 dimensions · 6 deep profiles · 435 claims checked

    Read alongside: Debunking cross-device myth · Signal Containerization: The Next Abstraction Layer for Agentic Advertising · Data clean rooms and zero-party data: Powerful solutions for a cookie-less future

    Plain-language version PDF
  2. Ad verification November 2026

    Verified Against What

    The State of Ad Verification in 2026

    Provisional: what DoubleVerify, IAS and the specialists sell, what MRC accredits product by product and environment by environment, what independent studies and the legal record show, and whether a verification verdict ever changes what a buyer pays. Products scored on the Verdict Quotient within three cohorts.

    vendors scored
    17
    dimensions
    12
    corrections logged
    20

    Verdict Quotient in bands · 7 deep profiles with SWOT · 792 claims re-checked

    Read alongside: Where the Surplus Settles · The Open Web Isn't Dead. It's Uninsured. · The Risk You Can Price · The First AI-Native Fraud Isn't a Fake — It's Laundered Authority

    Plain-language version PDF
  3. Retail media November 2026

    Receipt Required

    The State of Retail Media in 2026

    What retail and commerce media networks report as results, who produces the number, whether anyone accredits it, how off-site exposure meets purchase and what sponsored formats inside AI agents report. 91 retailer, marketplace, commerce media, enablement, optimisation and measurement services are scored on the Receipt Quotient within five cohorts, from 1058 source records and 737 checked research claims.

    service rows scored
    91
    dimensions
    12
    corrections logged
    609

    Receipt Quotient in bands within cohorts · 8 deep profiles with SWOT · 737 claims re-checked

    Read alongside: iROAS Is Not a Number, It's a Negotiation · The Open Web Isn't Dead. It's Uninsured. · Measurement for Agentic Commerce

    Plain-language version PDF
  4. Sell-side curation November 2026

    Who Decides on the Sell Side

    The State of Sell-Side Curation in 2026

    Who decides which inventory a buyer sees when selection moves to the supply side, what a curated deal contains and costs, whether its quality claims are measured, and who owns the curators. 21 curation products are scored on the Deal Quotient within three cohorts, from 389 sources and 454 verified claims.

    vendors scored
    21
    dimensions
    12
    corrections logged
    393

    Deal Quotient in bands · 7 deep profiles with SWOT · 446 claims re-checked

    Read alongside: The Great Repackaging · Where the Surplus Settles · The First AI-Native Fraud Isn't a Fake — It's Laundered Authority

    Plain-language version PDF
  5. Location advertising and intelligence October 2026

    Last Seen

    The State of Location Advertising and Intelligence in 2026

    What location data products sell, whether their visit counts match the ground, whether visit lift rests on a holdout, what FTC orders and state laws changed, and whether agent-driven buying can carry location. 47 visit-measurement, audience, foot-traffic, movement-data and SDK products are scored on the Ground Quotient within five cohorts, from 793 source records and 493 checked research claims.

    products scored
    47
    dimensions
    11
    corrections logged
    221

    Ground Quotient in bands · 8 deep profiles with SWOT · 493 claims re-checked

    Read alongside: Measurement, on the Street's Terms · The Context Economy Has a Governance Problem · The Metric That Was Never About You

    Plain-language version PDF
  6. Commerce and ai agents October 2026

    Who Does the Shopping?

    From mobile and social discovery to AI-assisted and delegated buying

    Which shopping tasks have moved to software, under whose authority, and with what verifiable value for merchants. Twelve buying routes across ChatGPT, Google, Microsoft Copilot, Perplexity, Amazon, Walmart, TikTok and Meta; the ACP, UCP and AP2 specifications read at pinned commits; a sandbox study of an AI agent shopping a test store; and a merchant contribution worksheet with displacement. Assistance is widespread, approval-based checkout is documented in a few assistants, and documented delegation is narrow and platform-bound.

    corrections logged
    673

    1010 atomic claims with passage provenance · 672 sources in the public register · 31 sandbox tests · 8 figures

    Read alongside: Measurement for Agentic Commerce · The Record Is Not the Relationship · The Customer's Agent Arrived First · The Intelligence Economy · The Shelf You Can Restock · Where the Surplus Settles

    Plain-language version PDF
  7. Data collaboration October 2026

    What Can Leave the Room?

    What buyers can receive, keep, use again and rebuild from nine data-collaboration products

    What a buyer can actually receive, keep, reuse and rebuild from nine data-collaboration products: Snowflake, Databricks, AWS Clean Rooms, Ads Data Hub, BigQuery clean rooms, InfoSum, LiveRamp, Amazon Marketing Cloud and Meta Advanced Analytics. Every product lets something leave; none documents a way to take the whole workflow elsewhere. Thirteen tests to run before signing.

