🧠

Understand the Algorithm

💡
You can explore your playlist recommendations on the My Mix page.
 

1. Keyword Search

  • The OKTIV.IO algorithm starts by searching Spotify for playlists relevant to your connected artist, using the keywords set in your dashboard page.
  • You can view and update these by clicking the edit icon next to the artist’s name in your dashboard:
    • notion image
  • When you connect an artist, this will default to the primary genres of the artist:
    • notion image
  • Experiment with different keywords using microgenres, locations, and vibes. The more accurate the keywords are for your particular artist, the better your results will be.

2. Audio Analysis

  • Now OKTIV.IO has found several thousand playlists that could be relevant for your artist, the algorithm builds a picture of what each playlist is looking for, and compares your artist’s latest song to each of those playlists to find the best fit.
  • There are many different features that describe the sonic characteristics of a song. How much the songs in a playlist vary on any given feature determines how important it is for your artist’s latest song to be close to the average value for that feature for the playlist.
  • For example, if I have a house music playlist designed for clubbing, all the songs in the playlist might be between 120 BPM and 125 BPM. There’s a very narrow range of acceptable tempos, because DJs need to keep people dancing at a similar pace. If my song is 100 BPM, there’s very little chance it will be placed in the playlist.
  • Another playlist might feature acoustic songs to be played in coffee shops. To get placed in this playlist, it will be really important for your track to feature mostly acoustic instruments, and not be too lively, but there might be more variation in tempo. Half of the songs in the playlist might feature vocals, and the other half might be instrumentals. In that case, it doesn’t matter too much whether your song has vocals or not. It just needs to be the right vibe.

3. Artist Analysis

  • Once the algorithm has ranked the playlists by how well your song fits into them sonically, it flags which of those playlists already feature the artist in question, and which feature the artist’s related artists (as determined by Spotify).
  • A curator that has already added one of the artist’s songs to the playlist is much more likely to feature them again, so those playlists will show at the top of your My Mix results.
  • Next, if your artist has a ‘fans also like’ section on their Spotify profile, OKTIV.IO will count how many times your related artists feature in a playlist. This is given the second priority in terms of sorting the playlists.

4. Sorting the Results

The final sorting order for your My Mix playlist results is based on:
  1. Whether the artist is already featured (yes/no)
  1. How many related artists are featured (sorted highest to lowest)
  1. How well your song fits into the playlist sonically (based on calculated similarity scores)
 
 
Did we miss something? Email our support team at support@oktiv.io✌️