Spotify Tells You How To Get In The Algorithm In 2026

One of the things I constantly hear from musicians is they have no idea how to make Spotify happy and get in the algorithm. You all hear me and well, some of the clowns who pretend to know about music marketing >> [music] >> telling you things then they don't always align with what I say and we contradict each other all the time. Well, we're going to fix that in this video since I got Sam Duboff who's the head of marketing and policy at Spotify for artists to tell you exactly how to win on Spotify and how to get in the algorithm. So this video I talked to him about how to get in the algorithm, what happens when you release too much music, how to get on editorial playlists, what to do with music videos now that you can put them on Spotify, and their newest features that are amazing for marketing your music. And in between his answers I'm going to add some of my thoughts on how you take what he says and do it to the fullest extent so you get your music as far as you can. So before we get started let me say the first thing we discussed is Spotify's labeling of AI musicians. And this is really telling about how they're tweaking the algorithm. So I actually suggest you listen to this. But if you just want to hear how to win, I have a time marker on the screen right now and a chapter marker that you can click so you can skip ahead to just hear about the algorithm if you can't stand to hear about AI for one more minute. Okay, let's talk to Sam. Sam, so excited to have you. Um, you have all these new different AI labels that you just rolled out at I feel like that's this has been the strongest reaction of positivity I've seen towards Spotify from musicians in a while. I feel like you guys really did something good here. But explain to me, you know, the big question people seem to have after the interview you did with Andrew Southworth which I actually recommend everyone watch, uh, is why not go all the way and get rid of the AI. A lot of people feel like this was trained on my data, they stole my stuff, you're giving safe harbor to a bunch of people who are gaining money off of things I created and theft from me. >> Yeah, thanks for having me. We, uh, you know, certainly know opinions and technology around AI music's going to evolve over time. So we figured our our best way to handle it is to try to talk on shows like this as much as we can, share how we're thinking about it, get feedback on Andrew's YouTube, I went through the comments and answered a bunch. I promise I'll look through the comments here, too. When it comes to AI persona labeling versus our AI credits, we're you know, trying to draw that distinction between how the music's made and you know, the artist identity and how authentically the artist is presenting themselves. When it comes to how music's made, you know, there's certainly a lot of opinions on the internet on on all sides of the issue, but what we're hearing from artists, songwriters, and producers is a desire to be in control of the decision themselves of if or how they incorporate AI. AI detection is good at some things, but has a lot of errors. I think you probably see the same post I see every day about distributors or other, you know, smaller streaming services who are relying on detection and the issues have with false positives and false negatives. >> Yeah. >> You know, for us, we never want to police the creative tools artists use and we're hearing from a lot of professional artists, songwriters, producers who are incorporating AI in all different interesting, authentic ways, you know, whether that's post-production elements, specific production elements, instrumentation, you know, to to tune up vocals for you know, ideas. We hear a lot of like you know, someone you have an idea for a melody and you might use AI to hear it a few different ways before you decide how you're going to produce it yourself. And so, our approach has been, you know, when it comes to how the music's made, leave that up to artists, songwriters, and producers. We're working on AI credits so it's more transparent when AI is playing a significant role, but there's a lot of authentic, legit, professional artists who are using AI in their workflows. To us, that's separate than the identity. When you're a listener and you're looking at an artist profile and you see a picture of someone and the name of someone or, you know, a group, a duo, you have an expectation that that's a real person that's behind the music that you could talk to in real life. When that's not the case, when someone's created an AI-generated identity, we want that to be clearly labeled. We feel like that's important. And from our user research, a ton of the kind of um you know, AI use cases that get listeners the most upset in a consistent way is that feeling when you see a profile, you think it's a real artist, you start listening, maybe you like it, maybe you don't, and then later you find out that wasn't a real person. That's what everyone agrees on. While there's a variety of opinions on um you know, where the line might be on the use of AI in the creative process. >> Got it. Yeah, I I mean, I think the most interesting thing of this is, you know, a lot of people say, "Why not just hit pause, shut the whole thing down? There's been no consent involved in this at all." What What is the logic to not hitting pause? Is it just the false positives? I mean, I know you guys famously my inbox uh has been home to a lot of people who have false positive for putting bots on their music where they're like, "Jesse, I swear on my dead