Reducing Post-Acceptance Driver Cancellations for Frequent Uber Riders
A PRD attacking post-acceptance driver cancellations for frequent Uber riders through reliability scoring and incentives
✳The problem they went after
Frequent Uber riders in urban markets keep hitting the same failure: a driver accepts, then cancels before arrival or never reaches the pickup point. Sixty-two percent of surveyed riders called driver cancellation a frequent issue and 70 percent said they switch platforms when it happens. The cause is an incentive mismatch. Drivers learn trip details after accepting, and a short distance, heavy traffic or an unwanted payment type makes the ride look not worth taking.
✳What they actually did
Two changes carried the highest RICE scores: a driver reliability scoring system that categorises drivers by post-acceptance cancellation history and weights matching accordingly, and an incentive model rewarding drivers with strong completion records. A complementary change closes the information gap that triggers the cancellation, showing drivers payment mode, estimated pickup time and trip margin before they accept. Ride completion rate is the primary metric, targeting 15 percent improvement over six months for frequent travellers, with post-acceptance cancellation rate secondary and matching acceptance rate watched so the stricter algorithm does not choke supply.
✳What you can take from it
Worth stealing
- Cancellation reframed as an information gap, not misbehaviour
- Reliability score weights matching, incentives reward completion
- Upfront trip details: payment mode, pickup time, margin
- Monthly audit of matching logic written into the PRD
Read this if
PMs changing a matching algorithm where the supply side can push back
The transferable bitCancellations are treated as a rational response to hidden information rather than as bad driver behaviour, so the fix works both sides: consequences through matching weight, better information before the decision. Committing to audit the new matching logic monthly and refine it if completion rates stall is a rare thing to write down.
✳The submission itself
Open the PDF →Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.
