Receipt rewards and shopper marketing
The Receipt Rewards Offer That Mistook Brand Affinity For Lift
A realistic measurement scenario showing how a useful advertising channel can still produce a misleading read when the comparison group is wrong.
Archetype: Receipt-scanning rewards app with personalized brand offers
Bias mechanism: Users who clip, activate, or submit qualifying receipts are already more engaged with the category and more likely to buy the brand. The action required for treatment reveals intent that the control group never had to reveal.
Business model pressure
Brands fund offers, shopper campaigns, or audience programs inside a rewards app where users submit receipts and earn points.
Advertiser proof claim
A case study can show that offer activators bought more of the brand. The relevant question is how much of that difference existed before the offer.
Statistical result
| Metric | Naive read | Stratified read | Modeled benchmark |
|---|---|---|---|
| Conversion lift | 15.8 pts | 1.1 pts | 0.7 pts |
| Incremental conversions | 9,139 | 640 | 406 |
| Incremental revenue | $283,314 | $19,827 | $12,580 |
| ROAS after media cost | 2.02x | 0.14x | 0.09x |
The naive analysis is 14.3x larger than the stratified estimate in this worked example.
The advertiser-facing story
The offer appears to create a large sales lift because buyers who interacted with the offer were far more likely to submit a qualifying receipt.
What broke
The treatment condition requires a behavior tied to category interest. A shopper who activates a cereal offer or submits a grocery receipt is not comparable to an app user who never saw, clipped, or acted on the offer.
Better design
Randomize offer visibility among eligible users, compare within pre-offer brand-affinity strata, distinguish activation lift from purchase lift, and include pre-period purchase balance.
Propensity-strata audit
The adjusted estimate compares treated and control users within similar treatment-propensity strata. That does not replace a randomized holdout, but it shows how much of the naive result was carried by who entered treatment.
| Propensity stratum | Users | Treated conversion | Control conversion | Within-stratum lift |
|---|---|---|---|---|
| 2 | 3,207 | 8.4% | 5.4% | 3.0 pts |
| 3 | 8,204 | 9.9% | 8.3% | 1.5 pts |
| 4 | 9,797 | 13.3% | 12.3% | 1.0 pts |
| 5 | 9,288 | 16.5% | 16.8% | -0.2 pts |
| 6 | 9,582 | 22.4% | 21.2% | 1.2 pts |
| 7 | 12,478 | 30.2% | 28.8% | 1.4 pts |
| 8 | 18,416 | 39.2% | 39.0% | 0.2 pts |
| 9 | 20,036 | 52.2% | 50.4% | 1.8 pts |
| 10 | 3,989 | 65.1% | 63.4% | 1.7 pts |
Takeaway
A strong advertiser report should not stop at attributed conversions. It should show how the comparison group was built, whether treatment users had stronger prior intent, and how much of the result survives a better counterfactual.
Use this case in a readout review
After reading the failure mode, move from diagnosis to a better question set with these practical tools.
- Claim confidence rubricScore whether the readout supports strong, qualified, or tentative claim language.
- Measurement method selectorChoose the method that best fits the question: holdout, matched market, MMM, survey readout, or QA.
- Retail media incrementality checklistTest whether reward activation measured brand affinity more than campaign-caused demand.