What Is a Retail Media Network? A Plain-English Explainer

What a Retail Media Network Actually Is

A retail media network is advertising space a retailer sells on its own digital properties — search results, product pages, the app, sometimes email — plus, often, placements the retailer's audience data can reach off its own site through a demand-side platform. The retailer owns the shopper; the brand pays to reach that shopper at or near the point of purchase.

Amazon Ads is the largest one. Walmart Connect, Target Roundel, Instacart Ads, and Kroger Precision Marketing sit in the next tier for US grocery and mass retail. Criteo doesn't own a storefront; it builds the technology that lets other retailers run their own network, which is why its own guide reads more like a platform pitch than a retailer's explanation.

Two things separate a retail media network from a normal ad network: the audience is a shopper mid-purchase-journey, not a browser, and the targeting runs on the retailer's own transaction data, not third-party cookies or panels.

How It Works — With Real Numbers, Not Just a Definition

Every retail media network reduces to the same mechanics: a brand sets a budget, bids into an auction (sponsored placements) or reserves programmatic inventory (DSP), the retailer serves the ad against its own audience, and the brand gets impressions, clicks and attributed sales back. The reporting is where most explainers stop. Here's what those numbers look like at scale.

Across 30 advertisers we managed on Amazon DSP in July 2026, the book ran at a 6.04x return on ad spend, delivering 78.4 million impressions at a $4.00 CPM and a blended $1.42 cost-per-click. That converted into 57,137 attributed purchases at a blended $5.49 cost-per-acquisition, 20.1% of them from shoppers new to the brand.

Work the math backward and you can see what a retail media network actually sells: reach, priced by CPM; intent, priced by CPC; and an attributed outcome, priced by CPA. Every retailer's network sells some version of that stack. What differs is the rate and the quality of the audience behind it.

The Question Every Definition Skips: Did the Ad Cause the Sale?

A retail media network reports last-click attribution by default — whoever's ad the shopper clicked last before buying gets full credit. That's fine for optimizing bids inside the platform. It is not proof the ad caused an incremental sale. A shopper who already had the product in their cart, or who searched the brand name directly, still counts as a conversion.

The honest way to check is a holdout: withhold ads from a matched group of shoppers, compare their purchase rate to the group that saw ads, and see if the difference is real. Reconciling DSP and sponsored ads in one measurement layer matters because a shopper can be served both, and platform-reported ROAS will double-count that same purchase in each channel's own dashboard.

If a reported ROAS looks too good — 10x, 15x on a network with modest reach — that's usually the tell. Check whether the number came from a holdout or from last-click before scaling the budget behind it.

Common Mistakes — Including Ones We've Made

The most common mistake is treating retail media network ROAS as if it settles the incrementality question by itself. It doesn't. It answers a narrower question — did people who saw this ad buy? — not whether they would have bought anyway.

• Budgeting on-site and off-site as separate channels. They usually reach overlapping shoppers, and two teams optimizing separately end up bidding against each other without knowing it.
• Scaling a campaign the week the number looks good. We've done this — moved budget into a placement on one strong week of last-click ROAS, before holdout data caught up and showed the real lift was smaller. The fix now is a measurement plan and a rollback trigger written down before the change ships, not after.
• Treating new-to-brand rate as a vanity metric. It's the number that says whether the network is growing the customer base or re-serving people who already buy from you.

When a number comes back wrong — a CPA that's crept up, a ROAS that collapsed after a holdout test — the fix is never spend more and see. It's isolate the variable that changed, re-run the comparison against a matched control, and only scale once the second read agrees with the first.

The Formats Inside a Retail Media Network

Most networks sell the same handful of formats, just under different names. Knowing the shape of the stack matters more than memorizing any one retailer's branding for it.

Where Dr. DSP Fits Into This

Dr. DSP is Amazon DSP — Amazon's demand-side platform for programmatic display, video and audio, not the Delivery Service Partner courier franchise — run as a managed product by Fable 5, out of Full Circle, which has managed more than $500M in revenue across 100+ brands. If you're deciding whether to run DSP as part of your own retail media mix, the honest question to ask any vendor, us included, is whether they measure with holdouts or hand you a last-click dashboard and call it done. We reconcile in Amazon Marketing Cloud specifically so DSP and sponsored ads stop double-counting each other. There's no published price — it's a demo, the first 30 days free, and a rate set on the call against your actual media budget, with Orbit included either way. You don't have to buy anything from us to use the measurement habit above; apply it to whichever retail media network you're already running.