Retail Attribution: How Retailers Prove In-Store, Store Visit and Product-Level Sales

Retail attribution is the practice of crediting a sale, online or in a store, to the ad that preceded it, and it is harder in retail media because most of a retailer's sales still happen in stores, where the ad and the purchase are recorded in different systems. E-commerce is only 17.1% of US retail sales, according to the US Census Bureau's seasonally adjusted estimate for the second quarter of 2026.

Osmos powers the in-store retail media network of Mr DIY in Malaysia, with more than 1,200 in-store screens live across more than 1,100 activated stores, where store sales are the proof brands ask for.

Osmos is a white-label retail media operating system for retailers and marketplaces. Its modular apps for onsite, offsite and in-store ads sit over your existing stack, so you can break your monetisation ceiling faster without rebuilding it.

Last updated: November 2026. Reviewed by Kunal Damgude, Growth and Product Marketing Manager.

Closed-loop attribution is the base layer: it matches an ad exposure to an order in the retailer's own transaction data. Attribution models, metrics and incrementality tests sit on top of it and together form the retail media attribution and measurement stack.

What Is Retail Attribution, and What Makes It Different in Retail Media?

Retail media attribution is more complex than search or social attribution because the sale can land in places the ad never touched: the retailer's website, its app, a physical store or another retailer's shelf. Each place records the purchase in a different system, often under a different identifier, and many shoppers buy with no identifier at all. On top of that, the choice of attribution window, the treatment of views and the product-level halo rule each change the return a retailer reports for the same campaign.

The Interactive Advertising Bureau (IAB), the US digital advertising trade body, and the Media Rating Council (MRC), the independent body that audits and accredits media measurement, frame attribution the same way in their Retail Media Measurement Guidelines: credit for a sale or a visit is assigned to a specific piece of marketing, ideally at the level of a shopper or household. Osmos's onsite and offsite apps report attributed sales from the retailer's own order data, and its in-store apps measure at store level.

In paid search or social, attribution mostly lives inside one platform: the ad is served, the click is tracked and the conversion is recorded in the same ecosystem. In retail media the ad server or supply platform delivers the ad, a commerce engine (or a point-of-sale system in a store) records the order, and a separate analytics layer produces the report. Joining them needs a consistent shopper identity, accurate product (SKU, or stock-keeping unit) data and aligned timestamps across every surface. If identity, product data or timestamps disagree between these systems, the attributed sales figure is wrong.

Brands lose confidence in a network whose attributed sales they cannot check. A survey of 166 retail media advertisers co-authored by Stratably, a retail media consultancy, and published on 4 February 2026 found that "Seven in ten brands report meeting or exceeding their retail media goals in 2025," yet "Only 15% of marketers say their organization is very or extremely effective at measuring retail media performance overall."

IAB Europe, the European digital advertising trade body, published its Commerce Media Measurement Standards V2.1 on 22 January 2026. The standard sets a 30-day lookback as the default reporting window and adds a formal definition of incrementality.

How Do Retailers Measure In-Store Attribution?

In-store attribution is the practice of linking an ad a shopper saw, on a store screen, in the retailer's app, on its website or offsite, to a visit or a purchase in a physical store. Retailers prove it to brands with store-level signals, because the shopper in front of a store screen is rarely identified and the till records the sale in a different system from the one that served the ad.

Retailers count the exposure before they attribute anything to it. The IAB and IAB Europe in-store retail media standards were written to "offer unified definitions, measurement standards, and guidelines for ad formats and store zones." The IAB/MRC guidelines (section 6.5) define the countable audience by place: "Establishing screen traffic counts of appropriate quality is essential. These counts determine the potential audience for each screen, focusing on individuals present within one or more designated zones where they have the opportunity for exposure to the screen." Sensor-verified impressions confirm that someone was present in that zone when the ad played; where no sensor is installed, an opportunity-to-see count based on footfall stands in. IAB Europe's in-store measurement draft, reported in September 2026 by ExchangeWire, an ad-tech trade publication, "favours footfall and the opportunity for visual contact with an ad" as the basis of an in-store impression; IAB Europe took public comments on it until 23 October 2026.

In-store media is booked and priced by place and time (store, day and hour), so its attribution is store-level by design. A retailer runs a dairy brand before noon or a beverage brand on weekend afternoons and measures the stores and dayparts that carried it. The screen plays the same ad to everyone who walks past it at that hour, so the proof a brand receives describes stores and hours.