mother, and you know, I've never done this. Somebody else did this to me. You guys false positive that." Is it just detection? Like that's what your answer sounded like to me. Can you give me more color there? >> Certainly, I should say in terms of the the legal issues, the entire history of Spotify is based on respect for copyright, right? Spotify was born out of [snorts] the you know, plight of music piracy and showed what you can do if you do things the right way, have a a license environment where people can listen to music, you make it so delightful people want to pay for it instead of resorting to piracy. You know, the whole company is premised on respect for copyright. As you know, any sorts of legislation, you know, get settled on a lot of these you know, really complicated issues around AI creative tools, we'll be following those those fully. I think, you know, that'll help the industry a lot to have clarity on kind of where some of those uh you know, where that legislation lands. So, you know, and some of that's, you know, you you wait and see and you want to see where the the law evolves. When it comes to detection, yeah, absolutely, you know, tons of music on Spotify from, you know, really big artists, emerging artists, all sorts of legit human artists have, you know, significant AI signatures in them cuz, you know, people might be using AI plugins to their DAW. They might, you know, generate a loop or a beat. Uh maybe they buy something on Splice and they don't realize it was AI generated. Um maybe they're using an AI tool for mastering that ends up regenerating a number of the stems through that mastering process. Uh and so, you know, from our point of view, a lot of the artists we talk about um talk to who are as legit as they come are experimenting with AI in their creative process. And so, you know, to do a, you know, full-on ban of music with significant AI signatures would really, you know, stifle the creative choice that that artists have. Um and, you know, privately and publicly we're hearing from a lot of artists who want the ability to incorporate it into their creative process. With all that said, transparency is that important pillar to it where, you know, we don't think it's up to us to police the creative tools that artists, songwriters, producers decide to use, but we think, you know, listeners have a right to know and to make their own choices about it. So, whether it's AI persona labeling or the tens of thousands of AI credits that are on Spotify every day, um you know, listeners can see the use of AI increasingly on Spotify. They can decide if they want to engage or not. Um as we announced uh earlier this month, when it comes to AI personas, by default, none of the music from AI personas is going to be in Spotify programming, whether that's editorial or algorithmic, unless the listener actively seeks them out, goes and follows the artist, tells us they want to go listen to it. So, our principle is about choice. Let artists, songwriters, and producers decide the creative tools they use, but also make sure that listeners have the ability to decide what they want to listen to. >> After this, I turned the conversation to asking what should an artist who wants to get in the algorithm or go deeper into the algorithm on Spotify do to get there. So, there's this viral post on the Suno forum that um I think a lot about that um it's a person who's like, "I put up hundreds of songs, not one single play." And there was this myth when Spotify started that we're going to test out everything. I don't know that you guys ever communicated that, but it was still this myth that was put out that every song gets tested. What I've mostly seen is that you get a little bit of attention, it does get tested out. What is the reality? Do you need inbound links? Do you need things? Like what what what are what is the reality? Like get give me popularity score like can you tell listeners what the road map that they should be doing to actually get into the algorithm that cherishes, that that grows listeners while they're sleeping? >> Yeah, you know, first to comment on the the slop side, you know, we announced last fall the launch of our music spam filter, which is a new algorithmic system we've built that identifies and tags people who are, you know, trying to game the system in all different ways, fully removes them from programming. One of the things we're looking at is mass uploads. Um you know, we know that for real human artistry it takes time. We never want to reward people for uploading low content. And over the past year, not just the music spam filter, we've made a bunch of changes and tuning to our algorithms to help elevate legit human artists and to remove slop and low effort content from programming. So, you know, I'd separate what that poster in the the Suno forum is saying is kind of a >> So, to do a little analysis, this seems to be a good thing to me. What Spotify is saying is if you are one of these artists from the school of I'm just going to post tons of music and see what sticks, they are actually saying that's not going to go in the algorithm. It's one thing if your song gets attention outside of Spotify and you're uploading tons and tons of music and if it starts to move numbers on Spotify from outside attention cuz it's good and you've promoted it, well, they may push it then, but they're punishing artists algorithmically that upload too much and they're not going to give them the benefit of the doubt at first. >> I got a question from question um >> Yeah, okay, so let's take it to the kid in his bedroom who slaved over