Product design for how Spotify chooses what to suggest to each listener
Company: Spotify
Role: Product Manager
Category: Product Design & Strategy
Difficulty: medium
Interview Round: Other
Spotify has asked you, as a product manager, to design its recommendation engine: the part of the product that decides which audio to put in front of each listener and when. Explain what the recommendations should achieve, decide which listeners and listening moments you are designing for, describe and prioritize what the engine and the listener-facing experience should do, and say how you would know it is working.
```hint Start from a listening moment
Pick one moment when a listener needs Spotify to choose for them, then list what the system knows at that moment and what a bad suggestion costs the listener.
```
### Clarifying Questions
- What should the engine recommend: songs, albums, playlists, podcasts, audiobooks, or a mix?
- Which surfaces are in scope: the home screen, what plays after an album or playlist ends, auto-generated personal playlists, or search results?
- Is the goal more listening time, better retention, more upgrades from the free tier to Premium, or more exposure for new artists?
- Should it serve every listener, or a group such as new sign-ups, free-tier users or long-time Premium subscribers?
### What a Strong Answer Covers
- A stated goal for recommendations and a scope choice (what is recommended and on which surfaces), each with a reason
- A chosen listener segment and the listening moments where today's suggestions fail them
- How the engine works at a product level: the signals it reads, how it weighs what a listener already likes against something new, and how it learns from their reactions
- Solutions prioritized with explicit criteria, including cold start for new listeners and the effect on artists and the catalog
- A primary metric that reflects listener value, with guardrails and a plan to test changes
### Follow-up Questions
- How would you recommend for someone who signed up five minutes ago and has played nothing yet?
- A skip can mean "wrong song" or "not right now". How would you tell the two apart, and does the difference matter?
- How would you keep the engine from steering everyone toward the same handful of popular artists?
Overview: A product management prompt about designing how Spotify chooses what to recommend to listeners. It tests setting a goal for recommendations, choosing a listener segment and listening moments, explaining the signals and feedback loop behind the engine, prioritizing solutions, and defining metrics with guardrails.