By Muqadas Batool
SEO Content Writer
Published: 26/09/2026
Two retail media networks can report the same campaign and still tell two different stories. One might report $100,000 in attributed sales using a 14-day window. Another might report $125,000 using a 30-day window, while also counting view-through conversions and brand-halo sales differently. Both reports can be internally consistent. They still do not create a clean comparison.
That is the core problem with retail media measurement in 2026. The industry can produce detailed reports, but identical-looking metrics can rely on different windows, exposure rules, buyer definitions, sales scopes, and attribution methods. A buyer can therefore receive more data and still have less certainty about what the numbers mean.
The argument is simple: retail media measurement is detailed, but detail does not automatically create comparability.
What retail media measurement actually delivers
Retail media measurement became attractive for a fairly obvious reason: the ad exposure can sit close to transaction data. A retailer can connect an ad interaction with a purchase on its own site, app, or, in some cases, in-store transaction data.
That creates useful visibility. Closed loop measurement can show whether exposed shoppers later purchased, how much attributed sales those shoppers generated, and whether they bought promoted products or related products. IAB/MRC guidance also calls for transparent reporting around methodology, attribution, limitations, and online versus offline outcomes.
But there is an important boundary.
A report can prove that a purchase followed an ad exposure. It does not automatically prove that the ad caused the purchase. That distinction matters because attribution answers “what received credit?”, while incrementality asks “what additional outcome did the advertising cause?”
This is where a trade budget shift can become difficult to evaluate. A buyer may move money toward a retail media network because its reported ROAS looks stronger, only to discover that the underlying measurement rules differ.
For buyers, the useful question is not simply whether a network can measure sales. It is whether the network explains exactly what it counts as a sale, when it assigns credit, and what sits outside the number.
The metrics, and where the definitions come apart
The most important gaps appear inside ordinary retail media metrics. The names look familiar, but the measurement rules underneath them can change.
| Metric | What it should mean | How definitions vary | Effect on comparability |
| Impressions and viewability | Ads delivered and, where applicable, actually viewable according to an agreed measurement standard | Some reports separate served, measurable, viewable, or opportunity-to-see impressions; in-store environments can use different exposure concepts | A larger impression count may reflect a different exposure definition, not better reach |
| Clicks | Recorded user interaction with an ad | Networks may report raw clicks, validated clicks, or different interaction events | CTR comparisons can become misleading if the numerator or denominator differs |
| Attributed sales | Sales receiving credit after an ad exposure or click | Product scope, online/offline sales, view-through rules, halo sales, and lookback windows can vary | One network can report more attributed revenue without generating more actual sales |
| ROAS | Attributed sales divided by media spend | Sales scope and attribution rules can change the numerator; fees and costs can also affect the denominator | ROAS is not comparable unless both sides use the same calculation |
| New-to-brand | Sales or buyers from customers considered new to the brand | Lookback periods can differ; Amazon uses a 12-month lookback, while Instacart documents a 26-week definition for certain reporting | A “new buyer” on one network may not meet the definition on another |
| Halo or offline sales | Related or in-store outcomes associated with advertising | Networks can include related SKUs, brand products, categories, or offline purchases using different matching methods | A broad halo can materially increase reported campaign value |
The differences are not theoretical. Amazon, for example, defines new-to-brand purchases using a one-year lookback, while Instacart documents a 26-week period for its new-to-brand reporting. Instacart also distinguishes direct sales from halo sales, while Amazon separately reports brand-halo activity.
So, can you compare two retail media networks? Yes, but only after you normalize the measurement rules.
At minimum, a buyer needs the same exposure definition, attribution window, sales scope, new-to-brand timeframe, halo treatment, and ROAS formula. Without those controls, the comparison may describe two different measurement systems rather than two different media performances.
For buyers evaluating RMN standards in practice, this is the real issue: a common metric name is useful only when the underlying definition travels with it.
Attribution windows, the quietest gap of all
Attribution windows can change reported retail media results without changing the campaign itself.
An attribution window defines how long after an eligible ad interaction a network can assign credit for a purchase. A longer window creates more opportunities for purchases to receive credit. A shorter window usually captures fewer later purchases.
Consider a simple campaign with $10,000 in media spend:
- A 7-day window identifies $20,000 in attributed sales.
- A 14-day window identifies $25,000.
- A 30-day window identifies $30,000.
The campaign has not changed. The reported ROAS has moved from 2.0x to 2.5x to 3.0x because the measurement window changed.
That makes attribution-window disclosure essential when comparing networks. Walmart Connect, for example, currently cites a 14-day attribution window for certain onsite search, onsite display, and offsite campaigns, while IAB Europe’s 2026 measurement standards use a 30-day lookback as the default while allowing flexible options.
The problem is not that one window is automatically right and another is wrong. The problem starts when buyers compare the resulting numbers without knowing the difference.
Incrementality, and why almost nobody runs it
An incrementality test in retail media asks a different question from attribution: would the exposed group have generated the same outcome without the advertising?
The basic design uses a treatment group that receives the media and a control group that does not. The buyer then compares outcomes between the two groups. A credible design needs a reasonable counterfactual, control over important sources of bias, and enough observations to distinguish genuine lift from normal variation. IAB’s 2025 incremental measurement guidance specifically emphasizes credible counterfactuals, bias control, and separating signal from noise.
