Andrew Southworth
Videos55:49

Spotify COMPLETELY CHANGED Playlists (here's how)

Discover how Spotify's recent overhaul of playlist ranking has reshaped music exposure, causing significant shifts for labels and artists alike.

Quick summary

Spotify has implemented a major change in how playlists are ranked in search results, impacting the visibility of tracks and artists on the platform. This shift has led to noticeable drops in streams for some labels, highlighting the evolving challenges of music marketing in the streaming era. The discussion explores the implications of this update, emphasizing the need for artists and labels to adapt their strategies to maintain and grow their audience. Understanding these changes is crucial for navigating Spotify's dynamic ecosystem and optimizing playlist placement effectively.

Transcript(English)

Austin, if you only had one minute to give music artists the best advice you possibly could, what would you say? One minute. >> Great question. Um, I would say that even though the music business now is the streaming business and consumption is very important. I would encourage all artists out there not to judge their self-worth [music] based on how many streams the records get. >> Perfect. You only needed 20 seconds. [singing] >> [music] >> which great great advice. And the funny part is I mean you you have a label that has done I think you said 500 million streams. Was that the number? >> It it is I know that seems like a big number probably out there for independent artists out there, but that is I would say these days a fairly modest record label. We've been in business now for a couple years. Warner Music ADA is nice enough to distribute us and uh as I mentioned before, we are primarily a cover song record label. So, uh you know, our thinking and I'm sure you've seen the trends right now in music around the rise of old music versus new, you know, catalog versus frontline. Um I've, uh we've noticed a trend over the past couple years of this kind of growing uh and being the case. And for us as some as someone who owns a number of large UGC playlists across DSPs, it makes a lot of sense to optimize around old records versus new. >> Absolutely. And and you know, we're we're kind of we're here today to talk about uh like third party play. I usually call them thirdparty playlist. You even call them UGC playlist, but the same thing. They're user created playlists. how starting September 1st, there was a massive crash in a lot of these playlists and it's there's there's there's a few like kind of doomsday markers for for people who have been investing a lot of money in growing playlists and we're going to get into that. Uh but first um can you just kind of tell us like how you got here? What's what's your backstory? A little more information on your label. I mean you just gave us a teaser with a >> cover song, but yeah. What's what's your story? Yeah. So, in college, I went to SMU in Dallas and in college I had a music blog website called freshtracks.com. This was very much back in the day when blogs were still a very big discovery mechanism. I'm not sure if you remember like good music all day. Know you had Tim Weber on a few weeks ago or this song is sick, but Fresh Your Tracks is very much in that vein. And a lot of it was focused on college music, you know, covering artists like Mike Stud, Sammy Adams, Huey Mack, Time Flies, etc., etc. And that was really kind of my force first foray into music and kind of how I got my feet wet in this business. >> Nice. Yeah, I think that was when Jason from SubmitHub was from as well, right? I for what was his blog called? Do you remember? >> I don't remember, but I It's funny as I know his name from the blog space and oftentimes I meet people in music, they still think of me as the Fresh Tracks guy. So, it still follows me around today. >> Yeah, I it's like I I know it but I'm just blanking right now in the moment. But but either way, like you know, it it's kind of interesting how much the world has changed because back then blogs were kind of king of the castle and there was even like a genre called blog core. >> Yes. >> And then that just kind of evaporated. I mean blogs are still a thing and they can still be useful, but um I would say immediately after that is is like playlisting became >> the the big riser. Um and you know a lot of people don't realize how how playlists work. So you you have you have a bunch of playlists with a bunch of followers, which is how you primarily promote your catalog of music, right? You have and a lot of labels do this because if you have a bunch of music and I know someone who has over a thousand songs in his catalog and running campaigns for each individual release is like one, it's a lot of work, but but two, if you can get one person to listen to like a hundred of your tracks uh as a label, that's a big advantage. You can have all these different playlists with different categorizations of your catalog and you can drive fans to them and then cause a bunch of repeat listening. You can then make playlists under under um alt accounts and promote them as if they weren't yours. They just look like a fan playlist, but secretly 80% of the catalog is your own. So, uh a lot of people watching this might not know about this like kind of ecosystem of playlists and like how you how you promote them and then how they start getting traffic for free. So, like talk a little bit about that world. Yeah, you're kind of touching on some of those functional music operators like Calmly for example, Strange Fruits, uh even Miracle Tones, for example, all you know, companies that are very smart that take a searchbased approach to playlisting. And we're really no