Retailers turn that exposure into proof of a visit or a sale in three ways, and each answers a different question for the brand:

1. Loyalty, card or login matching. Where the shopper was identified when the ad was served, in the retailer's app, on its website or offsite, a loyalty ID, payment card or app login links that exposure to a basket at the till. This is deterministic and proves the individual purchase. Its reach is limited by what section 3.6 of the IAB/MRC guidelines lists as the reasons a retailer cannot identify every buyer: cash transactions, card and loyalty capture rates, and privacy settings on devices and browsers. A retailer reports the matched share and the modeled share separately.
2. Store-visit measurement. Footfall counters and location panels show whether a campaign brought more shoppers into the stores that carried it. Footfall data shows the visit quickly, but to see what those shoppers bought, the retailer still needs loyalty matching or a store-level sales test.
3. Test-versus-control stores. A store-level lift test compares sales in stores that carried the campaign with sales in matched stores that did not, over the same period. It is the most defensible read when no click exists, because it measures sales directly and needs no shopper identity. Section 6.4 of the IAB/MRC guidelines adds that "store-level daily transaction counts can be reviewed and cross-referenced against data collected by sensors." The test takes careful design, because control stores need to match the exposed stores on size, region and trading pattern.

Out-of-home (OOH) attribution models work the same way for store screens: exposure from verified impressions or opportunity to see, visits from footfall, and sales from store-level comparisons.

Osmos In-Store retail media follows the place-and-time model, with targeting by store, city, zone or aisle and traffic-based dayparting. Osmos Footfall Analytics pairs each ad play with a sensor presence signal within plus or minus 5 seconds, from certified sensor partners such as Quividi, Advertima and RetailNext, to produce IAB-aligned verified impressions, with a footfall-based opportunity to see as a bridge where sensors are not yet installed. Osmos stores no raw video or biometric identifiers, only aggregated demographic buckets. Store-level point-of-sale sales ingestion in Footfall Analytics is in Early Access today, and Store Lift, exposed-versus-control sales lift across stores, is listed as coming soon on the in-store page. Each instrument in detail, from QR codes to brand lift, is covered in Osmos's five in-store measurement instruments, and the wider direction of store networks is in in-store retail media trends for 2026.

How Does Store Visit Attribution Work?

Store visit attribution credits an ad with a visit to a physical store. Door counters or sensors measure it at store level, by counting whether traffic rose in the stores and hours that carried a campaign against comparable stores or periods without it. Location panels measure it across devices, by comparing how often devices seen near an ad later visited a store with how often unexposed devices did. Ad platforms that report store visits model them: Google's help page for store visit conversions says they "use anonymous, aggregated statistics which are then extrapolated to represent the broader population of your customers," and reports them only where an advertiser has enough ad interactions and foot traffic to pass its privacy thresholds. The IAB/MRC guidelines treat a visit as an outcome attribution can credit, defining attribution as "the practice of assigning credit for consumer actions, like sales or visits, to specific marketing efforts." They also ask retailers to disclose the share of attributed outcome metrics, visits included, that is extrapolated.

A visit is a quicker and wider read than a sale, and it is often the only read where a store shares no transaction data. It is still a weaker outcome than a store sale, because a visit carries no basket and a panel covers only a sample of devices. Retailers therefore report visits as evidence of reach into the store, and pair them with a store-level sales comparison when a brand asks what the campaign earned.

How Do Retailers Measure Omnichannel Attribution Across Online and Store Sales?

Retailers measure omnichannel attribution by joining online orders and offline purchases to the same ad exposure, although the systems that record offline purchases rarely carry the same shopper identifier as the ad. The US Census Bureau put e-commerce at 17.1% of total US retail sales in the second quarter of 2026 (seasonally adjusted, released 18 August 2026), which leaves 82.9% of sales outside e-commerce. A retail media network that measures only online orders sees a minority of the sales its ads can influence.

The join between online exposure and offline purchase depends on a key the shopper carries into the store: a loyalty ID, a payment card or a logged-in app account. Loyalty-linked matching is deterministic, which is why it is the backbone of most cross-channel attribution for retail. Commerce media platforms measure online and offline sales together by matching online orders at the shopper level inside the retailer's own data, and offline sales at the shopper level where a loyalty or payment identifier exists and at the store level everywhere else. Where no key exists, part of the result has to be modeled.

