Metrics question

You have found out that 20% of Slack extension users uninstalled in one day. What are your assumptions, and what will you look into to find 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

Root-cause diagnosis for churn: can you generate concrete, testable hypotheses for a sharp uninstall spike rather than one vague guess.

Confirm the scope: 20% uninstall in one day is unusually sharp, so first check if it's tied to a specific release, event, or a single segment.
Hypothesis 1: a recent update introduced a bug, permission change, or performance regression that broke the extension for many users.
Hypothesis 2: an external trigger, like a security/privacy news story about extensions, or a browser update changing extension behavior broadly.
Hypothesis 3: a pricing or feature change (e.g. new paywall) that a segment of free users reacted to by uninstalling.
Segment the uninstalls: by browser version, extension version, geography, and user tenure to isolate which hypothesis fits the pattern.
Recommend the next step: check release/deploy logs and error rates around the exact day of the spike before proposing a fix.

Treats the sharp, single-day spike as a strong signal of a specific trigger event, not gradual organic churn, and reasons accordingly.
Names a concrete, plausible cause, a recent update or browser-level change, grounded in how browser extensions commonly break.
Proposes segmenting by extension/browser version first, the fastest way to confirm a release-related root cause.
Holds off recommending a fix until checking deploy timing and error logs against the spike date.

Treating a sharp one-day spike the same as gradual churn and proposing long-term retention fixes instead of investigating the trigger.
Not checking release/deploy timing against the uninstall spike, missing the fastest path to root cause.

What would you check first: deploy logs, or user-reported reviews/support tickets?
How would you win back users who already uninstalled once you find the cause?