different. Our thesis for a very long time, uh has been based on a uh 2019 research paper published by Spotify's Boston office. They have a research arm there. Uh it was I the the name of the exact article escapes me. It's a very long art title like developing evaluation metrics using mixed methods. The paper Andrew is an absolute snoozer. I don't recommend you read it. Uh but there were really two very key interesting findings that have really informed our business model. And the first finding was uh was that um is that 80 to 85% of people click on the first result in the ranked list that is recommended to them. So, in very plain English, when you're on Spotify, either on mobile or desktop, and you're searching, most people click on the first result that's recommended to them. I'm sure you do, too. >> Oh, yeah. >> So, very logical. And the second finding was is that most people only search up to six characters before making a selection from that ranked list. So, >> that's because essentially every the it's not like you type in a thing and click enter as you're typing. It's like generating the list. So like if if you're trying to search for I don't know um hard rock music or something. So you know H A R D and already it's popped up with hard rock cuz it's Spotify kind of guessed your intent. And so you're saying basically most people don't get to seven letters before they're already clicking on something that popped up. >> Yeah. They they usually choose up to six. And what's you you hit on something incredibly important there which is when it comes to search like the results do change based on the character you type. And that's something called neuro instance search which you're actually describing there. I'm not going to bore people with all that in terms of a search perspective, but you know, neural instance search really makes Spotify search architecture including rank. That's kind of a whole separate separate dimension that we can get into later. Um but really you know those in the playlist space that are that are building playlists for example which are most you know you look at the majors for example broke records big loud rock you know all these companies really really hyperfocused on building UGC playlists they're all doing it with search in mind. >> Right. Right. And and so um you can run ads on a playlist. Right. So just like you can run ads for running a song you can also drive people to a playlist. And depending on how that playlist is titled and has uh in the how the description is filled out determines what keywords it's eligible for, I believe. And then um the the mission is basically like give your playlist some traction with ads. Now your playlist is ranked for some keywords just like just like Google. If you type in how to rering my guitar, what you know, whatever it is on Google, there there's an article that won, right? it's at the top and there's a rank system there. And so Spotify works the same way. Um, what do you what kind of details do you have about how their their SEO system works and and how people will try to reach that number one spot? >> Yeah, I'm certainly not an expert. I've been only been doing this for 10 years now. Um, I can only speak from my personal experience of >> only been doing this for 10 years. >> Well, listen, Andrew, I'm learning every single day. I'm often wrong more than I'm right. [clears throat] But I what I've learned on the search side is that I really think there's two things that probably matter in terms of what Spotify is ranking up and ranking down. And that's number one, streams on a 28 day interval. So that makes sense. Everything on Spotify is 28 days, monthly listeners, discovered on section, you name it. Um and then two would be retention. So, you know, how many times is someone after they put on a playlist whether whether on mobile or desktop, you know, how long are they actually listening to that playlist? And I've I've heard, you know, internally there are like uh success or failure metrics that they're tracking on a playlist level. Um I don't have any more information than that, but I think if you really zoom out and just use common sense, it's in Spotify's best interest to recommend things with the most retention. And my guess would be those two metrics and perhaps CTR would probably inform what gets ranked up and what gets ranked down. But you made you made a really interesting point about running ads to playlist, which I think is a very popular thing at the moment. But what a lot of people don't realize, and you're an ads expert. I'm certainly not, but the cost, the CPA, uh, compared, you know, when when running ads through Facebook to a playlist compared to perhaps an original work is night and day. Often times it can be half or a third of the cost, >> which which drives this incentive to perhaps optimize around search by running ads. That's what I've done for songs that don't hit. If like let's say I'm trying to promote one of my own songs or a client song and you know we're like the best we're doing is 50 60 cents a conversion and it's like you know it's not that good. We could keep brute forcing it and maybe keep trying more ads but you know at some point you run out of ideas and it's it's it's like do I just give up on this song and and sometimes you know that is the right move. But what I've I've managed to do in some cases is let's create a playlist that the song fits perfectly in. Make ads with the song for that playlist. So it's the same exact audio in the video except the video instead of like having performance clips or stuff that's more related to the song. It's like scrolling through an old, you know, there's like an old uh new metal 2000s hits playlist, right? And then this song