How the Spotify algorithm works

This covers what Spotify itself publishes about its recommendations (its recommendations explainer, Spotify for Artists help pages, and newsroom and engineering posts), checked Sep 24, 2026. Anything Spotify hasn’t said is labeled as an unofficial claim.

How does the Spotify algorithm work?

“The Spotify algorithm” is shorthand for the recommendation systems that pick songs for each listener on Home, in Search, in personalized playlists like Discover Weekly and Release Radar, and in Spotify Radio and Autoplay. It isn’t one formula. Human editors curate editorial playlists separately, and the 0 to 100 popularity score is a different thing again. Spotify says its systems learn from what each person does (searching, listening, skipping, saving), plus basic personal details, trends across many users, and facts about the song like genre and release date. Spotify doesn’t publish the weights, and it doesn’t publish any threshold that switches recommendations on.

For you as an artist, the practical version is simpler. The systems don’t look at your song and decide it’s good. They watch what listeners do with it and look for more people who’d probably do the same.

That’s why two artists with the same follower count can get very different results. One has 500 followers who save and replay every release. The other has 500 followers who played one song once in 2023. The first artist is sending Spotify useful information. The second mostly isn’t.

A quick split before going further, because these get mixed up constantly:

• Editorial playlists (the ones with the Spotify logo) are picked by people. You reach them by pitching.
• Personalized playlists and mixes (“Made for” a listener) are built by the algorithm. You can’t pitch them.
• The popularity score is a number in Spotify’s developer tools that follows recent plays. It’s explained on its own page: Spotify popularity index explained.

Spotify has actually written down what goes into its recommendations. It’s a short list.

What Spotify says the algorithm looks at

Spotify says its recommendations use four kinds of input: a listener’s taste profile, basic personal information, trends across users, and the characteristics of the content itself. The list comes from Spotify’s page on understanding recommendations, last updated Mar 12, 2026. The taste profile is built from actions like searching, listening, skipping and saving to Your Library. Spotify doesn’t say how much any one of those actions counts. The same page says Spotify combines editorial curation by people with algorithmic recommendations.

Here’s the list with Spotify’s own examples:

Two more official statements fill in the picture:

• Playlist adds count. Spotify’s page on playlist types says that when fans add music to their playlists, it tells Spotify what they like and what to recommend.
• Similar listeners matter. Spotify’s head of personalization described it like this in a 2021 newsroom interview: if two people share four top artists but have a different fifth one, each might like the other’s fifth artist.

And one commercial input. The recommendations page says Discovery Mode lets artists and labels flag priority songs, which makes them more likely to be recommended, and Spotify takes a commission on those streams.

Notice what’s not on the list: a save-rate target, a skip-rate cutoff, a minimum number of streams. More on those below.

Those inputs feed a handful of specific places your song can land.

Where the algorithm puts your music: Release Radar, Discover Weekly, Radio and more

Spotify’s algorithmic spots for your music are Release Radar (new releases, updated every Friday), Discover Weekly (updated every Monday), Spotify Radio, Autoplay, and the Daily Mix and other Mixes. There are also algotorial playlists, where Spotify editors choose a pool of songs and an algorithm picks and orders them for each listener. Each one works a little differently, and Release Radar is the only one with published rules you can plan around: deliver and pitch at least 7 days before release, and your followers get the pitched song. You can see which of these sent you plays in Spotify for Artists under Source of streams.

Release Radar has the most detail. Spotify’s Release Radar page says a song stays for up to 4 weeks if the listener hasn’t heard it, each listener gets one song per artist per week, and re-releases, Various Artists releases and some alternate versions (acoustic, karaoke, live) aren’t included. Remixes are. Pitch at least 7 days before release day to choose which song goes in.

Discover Weekly is Spotify’s best-known discovery playlist. In its 10-year anniversary post, Spotify said it drives 56 million new artist discoveries a week, with 77% of those from emerging artists.

Spotify Radio and Autoplay are described in the listener help pages only in one line each: Radio builds a collection from any artist, album or song, and Autoplay plays similar songs when an album or playlist ends. That’s quiet but steady exposure. If your song sits next to a bigger artist’s songs, their listeners’ Radio and Autoplay sessions can reach you.

Algotorial playlists are the ones that look editorial but play differently for everyone. Spotify’s engineering blog says editors build the pool and the algorithm picks and orders tracks per listener, with listening, skipping and saving training it.

To see which of these reached listeners, open Source of streams in Spotify for Artists. Plays from Discover Weekly, Release Radar, Radio, Autoplay and the Daily Mix are grouped as “Personalized playlists, autoplay, and mixes.” Our guide to Spotify monthly listeners covers the listener side of those stats.

That’s the official picture. The unofficial one is a lot louder.

Spotify algorithm myths: save rates, skips and magic numbers

Spotify has never published a save rate, skip rate, stream count or popularity score that triggers algorithmic playlists. The numbers you’ll see online, like “get a 20% save rate” or “hit 10,000 streams and Discover Weekly kicks in”, come from blogs, courses and promotion sellers, and they don’t agree with each other. What Spotify does say is narrower: saving and skipping are among the actions that shape a listener’s taste profile, and playlist adds tell Spotify what to recommend. That tells you which direction helps. It doesn’t give you a finish line.