that song. >> Yeah, exactly. So you know, that that's on that side. When it comes to um artists who are, you know, spending the right amount of time with their their art and being intentional about what they release and they're not spamming uh or impersonating or deceiving, that's exactly who the Spotify algorithms are built to to elevate. As people can imagine, with this many uh tracks getting released to streaming services every day, not everyone is going to get the same chance and what we're um you know, what I hear often is that the hardest part is those first streams, those first listeners. We probably Spotify is not the best platform for people trying to get that initial, you know, zero to one uh in their music career. There's other social platforms or, you know, touring live or, you know, peer-to-peer word-of-mouth marketing where you kind of need to get that first critical mass >> So so I I know what they're saying. They're saying is that 10,000 monthly listeners like what is it what what what is that? >> Oh yeah, definitely no hard lines like that. I just uh you know, what we've been trying to build is that you kind of earn trust and reputation over time on the platform. And so, you know, when it comes to Fresh Finds our um indie playlist ecosystem where we playlist uh you know, 10,000 indie artists a year, a lot of them have under five releases, a lot of them, you know, it might be the very first release uh and our editors kind of go through Fresh Finds and program a lot of it. And once you're on fresh finds, over the following few months, you on average double your royalties and audience. You know, that's 10,000 a year. There's way more artists than that trying to get that first start. But, you know, that's one way to to build a reputation. Another is you start to build your audience, you get more followers. When you have those first followers, your new songs going to go into their release radar. The way our algorithms work is the way all algorithms work, which is you're looking for enough data to be able to say if this listener liked this artist or song, this other person has similar taste and they will as well. So, there's no magic number, I promise, right? Once you get over this many streams, it clicks into some different sphere. That's not how it works. It really is brick by brick at every stage. So, even if you only have 10 listeners, that starts to give us a little bit of data of like, okay, what's true about these 10 listeners? What context do they enjoy this music? Were they saving? Were they adding it to a playlist? Did they listen once or did they listen a bunch? That helps us find a few more. And then you get that kind of scale as things grow. And as you have 10,000 listeners, that gives us data on 10,000 people who loved your music. When you have 100,000, that gives us even more. And so, that's the way our algorithms are tuned, where the hardest part is just getting that initial critical mass. And then we're looking at every signal you can imagine. And this is how machine learning algorithms work and how generative algorithms work, where you know, we look at time of day and the set and the sorts of taste profile that each listener has, how they discovered the song. And we kind of, you know, when I say we, our algorithms, you know, that are are built to say, okay, if you if these people liked in these ways, these other people will, too. And it and it builds from there. >> So, this is important. What Sam is confirming here is a thing we've been discussing on this channel for nearly 7 years, which is Spotify needs to see people listen to your music, and then they learn to recommend you to fans of other artists after they see who's come in to listen to your music. If this isn't making sense to you, watch the video in the description on how algorithms work that I linked below cuz I explained this very thoroughly in that video. >> You can confirm a thing that Andrew and I have had a long thing of it's not just plays, it's skips matter. Like one of the jokes I've always had to Andrew is like, "What if we advertised a noise song where there's somebody screaming to Swifties? Is it still going to go with the algorithm?" Or and I'm like, "I can't imagine you guys don't go, 'Wow, they all hit stop and run like hell from this.'" >> That's right. All the things you can, you know, if you if your listeners have seen the the sorts of analytics we've added to Spotify for artists the last few years where, you know, we're trying to show you more of the factors that matter most in the algorithm. And so that's streams per listener, that's active streams, not program streams, that's adding to playlist or saving, right? Those uh if you find some, you know, spammy way to find a bunch of listeners who then skip, don't save, only listen once, that ends up hurting you. Um so it's all about finding that right audience that connects with your music. A num uh you know, a pure play count really isn't factored in much at all. It's about repeat listens, active audience, super listeners. We're trying to introduce listeners to music that they're going to love for the long term. So any signal we have that the song doesn't connect with a certain type of person, we're not going to recommend it to those sorts of people anymore. And so, yeah, definitely artists shouldn't be thinking in terms of like stream count or listener count all being created equal, all mattering the same. Fandom signals are what all of our algorithms are tuned for. >> So many people tell me their music marketing team tells them if they just hit a number of plays the algorithm will go