Assessing the health of UberX rider-driver matching with marketplace metrics

You are a product manager at Uber. The interviewer asks you to describe the marketplace that pairs UberX ride requests with available drivers, and then to explain how you would judge how well that marketplace is performing. Cover both halves: what the matching system does, and how you would assess its health. ```hint Look from three seats A match can look fast in the dispatch logs and still be bad for the rider waiting at the curb or for the driver crossing town to reach them. Think about what each side experiences between the request and the start of the trip. ``` ### Clarifying Questions - Does "the marketplace" mean only the dispatch step that pairs a request with a driver, or the wider system that also includes dynamic pricing, driver positioning and pickup-time estimates? - Is this a standing health review, or a reaction to a specific concern such as rider complaints about long waits in some cities? - Should driver outcomes, such as earnings per online hour and idle time, weigh as much as rider outcomes in the verdict? - Should performance be judged across all markets at once, or city by city where supply and demand conditions differ? ### What a Strong Answer Covers - A clear description of the matching loop: requests, available drivers, the dispatch decision, driver acceptance, the pickup, and how pricing and positioning feed into it. - Rider-side measures of match quality (request-to-trip conversion, time to match, actual pickup wait, unfulfilled demand) and driver-side ones (utilization, empty driving to pickups, earnings, acceptance and cancellation). - One headline view of marketplace health, with a reason it beats average match time alone. - Segmentation by city, hour and supply-demand state, plus guardrails that catch one side being sacrificed for the other. - A way to tell a matching problem apart from a supply shortage or a pricing problem. ### Follow-up Questions - Average time to match fell after an algorithm change, but driver earnings per hour also fell. How do you judge the change? - How would you run an experiment on a matching change when riders and drivers in the same city compete for the same pool? - Riders in one city see no cars available far more often late at night. How do you decide whether that is a matching problem or a supply problem?