The Spotify algorithm is made up of three main parts:
- Collaborative Filtering
- Audio Analysis
- Web Crawling
Collaborative filtering sounds fancy but its actually quite simple. Imagine if 100 people like Ed Sheeran and Justin Bieber, but 10 of those people love Shawn Mendes while the other 90 have never listened to him. It is reasonable to think those 90 people might also like Shawn Mendes. This is what Spotify is doing all the time, for tens of millions of people every day.
Spotify uses collaborative filtering to create personalized content for its listeners. The algorithm analyzes what users are listening to, how frequently they’re playing certain songs, and how long they’re listening for. Based on this data, the algorithm creates playlists that other people with similar music tastes might enjoy.
For every song on the platform Spotify performs an audio analysis to determine the key, tempo, danceability, emotion, liveness and more of the song. You can see what Spotify thinks about your song using a free tool like Musicstax. They use this information in ordering songs on playlists, but also as a way to help classify the style of music a given song is, especially before there is enough collaborative filtering data.
Spotify is constantly monitoring (crawling) millions of websites online to see which taste makers are talking about up and coming artists, which artists are touring together and which artists are on the rise. Spotify wants to be the place you discover new music and by crawling the internet they can discover hits before people know they’re hits.
Learn even more about this in my Spotify Algorithms book here: https://listengenera.com/spotifyalgorithms.OYD