crazy. But this confirms that is not the least bit true, and that Spotify is measuring authentic enthusiasm for your songs. This also allows them to cut down on botted plays and promote music that is more likely to keep users using their app. It makes sense. If they're just pushing the botted garbage that gets pushed, people are going to leave the app because it's playing songs they don't like. So, it makes sense that they're measuring all these authentic things that show that people actually really do like these songs. >> Amazing. That is so such great clarity. I If I see one thought that's festering the air at the most is there ain't no way a human's reading the Spotify editorial playlist. You guys now just have a video editorial playlist where you're saying, "Please pitch us." A lot of people are saying this pitching is annoying. I don't like writing it. Can I just bulk upload these pitches?" All these things. Talk to me about what this looks like in 2026. There's also all these artists who say, "I get on playlists all the time. I've never written a pitch in my life. What are you talking about?" >> Yeah. It's It's It's really tough. Everyone has their own personal experience. Uh Our editorial team's amazing. All they do all day is read and listen to music. There absolutely are way way way more people pitching than spots that can be in these playlists. And that's definitely part of the reality. You know, when you think of the overall value Spotify can bring to you as an artist, your algorithmic recommendations are always going to be bigger than the pure editorial ones. So, you know, a third of all artist discoveries on Spotify are from those personalized streams. So, sometimes I'll see artists who have huge amounts of streams on in mixes um in Discover Weekly, on Release Radar. And you kind of take that for granted. You're like, "Oh, well, of course Spotify does that for me. Why aren't I getting you know, these editorial playlists that that that I want to?" And editorial matters, you know, getting that sort of human validation really means something to to artists and to listeners, but it's just one piece of the puzzle. Many tens of tens and tens of thousands of artists are playlisted every year. Unfortunately, there's like hundreds of thousands releasing music every year and you know, the the playlist pitching process is our best way to try to make it fair. Every artist, biggest artist in the world down to the newest artist getting started uses the same tool and we have specific playlist ecosystems like fresh finds that are just for emerging artists. With fresh finds we can get 10,000 artists a year playlisted. That's actually a ton if you think about how much music does editors have to go through to get to 10,000 new artists a year and fresh finds has discovered Roll Model and Doja and Megan Moroney and Langley and you know, that's because >> The list goes on, yeah. >> Yeah, these these editors are amazing, but there is just a reality where you know, certainly no one can lie about the numbers. We can get 10,000 artists on play on fresh finds every year. There's you know, hundreds of thousands of of new artists trying to make it in music every year. So, I always hate hearing from artists who pitch a bunch and never get playlisted and I just hope they know their pitches are being read and the song's being listened to. You know, we're trying to help them in every way we can build that career whether that's editorial play listing or all the other tools we offer. >> So, I I glazed over it. Videos getting playlists, you're asking for a lot more videos. I get asked every day what type of videos should be getting uploaded here. I know you guys have some language around this, but you know, people are saying to me like should I be putting my daily vlog up is it going to help if I do this? It doesn't sound like you really want a spam. You want the high quality content, am I right there? >> Yeah, we're all in on video. Artists should definitely be getting high quality music video on the Spotify. There's lots of new kind of programming opportunities for you if if you do that. We see listeners turning to a lot. You know, if you have a a podcast or talk content, you can set up a podcast profile for our Spotify for artists video. We're focused just on music content. Um so, you know, we know how hard it is when you're uploading video content to Instagram, Tik Tok, shorts, bright social feeds, you're competing with all the video content in the world. When you're uploading video on on Spotify, it's just about the music and we want official music videos, we want performance videos and we want covers. If you do high quality covers, it is going to live on your profile and with your catalog um indefinitely. This isn't about get 48 hours of views on a feed and it goes away. So, it should be at the quality of bar where you kind of want this to be on your profile and in your catalog for the long term. Um performance videos are performing particularly well because there hasn't really been a space for that on Spotify before and we've built it so listeners can save that video and they can listen to just the audio on their playlist, too. So, you're actually, you know, making a new song, that live version becomes this new kind of entity that listeners can engage with on platform and you know, we've got new video algorithmic feeds. Uh we've got new um video editorial playlists and then all over your artist profile and catalog, we help listeners rabbit hole and go deep into your world and your artistry. >> So, I think some of the important language Sam used here is they want