Metrics question

Orders cancellation increased by 15%. How do you diagnose the root cause?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Tests structured root-cause diagnosis for a delivery marketplace metric spike, moving from data segmentation to a validated hypothesis.

Confirm the number and window: is the 15% cancellation increase week-over-week, and is it platform-wide or concentrated in specific regions or restaurant partners.
Check for instrumentation issues first: a tracking change or new release that might be over-counting cancellations rather than a genuine behavior change.
Segment by who cancels (customer, restaurant, or courier) and by stage (before or after courier assignment), since the fix differs completely by cause.
Check for external factors: a competitor promotion pulling couriers away, weather disrupting delivery times, or a specific restaurant chain having operational issues.
Once isolated, validate the leading hypothesis with a quick data cut, e.g. comparing affected versus unaffected regions, before proposing a fix.

Segments cancellations by actor (customer, restaurant, courier) and by timing, which localizes the root cause instead of guessing broadly.
Rules out measurement artifacts before jumping to a behavioral explanation, a discipline interviewers specifically look for on drop/spike questions.
Considers courier supply shocks (competitor pull, weather) as a real, common cause in gig-delivery marketplaces, showing domain knowledge.
Proposes a concrete validation step, comparing affected and unaffected segments, rather than stopping at a guessed cause.

Jumping to a fix before segmenting who is cancelling and at what stage.
Ignoring courier-side supply issues, which are a very common real cause of cancellation spikes in delivery marketplaces.

What would you do in the first few hours if this were an active incident?
How would you tell a courier-supply problem apart from a restaurant operational problem?