The experiment was simple and the retailer generous. For six months, shoppers at the checkout of a mid-sized grocery chain were offered one of two propositions, at random. Half were told that paying a small fee would secure a discount on a future visit. The other half were told that declining the same fee would forfeit a discount they had already earned. The discount was identical. Only the framing changed.
Across 40,212 baskets, the second group paid the fee far more often — at every basket size, in every store, in every week of the trial. The gap narrows as baskets grow, which is itself interesting: the loss of a small discount stings more when the whole shop was small. But it never closes.
What the numbers show
The size of the effect is roughly what the laboratory predicts, which is reassuring for the laboratory and mildly alarming for the rest of us. A premium of about 12 percentage points, paid to avoid a loss that is economically identical to a foregone gain, is a great deal of money once it is multiplied across a country's tills.
We are not publishing the retailer's name, at their request, and we have removed anything that would identify a store. The full method, the anonymised data and the code that produced every chart in this piece are linked at the end. Check our sums; that is what they are for.