is number one on there, but then they're scrolling in the ad. is scrolling through the playlist and it's showing, you know, here's Lincoln Park and corn and deaf tones and whatever. So, um, because you're prone to the ad from the customer, you customer fans perspective, whatever you want to call them. Um, they're it's easier for them to say yes because you're not giving them like one song they haven't heard of. You're sending them a playlist they already know they like and, uh, they're seeing things they already know, so they're more likely to. So, therefore, all of a sudden, cost gets cut in half. song campaign kind of saved in a way. >> Um there's trade-offs of of course, but it's it's a it's a whole thing. Um which which kind of brings us to September 1st where you you just explained how you know generally what we understand from from this and and I can confirm like basically everything you just said from all the playlist curators I know all the labels I know that this is how their algorithm works from what we can tell. It's basically the same way that YouTube and Tik Tok and Instagram, etc. work. They sort things by retention and clickthrough rate and all that. Um, you would expect things that are doing equally well to keep ranking high and to keep getting a consistent result. And then all of a sudden, September 1st comes around and I'm getting I'm hearing whispers through the grape vine of like, why did my playlist just instantly stop gaining followers? It was gaining like 10 a day and now it's losing 10 a day. and and um all of a sudden this playlist curator's playlist is just like not getting the streaming volume it used to be and um and I heard it from multiple people and then eventually uh we chat and and then you know and then so uh what's been happening since September 1st now that everyone has that context >> panic on the street just panic no to be serious but you know I I like you probably look at all the same data tools thirdparty visualizers like artist.tools tools and chart metric and you name it and you looking at not only our own playlist to see how they're doing from a performance perspective looking at the growth curves for example um but also looking at others too you know you trying to see how the space is doing and what I observed on the 31st were that I started to see hundreds of very historically you know reliable popular user playlist which all by the way ranked as the top result either were totally removed from search started ed to lose followers uh or started to lose streams precipitously. And what I observed first is the hardest people that were initially hit were the selfserves which was Groover, Playlist Push, SubmitHub, Youugrow. You know, a lot of these companies uh you know are essentially that the promise is you pay for consideration. They're essentially open air playlist bazaars putting together artist and curator. And um I was a little surprised that they were hit first, but over the course of it's now September 30th when we're doing this interview. Over the course of 20 days or so, I started to see that initial uh issue with those self-s serves spread to other companies, other independent record labels, other folks, for example, that own popular playlist as well that perhaps weren't plugged into these platforms or perhaps weren't violating the 2s. And uh it's concerning. I'm I'm seeing um spread not only to playlist that Spotify, you know, I don't want to jump ahead too far. U I I know we have we have a big big revelation here to share with with some evidence, but uh to keep it simple, but to to summarize here, I'm seeing this spread uh to very established legitimate networks, and it's concerning. >> Yeah. And and the first big thing I heard because I I hear people tell me all the time that the algorithm is crashing and dying and nothing's happening anymore. And there's been a few big ones that actually happened recently. Like for example, back in like February, March, a lot of AI music was just completely throttled in the 90% reductions in algorithm streams. And then around early August, there there seems to have been human artists who have been getting lumped up in something with that too, which which I covered both of those instances on my channel in sometime in the last month. And then a couple weeks ago, Magic Records put out this Instagram post where they're like, "All of our playlists are like I don't know if they were locked out or they've all they stopped growing." thing and I think it was the second case where they weren't locked out. It's just all the players weren't doing anything anymore. Um, and like those guys um when I went to Indie Week in New York, I I met them and they they're talking about how we get that lovely thumbs up thing from Apple there. They were talking about how like how much they spend on ads from their playlist for their their label and like like they literally like showed me like this is what we do, right? And um that's the legitimate way to do it. So like I I can I can't confirm everything they're doing, but from everything I've seen, they're about as legit as it comes, right? They're they they make music. It's humanmade. They run ads to it. They get real listeners. They have a bunch of actual fans. They've got like three billion streams or something crazy like that. So, if it can happen to them, [laughter] it's it's it's kind of like who like who are the people who are affected by this that don't have the power to have thousands of people yell on the internet on their behalf cuz their post made a big stink online. Like there were it got it made waves like I reposted it. A bunch of other people who who know them and know that they're legit reposted it. Tagged Spotify >> and they got