Metrics question

Events usage, specifically, the "Yes, I'm going" usage is down significantly in the past three days. What would you do?

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

Rapid metric diagnosis under a short, sudden decline window, structuring internal and external hypotheses quickly.

Confirm the drop is real and check the measurement pipeline first, since a three day window is short enough for a tracking glitch.
Check for recent product changes shipped in the last week, like a UI change to the RSVP button or Events flow.
Segment the decline by platform and region to see if it is isolated or broad, which narrows the likely cause.
Check for external factors, like a major news event or holiday period reducing event planning activity broadly.
Prioritize the strongest hypothesis and propose an immediate mitigation, such as rolling back a recent change if it is implicated.

Moves fast given the short three day window, prioritizing a check of recent releases before deep analysis, since sudden drops often trace to a recent change.
Segments by platform to catch a scenario like the RSVP button breaking on one app version.
Proposes an immediate mitigation, like a rollback, rather than only a long term investigation plan given the urgency.

Treating a sudden, sharp drop the same as a gradual decline, missing that a recent change is the likely cause.
Spending too long on broad analysis instead of first checking recent releases given the urgency of the signal.

How would you decide whether to roll back immediately or investigate further first?
How would you prevent this kind of regression from shipping in the future?