Multi-banner groups add one more join, because the same shopper may hold separate accounts across banners; Osmos's multi-banner retail media guide covers that shared identity spine, and how shoppers move between those online and offline paths is set out in Osmos's guide to optimizing the shopping experience.

Osmos groups the five capabilities retailers combine to measure attribution across channels by function:

1. Retailer closed-loop reporting, which matches exposures to orders in first-party transaction data, onsite and offsite.
2. Store-level point-of-sale ingestion, which brings in-store sales into the same reporting as online orders.
3. Clean rooms, which let a brand and a retailer combine data for analysis without either side exposing raw customer records.
4. Incrementality testing, which uses holdout groups to show what the ads caused.
5. Media mix modeling (MMM), which works from weekly or monthly totals, so it still runs where shopper-level identity is missing.

Machine learning handles deduplication and modeling inside these capabilities and depends on an identifier to match shoppers. The February 2026 survey of 166 advertisers co-authored by Stratably shows how much brands want this solved: the measurement areas advertisers ranked highest were incrementality at 75% and cross-channel measurement at 59%.

What Is Retail Product Attribution, and How Do You Attribute Sales at the Product Level?

Retail product attribution is the practice of deciding which product sales a campaign claims: only the exact SKU advertised, its variants, the rest of the brand, or the wider category. A shopper who clicks an ad for one beverage flavour and buys another, or a larger pack, still responded to the ad, and the halo rule decides whether that sale counts.

The IAB/MRC guidelines (also section 3.6) set SKU-level reporting as the recommended minimum because most spend promotes specific products, and ask retailers to be open about any halo products they count. Incremental, a measurement vendor, found in a June 2026 study reported by PPC Land, a digital advertising trade publication, that 93% of the measured impact of retailer search campaigns occurred within the same retailer's environment, which means the retailer's own SKU-level order data captures most of what onsite search ads do.

Fashion product attribution adds a variant problem. One style may exist as dozens of SKUs across sizes and colours, and stock-outs push shoppers between them, so a same-SKU rule undercounts. Grouping variants at the style level gives brands a truer read. For onsite listings, the creative is assembled from the product page itself (image, title, price and rating), so the lever a brand controls is listing quality. Osmos Product Ads reports back to SKU with brand halo included, which gives brands both the strict and the wide view in one report.

How Do Click, View and Window Choices Change What a Retailer Can Claim?

Click, view and window rules decide which purchases a campaign is allowed to claim, so the same campaign can report very different results under different settings. Campaign attribution in retail media is built from these choices, and each has a use and a limit.

• Click-through attribution credits a purchase only when the shopper clicked the ad first, which makes it hard to dispute and is why many networks start there; it leaves out display and video ads that shoppers saw and acted on later without clicking.
• View-through attribution also credits purchases after an ad was seen but not clicked. It reflects how awareness formats work in practice. Report views and clicks separately, so a brand can see how much of the result rests on exposure alone.
• Attribution windows set how long after the ad a purchase still counts. Section 3.6 of the IAB/MRC guidelines lists 3, 7, 14, 28 and 30 days as the common digital windows, prescribes none of them, and asks the retailer to disclose its choice ahead of the campaign and again alongside the results. A short window suits frequently bought grocery items; considered purchases such as electronics or furniture need a longer one.
• Single-touch and multi-touch models decide how credit is shared when a shopper saw several ads. Last-touch gives everything to the final interaction and is simple to run at scale; multi-touch (linear, time-decay or position-based) spreads credit across the path but needs cleaner identity data across every surface.

Any of these settings is defensible when the retailer discloses it before the campaign and keeps it the same from one campaign to the next. The full model list, with worked weightings, is in the measurement guide.

How Do Attribution Models Affect ROI for Retail Media Campaigns?

Attribution models affect reported ROI because they change the numerator: a wider window, view-through credit or a broad halo rule adds attributed sales, while a narrow click-only rule removes them, even though the campaign itself is unchanged. Return on ad spend (ROAS) is attributed sales divided by ad spend, so every rule upstream of that division moves the headline figure.

