Teaching Spotify's Algorithm

Associating your song with similar sounds and artists allows you to "teach" the Spotify algorithm where to place your music. The core of this strategy revolves around Spotify's "Fans also like" section.

Use playlists to guide Spotify's "Fans also like" section

This is a section on Spotify that showcases artists who are similar to the current artist you're listening to.

Spotify determines who to show in this section using two primary factors:

1. How many listeners you share with that artist

• This is where SubmitHub's playlists can help: by exposing you to listeners who enjoy similar artists

2. Online content about your music

• Spotify searches the internet to see what blogs have written about you, who you've toured with, and the acts you've referenced as inspiration in the bio of your own Spotify profile
• Pro tip #1: you can "@" other artists when writing your Spotify bio
• Pro tip #2: you can make your own "artist playlist" that includes the artists you want to be compared to and references them in the playlist description

Spotify gets into detail here and here.

As an example, below is a screenshot of the "Fans also like" for The Beach Boys:

There are two really cool benefits of being associated with similar artists:

• New releases by artists you're associated with will also cause a bump in your listeners
  • Let's say you're The Beach Boys, and The Monkees release a new hit single
  • Suddenly they're getting millions of new listeners, and Spotify needs to figure out who to play next when those listeners decide to play "The Monkees Radio"
  • Because you're listed as a similar artist, they're gonna start playing your songs - and so you'll get a nice little bump in listeners as well
• When you release a new song, Spotify will expose it to users who enjoy the artists listed in your "Fans also like" section
  • Depending on how well your song performs, Spotify will dial up your exposure to this overlapping fanbase, and possibly even feature you in their Editorial playlists
  • Factors that impact "performance" include: how many people saved the song? Did they listen the whole way through? Did they listen on repeat?

How SubmitHub can help

About two years ago I wrote an article about this, but to sum up: I believe that one of the most-effective Spotify promotion strategies is to focus on playlists that feature artists you want to be associated with. This starts with good genre targeting.

On SubmitHub, we've put a lot of effort into helping you identify the right playlists. And we will keep getting better at this over time as we collect more data.

Getting added to a playlist that's well-aligned with your style means that Spotify's algorithms "learn" to associate your song with the other songs in that playlist. On the flip side, if your song is added to a playlist that features a variety of different genres, Spotify's algorithm will struggle to extract an accurate picture of what your "Fans also like" section should look like.

A well-focused Spotify playlist can also help improve your "performance metrics," which are an important underlying component of Spotify's algorithm. These metrics include skip rate (ie, did they listen to the end?) and repeat listeners.

Finding the right playlists

Accompanying most playlists on SubmitHub you'll see a "genre focus" rating. The purpose of this rating is to convey whether a playlist's songs focus on specific genres, or take a broader approach.

These ratings allow you to focus on playlists that will help expose your song to songs by similar artists.

Methodology

For those who are curious about what's going on behind the scenes, there are two main factors that we use to determine the consistency of a playlist:

1. The genres tags on songs that have been approved and shared to that playlist via SubmitHub

• On the back-end, certain genres are "associated" with other similar genres. For example, Light Drum & Bass and Mainline Drum & Bass are considered to have a strong overlap, so seeing them together on a playlist sends a strong positive signal
• Typically, a "consistent" playlist sees about 75% of songs cluster around 3 genres (the screenshot above has a 84% of songs fall into 3 similar genres, which is a pretty strong signal)
• An "inconsistent" playlist features many different genres, with no obvious stand-out combination of "associated" styles

2. An in-depth audio analysis of the playlist, as provided by Spotify's API

• This approach considers things like BPM, valence, energy, danceability, and instrumentation
• It takes each individual song in the playlist and scores them in each of these categories. These songs are then compared to all the other songs in the playlist, and an average "deviation" is calculated
  • For example, a playlist that features tracks in a range from 80-160 BPM would have more deviation than a playlist that features tracks in the 125-130 BPM range

Conclusion

Summing this all up as cohesively as I can:

You can use playlists to teach Spotify's algorithms who you sound like. Effective targeting will boost your algorithmic performance, which may in turn lead to more editorial placements.

There are a variety of other well-tested methods you can use, including Instagram advertising and social media activity. As with all things, we recommend taking a "basket" approach where you try to tackle the problem from many different angles. There's no right answer here, and what works for one person might not work for another.

Thoughts? Feel free to share a comment :)