Strategy question

You are the product manager at Google Hangouts. After launch, you have an avg. review rating of 3 - detail (40% rated 1, 40% rated 5, 20% rated 3). What will 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

Tests diagnosis of a bimodal satisfaction distribution: recognizing that an average score hides a split user base with different underlying problems.

Recognize the key insight: a bimodal split (40 percent rating 1, 40 percent rating 5) means the average of 3 is misleading, there are two very different user experiences, not one mediocre one.
Segment the 1-star and 5-star groups by usage pattern, device, use case (1:1 calls versus group video), and possibly by when they started using the product.
Hypothesize causes for the 1-star group: likely reliability issues (call drops, poor video quality) or a specific device or network condition causing a bad experience.
Hypothesize causes for the 5-star group: likely a specific well-executed use case, such as reliable 1:1 calls for a particular device and network combination.
Investigate root cause for the negative segment through qualitative feedback (reviews, support tickets) since quantitative data alone will not explain why.
Prioritize fixing the specific driver of 1-star reviews over broad feature additions, since removing detractors, not adding more happy users, is likely the highest leverage move.

Identifies the bimodal distribution as the real story, explicitly rejecting the average rating of 3 as a misleading single number.
Proposes segmenting the two extreme groups separately rather than treating all reviewers as one undifferentiated population.
Prioritizes fixing the negative segment's root cause first, since eliminating a bad experience for detractors is often higher leverage than adding features for already-satisfied users.
Uses qualitative data (reviews, support tickets) to explain the quantitative pattern, recognizing ratings alone will not reveal the mechanism.

Treating the average rating of 3 as the real signal instead of recognizing the bimodal split underneath it.
Proposing broad new features without first diagnosing why 40 percent of users are giving the lowest possible rating.

What would you do first if the 1-star reviews turned out to be concentrated on one specific device or network type?
How would you validate the root cause before committing engineering resources to a fix?