Ecosystem Surface Deep Dive

Retail Media Network.

Where first-party purchase data becomes closed-loop, shoppable media.

Retail media turned the shelf into an ad network. Retailers and marketplaces have the one thing the open web lost — logged-in, consented, purchase-level first-party data — and they sell access to it on-site, off-site, and increasingly in-store. The opportunity is closed-loop measurement against real sales. The problem is that every network defines the loop differently.

A retail media network is not a media buy. It is access to first-party purchase data — and the strategy is comparability and governance across networks that all measure differently.

Scope this playbook → Back to Ecosystem Surfaces →

Retail Media Network — the lit node in the activation flow: governed data in, a governed decision out. Hover a node for detail.

Decision fit

Fast read.

Best when: You want media measured against real purchases and can run across more than one network.
Not when: You need one comparable cross-network metric today, or you have no first-party / rights story.
Primary buyer: Commerce, shopper-marketing, performance, and brand-media leaders.
Primary output: Closed-loop, sales-attributed media — on-site, off-site, and in-store.
Main risk: Accepting each network’s own metrics as truth and losing comparability.
Best next step: Define one incrementality / comparability framework before scaling spend across networks.

Why now

Market context: from trade spend to media network.

Retailers turned logged-in, purchase-level first-party data into a high-margin media business.
Spend is shifting from trade / shopper-marketing budgets into measurable media.
The surface expanded from on-site search and sponsored products to off-site (DSP, social, CTV) and in-store (screens, audio, DOOH).
Closed-loop measurement against real sales is the draw; clean rooms are how brands verify it.
“Commerce media” extends the model beyond grocery and mass to marketplaces, delivery, travel, and even financial services.
The unsolved problem is standardization — every network measures, defines, and reports differently.

Evolution

Retail media evolution.

Retail media evolution

01
On-site search
Sponsored products and search ads on the retailer’s own site / app.
02
On-site display
Banners, brand pages, and category takeovers across owned properties.
03
Off-site extension
Retailer audiences activated off-property via DSP, social, and CTV.
04
In-store + omnichannel
Screens, retail audio, sampling, and DOOH tied to the same data.
05
Closed-loop + clean rooms
Sales-attributed measurement and privacy-safe verification with the retailer.
06
Standardization + agentic commerce
Comparable metrics and automated buying across many networks.

Landscape

Who plays here — examples, not a ranking.

Named as examples, not a ranking. Retailer networks and the tech that powers them are distinct — a network is the seller of access; enablement tech runs the auction underneath. Validate current names and “powered by” relationships.

Grocery / mass / club networks

Amazon Ads (+ AMC clean room)
Walmart Connect
Target Roundel
Kroger Precision Marketing
Albertsons Media Collective
Costco
Sam’s Club MAP
Instacart

Specialty / vertical networks

Best Buy Ads
The Home Depot (Orange Apron Media)
Lowe’s Media Network
CVS Media Exchange
Walgreens Advertising Group
Ulta Beauty (UB Media)
Macy’s Media Network
Chewy Ads

Commerce / delivery / financial media

DoorDash Ads
Uber Advertising
Instacart Carrot Ads (enables other retailers)
Chase Media Solutions (financial services)

Enablement tech (powers RMNs — not networks)

Criteo Commerce Media
Epsilon Retail Media (formerly CitrusAd)
Koddi
Topsort
Moloco Commerce Media
Pacvue
Skai

Capability map

What it does — and where it quietly fails.

What to weigh — and where it bites. Validate current support per network.

