Spotify lets you text your recommendation algorithm

Anyone who streams music knows the frustration of an algorithm that takes things too literally. Fall asleep to white noise or let someone else queue songs on a road trip, and your homepage gets derailed for weeks.

Spotify began rolling out a beta feature called Taste Profile to Premium subscribers aged 18 and older in the U.S. First shown off in March at SXSW by co-CEO Gustav Söderström and tested in New Zealand, the tool is now live in Spotify's biggest market.

Peeking inside the recommendation box

Until now, Spotify's algorithm was entirely a black box that dictated your Made for You mixes, Discover Weekly, and annual Wrapped. If you listened to something, the system assumed you wanted endless variations of it, often trapping you in an echo chamber of overly similar tracks.

Taste Profile opens that box up. By tapping your profile picture on mobile, you can read Spotify's brief summary of how it views your taste across music, podcasts, and audiobooks. More importantly, you can talk back to it using plain text.

You can ask to explore 90s alt-rock, demand more energetic morning playlists, or tell it to prioritize tech podcasts while banishing genres you are tired of. Spotify says the updates take effect on your Home feed within a few hours, and you can edit or delete your notes at any time.

Moving past blunt exclusion toggles

Spotify previously introduced small fixes, like excluding individual playlists from recommendations or toggling off kids' music so children's songs would stop hijacking parents' feeds. Taste Profile takes a much broader approach by treating user guidance as proactive steering rather than passive cleanup.

For now, the beta remains limited to paying adult subscribers in the U.S. and New Zealand. Spotify has not shared a timeline for a wider global rollout.

Why it matters

Recommendation feeds usually suffer from having great memory but zero intuition. Manually retraining an algorithm by skipping tracks and liking new artists takes days of deliberate effort. Letting people type a quick note to reset their feed saves genuine hassle, especially for anyone whose daily listening bounces between focus tracks, family road trips, and podcast catch-ups.

We think this is a clever hybrid approach to personalization. Instead of replacing the recommendation engine with an unpredictable chatbot, Spotify is using natural language as a trim tab on top of its core algorithm. It keeps the discovery engine intact while giving listeners an escape hatch when suggestions get stale.

This will likely pressure competitors like Apple Music to rethink how much visibility and direct control they give users over their own algorithmic profiles. The catch lies in responsiveness: if processing takes several hours, it cannot cater to immediate mood swings, and any misinterpretation of nuanced text requests could easily turn a promising control panel into another ignored settings menu.