Here’s the usual list, labeled honestly:

Where do the numbers come from? Mostly from people reading their own dashboards. Someone sees a song with a high save rate land in Discover Weekly and turns one case into a rule. Some of it is honest pattern-spotting. Some of it is marketing for a course that promises to teach the “real” numbers.

The fair takeaway: saves, playlist adds and full listens are real signals Spotify names. More of them from the right listeners is good. A target percentage isn’t something Spotify has ever given you, so don’t pay anyone who claims to know it.

So if there’s no number to hit, how should you think about the algorithm?

Can you “beat” the Spotify algorithm?

TheDebutRoom’s view: you can’t beat the Spotify algorithm, but you can feed it. Spotify describes its recommendations as built from what listeners actually do with a song (play it, save it, skip it, add it to a playlist) and from what similar listeners do. So the system amplifies real behavior. It doesn’t reward tricks. Spotify’s own numbers also say a small audience isn’t a lockout: it reports that 77% of Discover Weekly’s weekly artist discoveries are emerging artists. The job is getting the right listeners to do real things with your song, then letting Spotify find more people like them.

“The right listeners” is doing a lot of work there. A hundred people who like your genre and save the song send a clear message. A thousand random plays from an untargeted ad or a mismatched playlist can send a muddled one: lots of skips from people who were never going to like it. Spotify doesn’t publish how it weighs that, so treat this as reasoning, not a rule. But it follows directly from the “similar listeners” logic Spotify describes.

If you’ve released a few songs and none of them got picked up, that stings, and it’s easy to decide the algorithm has it in for you. It doesn’t. It usually means it hasn’t had enough real listener behavior to work with yet. That’s fixable, and mostly without money.

Here’s what feeding it looks like, step by step.

How to get the Spotify algorithm to recommend your music

Getting recommended by Spotify comes down to five steps: deliver early, pitch one song, send real fans to that song, get it into listeners’ own playlists, and keep releasing. You need your distributor’s upload done at least 7 days before release day and access to Spotify for Artists for your profile. None of this guarantees a placement. Each step lines up with something Spotify has said about how its recommendations work.

1. Deliver at least 7 days early, with the right genre. Spotify says to deliver at least 7 days before release to reach Release Radar in the first week. Genre and release date are among the content characteristics Spotify lists, so pick the genre you’d honestly file the song under when you upload. If you’re still choosing where to upload, our list of the best music distribution services compares them.

2. Pitch one song. Spotify’s pitching page says you can pitch one unreleased song at a time. It says pitching doesn’t guarantee editorial placement, but a pitch sent at least 7 days early is the song that goes to your followers’ Release Radar.

3. Send your real fans to that one song. Release week is when your followers are most likely to play, save and replay. That’s the taste-profile behavior Spotify names. Point every post, story and email at the same link.

4. Ask for playlist adds, not just plays. Spotify says fan playlist adds tell it what to recommend. “Add it to your running playlist” is a better ask than “stream my song.”

5. Release again. A Release Radar slot lasts up to 4 weeks for listeners who haven’t heard the song. Releasing every month or two keeps something new in front of your followers. That’s arithmetic from Spotify’s rules, not a secret cadence rule.

Two paid options sit alongside this. Paying curators to review your song is allowed if nobody promises plays or placements. Playlist Push is one example: you pay for curator reviews, and its site says there are no guarantees. It fits best when your song is finished, genre-clear, and you have a budget you won’t miss.

Discovery Mode is Spotify’s own paid lever. Spotify’s Discovery Mode help page says it takes a 30% commission on recording royalties from streams in Discovery Mode contexts, and teams need at least 25,000 monthly listeners and songs released at least 30 days. Those contexts are Spotify Radio, Autoplay and Mixes. Our Spotify Discovery Mode guide covers whether it’s worth it.

Doing all five right can still go nowhere if Spotify decides your uploads look like spam.

What stops Spotify recommending your music

Spotify says its music spam filter tags uploaders and tracks that use tactics like mass uploads, duplicates, SEO hacks and artificially short tracks, and stops recommending them. Spotify announced the filter on Sep 25, 2025 in its post on AI protections, saying it would roll out gradually and add new signals over time. Artificial streaming is the other trap. Spotify’s artificial streaming page says it can withhold royalties, correct public stream numbers, remove songs from playlists and remove manipulated content.

In plain terms, don’t do these:

• Upload the same song twice (new title, same audio) to get another shot at Release Radar.
• Stuff titles with other artists’ names or trending search terms.
• Pad a release with very short tracks to farm plays.
• Buy streams, or pay any service that guarantees streams, followers or playlist placement. Spotify names those guarantees as the warning sign.

The worst version of this doesn’t feel like a scam at the time. It’s a friendly DM: “Love the track, we can get you on playlists with 50K followers.” Then the plays come from accounts that don’t exist, royalties get withheld, and a song that could have found real listeners gets flagged instead. You lose the money and the release.

Everything else about Spotify for Artists, from pitching to profile tools, is in the Spotify for Artists hub.