How many of your coupon redemptions are discounts on sales you already had?

How many of your coupon redemptions are discounts on sales you already had?

A subscription retailer's promo codes showed a glowing return — every redemption tied to a sale. The skeptical question: how many of those buyers were going to purchase at full price anyway, making the coupon pure margin leakage?

The test. The team ran an incrementality experiment: randomly withholding the discount from a control segment at checkout-intent and comparing purchase rates against a treated segment that received it. This converts "redemptions" (correlation with sales) into "incremental sales" (caused by the offer).

What such tests find. Across published retail and subscription incrementality studies, a large share of redeemed promotions go to already-decided buyers. Documented cases report that 40-60% of coupon redemptions were non-incremental — the customer would have converted without the discount — meaning the headline ROAS overstated true profit-positive lift by roughly 2x. The incremental subset was real but far smaller than redemption counts suggested.

The nuance. A redemption is an observed correlation between offer and purchase. Incrementality is the counterfactual: would this person have bought anyway? Only the holdout distinguishes them, and the difference is the entire profitability case for a promo.

Bottom line for practitioners: never judge a promotion by redemption volume or redemption-tied ROAS — both reward cannibalizing your own full-price sales. Run a randomized holdout on the offer itself; the gap between treated and control purchase rates is the only number that reflects incremental margin. In the documented cases, half the discount was buying nothing.