A food delivery platform's median delivery time in one city rose from 28 to 39 minutes over six weeks. Order volume is flat. Rider count is up 4%. Restaurant preparation time is unchanged at a median of 11 minutes. The rise is concentrated in one of the city's five zones, where median time went from 26 to 51 minutes. That zone added 40 new restaurant partners in the period, mostly in a single high-rise commercial complex. Rider acceptance rate in the zone fell from 88% to 61%.
Isolate where the problem is, give the two or three causes worth testing, say which the data supports, and recommend what to do next.
100 points, 60% to pass.
The isolation is handed over: one zone, not the city. The interesting figure is acceptance rate collapsing to 61% — riders are declining these orders, which means something about them is unattractive. Forty new restaurants in a single high-rise commercial complex is the likely mechanism: tower pickups mean lifts, security desks and long walks that are unpaid waiting time for the rider. Preparation time being unchanged rules out the kitchens. Rider count up 4% rules out raw supply. Strong answers test the hypothesis with pickup-to-dispatch time inside that complex specifically, and recommend paying for the waiting rather than adding riders.