Attribution is not incrementality.
If someone redeems a promotion, that does not prove the promotion caused the sale. The customer might have visited regardless. That distinction matters when you evaluate a campaign’s profit.
Choose one primary behaviour
For a local shop, it may be a verified return visit within 30 days. For a hotel, it may be a completed second booking. For e-commerce, consider incremental orders and contribution margin per eligible session.
Use a holdout when feasible
A randomized group that does not receive the offer can help estimate what would have happened without it. When that isn't feasible, use a transparent observational comparison and note competing explanations such as seasonality, inventory or paid media.
Keep a plain-language scorecard
- Eligible audience and accepted entries.
- Event outcome, promised rewards and delivered rewards.
- Repeat rate or qualifying order rate in both groups.
- Incremental gross margin before and after relevant campaign charges.
- Customer support issues and unfulfilled benefits.
Numbers that look smaller but are attributable to a credible method are worth more than impressive ones without evidence.