a lot of the issue resolved. But I think they're still having problems. But, uh, this this is kind of like it if it it's this is like a potential doomsday scenario for a lot of these, um, self-s serve players, but also some labels who have invested a lot in growing their Spotify playlist. >> So, so a lot to say there. So, first of all, Magic Music, the reason that they, so they had access for about two weeks or so, they were essentially logged out of their their playlist accounts. And the reason that it turned out that was the case, and by the way, I have a ton of respect for Jeff and Ben. They're incredibly smart people, was that someone was impersonating them online, essentially, you know, reaching out to independent artists and record labels saying, "Hey, I'm Magic Music or I'm an ANR at Magic Music. I can get you in these playlists for 100 bucks. Which if you, you know, if you were to look at Luminate, for example, and see the kind of consumption that Magic Music is doing every single day, and you did the math on the master and the composition, you know how ludicrous it would be for them to to to prospect and ask anyone for 100 bucks. So, you know, the scam was a silly one. They eventually, I believe, worked with uh a lawyer and Spotify for Artists and their distributor to get those to get those accounts back, which is a great result. But you're hitting on a very important point about you know people running playlist as a business model using it essentially as instant distribution uh you know instead of trying to rely on solely algorithmic or editorial. >> So you know the smartest operators in music right now and I would lump magic records in there. Um Seven Colt 2 broke records you know these are people that are really doing the math and by the math I mean they're looking at what's their cost of acquisition. So that's that's CPA through Facebook. You know, what does it cost you to get a click or a follower on your on your Spotify playlist and then in relation to how many pe how many people are actually choosing to reream that playlist, which is called the stream to listener ratio. And as we know, every stream beyond 30 seconds triggers a royalty on the master and the composition. And if you can get enough streams or rereams or if the stream to listen ratio is high enough in relation to the click, you can actually make make a profit. And again, like everyone that's smart in music is doing this math and that's why what magic is doing is is so interesting and important and um Yeah. So yeah. Yeah. No, absolutely. I mean I know people who are dropping like 10 20k a month into promoting their playlist and that's by no means the cap. That's just what I like currently running. [laughter] >> Right. So, and there's there's people who have done hut. I know people who have spent over two grand a day on specific playlist and ads because the they're making money and they, you know, they they saw that thing of if I can get number one for all these keywords, we're going to get the stream. They're not just getting the ads anymore. They're getting the free organic spillover effect. Um, you sent me some some some screenshots and reports and I think it would probably make more sense for you to share your screen than me because you're more thorough with it. Um, so I probably I think at this point it'd be good to look through some of these examples. People can see what we're talking about with with how these playlists are shrinking. >> Yes, I think that'd be very helpful. I think I sent some some screenshots via artist.tools, for example. Andrew, here's my first example of a user playlist that was hit with this uh designation or spam filter on or system as you want to call it on August 31st. If you toggle here the last 90 days for example, you can see this playlist historically has seen very very consistent growth. Let's look beyond in 90 days just to confirm that that theory. Here's 6 months. >> Yeah, >> this also clearly to me looks like a playlist that was ad promoted just like like it's getting this constant flux of um ad spend going to per day. And then I if you go back to last 365 I think you had it in the graph there was like a point where it um where it increases like I'm guessing that was just them like they doubled their ad spend or something to see >> well I hate to differ with you on this but this is a playlist that actually appears as the recommended top result on desktop and mobile and perhaps they're running ads on this playlist but if they if they were running ads let's say on the 31st from Vermont on why is the playlist losing followers? Right. That's true. That's true. Cuz I guess what would explain that increased slope there back in like December, January? >> Yeah. Good, good point. I'm not sure there's always I mean there's always some variation in search. >> Yeah. >> As as you know, like playlists do gain and lose popularity based on the seasonal search, variety of factors that are probably >> probably above my pay grade. But you can very clearly see on August 31st that it went from growing to to dying. >> Yeah. >> And I don't see it hasn't for the past gosh almost 22 days. It hasn't seen one new follower. It's only lost. And that's why, you know, that's that's would be my first example of a playlist that's been hit with this this confirmed spam filter by an unnamed ex unnamed Spotify executive. But White Girl Banger is like it's a very popular UGC playlist title. It's all mid-2000s music. It's Pitbull. It's Kesha. It's Rihanna. >> I see a Katy Perry on there. >> Yeah. [laughter] Yeah. It's it's all the the millennial hits for people that dance with their hands in the air. So, um let me toggle the chart view here. This one you're going to see is not as