A new Slop University paper treats a supermarket express lane's posted item limit as a threshold rule and finds it sorts baskets sharply while leaving lane service time where it was, even after nine stores lowered the number on the sign.
A regression-discontinuity study across 22 supermarkets finds shoppers adjusting their baskets at the limit and the transaction taking just as long
The School of Continuous Improvement has spent the year on measures that stand in for the things they measure. Its latest paper picks one that hangs from the ceiling of almost every supermarket in the country.
The number on an express-lane sign is understood to be about speed. Working from lane-level records for 38,412 checkout transactions across 22 stores of a single grocery chain, the team compared what happens either side of the twelve-item limit. The limit sorts, and sorts sharply: baskets over it almost never reach an express lane. The transaction itself does not notice. Service time either side of the cutoff differs by 1.7 seconds against a mean of 96.4, and the things that do move the clock — how the customer pays, whether anything has to be weighed — are not the things a sign can carry.
Shoppers, meanwhile, adjust to the number. Baskets cluster at exactly twelve, with a matching gap immediately above it. Nine of the stores lowered their posted limit from fifteen to twelve partway through the study: the clustering moved within three weeks, the service time did not, and the chain’s measure of express-lane compliance rose from 88.1 to 96.4 per cent.
A number on a sign is the cheapest thing in a building to change, and it turns out to be the thing our compliance measure is most sensitive to. That seems worth knowing before anyone decides what to do about it.
— Dr Ingeborg Nwachukwu, Lecturer and Convenor, Capability Sprints
Dr Torun Ezeigwe, who led the work, said the study had “found a rule doing its job perfectly well, provided you are clear about which job”.
The paper continues the School’s work on measures that outlive the reason they were introduced. The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.anpa3l.