Interview Questions › Metrics

Airport pickup wait times feel worse: what do you measure?

Uber redesigned airport pickups with a digital queue for drivers. Riders say waits feel longer. Which metrics would you use to decide whether the redesign worked?

Context

Drivers wait in a holding lot and are dispatched in order. The redesign aimed to make driver earnings fairer.

Clarifying questions to ask first

• Actual or perceived wait?
• Which airports and times?

How to approach it

1. Both sides: rider wait and driver fairness
2. Perceived vs actual wait
3. Supply in the lot by hour
4. Trade-off and decision

Framework: Goal of the product -> Users and value exchange -> North Star -> Supporting metrics by funnel stage -> Guardrails and counter-metrics -> Decision or target

Sample answer

• P50 and P90 request-to-pickup time
• Driver earnings per hour in the airport queue
• Rider cancellations after match

1. P50 wait is unchanged at 7 minutes; P90 rose from 14 to 21 minutes
2. The P90 rise happens 11pm to 2am when few drivers are in the lot
3. Driver earnings per hour at airports rose 9% and queue-jumping complaints fell 70%

Keep the queue: it fixed driver fairness and median waits held. Fix the late-night P90 problem with targeted supply incentives. Measure P90 wait by hour as the primary rider metric, with driver earnings per hour as the guardrail.

What interviewers look for

• Rider request-to-pickup time
• Driver earnings per hour at airports
• Cancellation rate