Product design question

You work at Yelp. How would you create a system that would provide good advice to users of new restaurants to try?

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 recommendation-system product design for local discovery, balancing personalization with trustworthy signal quality.

Clarify the goal: help users decide where to eat right now with confidence, not just list nearby options.
Identify pain points: inconsistent review quality and fake reviews, and no easy way to match a specific craving or occasion.
Design occasion-based filtering, quick, date night, group, plus a trust-weighted rating that discounts likely fake or incentivized reviews.
Prioritize review-trust signal quality first, since fake reviews are Yelp's most-cited credibility problem.
Define success as click-to-visit conversion, repeat use rate, and user-reported satisfaction with the recommended restaurant.

Names review-trust and fake-review filtering as the priority problem, grounded in a well-known real Yelp credibility issue.
Adds occasion-based filtering as a concrete, differentiated mechanism beyond simple star ratings.
Ties success to actual visits, not just clicks, matching the real user goal.

Proposing only more personalization or ML ranking without addressing the trust and fake-review problem.
Giving no success metric tied to actual visits, only clicks.

How would you detect and discount fake or incentivized reviews at scale?
How would you measure whether a recommendation actually led to a satisfying visit?