Recent updates

I was on a panel discussion over the weekend that centered around Spotify playlisting, and one of the questions that naturally came up was: "What can artists do to improve their chances of triggering the Spotify algorithm?"

Now, I certainly don't have access to Spotify's algorithm and thus can't say anything with absolute 100% confidence. That said, I think this idea is something that makes sense.

The theory

Spotify's "algorithmic" playlists are primarily focused on helping expose users to songs that are similar to the style of music they already enjoy. If a user listens to techno all the time, Spotify's going to push more techno down their throat. Similarly, if they love country music, their list of recommendations will likely be filled up by other country songs.

And so it follows: if you make techno music, you need "teach" Spotify to suggest your song alongside other techno songs.

One way to do this is to make sure that your song is included in playlists that focus primarily on techno. If your song is added to a playlist that covers pop, acoustic, metal, and... techno, that's going to confuse Spotify's algorithm. They'll see users listening to your song alongside others from completely unrelated genres, which will make it harder for them to target users who might like your song.

Helping your find your niche

Right, so now you have my theory: get your song into playlists that "align" with your sound and Spotify will know who to "algorithmically" recommend your music to.

But Jason, you say, there are 1,000+ playlists on SubmitHub! Help me figure out who to send to!

Good news: I've been trying to do just that. Over the years we've been honing our "genre match" score, which helps align you with curators who tend to gravitate toward your genre. This week, I took that a step further.

You'll now find two new pieces of information accompanying most playlists:

1. Their "top 3 genres", based on historical approvals that took place on SubmitHub
2. A "nicheness" score, which I'll get into below

These two data points should help you avoid playlisters who seem to *not* align with your personal profile.

Wait, what's "nicheness" and how do you come up with that?

Yeah, this one was tough, and I'll readily admit that it's pseudo-science made up by yours truly.

The first thing I do is look at the "top 3" genres of each playlist (based on their sharing history). If those 3 genres cover ~85% of the songs in that playlist, that indicates it's quite focused. On the flip side, if those "top 3" only make up ~30% of the songs in that playlist, it's a strong signal that there's quite a bit of genre variation.

That data point alone isn't enough for me to confidently (and algorithmically) make an assertion about the playlist's nicheness or not, though. So I've brought in a second variable...

Spotify provides a whole bunch "audio analysis" for each song that gets uploaded to Spotify. This includes things such as bpm, energy, danceability, loudness, "speechiness", duration, and more. Using a whole bunch of clever code I can take a "snapshot" of all the songs in a specific playlist and try to determine how similar or divergent they all are. Again, this is a very inexact science, but I've found that when coupled with the above "top 3 genre" analysis, you actually end up with a pretty strong signal regarding whether a playlist is "niche" or "broad" in its genre focus.

What to do with this info

I'll make it short and sweet: if you're sending out a song in a very specific genre, try to almost exclusively target playlisters who focus on that genre. Look for the following:

1. A high genre match score
2. "Top 3 genres" that include your genre
3. "Genre variation" (nicheness) notes that say either "Low (consistent/niche)"