What Changed
On July 7, Amazon Ads began showing DSP advertisers something it had withheld for years: which specific retailers convert their ad exposure into a sale. Omnichannel Metrics, the measurement product Amazon uses to track purchases that happen off Amazon after a shopper sees a DSP ad, now breaks results out by named retailer instead of a vague "beyond Amazon" bucket. An advertiser running a DSP campaign can see that a shopper who saw an ad on Amazon bought the product at Target, at a regional grocery chain, or at Walmart. That breakout is available at both the study level and the individual campaign level.
This did not appear out of nowhere. Amazon added product-category breakouts to Omnichannel Metrics in April, using the same exposure-based method to show which categories, not which stores, ad-exposed shoppers were buying into. July's update is the second step in a build-out that started as "some of your ad exposure drove sales somewhere else" and is now specific enough to name the somewhere else.
The Infrastructure Behind It
None of this works without a way to get a brand's own customer data into Amazon's systems in the first place, which is what Ads Data Manager (ADM) is for. Amazon walked through ADM's mechanics in a technical session on May 1. Brands create a data room, configure a dataset, upload hashed audience data, and push it out to DSP, Amazon Marketing Cloud and other Amazon ad products from one interface. ADM sits apart from DSP and the ad console rather than inside them, and it accepts data from connected CDPs directly. In effect, Amazon has built a front door for first-party data specifically so it can be reused across every measurement and targeting product Amazon sells, on Amazon's terms.
Instacart Is Doing the Same Thing
Amazon is not alone. Instacart launched Data Hub, its own clean room, at CES in January, giving CPG brands and agencies a way to join their first-party data with Instacart's grocery purchase signals inside a controlled environment. The tool lets brands build audience segments for activation off the Instacart marketplace, analyze product affinities, and measure campaign impact with more than one attribution approach. It has been in pilot with select agencies and is expanding through the year.
Two of the largest retail media networks in the country independently built the same kind of product in the first half of 2026: a clean room that lets a brand's own data meet the retailer's purchase data, feeding attribution that reaches past the retailer's own site. That is not a coincidence. It is retail media catching up to a demand marketers have been making since third-party cookies started disappearing: prove the ad worked somewhere other than inside your own walled garden.
What It Does Not Prove
Read what Amazon actually says about the method before treating a retailer breakout as gospel: it is exposure-based measurement, meaning it counts a sale that happened after a shopper saw the ad. That is a real signal, but it is not the same as proof the ad caused the sale rather than simply being present near a purchase that would have happened anyway. The industry has spent the last two years arguing about the gap between reported and incremental return on ad spend for exactly this reason. A retailer naming Target instead of "another retailer" is a genuine improvement in specificity. It is not a substitute for a marketing team's own incrementality testing.
There is also the obvious incentive problem: the company building the measurement tool is the same company selling the media. A more detailed report from Amazon is still a report from Amazon about Amazon's own ad exposure.
What to Do Differently
Marketing teams running DSP or Instacart budgets should treat the new retailer- and category-level breakouts as one input into channel mix decisions, not the deciding one. Keep running independent lift studies or marketing mix modeling alongside them. Do not retire in-house measurement just because the platform now offers more detail.
Anyone connecting a CDP to Amazon should treat ADM as a one-way door. Once first-party data is uploaded into a retailer's clean room, it gets reused across that retailer's own targeting and measurement products. That decision belongs with whoever owns data governance, not just whoever owns the media buy.
Any brand evaluating retail media budgets should ask Amazon, Instacart, and any other network making this pitch for the same thing: the raw exposure and purchase counts behind the breakout, not just the summary number. Get that before deciding whether a claimed retailer-level lift changes how many dollars a channel gets next quarter.