pronounced, but you can see >> definitely a change though. You know, it goes from on some days Saturday, which you and I both know Saturday, Friday, Saturday are usually the highest consumption days on Spotify, that it goes from seeing a pretty consistent like, thousand followers a day down to half, 513, 437, and it goes back up as the weekend approaches, 6 >> Saturday and Sunday are actually the two worst streaming days I generally see. >> Yep. >> Um Yep. But but yeah, this this is starting like the August 31st, September 1st threshold is that's where the followers what did they cut in half. >> Yes, >> potsentially. >> Yeah. >> And one thing I haven't mentioned yet is um there are people who have been winning because of this change. I talked to someone last night where um I hadn't talked to him in a in a couple months, so I I hadn't gotten a chance to see what happened to him with this change. He he also has most of his volume is coming from these these old playlists that he's made and a lot of them they do very good organically. Like we're talking millions of streams organically per month. Very very minimal ad spend. Um and I was honestly kind of worried. I was like we're going to hop on this call. It's going to be like tears doomsday whatever. And we look and starting September 1st all of his playlists almost have doubled in streaming volume. So there is there can be winners on the receiving end of this where certain playlists have lost their their streaming volume because they got deranked and then other people have been swapped up to it. And fortunately for him um he's he's an example of someone who's always done everything the right way. You know human catalog of music he's made with his bare hands. Um playlist he owns he plays fair. he only uses it for his own catalog even. So it's like the most extreme example of like there's nothing he's ever done remotely wrong, >> right? >> Um and he's winning from it and I I was just happy to see that. But you know that's the opposite side of this that I just be worth pointing out. >> There there are always winners and losers. Okay. So we'll go on to our third example here. This one will be a little more extreme. That is night and day difference as of August 30th. >> And I I think a lot of a lot of people are probably going to say in the comments were like, "Well, all these playlists are just they're just bots." But the first example we showed, the line was quite straight. But the the if someone's not familiar with these graphs, the the bar charts below the line are the actual daily metrics. So even if a line looks like pretty straight, like you can see that the daily followers are kind of all over the place. So that's not like proof that it's not bots cuz like you could you could program a bot script to to do that if you wanted to, but but generally people that are botting stuff. Uh they're not going they're not trying to fool smart people or smart's the wrong word. They're not trying to fool people who are fully aware of how to find the hints of boss. They're trying to get the easy sale to rip someone off and move on cuz there's, you know, they they could try to fool like fooling someone like you. Like you've been in this for 10 years. You know what a real playlist looks like. And it's going to take a lot of work and a lot of money for someone to fool you into thinking a playlist is real when it's not. When there's a thousand artists who have never even looked at a playlist growth chart before that are going to be way easier targets, >> right? I I often I often just to just to I mean there's a lot of tools that I use for example some first party some not but oftentimes I simply look at the search position and if a playlist isn't you know the top result or the second result on mobile or desktop that's always to me a very negative indicator and then you have to ask the question well where is the where where is the consumption or the engagement coming from then if it's not searchable so that's usually what gives away for oddsters when they try to do this none of their playlist titles are ever all that searchable and they never really rank Yeah. And and you know, shout out to Artist Tools here. This is the site we're on, but you can um on Artist Tools, you can actually click SEO and you can go and you can see the search position country by country. So, you could see, you know, this playlist is ranked number one for this keyword in the US, it's ranked number three in Brazil, it's number two in the UK, whatever. And you can just look up every keyword in every country that you're driving volume to to see where it's where it's ranking. Um, so if you want uh 30% off Artist Tools, the link in my description is uh Aaron over there is a a genius. Aaron from Artist Tools. [laughter] It's great platform. >> He's got a hard job. He's doing an excellent job with uh with that platform and I use it. I use it constantly. >> But you're you make a really important point which is a very niche point. But search results are never universal. They do change based on kind of the device you have, what plan you have, premium or, you know, ad supported for example, you know, what country you're registered in. Like search results again are not universal and can really differ based on on your location. So that's a very nuanced point you're making there. And you're absolutely correct. >> And so like why do you think this is happening? And this is kind of when we're getting to a little more, you know, we put our tin foil hats on a little bit, so to speak. But but there's data to back to to back, you know, like we know what's happening. A lot of playlists, >> generally larger ones, but but there's also ones at every size point. Playlists that