Spotify Product Design: Turning One Seed Song Into a Personalized Playlist

Quick Overview

A product design question asking how a music streaming app such as Spotify should turn the one song a listener just played into a playlist. It tests reading an ambiguous seed signal, choosing a listener segment, ordering and adapting the playlist, and defining success metrics with guardrails.

Spotify Product Design: Turning One Seed Song Into a Personalized Playlist

Company: Spotify

Role: Product Manager

Category: Product Design & Strategy

Difficulty: medium

Interview Round: Other

You are a product manager for a music streaming app such as Spotify. A listener has just played one song. Design how the app should turn that single song into a playlist for this person: what the playlist should contain and in what order, how the listener experiences it, and how you would tell whether it works. ```hint Decide what one song tells you A single song is a weak and ambiguous signal: the listener may care about its artist, its genre, its language, its mood or the moment they are in. Settle which of these you will trust, and how the playlist corrects itself when the guess is wrong. ``` ### Clarifying Questions - Is the listener new to the app, so this song is nearly all we know about them, or an established listener with a long history? - Does the playlist start automatically when the song ends, or does the listener ask for it and expect something they can save and come back to? - Should the playlist hold only songs, or may it mix in podcasts and other audio? - Do free-tier limits on skipping and picking tracks apply to this playlist, or is it designed for paying subscribers? ### What a Strong Answer Covers - A stated goal for the feature and a chosen listener segment or listening context, with the ambiguity of a single seed song acknowledged. - The signals used to find candidate songs from the seed, including how the design copes when the listener or the song has little history. - How the playlist is ordered and kept varied, and how it adapts as the listener skips, likes or saves tracks. - A prioritized first version and the experience around it, such as how the playlist is named, saved and steered. - Success metrics with guardrails, and risks such as repetitiveness, popularity bias and unwanted explicit content. ### Follow-up Questions - The listener skips the first two generated tracks within a few seconds each. What should the system conclude, and how should the rest of the playlist change? - How would you keep seed-based playlists from steering everyone toward the same handful of popular tracks, and how would you measure whether it worked? - How would the design change if the seed is a new release that almost nobody has streamed yet?

Overview: A product design question asking how a music streaming app such as Spotify should turn the one song a listener just played into a playlist. It tests reading an ambiguous seed signal, choosing a listener segment, ordering and adapting the playlist, and defining success metrics with guardrails.

Spotify logo
Spotify
Sep 3, 2025
mediumProduct ManagerOtherProduct Design & Strategy
0
0

You are a product manager for a music streaming app such as Spotify. A listener has just played one song. Design how the app should turn that single song into a playlist for this person: what the playlist should contain and in what order, how the listener experiences it, and how you would tell whether it works.

Clarifying Questions Guidance

  • Is the listener new to the app, so this song is nearly all we know about them, or an established listener with a long history?
  • Does the playlist start automatically when the song ends, or does the listener ask for it and expect something they can save and come back to?
  • Should the playlist hold only songs, or may it mix in podcasts and other audio?
  • Do free-tier limits on skipping and picking tracks apply to this playlist, or is it designed for paying subscribers?

What a Strong Answer Covers Guidance

  • A stated goal for the feature and a chosen listener segment or listening context, with the ambiguity of a single seed song acknowledged.
  • The signals used to find candidate songs from the seed, including how the design copes when the listener or the song has little history.
  • How the playlist is ordered and kept varied, and how it adapts as the listener skips, likes or saves tracks.
  • A prioritized first version and the experience around it, such as how the playlist is named, saved and steered.
  • Success metrics with guardrails, and risks such as repetitiveness, popularity bias and unwanted explicit content.

Follow-up Questions Guidance

  • The listener skips the first two generated tracks within a few seconds each. What should the system conclude, and how should the rest of the playlist change?
  • How would you keep seed-based playlists from steering everyone toward the same handful of popular tracks, and how would you measure whether it worked?
  • How would the design change if the seed is a new release that almost nobody has streamed yet?