Product design question

Design a recommendation engine for Spotify.

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

Product design for a recommendation system, including the trade-off between discovery and listener comfort.

Clarify the goal: more discovery of new artists, higher session length, or higher save-to-library rate.
Segment users: casual listeners who want familiar hits, and explorers who seek new music, since each needs a different mix.
List signals available: listening history, skip rate, playlist adds, time of day, and social follows.
Design the surface: a Discover Weekly style personalized playlist versus in-session next-track suggestions.
Balance the algorithm between exploitation (songs similar to history) and exploration (new artists) with an explicit mix ratio.
Define success as skip rate on recommended tracks and long-term artist diversity per listener.

Separates casual listeners from explorers with a concrete signal, like skip rate under 20 percent for casual, over 40 percent tolerance for explorers.
Proposes an explicit exploration ratio, for example 80 percent familiar and 20 percent new, marked as an assumption to tune.
Names the cold start problem for new users and proposes onboarding via genre and artist picks.
Picks save rate and 30 day retention as success metrics instead of only click through rate.

Designing one recommendation model for all users instead of tuning by listener type.
Ignoring the cold start problem for brand new accounts.

Chapters of the AI PM course that teach what this question tests.