A Slop University paper classifies 1,148 attempts to leave, cancel, unsubscribe or dispute across four everyday platforms and finds escalation to a person, not automation, behind the largest share of requests that are not granted.
A four-cell typology finds escalation to a person behind most requests that don't get granted
A group chat, a neighbourhood forum, a booking app and a review page have, on the face of it, little in common. A paper released today by the School of Emergent Priorities finds that once someone tries to leave, cancel, unsubscribe from, or dispute something on any of the four, the same pattern takes over.
Doctoral candidate Tove Ilesanmi, working with Professor Verity Marris and Dr Lindiwe Achterberg of the Trajectory Analytics Group, had 214 participants log every such attempt over fourteen months, then sorted the results into four outcomes by two questions: was the request handled by a person or a system, and did the platform actually do what was asked. The category the team calls Retention Hold — a person gets involved, and the request still doesn’t go through — turned out to be the single largest outcome, on every platform, including the two whose own help pages recommend contacting a person as the surer route.
Retention Hold isn’t a failure of the automated systems. If anything, it’s what happens once a person gets involved.
— Tove Ilesanmi, Doctoral Candidate, School of Emergent Priorities
Professor Marris said the four-cell framework gives the School “a single measure it can now ask of any process with an exit or an appeal built into it, whatever the process happens to be called.” The finding has already prompted a change closer to home: any renewal of a student-facing app contract must now disclose which of the four cells its own exit and appeal handling falls into before the University will sign.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.r7bp15.