Closed-Loop Attribution in Retail Media

Closed-loop attribution is the practice of matching shoppers’ ad exposure to their purchases in a retailer’s own sales records, so a retail media network can report which sales followed its ads. Also called closed-loop measurement, it works because the retailer sells the ad and records the sale, online and in stores.

Learn more

By Mateusz Drela, CEO of RMIQUpdated · 5 min read

How closed-loop attribution works in retail media

The loop has four parts.

  1. Know who saw the ad. Networks identify exposed shoppers through login data, user IDs and loyalty programs, which IAB/MRC call highly accurate deterministic data.
  2. Record the purchase. The retailer logs the sale, online or in store, against the same shopper or household.
  3. Match inside a window. A purchase counts only if it lands within the attribution, or lookback, window. IAB/MRC leave the length to each network. They call 3, 7, 14, 28 or 30 days typical in digital measurement and require the window to be disclosed before the campaign runs.
  4. Model the gaps. Cash payments, limits on card and loyalty capture, and device privacy settings leave some buyers unidentified. Many networks extrapolate from identified shoppers and should disclose the extrapolated share of attributed sales.

The output is attributed sales, which feed return on ad spend (ROAS).

Closed-loop attribution at Walmart, Kroger and Amazon

Walmart Connect says its closed-loop measurement ties ad exposure across Walmart's digital properties and offsite channels to sales on Walmart's site and in its stores. Walmart had more than 4,600 U.S. stores as of February 2025. Walmart Connect says advertisers read the results in its Ad Center, alongside incrementality and brand impact reporting.

Kroger Precision Marketing builds its loop on Kroger's loyalty ecosystem, which its FAQ page put at more than 62 million verified households in September 2026, and measures media against real household purchases in stores and online.

Amazon's version is a clean room. In Amazon Marketing Cloud, advertisers query pseudonymized, event-level Amazon Ads signals such as ad-attributed impressions, clicks and conversions, can join their own inputs, and receive only aggregated outputs. These consoles and clean rooms are the tools brands use to read closed-loop results.

Onsite, offsite and in-store: where the loop is tight and where it is modeled

The loop is tightest for onsite ads, where the retailer serves the ad and records the purchase itself, the direct shopper-level connection IAB/MRC describe as ideal. Offsite ads run on inventory the retailer does not own, so exposure has to be matched to shoppers through identity data. Where deterministic data runs short, IAB/MRC accept probabilistic data and ask for the method to be disclosed. One agency executive told Digiday in December 2025 that advertisers in new offsite shoppable partnerships will want a closed-loop measurement solution.

In-store sales close the loop only when the buyer is identified at checkout through a loyalty card or payment data. Kroger matches in-store purchases to loyalty households, and unidentified store buyers fall into the extrapolated share. IAB/MRC also warn that digital-only attribution can credit sales that television helped drive.

What closed-loop attribution cannot tell you

Closed-loop attribution shows that a sale followed an ad. Whether the ad caused it is a separate question. IAB assigns that question to incrementality, which tests what marketing caused, and leaves attribution and ROAS to describe what happened.

Attribution has other blind spots. Without strong off-platform reporting, it cannot separate a new customer from one who already intended to buy, as AdExchanger reported in 2023. It may miss other drivers of a sale, such as price, promotions and competition. It sees only the retailer's own sales, and IAB/MRC ask networks to disclose any modeling of sales at other retailers. The results also come from the network that sold the ad, a setup one analytics executive told Digiday in January 2025 leaves retailers "grading their own homework". Measuring cause takes a control group, the basis of incrementality and incrementality testing.

Why it matters for brands and sellers

Attributed sales are the numbers retail media budgets get argued with. Before comparing ROAS across two networks, check three things behind it:

  • The attribution window, which each network sets and must disclose.
  • The extrapolated share, the part of attributed sales modeled for shoppers the retailer could not identify.
  • The SKU scope, meaning whether attributed sales cover only the advertised SKUs or also halo products. IAB/MRC set SKU-level reporting as the minimum where relevant and want halo products disclosed.

