What an In-Store Ad Can Honestly Claim

The subject of this series is the causal evaluation of in-store retail media, with digital signage at the point of sale as the focal case. The final question is not whether an in-store ad can influence behavior; the evidence says it can matter under reported conditions. The question is how far a claim may travel—from a message rendered, to an opportunity to see, to a featured-product purchase, to incremental sales, and finally to an economic decision. The useful judgment is to stop each claim at the strongest boundary the evidence actually crosses.

That boundary is not a sign of failure. Retail media is valuable partly because a retailer can connect inventory, store context, exposure estimates, and transaction data more closely than many other advertising environments. But connection is not causation, and causation is not profitability. A disciplined report keeps the links visible so that a retailer, brand, or analyst can improve the next decision rather than borrowing certainty from a later step.

A Claim Is a Chain, Not a Number

Start with the object being bought. The IAB defines in-store retail media as advertising inventory that uses retail data for planning, execution, and measurement in a store and is available to brands. The current standards focus on digital screens, audio, and connected shopping formats, while recognizing a wider field of static and physical activations. This is a retailer-controlled medium, not simply a generic screen in a public place.

The first context is the store zone. Exterior, entrance and out-of-category, checkout, in-aisles, and other or connected-store locations put a message into different relationships with movement, dwell, product proximity, and purchase timing. A screen beside a featured product may create a more direct path to a choice than a screen near an entrance. A checkout message may meet a shopper after many decisions have already been made. A zone is not a performance ranking, but it is a condition that must remain attached to the report.

The next event is delivery. An ad play says that a unit was displayed or rendered on a particular format. It can tell a retailer whether the booked creative ran and whether the touchpoint complied with its schedule. It cannot tell the retailer whether anyone was nearby, whether the product was available, or whether the shopper would have chosen the product without the message.

Exposure measurement adds people and conditions to the delivery record. Gross impressions can use the number of people present in a display exposure zone while the display is functional. Opportunity to see can describe a potential chance to view or hear the activation. Presence, dwell, and viewability can make that opportunity more plausible. A likelihood-to-see adjustment can add sensor- or analytics-based refinement. Each is useful for a different question, but an opportunity remains an opportunity. It is not confirmation of attention and is not a sale.

The sales outcome is another step. It can concern purchase incidence for the featured product, spending among purchasers, total brand sales in the same category, category sales, new-to-brand or new-to-category shoppers, repeat shoppers, or the timing of purchase. These outcomes share a campaign but not a denominator. The phrase “sales lift” is incomplete until the product, brand, category, shopper group, and time window are stated.

Only after the outcome is defined does incrementality enter. Incremental sales ask what would have happened without the media. That counterfactual cannot be read from the fact that a message preceded a transaction. It must be approached through a credible comparison or model that addresses other activity, price and promotion, store format and size, traffic, timing, and the coverage of identified shoppers. An ad play is therefore not “one step away” from an incremental sale in a simple funnel. It is one observation in a chain whose later links require additional evidence.

flowchart LR A["Retail-media inventory and store zone"] --> B["Ad play: message rendered"] B --> C["Exposure estimate: presence, dwell, viewability, or OTS"] C --> D["Observed purchase outcomes"] D --> E["Featured-product purchase"] D --> F["Same-brand or same-category outcomes"] D --> G["Spending and purchase timing"] E --> H{"Comparable counterfactual and relevant time window?"} F --> H G --> H H -- yes --> I["Incremental-sales or consumption claim"] H -- no --> J["Observed or presumed lift only"] I --> K{"Cost, margin, coverage, and privacy addressed?"} K -- yes --> L["Economic decision"] K -- no --> M["Behavioral evidence; ROI not established"]

The diagram's gates answer a common reporting error. A campaign can have a functioning screen, a large exposure estimate, and a positive transaction result while still lacking a credible counterfactual. Conversely, a campaign can have a useful behavioral signal without yet supporting a profit decision.