A new Slop University paper describes GONDOLA, a domain-specific language compiled to electronic shelf-edge-label firmware, and reports its adoption by a strip-mall trader co-op as the binding basis for supplier order quotas.
GONDOLA compiles markdown rules to label firmware wired to a store's till, and a trader co-op has adopted its waste-risk flag list to set next cycle's supplier quotas.
A punnet of strawberries used to get marked down when a shift worker decided it was time. At twenty-two stores of one strip-mall trader co-op, that decision now belongs to the price label itself.
The School of Continuous Improvement’s Adaptive Metrics Lab built GONDOLA, a small rule language compiled to run on the same battery-powered labels stores already use for ordinary price changes, and wired its evaluation to the store’s live EFTPOS feed so a rule can react to what is actually selling, not only to a use-by date. At the close of each ordering cycle the system also produces a ranked list of stock it judges at high risk of being written off — a list the co-op’s supplier-quota committee has since begun reading as the basis for the following cycle’s order allocation.
For the Adaptive Metrics Lab, the deployment extends a research programme built on taking store-floor judgement calls and giving them somewhere to run.
A shelf edge that can already display a price can just as easily decide one. We wanted a rule for that decision to live at the shelf, not only in a manager’s head.
— Associate Professor Casimir Beng, Lead, Adaptive Metrics Lab, School of Continuous Improvement
Dr Solveig Adeyemi, a co-author on the study, said the fourteen-week trial had shown the Lab “how quickly an operational tool can be read as a policy instrument, once someone downstream decides to read it that way.”
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.emqbbr.