Incrementality in Retail Media: Meaning and Evidence
Incrementality estimates what advertising changes beyond the counterfactual. Learn why exposure comparisons, attributed ROAS and sequential periods need care.
Incrementality in retail media
Incrementality is the difference an intervention makes compared with what would have happened without it. In retail media, that may concern sales, orders or contribution. A credible estimate requires a defensible counterfactual; shoppers who happened to see an ad are not automatically comparable with shoppers who did not.
Ask what changed because of the intervention
A campaign can be associated with purchases without causing all of them. Incrementality asks about the additional or displaced outcome relative to the alternative experience. Define the metric and population before deciding whether a reported improvement answers that question.
The IAB incremental-measurement guidance discusses experimental and non-experimental approaches. Their assumptions and limitations differ. A randomized holdout is a useful design when feasible; an observed before-and-after comparison does not inherit its protections simply because it has a “control” label.
Compare an explicit counterfactual
In a synthetic illustration, equal-sized assigned groups produce £12 and £10 net sales per eligible shopper. The estimated difference is £2 per shopper under that design. Whether it is reliable or commercially meaningful still depends on uncertainty, implementation quality and the selected outcome. The arithmetic alone is not evidence from a real campaign.
Exposure selection matters. Shoppers who browse more may both encounter more ads and buy more. Comparing them with less active shoppers can mistake existing differences for advertising impact. Preserve assignment and eligibility rules when estimating a randomized treatment effect.
Keep scope and evidence class visible
An advertised product can gain sales while another product loses them. Examine the agreed product, brand and category scope to understand substitution. A positive product effect is not automatically incremental category profit.
The Intersport case states that its Bayesian comparison used sequential periods, not a randomized holdout. Its attribution evidence should remain labelled accordingly. For the practical design steps, use the retail-media incrementality guide.
Illustrative example
SYNTHETIC equal-size assigned groups:12 GBP net sales per shopper minus10 GBP=2 GBP estimated difference; uncertainty not calculated.
Sources
- Guidelines for Incremental Measurement in Commerce Media, checked 2026-10-25
- Intersport Austria retail-media case, checked 2026-10-25
How Particular Audience does it
In production
- Intersport Austria: Intersport Austria: five-week onsite-media launch and attributed brand-halo reporting (17.3× Reported attributed brand-halo ROAS (sequential periods, not a holdout))
Frequently asked questions
Does an exposed-versus-unexposed comparison measure incrementality?
Only if the design and assumptions establish a credible counterfactual. Exposure often depends on activity, eligibility and campaign delivery, which can also relate to purchasing. A simple comparison of those groups is not automatically causal, even when their headline demographics look similar.
Does incrementality always mean additional sales?
No. Define the outcome being estimated. A study can examine orders, revenue, contribution or another meaningful measure, and its effect can be positive, negative or uncertain. Do not substitute a sales effect for a profit effect without the required costs and accounting scope.
Can a before-and-after result still be useful?
Yes. It can describe rollout performance and help form hypotheses. State that promotions, seasonality, assortment and other changes may explain part of the difference. A causal interpretation requires a justified method and explicit assumptions, not merely sequential periods labelled test and control.
Related guides
- Retail media measurement: incrementality, attribution and the metrics that matter: How to measure retail media: closed-loop attribution, holdout incrementality, halo, new-to-brand and the effect of ads on the retailer's own conversion.
- Holdout tests in advertising: A holdout assigns an eligible group to a comparison experience. Learn how stable assignment, valid outcomes and uncertainty support advertising measurement.
- ROAS in retail media: return on ad spend: Understand retail media ROAS, its sales and spend definitions, and why attributed return differs from profit and incremental return on advertising spend.
- How to measure retail media incrementality: Plan a retail media holdout with stable assignment, comparable outcomes and visible uncertainty. Separate incremental sales from attributed ROAS and brand halo.