[FAQ]

Hi,

I’ve been using the new AI “DJ” feature on Spotify a lot, and it’s pretty impressive. It’s great at mixing in songs I already love with new tracks that, for the most part, are a really good fit for my taste.

This got me thinking from an artist’s perspective: how is this thing actually working? It’s clearly more advanced than a standard algorithm, especially with the AI voice giving context. Where does it pull its data from to decide what to play next?

And more importantly, is there anything we as artists can do to influence it? It feels like getting a song recommended by the DJ could be a massive discovery moment, similar to landing on a big editorial playlist. Can we ‘pitch’ to the DJ, or is it 100% out of our hands?

Cheers!

[Jeff Bullas]

That’s the right question to be asking about the platform’s next evolution.

Short Answer: The AI DJ recommends audio content by combining a user’s personal listening data with human editorial insights. Artists cannot pitch to it directly, but can influence it by improving their overall algorithmic data, especially by using tools like Discovery Mode.

Think of the DJ not as a single playlist, but as a real-time system that analyses both user data and your track’s own content profile.

The DJ feature pulls its recommendations from several data sources. First, it deeply analyses a listener’s personal audio history, looking at their streams, saves, and playlist additions to understand their taste. Second, it incorporates text-based and editorial insights from Spotify’s human experts, which the AI voice then uses to provide context about the audio it plays. Third, it pulls from the same pool of algorithmic data as your other recommendations. You cannot influence this with a direct text-based pitch as you would an editorial playlist. However, you can provide the system with stronger signals by ensuring all your audio content’s metadata is correct, encouraging fan engagement like saves and shares, and, most directly, by utilising Spotify’s Discovery Mode tool to signal that a specific audio track is a priority for discovery.

Cheers,

Jeff

[Author]

Posts