A Slop University research poster tests whether showing a shopper's loyalty-points balance before payment changes checkout spending, finds no effect across 4,118 transactions, and reports partner sites ordered to withdraw marketing language the trial couldn't support.
A trial across fourteen supermarkets finds no spending effect, and partner sites must now drop the prompt's "savings nudge" label.
A shopper’s loyalty-points balance, shown on the terminal screen in the seconds before payment, changes nothing about what lands in the basket. A research poster from the School of Emergent Priorities tested the prompt against a matched control across fourteen supermarkets, and the two conditions came back statistically indistinguishable.
The Horizon Register doesn’t need a positive result to act on. It needs a margin narrow enough to trust, and this is the first checkout signal that’s cleared ours.
— Professor Verity Marris, Director, Trajectory Analytics Group
The trial randomised which of 4,118 transactions saw the balance before or after payment was complete, holding till staff blind to the condition running on a given day and fixing the equivalence margin before any data came in. Ilesanmi welcomed the attention the null result had already drawn from partner retailers, saying the project “raises exactly the questions a checkout screen should have to answer before anyone puts a claim on it.”
The project sits inside the School’s wider audit of anticipatory signals at the point of sale — an EFTPOS timestamp, a rewards balance, a queue log — each tested against a pre-registered ground rather than assumed useful because a screen already shows it. On the strength of this result, every partner site describing the balance prompt as a “rewards reminder” or “savings nudge” has one review cycle to withdraw the wording, with future checkout-screen behavioural claims held to the same trial before they can carry promotional support.
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.jdanbw.