What to expect, what they’re really testing, and what a strong answer looks like — scored.
Two-sided marketplace dynamics, supply/demand balancing, real-time operations, and metric trade-offs between riders and drivers. Uber PMs must understand that optimizing for one side of the marketplace often hurts the other.
The question below was asked by Uber interviewers. The answer is graded on the five dimensions real PM interviewers use: structure, specificity, reasoning, decision quality, and delivery.
“Design a metric to measure marketplace health on Uber.”
Marketplace health is a two-sided concept, so any single metric will be misleading. I'd design a composite health index with three components: supply sufficiency, demand fulfillment, and price efficiency.
Supply sufficiency: percentage of rider requests that have a driver available within 5 minutes at the moment of request. This measures whether there are enough drivers to cover demand without excessive wait. I'd set a threshold — say, 85% of requests matched within 5 minutes — and track deviations below this as health signals.
Demand fulfillment: completion rate — the percentage of trips requested that result in a completed ride. Cancellations (both rider and driver initiated) are the most visible sign of marketplace failure. I'd break this down by who cancels and why: driver no-shows (supply quality), rider cancellations after dispatch (price or wait expectations), and rider cancellations before dispatch (changed mind or found alternative).
Price efficiency: the spread between the 10th and 90th percentile fare for the same route at the same time. A healthy market has tight surge pricing — users pay close to the 'true' price for the trip. A wide spread means the market is clearing through price volatility, which signals chronic imbalance.
If I had to pick one North Star: completion rate. It captures both sides — drivers who don't show up and riders who cancel after waiting — and directly correlates with revenue. A 1% drop in completion rate at Uber's scale is a meaningful revenue impact.
I'd track all three metrics weekly by city, with daily alerts if any city drops below the threshold on completion rate.
Three-component framework covers both sides of the marketplace clearly; commits to a North Star at the end.
Names specific thresholds (85% in 5 min, 10th-90th percentile spread), granular cancellation breakdown, and city-level tracking cadence.
The 'single metric will be misleading' framing is correct and shows two-sided market understanding.
Forces a North Star decision (completion rate) with a clear justification.
Dense but purposeful; every sentence adds signal.
This is a high-quality answer because it treats marketplace health as genuinely two-sided and resists oversimplifying to a single engagement metric. The three-component framework is intellectually honest — supply sufficiency catches driver shortages, demand fulfillment catches transaction failures, and price efficiency catches chronic imbalance. Committing to completion rate as the North Star is correct and defensible. The one thing missing is acknowledgment that completion rate can be gamed (drivers accept then cancel), which would be the first follow-up from a sharp interviewer.
Add one sentence noting that completion rate can be inflated by driver gaming (accept-then-cancel to preserve their acceptance rate) and name the counter-metric you'd use to catch it.
Paste any Uber PM interview question and your answer. Get scored on the same five dimensions — instantly, free, no signup.
Grade my answer free →First grade is free. No account needed.