were ranking really high are all of a sudden tanking, right? >> There's winners and losers. So why would Spotify do this? There there's probably a lot of legitimate reasons why they would. I'm sure I mean Spotify is a very smart company and they probably have data internal metrics that back up every change they make. I think, you know, reading them from their messaging lately to the interviews their executives do, my perspective is that they really are trying to clean up clean up fraud, AI spam, slop, lawyer for content. We don't know what how they define spam or lower lawyer for content, which is a little concerning. Um, but it's probably a reasonable definition. Um but you know from my perspective um I I believe that a lot of this I I believe that Spotify understands that search is probably the most important and most powerful force on Spotify and it's like the movie Dune control the spice control the universe right [laughter] control >> I knew you'd love that science fiction reference >> uh you know so it's the same idea on Spotify control surge control the universe And um if you're Spotify, like there are financial incentives to directing traffic away from historically very popular UGC playlists that either, you know, people have created through search or got gotten lucky on organically, away from those UGC playlist to their own editorial and algorithmic surfaces. And you and I both know there's been a lot of reporting, for example, around perfect effect content. Spotify, for example, commissioning custom music for their playlists, even though they again, they're always clear. They don't own the music. That's correct. But they do they do, you know, commission it and license it, and often times pay much lower royalty uh in comparison to regular rights holders for this music. So, that would be my first point on on why they would direct traffic away. two. I mean, if you look at a lot of these algorithmic niche mixes, which I've noticed over the past 38 days, I'm just simply searching for some of these popular playlist titles that have all of a sudden disappeared. For example, our sleep music playlist, which is one of the biggest UGC playlist in Spotify, I noticed that the day that it disappeared from search on mobile and desktop, the first result was Spotify's editorial fall asleep instantly playlist. And the second playlist was gentle quiet mix which as you know Andrew Spotify has a program called discovery mode where they'll recommend you inside these algorithmic niche mixes in exchange for you taking a 30% royalty reduction. And essentially what Spotify is doing is they're doing their own version of thirdparty playlisting but they're just doing it algorithmically. So, you know, I think really the the main driving factor here of why they would turn off search for the UGC side and turn it on for their own is simply just because Spotify is a business and they're interested in in making as much money as possible. It's very logical. >> Yeah. And then when you say niche mixes, so Spotify has had your daily mix for forever and then you you might have anywhere from like two to six your daily mix playlist, which is depending on how many different discreet types of music you listen to. like I might have like a kind of a progressive metal mix, right? And then I'll have another one that's more like a modern alternative metal core mix and then I'll have like a EDM one that, you know, so each distinct state here. And now they have ones where if you look up like I'm just going to do this test right now and see what happens. Like if I go on and type in like dinner time or something stupid, uh I get Oh, actually that was I was going to say this a stupid play. was actually one of my playlist. [laughter] Lowi dinner party. Oh, nope. That was my stupid idea. >> Spam. Sorry. >> Spam. Yeah. And but there's always one. Good vibes dinner mix. So that's that's what I'm talking about here. I just type in a a random keyword dinner time. And I there's actually multiple. There's cozy dinner mix as well, >> right? >> Um and you know these playlists are I presume completely algorithmically generated. So they're seeing what people are searching for and they're algorithmically generating a mix for that that's tailored for the individual and um that mix because it's a Spotify when it's tailored to you is going to show up pretty high in their search results. And so it it does make business sense that, you know, mixes are part of discovery mode. And if we have these playlists that are pretty high in the ranking where artists can can give us 30% of their royalties to be more present in these playlists, um it's going to save them a bunch of money in royalties. Um >> Spotify calls this music with improved margins. [laughter] Is that is that something you coined or is that something >> I know I'm actually quoting Liz P. She had a really great book on this. Uh Liz P is a journalist. She wrote a book about exposing some of this perfect fit content. Uh uh a book a couple years ago, but that was one of the internal uh I guess the language that Spotify was using internally about how they describe this PFC content. So >> now this this actually a good topic to spend a couple minutes on. Um, the perfect fit contact thing by by Liz. Um, I still need to read it, so forgive me everyone. U, but [laughter] I is there like concrete evidence in that book this was happening? I I guess I like I haven't seen it and you know, is there like actually like proof that that's what they were doing outlined in that book? And if so, like what are the kind of things that she outlined as as evidence of what was happening? Yeah, it's been a while since I've read the book. I I I should have brought that. I should have