Metrics question

Few customers are unable to do refund on the Shopee Platform. How would you go about finding the right 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

Root cause analysis skills: methodically narrowing a vague complaint to an actionable technical or process failure.

Clarify: quantify how many customers are affected, and whether it is all refunds or specific types, like partial refunds or return-based refunds.
Segment: break down by platform (iOS, Android, web), payment method, and region to see if the issue clusters somewhere specific.
Check the funnel: instrument each step of the refund flow, request submission, approval, and payment processing, to find where users drop off or fail.
Check recent changes: correlate the timing of complaints with any recent deploys, payment provider changes, or policy updates.
Talk to affected users directly or read support tickets for the specific error message or behavior they experienced.
Prioritize the fix based on where in the funnel most failures cluster, and validate with a small test before full rollout.

Segments by platform, payment method, and region first, rather than treating the failure as one uniform problem.
Proposes funnel instrumentation to isolate the exact failing step rather than guessing.
Ties the investigation to a recent change correlation, a practical, high-signal diagnostic step.

Jumping to a fix without first segmenting where the failures are concentrated.
Not checking recent deploys or provider changes as a likely correlated cause.
Relying only on aggregate numbers without reading actual user reports for specifics.

Chapters of the AI PM course that teach what this question tests.