your music videos you put effort into and one video per song. In my studying, K-pop artists are often the artists who make the most videos per song. Sometimes 13 or more videos for each song and even they only have up one video per song. So, that seems to be the move in 2026 on Spotify for now. But, we have to remember, these videos help build a relationship with you and your fans and help those fans become those super listeners that help get you deeper into the algorithm. As Sam confirms in this video, is that the super listeners help you be measured more highly to be put more in the algorithm. >> What in the dashboard could people see and actively look at and go, "Okay, I should now do this with my marketing?" >> You know, I think people mostly know all the the basic analytics that have always been in there. Number one thing I would look at is the segments section. That's where you really get details into the listing behavior of your fans, where you can see how many of them are just programmed listeners, where they're only listening to your music when it's being recommended to them by Spotify or by a third-party playlist. Um versus your active audience, which is the listeners who actively choose to listen to you from their own playlist, their own liked songs, your artist profile, your catalog page. And active audience is the best predictor of long-term fandom. We kind of then break down active audience into light, uh moderate, and super listeners. Super listeners stream you actively more than 15 times a month. They're 2% of your audience on average, but 18% of your streams. They are more than 50% of the merch sales and concert sales you get from Spotify. So, you know your marketing plan and all the marketing activities you've done, you should be mapping that against the segments data in Spotify for artists and see what are the things you're doing that aren't just getting program streams where people try you out. Maybe you get light listeners, but they listen once or twice. You want to be doing the things that get you moderate listeners, super listeners in your active audience that have high stream per listener counts and stay long-term. Our algorithms are tuned to reward that. That's when we're going to recommend it to more people. And that's the sort of fans that are going to power your business. >> I know what he's saying here is a bit dense and reading analytics is no fun. I actually have in my members video a guide on the analytics that matter, which you can get access to for $5 a month where I teach things like that and give you access to knowledge like what's in this video with over 170 other videos I've made on how to grow your fan base. So, you can find that video and join our member feed with the link in the description for the low price of $5. Playlists now have comments uh that you could use. What are people going to use that for? I have my own ideas, but I'd love to hear from uh where that came from. >> Yeah, I'm really excited about this. They're not uh comments in the traditional sense like you'd see on on social platforms. They're comments and commentary from the creators of the playlist, whether that's Spotify editors on Rap Caviar or a user curator or a collaborative playlist where friends make it together. Really cool. We've always gotten this request. You want to tell the story of the playlist, you mean? People should try it out, uh you know, on any Spotify the big Spotify playlists, a lot of them now have it. Today's Top Hits or New Music Friday or a Rap Caviar where our editor can see their face, who they are, and they'll share more about why they play listed it. And you get this whole new dimension instead of that passive listening of a playlist, you're saying, why did this human curate it and what should I be listening for in it? Um and then we wanted to give that to users, too, for the amazing playlists that they make. So, when you're sharing things with your friends or if you have a big community around your playlist, um you can really show your taste in this whole new way and give people a deeper listening experience. Uh so, we'd only just rolled it out. I'm really excited to see the data come in of of how people are using it. >> Yeah, I mean, I think of it for everything from telling fans, like, this is how my album was inspired to, you know, even for me, like, I have playlists of like my favorite drum sounds for when I'm mixing things. I'd love to be like, this is great for cymbals. And write what I was thinking and just, you know, >> And I'm glad you said that cuz you're my one of the things I'm most excited about is if artists will make user playlists using it and then you can make it an artist playlist on your profile and you could, you know, have select songs in your catalog and you annotate the inspiration or your songs mixed with other artists' songs and the ones you're inspired by and then the ones you made yourself. There's some really cool artist use cases. >> Zev, thank you so much for being so giving with your time and with your knowledge of all this. It really, really, I feel like uh is going to help people achieve their dreams, so I'm very grateful for the work you're doing. >> Of course. It's so great to meet you. >> So, here's the thing. While you just learned this, if you really want to grow your fan base, you need to understand how to grow your music with Spotify artist playlists, which is the best way to to a good algorithm and get your music into the algorithm right now. So, I would really click play on that, which is on the screen right now where our link to the description. Thanks for watching.