The direction of the error depends on the model: generous rules absorb sales that would have happened anyway, which flatters ROAS, while siloed last-touch rules can understate it. Incremental argued in its June 2026 white paper "The Cross-Retailer Halo Effect," covered by PPC Land, that siloed last-touch attribution "systematically understates the return on investment for upper-funnel retail media channels, in some cases by a factor of three to five." Awareness formats lose most under last-touch, because their effect arrives days later and through other touchpoints.

In commerce media platforms that publish their method, attribution works at the order level, and the window can be configured per format or channel. Osmos Product Ads, for example, gives order-level reporting with SKU, category and keyword performance, includes brand halo and supports customizable attribution windows. Osmos Offsite deduplicates conversions at SKU level across channels, with the window set channel by channel (for example 7, 14 or 30 days after a click). In commerce media, the wider category that includes payments and delivery platforms, the same mechanics apply. A disclosed, stable method lets brands compare performance across quarters.

How Do Retailers Attribute Mobile Sales Now the Device ID Is Gone?

Retailers attribute mobile sales through the shopper's own login, because shoppers move between an app, a mobile browser and a desktop, and the device-level identifiers that once tied those sessions together are now mostly unavailable. Since Apple introduced App Tracking Transparency (ATT) with iOS 14.5, apps need the user's permission to track across other companies' apps and sites. Adjust, a mobile measurement company, measured industry-wide opt-in at about 35% of users shown the prompt in Q2 2025, which is 2025 data and counts only users who saw the prompt.

Retailers now use the shopper's logged-in account, rather than a device ID, to connect purchases across devices. A retailer's app account or loyalty login is first-party, stable across sessions and consented within the retailer's own relationship with the shopper, so it works without cross-app tracking. Its limit is coverage: it sees only sessions where the shopper signed in. Guest browsing and logged-out mobile web still need modeled attribution, disclosed as such.

Retail media identity resolution is the process of linking a shopper's app, web, loyalty and store records into one profile, using deterministic keys such as login and loyalty IDs first and privacy-safe matching where those are missing. Osmos uses retailer-owned first-party data for targeting with no third-party cookies, which is the same identity layer attribution reads from. How that identity feeds ad decisions is covered in Osmos's guides to AI-powered ad serving and behavioral targeting and ad serving, targeting and attribution.

How Do Retailers Prove Offsite Results Inside Walled Gardens?

Retailers prove offsite results by measuring them against their own sales data, because platforms such as Meta, Google and TikTok report conversions inside their own environments, while the retailer's sales sit in its own transaction data, and neither side hands the other raw records. A brand that buys offsite through the retailer therefore receives two different stories: platform-reported conversions and retailer-recorded sales.

Offsite retail media reaches the retailer's known shoppers on the channels where they spend time, using the retailer's first-party audiences. Its result should be measured against the retailer's orders as well as the platform's own conversion count, because a platform-reported figure is not independently verified.

Osmos Offsite retail media closes that loop to the sale by tying offsite spend to the retailer's transaction data. Its live channels are Google, Meta and TikTok, with DV360 and Pubmatic also connected, and by default every campaign runs on the retailer's own seat in each channel. The channel choices behind an offsite program are set out in Osmos's four offsite retail media channels guide, and the measurement work that comes with adding surfaces is in expanding retail media beyond onsite into offsite. How the largest US networks report their own closed loops is compared in the Walmart, Amazon and Instacart deep dive.

Attribution or Incrementality: Which One Shows the Ad Caused the Sale?

Incrementality testing is the method that shows whether an ad caused a sale, by comparing purchases in an exposed group with a control group that did not see the ad. Attribution assigns credit for the sale across the ads the shopper saw before buying. Chapter 4 of the IAB/MRC guidelines describes incrementality as the potential causal impact of marketing, found by isolating results from every other business factor such as price, promotion or season.

To isolate retail media lift, a test holds back a control group that does not see the ad and compares its purchases with an exposed group over the same period. Attribution still matters, because it explains the path and distributes credit across formats; incrementality checks whether that credit reflects real additional sales.

Osmos Incrementality is in Early Access today: it holds back a control share of shoppers on onsite sponsored display campaigns and compares views, add-to-carts and purchases against the exposed group, over a window of at least eight weeks. Test design, sample sizes and holdout choices are covered in Osmos's incrementality analysis guide.

What Should a Retailer Building a Media Network Put in Place First?

