Retail media networks — advertising sold by retailers on their own sites, apps, in-store screens and increasingly across the open web — captured roughly $128 billion in advertiser spend in 2026. The reasons are structural: logged-in shoppers provide first-party data that survived cookie deprecation, ads sit at the point of purchase, and the retailer can close the loop between an impression and a sale with its own transaction data.
Amazon remains the largest network, followed by Walmart Connect, and then a long tail of grocers, pharmacies, marketplaces and travel platforms. For CPG brands retail media is already the biggest digital line: 67% of CPG advertisers rely on retail media first-party data and report an average 2.8x higher conversion rate than off-platform targeting.
What makes retail media different
- Targeting on real purchase history, not inferred interests.
- Closed-loop attribution: the retailer knows who bought, online and in store.
- Placements where intent is highest: search results, product pages, cart, checkout and receipts.
- Off-site extension: retailer audiences used to buy CTV, social and display across the web.
The formats that matter
Sponsored products in search remain the workhorse because they capture demand with a clear ROAS. Sponsored brands and display build category share. Off-site and CTV buys using retailer audiences are the fastest-growing segment, because they combine broad reach with purchase-based targeting and still report back to sales. In-store digital screens are emerging as the bridge between e-commerce data and physical shelves.
How to buy well
Start with the retailers where you already have share
Retail media works best where your products are stocked, rated and competitively priced. Ads cannot fix out-of-stocks or bad reviews; they amplify them.
Separate harvesting from growth
Branded and category-leader keywords harvest existing demand and show high ROAS; competitor conquesting and new-category terms drive growth at lower ROAS. Budget them separately with different targets or the algorithm will starve growth.
Feed the networks better data
Clean product content, complete attributes and rich media raise organic rank and quality scores alike. The same feed discipline that agentic commerce demands pays off here.
Use clean rooms for cross-channel questions
Data clean rooms let a brand and a retailer match their first-party datasets in a privacy-controlled environment to measure reach overlap and incremental sales without sharing raw records. They are the only reliable way to compare retail media with other channels on equal terms.
Measurement: beyond ROAS
Retailer-reported ROAS is high partly because the ads reach shoppers who were about to buy. Insist on incrementality tests — geo or audience holdouts — and on new-to-brand metrics. Feed retail media into marketing mix modelling alongside other channels; 46.9% of US marketers are increasing MMM investment precisely because platform attribution cannot settle budget questions across walled gardens.
Common mistakes
- Judging every network on the same ROAS target.
- Running retail media in isolation from trade and promotion plans.
- Ignoring off-site retailer audiences because they look like display.
- Letting the retailer’s managed service set bids without margin data.
What to expect next
Consolidation of smaller networks into a few ad-tech platforms, standardised measurement under industry bodies, retail media inside agentic shopping assistants, and more in-store inventory. Brands that build retail media capability now — people, data feeds, measurement — will be buying the most accountable media in the market.
Keep reading: Amazon Ads · Advertising