Metrics question

You’re the PM for a social media company. You are owning a new feature that allows video sharing (like YouTube ). What would you pick as your success metric?

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

Metrics design for a new feature launch, checking you propose a metric tied to genuine value delivered, not just usage volume.

Clarify the goal of the video sharing feature, likely increasing content diversity and engagement without just chasing raw upload counts.
Propose a primary success metric, video view completion rate or watch time per video, since it reflects whether shared videos are actually being consumed, not just posted.
Add a supply-side metric, number of unique users sharing video content, since a healthy feature needs broad participation, not just a few power users.
Add an engagement quality metric, comments or shares per video view, distinguishing passive viewing from active engagement.
Include a guardrail metric, checking that video sharing isn't cannibalizing other valuable content types on the platform.
Combine into a short scorecard and confirm whether the interviewer wants a single north star metric or a fuller tree.

Chooses completion rate or watch time over raw video count, since a feature succeeds only if shared videos are actually watched, not just uploaded.
Adds a supply-side breadth metric, unique sharers, to ensure the feature isn't just used by a small power-user segment.
Names a cannibalization guardrail explicitly, checking video sharing doesn't just displace other content types the platform already relies on.
Distinguishes passive watch metrics from active engagement metrics like comments and shares, since both matter differently.

Proposing raw video upload or view count as the primary metric without addressing whether it reflects real value.
Ignoring supply-side breadth, treating a feature dominated by a few power users as automatically healthy.
Not including any guardrail against cannibalizing existing content types.

How would you detect if video sharing was cannibalizing photo or text post engagement?
What would you do if watch time was high but very few unique users were sharing videos?