tabbed some some notes here to be prepared for this to give you a good good response on that. I don't want to put words in her mouth and >> perhaps there was reporting that didn't make the book, sources, for example, that didn't want to be named or, you know, just didn't make the book, for example. So, I just I don't want to speak out of turn there and give you a pin on some I don't know much about. >> So, that that makes sense. I mean, like I I I believe people have asked Spotify about this and they've just outright denied it. And so like I didn't know if there was something where it's it's like a he said she said thing and so we don't really know or if it's something that there's more concrete proof. I mean I can't think of a reason why a a reputable journalist would fabricate evidence. So, like there's that like Spotify has a lot more to lie about here than Liz, but >> right Spotify response though when they when they respond to these allegations, it's always a very lawyer response, which is they're always splitting hairs saying things like, "Well, we don't own this music." Uh, for example, a lot of artists when they when they critique this PFC content program, they always say, "Well, Spotify owns these songs and they're playlisting themselves." No, no, they're not. They don't own the songs, but they are purposely commissioning it from song houses like Epidemic Sound, Firefly AB. I mean, there are many of these companies, all of which curiously are are in the Nordics, for example, some of them in Sweden, uh, which there perhaps is a connection there. Um, but, you know, a lot of them Spotify denies owning the music, but they never clearly deny that they're not commissioning it and putting in their playlist. So that's that's I know I know perhaps I'm splitting hairs there, but I always feel like are a little >> makes sense. A lot of times have people heard Spotify or a lot of times I've heard people call out Spotify for for that. >> So Spotify is commissioning people to make music that then they own or they license for a lot lower rate and they don't have to pay that out, >> right? Um, my counterpoint is like, well, I can't like prove that or deny, but what I can confirm is there's a whole lot of labels who do that because I know people who have been hired by a label to to commission a bunch of songs and the the label released it under a fake artist name. So, they got they got this guitarist, they got this vocalist, they paid them a pretty low rate because they're in a cheaper country and to make this really amazing music and then they released it. didn't I don't remember if they credit them or not in this as a songwriter or performer. I forget that detail. But but either way, they're not getting paid anything from that music because they were paid up front, which is a normal work for hire thing. There's there's nothing there's nothing wrong with that. But the point is people would call that out and be like, all these are fake artists. And it's like, yeah, but are they owned by Spotify? because I know several labels who do this and I know several musicians who have been hired by these labels that do this and in some cases what those individuals do later is they get the money they made from being hired by those labels to go start their own >> fake artist company. So there's also a real like man the thumbs up thing again. Um there's also a whole bunch of um there's a whole bunch of uh actual companies >> that are going out of their way to do this. Of course it's not going to do it now that I tried it. >> Wow, look at that. Yeah, Apple with their, you know, trillion dollar company over here. Um, but you know, the point is there's real people doing it, too. And yeah, I don't I don't think Spotify for the, you know, just to be clear, I'm not an anti-SP Spotify person. I think a lot of times the criticisms that get lobbied against them are often very hollow. It's often by people that have no real businesses of their own in music. You know, they don't understand how the business works. They couldn't define stream share. uh they couldn't tell you they're between ad supported versus premium. You know, there there people often times that still think that Spotify pays on a per stream basis. So, you know, I'm really not just for the listeners and the viewers out there, I'm really not an anti-Spot person. I believe actually that Spotify perhaps saved the music business. They've ushered in this incredible age of uh of consumption, discovery, and success for people. And I would say that I'm probably, you know, I've had very modest success as an independent record label, but Spotify is a huge part of our business. And I'm not even sure I'd have one without them, which is why these changes to SEO playlist search are just so distressing. Yeah. And and even though there's winners off the backs of this, um to me, the reason why this would be so scary being in someone in your position or any other labels or playlist creators or whatever is isn't that it's not that it necessarily has happened. It's that it could what could happen next or what could continue to happen right if you're looking at the prospect of possibly spending five grand 20 grand 50 grand on growing spending money on Spotify playlist now you in the back of your mind there's a well what if they kill the SEO rankings of my playlist do you bother anymore and obviously there's going to be most people are going to keep on keeping on because like Spotify is the biggest music platform But it is now it increases the risk factor of all this like a lot. Um to know that they would even touch this stuff that for years we've all considered like the bedrock of Spotify. If I agree I mean if they've implemented a technology for