Digiday tracked mounting scrutiny of network-reported numbers through 2025. Brands that read results by SKU, with each network's window and modeling in view, make better calls on where the next dollar goes. Compare how each network measures in the retail media network profiles.

Frequently asked questions

What is an example of closed-loop attribution?

Walmart Connect is a clear example. A shopper sees a brand's ad on Walmart's site, app or an offsite channel, then buys the product online or in a Walmart store. Walmart matches the exposure to the purchase in its own records and reports the sale.

Which retail media networks offer closed-loop attribution?

Any network whose retailer records the purchase can close the loop. Walmart Connect ties ads to sales on its site and in stores, Kroger Precision Marketing matches ads to loyalty households' purchases, and Amazon provides event-level ad data through its Amazon Marketing Cloud clean room.

Is closed-loop measurement the same as closed-loop attribution?

Yes, the terms are used interchangeably. The IAB/MRC guidelines describe event-level attribution as "also sometimes referred to as closed-loop attribution", and Walmart Connect and Kroger Precision Marketing call the same capability closed-loop measurement. Both mean matching ad exposure to recorded purchases.

Does closed-loop attribution prove that an ad caused a sale?

Closed-loop attribution only shows that a purchase followed an ad within the attribution window. Proving that the ad caused the purchase requires comparing exposed shoppers with a control group that did not see the ad, which is how incrementality is measured.

Does closed-loop attribution work for offsite and CTV ads?

Partly. Offsite, connected TV and streaming ads run outside the retailer's own site, so the network has to match exposure to shoppers through identity data or statistical modeling. IAB/MRC accept probabilistic data when deterministic data is insufficient, provided the method is transparent and disclosed.

How RMIQ can help you with Closed-Loop Attribution

RMIQ enables true closed-loop attribution for retail media by unifying planning, activation, and measurement across Walmart, Instacart, Amazon, Sprouts, Thrive Market, Target, Uber, and more, eliminating siloed dashboards that obscure the link between spend and sales outcomes. Its multi-agent AI architecture operationalizes attribution signals in real time: autonomous agents ingest SKU-level performance, keyword trends, and cross-network learnings, then adjust bids, reallocate budgets, orchestrate A/B tests, and refine strategy to maximize incremental return with minimal manual oversight.

Because the platform covers over twenty retail media networks—reaching up to 85% of the U.S. retail audience—brands can attribute impact consistently using a single taxonomy and consistent identifiers, track against unified KPIs, and compare like-for-like performance without manually stitching datasets. Closed-loop feedback is embedded directly into execution; sales and ROAS outcomes flow back into the agents’ models to continuously optimize toward profitable growth, with customers reporting an average ROAS lift exceeding 50% and up to five dollars in new sales per dollar invested. The consolidated interface centralizes reporting, experiment results, and campaign workflows, enabling teams to quantify incremental lift at the SKU, keyword, and audience levels, validate hypotheses rapidly, and redeploy budgets to the highest-yielding channels and tactics.

For enterprises managing thousands of SKUs, RMIQ scales attribution granularity without sacrificing speed, while smaller brands benefit from rapid onboarding—often in five minutes—and strong support to accelerate time to value. By fusing AI-driven optimization with end-to-end measurement, RMIQ turns attribution from a retrospective report into a real-time control system, empowering B2B marketing and commerce leaders to prove media impact, reduce wasted spend, and systemically compound performance across the retail media landscape. This unified approach streamlines reporting, enables forecasting and scenario planning, aligns agencies and internal teams on a single source of truth, and de-risks investment decisions by linking every tactic to sales velocity and margin contribution.

Our Current Partners

We are already helping leading retailers and platforms grow their retail media businesses, including:

Become a partner

Drop your email 

and we’ll show you how to double your retail media ROAS – no strings attached