A School of Continuous Improvement paper sorts 14,760 café loyalty-card histories into a tempo-and-completion quadrant and finds the resulting refurbishment-funding priority barely tracks which outlets are actually losing custom.
A tempo-and-completion quadrant sorted from an eleven-outlet café chain's stamp cards is set to decide next year's store-refurbishment budget, though it barely tracks which outlets are losing custom
An independent café chain’s paper loyalty card was never built to answer questions about a store’s future, and the School of Continuous Improvement has spent eighteen months teaching it to anyway.
A paper led by Dr Bram Ntuli, with Associate Professor Casimir Beng and Senior Lecturer Thandiwe Solberg, barcoded loyalty cards across the chain’s eleven suburban outlets and timed every stamp rather than only counting it, sorting 14,760 stamp histories into a two-by-two quadrant of pace and completion: Sustained Engagement, Front-Loaded Enthusiasm, Arrested Progress, and Silent Attrition. The chain’s executive committee had proposed weighting each outlet’s share of next year’s refurbishment budget by its Front-Loaded Enthusiasm reading, taken as an early signal of demand worth resourcing.
What a stamp card decides to remember about a customer, and what it quietly leaves out, is exactly the kind of question the School likes to sit with a little longer than the answer requires.
— Senior Lecturer Thandiwe Solberg, Convenor, Demo Quarter
Checked against each outlet’s own till receipts over the same eighteen months, the proposed allocation tracked actual footfall decline only weakly, and if anything in the wrong direction. Associate Professor Casimir Beng, Lead of the Adaptive Metrics Lab, said the instrument had “done exactly what a two-axis measure is supposed to do,” and that what the executive committee chose to do with the reading afterwards was “a decision the committee is entitled to own outright.”
The University frames the study as part of a broader interest in watching an instrument closely enough to see what it was never asked to do, and has recommended the chain adopt the quadrant regardless, alongside a published register of the readings behind it.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.lb797o.