Technology for Product Managers: How does Spotify recommend music to you?
--
“Technology for Product Managers” is a series intended to up-skill current and aspiring PMs on popular technologies used in building the next-generation products used and loved across the world. This series will also help you prepare for technical interviews at top technology firms.
Spotify is in a tough business. It has to compete against global technology giants, facing competition from Apple Music, YouTube Music, Amazon Music, and several others. Further, music is getting increasingly commoditized. All popular streaming services have huge acquired huge music libraries, and it is incredibly difficult, rather impossible to differentiate based on the size or quality of your music library.
How does Spotify lead this market? One of the most popular features driving incredible growth and amazing retention is its recommendations. People love Spotify’s Discover Weekly and Daily Mix playlists. It almost eerily recommends music exactly to your tastes. Let’s try to understand how.
What does Spotify know about you?
Spotify observes the following behavior about you when you use their app:
- What songs did you listen to?
- What songs did you skip within the first 15–30 seconds?
- What songs did you like enough to hit the ❤️ and add to your liked songs?
- What songs did you add to a playlist?
Keep in mind that this is a simplified version to explain the concepts. In reality, hundreds of other parameters, such as time of use (i.e.: Joe listens to his workout playlist at 7 am), time of snooze (i.e.: Joe listens to his sleep playlist at 9 pm with a 30 min snooze), and many others are tracked to improve recommendations.
How can Spotify use this information to recommend music to you?
Technique 1 — Collaborative Filtering
