Thanks to the rise of music streaming services, it’s never been easier to find new music — so why do I keep listening to the same three albums on repeat?
Part of the problem is that Apple Music’s recommendation algorithm (AKA the ‘For You’ tab) isn’t very good. Spotify’s Discover Weekly playlists are a lot better, but they’re not perfect, either; the same goes for other services like Pandora and Google Play Music.
If you’re struggling to find new music and you’re app of choice isn’t helping, don’t give up. There are a bunch of simple ways to improve those recommendations. Here’s how to do it.
You might think that simply tapping Love or Dislike on songs in Apple Music would be enough to guide the app’s recommendation algorithm, but it doesn’t do much. That’s because Apple keeps a record of all the music it thinks you like, even the stuff you haven’t listened to in years.
Mac Observer’s Andrew Orr recently figured out the trick to fixing Apple Music’s For You section. When you first sign up, Apple Music asks for a list of artists and genres you’re into. After that, every time you add a new artist to your library they’re also added to that list. So all the old music you used to like is still guiding Apple’s recommendations.
The solution is simple. Just open the app and tap on the head-shaped icon (it might also be a picture of you) in the top right corner. Then hit “Find More Artists and Curators.” From here you can remove any artists you no longer like, which should improve your recommendations moving forward.
If you share access to your Apple Music account, like if you have a HomePod that multiple people use, you may also want to switch off “Listening History.” That way Apple won’t factor your roommate’s crappy music taste into your recommendations. To disable the feature open the Settings app in iOS. Then head to Music, select “Listening History,” and toggle it off.
Spotify’s Discover Weekly is usually pretty on the money, but if you’re getting bad playlists there are a few different things you can do.
The streaming service tracks every song you skip and registers that you don’t like it. So make sure to quickly skip any annoying songs that come up, and try not to skip a track you do like just because you’re not in the mood to hear it.
You can also try adding the music you want to hear more of to a playlist, which should give Spotify’s algorithm a little more guidance. As an added bonus, if you make a playlist around a specific genre or style, the streaming service will add some extra recommendations at the bottom of the playlist that match your theme.
Finally, if you’re trying to show Spotify that you like a new genre, don’t give up after a few hours. It takes a day or two for the app to register your new interests. Anything less than that and Spotify just assumes you let someone else borrow your account.
Google Play Music
Google’s huge investment in machine learning should mean great recommendations in Google Play Music, but that’s not always the case. If you’re coming up short, head into the app’s Settings menu, tap on General, and then Improve Your Recommendations. From here, Google will ask a few questions to get a better sense of your musical interests.
If that doesn’t work, Google Play Music also offers the nuclear option. Head back to the main Settings page and select “Delete recommendation history.” This will clear all your data, giving you a fresh start to show Google what kind of music you actually want to hear.
Pandora was one of the first companies to offer algorithm-based music suggestions, and the internet radio service is still going strong. However, if you’ve ever used Pandora you know how easy it is to ruin a good radio station by hitting thumbs up or thumbs down on the wrong songs.
It turns out the trick is to avoid hitting the thumbs up button at all. That’s because a positive response will tell Pandora to play more of the same, quickly reducing the variety of music you hear on that station. Instead, stick to shooting down the songs you don’t like, which will push the service to introduce you to new genres and artists instead.