  8. Holding companies October 2026

    The Evolution of Holding Companies

    Control, client value and three paths for marketing work

    Where value moves as software takes on more marketing work: who sets goals, controls actions, proves growth, earns the fee, carries risk and keeps the learning. Twenty years of holding-company history read as a contest over control, three conditional paths for running the work, and a task-cost model a CMO can test.

    corrections logged
    255

    210 atomic claims in a public ledger · 51 sources in the public register, none re-opened in this review · 104 numerical and comparability checks · 21 figures (6 added 6 October 2026, 1 on 7 October)

    Read alongside: The Great Repackaging · Where the Surplus Settles · Nobody Sells an Outcome · The Answerer of Record

    Plain-language version PDF
  9. Attention measurement October 2026

    Paying Attention

    The State of Attention Measurement in 2026

    What attention metrics measure, which hold accreditation, whether they predict memory, brand or sales beyond time in view, and how attention is traded, bought and carried into agentic buying. 20 media and creative attention products are scored on the Focus Quotient within two cohorts, from 985 sources and 584 verified claims.

    products scored
    20
    dimensions
    12
    corrections logged
    260

    Focus Quotient in bands · 8 deep profiles with SWOT · 584 claims re-checked

    Read alongside: The Metric That Was Never About You · Attention Isn’t Enough—The Future of Marketing is Relevance · Relevance Realization: The Cognitive Science of Attention in the Age of AGI and Advertising · Nobody Sells an Outcome

    Plain-language version PDF
  10. Gaming advertising October 2026

    The State of Gaming Advertising in 2026

    Who sells brand advertising in games, what the standards cover, and how much of the pitch the public record supports

    Gaming advertising is several markets sold under one name. The only public US dollar series found counts ads placed in games, mostly on mobile, and EMARKETER says developers buy most of the ads served in games; MRC's record on 29 September 2026 listed no accredited intrinsic in-game metric; and every published field result found for a live brand campaign comes from a party with a stake in it. The paper scores in-game networks, platform and publisher ad products, and game-adjacent media within separate cohorts.

    products scored
    23
    dimensions
    12
    corrections logged
    143

    Play Quotient, in bands where a cohort is large enough · 8 deep profiles with SWOT · 807 claims checked by an AI verifier

    Read alongside: The Evidence Premium · The Metric That Was Never About You · Business Outcomes Isn't a Number

    Plain-language version PDF
  11. Audio advertising September 2026

    Counted, Not Heard

    The State of Audio Advertising in 2026

    An audit of the public proof behind US audio advertising. The certified metrics count downloads and ad deliveries, not listening; most attribution products that document a method match IP addresses, households or pixels; none of the attention or in-game retention claims this review coded has independent support.

    vendors scored
    34
    dimensions
    12
    corrections logged
    74

    Listen Quotient in bands · 7 deep profiles with SWOT · 640 claims re-checked

    Read alongside: The Evidence Premium · iROAS Is Not a Number, It's a Negotiation · Measurement, on the Browser's Terms

    Plain-language version PDF
  12. Native advertising September 2026

    Borrowed Trust

    The State of Native Advertising in 2026

    Open-web native, as the recommendation and in-feed networks sell it, is a supply-rights business sold as a performance product. What can and cannot be sized, what the causal evidence shows, and what a buyer should test. A scored vendor field, the Native Quotient, and every correction from the review rounds are published with it.

    vendors scored
    24
    dimensions
    12
    corrections logged
    275

    Native Quotient in bands · 9 deep profiles with SWOT · 317 claims re-checked

    Read alongside: The Open Web Isn't Dead. It's Uninsured. · The Risk You Can Price · The Evidence Premium

    Plain-language version PDF
  13. Mobile app growth September 2026

    What the Machines Can See

    The State of Mobile App Growth in 2026

    What the app-growth engines optimize, what attribution can and cannot show, and what durable growth depends on: the counterfactual, the store's cut and value after a year. A scored field of eleven app-growth platforms, the App Quotient, and every correction from three review rounds are published with it.

    app-growth platforms scored
    11
    dimensions
    11
    corrections logged
    119

    App Quotient in bands · 272 findings re-checked

    Read alongside: Is Apple harvesting adtech data? · The Loop Closed Inside the Wall · One Event, Three Machines: Orchestrating Conversions Across PMax, Advantage+, and OpenAI Ads · iROAS Is Not a Number, It's a Negotiation

    Plain-language version PDF
  14. Contextual advertising August 2026

    Proof and Story

    The State of Contextual Advertising in 2026

    How meaning, emotion and narrative are derived from content, who can actually do it, and the two capabilities the entire vendor field is missing. The interests declared and every first-pass claim the verifier rejected are published with it.