A retailer should put a disclosed attribution method, a shared shopper identity and a holdout plan in place before adding channels, because together they decide whether brands believe the numbers that follow. Osmos recommends these six retail attribution strategies, in this order:

1. Publish one attribution method per format. State the window, the click and view rules and the halo level before each campaign starts, as IAB/MRC section 3.6 requires.
2. Link the loyalty or account ID to every channel. Onsite, offsite and in-store should read from the same first-party identity, with the matched and modeled shares reported separately.
3. Measure stores at store level. Use footfall, verified impressions and store-level sales comparisons, and plan in-store media by store, day and hour.
4. Report at SKU level, with the halo shown separately. Give brands the strict number and the wide number side by side.
5. Hold out a control group. Run incrementality on the formats where brands question attribution most, starting with awareness formats.
6. Keep the method stable as the network grows. Osmos's guide to expanding retail media without breaking measurement covers how to add channels without resetting the baseline.

Brands renew when they can compare reports across quarters under an agreed method, trace sales to their own products, read store-level data for in-store spend and see a lift result for part of the plan. The IAB/MRC guidelines put the standard this way: "The overarching themes are transparency, empirical support, and ongoing validation." Osmos Measurement brings these reads together for retail media networks, and a demo shows how they work on a retailer's own data.

Frequently Asked Questions

Which retail media networks provide closed-loop attribution?

Most large networks report from their own transaction data, but they differ in windows, identity coverage and in-store reach. The Walmart, Amazon and Instacart deep dive compares how three of them do it.

What agents or tools assist with attribution in commerce media?

AI agents in commerce media mostly help with the work around attribution: pulling reports, explaining why a campaign's attributed sales moved, and recommending budget or bid changes. They work on top of the retailer's attribution data, which stays the source of record. Why coordinated agents matter more than any single model is argued in Osmos's piece on multi-agentic AI orchestration in retail media.

Do third-party attribution tools replace a retailer's own attribution?

No. Brand-side analytics tools combine data from several retailers and media channels for the brand's planning, but they depend on what each retailer reports. The retailer's order-level data remains the source those tools read, so a clear, disclosed method at the retailer improves every downstream tool.

Can retail media attribution be real-time?

Exposure, click and online order data can be reported close to real time, so pacing and bids can be adjusted mid-flight. Attributed sales are only final once the attribution window closes, and in-store sales usually arrive as daily store totals, so final ROAS is a post-window number.

Does attribution work differently for restaurant and food-delivery advertisers?

Yes. On food-delivery platforms the order is placed and fulfilled inside the platform's app, so attribution runs on onsite and offsite channels, and windows are usually shorter than for considered purchases because a meal order follows the ad closely. How delivery platforms build that ad business is covered in Osmos's guide to food-delivery retail media ads.

What does a retail sales attribution example look like?

A shopper clicks a Product Ad for a 500 ml shampoo on Monday and buys the 1 litre bottle of the same brand in the app on Thursday. Under a 7-day click window with same-SKU matching, the sale is not attributed; with variant grouping it is attributed to the campaign; with brand halo it would also count other products from that brand in the basket. Because the same order can produce three different results, the retailer and the brand agree the matching rule before the campaign goes live.

Sources

1. IAB and Media Rating Council, Retail Media Measurement Guidelines, January 2024, sections 3.6, 4, 6.4 and 6.5.
2. IAB and IAB Europe, In-store retail media standards (public comment release), 3 December 2024.
3. IAB Europe, Commerce (Incl. Retail) Media Measurement Standards V2.1, 22 January 2026.
4. ExchangeWire, IAB Europe Advances In-Store and Travel Media Measurement with New Proposals Open for Public Comment, 17 September 2026.
5. US Census Bureau, Quarterly Retail E-Commerce Sales, 2nd Quarter 2026, released 18 August 2026.
6. PPC Land, Retail media's hidden ROI: how siloed attribution misses half the picture, 15 June 2026, reporting Incremental's white paper "The Cross-Retailer Halo Effect: What Siloed Attribution Misses."
7. Adjust, ATT opt-in rates: 2025 data and benchmarks, Q2 2025 data.
8. Skai with Stratably, "The 2026 State of Retail Media Measurement and Incrementality," 4 February 2026, survey of 166 retail media advertisers. skai.io/blog/the-2026-state-of-retail-media-measurement-and-incrementality/
9. Google, About store visit conversions, Google Ads Help, retrieved 12 November 2026.