Capability	What it means	Why it matters	Watch-out
On-site search / sponsored products	Paid placements in retailer search.	Closest to purchase intent.	Auction dynamics and incrementality.
On-site display	Banners, brand pages, takeovers.	Upper-funnel on owned property.	Measurement vs sponsored products.
Off-site extension	Retailer audiences via DSP / social / CTV.	Reach beyond the site.	Attribution back to sales is harder.
In-store media	Screens, audio, sampling, DOOH.	Point-of-purchase influence.	Exposure and measurement maturity.
First-party purchase data	Logged-in, SKU-level signal.	The reason RMNs exist.	Rights and what leaves the platform.
Closed-loop measurement	Ads tied to real sales.	Outcome proof.	Each network defines the loop.
Incrementality / lift	Causal sales impact.	Separates real value from harvested demand.	Few networks default to it.
Clean-room measurement	Privacy-safe verification with the retailer.	Independent read on outcomes.	Output policy and matchable rights.
Audience extension	Bring / build audiences across channels.	Targeting beyond on-site.	Match quality and rights.
Comparability / standards	Common metric definitions.	Compare networks fairly.	Still fragmented industry-wide.
Self-serve vs managed	Buying model.	Control vs effort.	Maturity varies widely.
API / automation	Programmatic + bulk ops.	Scale across networks.	Coverage is uneven.
Brand safety / quality	Placement and content control.	Protect brand.	Off-site paths vary.

First-party data

How your first-party data plugs into an RMN.

On the brand side, RMNs are a place to verify and extend — not just buy. These connections decide whether you can measure independently.

From the CDP
- Bring consented audiences for targeting / suppression
- Suppress existing customers to chase incremental sales
From clean rooms
- Overlap your data with the retailer’s for verification
- Privacy-safe incrementality and audience build
From the network
- SKU-level sales and conversion feedback
- Closed-loop reporting back into BI / MMM
To measurement
- Feed outcomes into a single comparability framework
- Reconcile network metrics against your source of truth

How it connects

What feeds an RMN — and the catch.

Input	How it feeds the RMN	Watch-out
CDP segments	Audiences for targeting and suppression	Activation rights and match quality
Clean room	Overlap, verification, incrementality	Output policy and what is matchable
Retailer purchase data	Closed-loop, SKU-level attribution	Definition differs by network
Off-site DSP	Extends audiences beyond the site	Sales attribution gets weaker
Network reporting	Outcome metrics back to the brand	Not comparable across networks
BI / MMM	Reconciles RMN spend vs total sales	Needs a common metric framework

Agentic shift

What agentic commerce media changes.

Agentic buying across retail media promises one operator managing many networks — planning, bidding, and optimizing toward sales. The bottleneck is not automation; it is whether the metrics underneath are comparable enough to optimise against.

Agent use cases

Cross-network planning and budget allocation
Automated bidding on sponsored products
Audience build and suppression across networks
Incrementality-aware optimization
Anomaly and overspend detection
Reconciliation against a single source of truth

Agent risks

Optimizing to each network’s favourable metric
Chasing harvested demand, not incremental sales
Activating purchase data without rights
No comparability across networks
Opaque off-site attribution

Governance needed

One comparability / incrementality framework
Activation rights and suppression rules
Measurement source of truth
Network-by-network metric mapping
Human approval on budget shifts, audit log

Comparability is the unsolved problem

Measurement standards now exist and are converging — the IAB / MRC Retail Media Measurement Guidelines in the US and IAB Europe’s commerce- and in-store-media standards. But adoption is uneven, so in practice every network’s “ROAS” still means something slightly different. The operator’s job is to impose one comparability and incrementality framework across networks, rather than accept each network’s own scorecard. (Validate current standard versions and adoption.)

Strategic read

SWOT.

Strengths

First-party purchase data
Closed-loop measurement
High purchase intent
Expanding on / off / in-store surface

Weaknesses

No cross-network comparability
Metric definitions differ
Incrementality rarely default
Off-site attribution is weaker
Fragmentation and operational load

Opportunities

Clean-room verification
Incrementality as standard
Commerce media beyond grocery
Agentic cross-network buying
In-store + CTV convergence

Threats

Self-graded metrics
Walled measurement
Data-rights missteps
Fee and take-rate opacity
Standardization stalling

Reference architecture