That makes incrementality more useful for causal questions, but also harder to run than ordinary campaign reporting. A test can require enough spend, time, geographic or audience separation, and cooperation between the advertiser and measurement provider.
It would be too broad to say that almost nobody runs incrementality tests. Major retail media platforms increasingly offer sales-lift or incrementality approaches, and Walmart Connect, for example, documents both sales-lift studies and incrementality testing.
When a full test is not possible, a buyer can still ask for the network’s causal-measurement methodology, available test options, control design, minimum requirements, and a clear distinction between attributed and incremental outcomes.
That distinction matters. A sale that followed an ad is not automatically a sale the ad caused.
In-store measurement, the weakest link
In-store measurement creates another layer of uncertainty because the path from exposure to purchase often involves more inference.
Some outcomes can be measured directly through transaction records. Others depend on matching exposed shoppers to purchases, estimating exposure from store media delivery, or modelling relationships between media activity and sales.
The distinction should stay visible:
- Directly measured: a transaction or exposure event exists in the underlying data.
- Modelled: a statistical method estimates an outcome that the available data cannot observe directly.
- Inferred: the reported result follows from assumptions or relationships rather than a directly observed transaction.
IAB and IAB Europe’s in-store guidance recognizes multiple exposure concepts, including ad play, gross impressions, opportunity to see, and likelihood to see. It also recommends defined reporting periods around exposure and sales measurement.
The same issue appears when comparing onsite and offsite activity. An onsite ad may sit close to a retailer’s own transaction data, while offsite exposure may require additional identity, matching, or measurement processes.
The important point is not that one method is automatically unreliable. It is that buyers need enough disclosure to understand what the number actually represents.
What a buyer can demand right now
Buyers do not need to wait for perfect market alignment before asking for comparable reporting.
The Retail Media Measurement Contract Test
Use this checklist before committing budget:
- Put metric definitions in writing.
Define impressions, clicks, attributed sales, ROAS, new-to-brand, halo sales, and offline sales before the campaign starts. Ask for the formula, eligible products, exclusions, and data source. - Require attribution-window disclosure.
Record the click-through and view-through windows separately. If the network can customize them, specify the exact window used in reporting. - Agree on a test before signing.
If incrementality matters, decide in advance whether the campaign will use a holdout, geo test, match-market design, or another approved methodology. IAB guidance lists experiments, model-based counterfactuals, econometric models, and hybrid approaches among available methods. - Demand useful reporting detail.
Ask for enough breakdown to reconcile spend, exposure, conversions, attributed sales, and other outcomes. IAB/MRC guidance specifically emphasizes granular reporting and disclosure of methodology, limitations, biases, and non-measured impressions. - Use third-party verification where available.
Where an independent measurement or verification provider can validate delivery or outcomes, establish what it will verify and whether its methodology differs from the network’s internal reporting.
This is a practical measurement agreement, not a request for unlimited data.
The goal is simple: make sure that when the campaign ends, the buyer can explain what happened, how the number was calculated, and what the number does not prove.
The read
The measurement gap will probably not disappear simply because the industry has more guidelines. Alignment can help, but buyers ultimately create pressure through procurement.
When a buyer asks two networks to report the same metric using the same definition, attribution window, sales scope, and test methodology, the commercial conversation changes. The trade budget shift then rests on comparable evidence rather than whichever dashboard produces the largest ROAS.
A separate article can cover the wider standards debate. Here, the practical issue is narrower: buyers need comparable definitions before they need more metrics.
The action for this quarter is straightforward: take the measurement section of your next RMN agreement and turn every important metric into a written definition before you approve the spend.
FAQ
Is retail media measurement reliable?
Retail media measurement can provide strong visibility, especially when a network can connect ad exposure with its own transaction data. But reliability depends on the metric definition, attribution method, data quality, and disclosure. A reported attributed sale does not automatically prove incremental sales.
What is closed loop measurement?
Closed loop measurement connects advertising exposure with downstream retail outcomes, usually through a retailer’s first-party transaction data. It can show which exposed shoppers later purchased and what sales received attribution. It does not, by itself, establish that advertising caused those purchases.
What does new to brand mean in retail media?
New-to-brand identifies shoppers or purchases considered new to a brand under a defined historical lookback period. The timeframe matters. Amazon currently documents a 12-month lookback, while Instacart documents 26 weeks for certain new-to-brand reporting, so buyers should never compare the metric without checking the definition.
How do attribution windows change reported results?
Attribution windows determine how long after an eligible ad interaction a purchase can receive credit. A longer window can capture more attributed purchases and therefore increase reported sales and ROAS. Comparing campaigns with different windows can make one campaign appear more efficient even when the underlying demand is similar.
Can you compare two retail media networks directly?
Yes, but you need to normalize the measurement rules first. Compare the same attribution window, exposure definition, sales scope, new-to-brand timeframe, halo treatment, and ROAS calculation. If those inputs differ, the resulting figures describe different measurement systems rather than a clean performance comparison.
What is an incrementality test?
An incrementality test estimates the additional business outcome caused by advertising rather than simply counting outcomes after exposure. It usually compares a treatment group that receives advertising with a control group that does not. A credible test needs a defensible counterfactual and controls for meaningful sources of bias.
Subscribe to the daily brief for concise updates on retail media, commerce, measurement, AI, and the changes that matter to marketers and buyers.