example that you and I have seen evidence for example of an executive unnamed executive that confirmed in writing that this spam filter system is making inferences on user playlist for example and the content inside them and based on those inferences are deciding to either rank them or recommend them more or less. Uh then then you raise an excellent point which is where's the opportunity there for the independent music space? you know, why would anyone try to build a user playlist? I'd also add too that like user playlists from my perspective are one of the biggest differentiating factors between Spotify and other DSPs. I mean, Apple and Amazon are primarily just pitch only. And you can certainly make your own user playlist there. But I think as a user, often times that I'm searching for something quickly up to six characters to listen to to listen to like I often land on a user playlist and they're often like expertly curated and they're great used listening experience and they've got high retention. So, you know, that's what makes it all the more surprising that Spotify is just all of a sudden just like turning off searchability for so many great user playlists. Uh, you know, it's it's it's it's a surprising move. So, >> yeah. And there's one thing I've thought about before, which is like if Spotify were to just get rid of all third party playlists on the platform entirely, um, it would solve a lot of problems, right? like the whole like playlist scams and people adding people's music without their permission to playlist and then their music getting pulled because their playlist is botted and you know there's a lot of problems that solve from it but you're right I don't know if Spotify would be where they're at today without user playlist being such a thing you know because everyone shares playlists all the time no one shares Apple Music playlists because Apple Music playlists kind of suck >> [laughter] >> Um, so >> Andrew, you're making too much sense here. >> Yeah, [laughter] I guess. But but uh there was something that um you tried running ads to to one of your affected playlists where >> um I believe you were you were curious like like why isn't this playlist gaining followers anymore? So, like can you talk a little bit about this experiment you ran on one of your now dying playlists? >> Yeah, there there there's a couple now. So, you know, when I noticed some of our playlist started to just all of a sudden lose followers, I decided to run some ads, run some conversion campaigns inside FBA, um, and test across different genres, for example, country, rock, pop, etc. to see if I could, you know, change the direction of the playlist losing followers back to growing again. And if you'd like me to, I can share my screen here. We can kind of get into uh what I saw. Uh but >> really what what was what was uh concerning was that I've sent some of this data to you ahead of time to kind of review. But if you compared the hyped edit link clicks, you know, the daily number of link clicks inside FBA compared to the number of followers, for example, I was getting on my sleep playlist through artist.tools, the conversion rate on those, let's say, 1300 clicks on one day was like less than 10% of the followers that I was getting on artist.tools. And you know, from experience, you're you're the expert here, not me. But if you're spending, you know, 250 to 500 bucks a day, you know, running driving traffic to a playlist, uh, playlist, for example, how many followers would you estimate you would get you just in general? >> Uh, if you're spending 500 a day and you're getting a, you know, getting three conversions per dollar, you're getting 1500 conversions. I would assume 750 followers, >> right? >> So, basically for everyone, everyone that's had never run these before, >> right? >> Typically about half your conversions will follow. somewhere between 40 and 55% I would say. >> Well, I was getting 128. >> So, this this raises a question in my mind, which is if these playlists all of a sudden, let's say they aren't spam, right? Like, we don't know what Spotify's definition of spam is. It seems to be very opaque at the moment. Perhaps they're painting with too wide of a brush. But these playlists that do have this designation at the moment, um, it raises a question of are the playlist followers being throttled or are they simply not being counted correctly? And if it's the latter, like they're not being counted correctly, that then raises a more significant question, which is, are the streams or the consumption that's in these playlists, is it being accounted for correctly? and therefore our our rights holders, the people that own the masters in the composition. Are these people actually getting paid correctly because of this AI spam system? So, it's much deeper than just playlist followers. It actually uh speaks directly to are royalties even being calculated correctly here. >> That's also a concern and I've shown you a couple screenshots now to prove that up to support my point. Um but it's it's very concerning and right now there are more questions than answers. >> Yeah. Cuz the the follower thing you you know that that's a that is a tactic that I've social media platforms take where for example if you make a new account on Tik Tok and you start immediately going and following accounts like day one. If Tik Tok doesn't trust your account yet, they will let you click the follow button. They'll let you like post. They'll let you comment. and then you go check 10 minutes later and refresh the app, whatever you and suspiciously you're not following any accounts and you've never liked a post and you've never left a comment. So they they let you think you took that action but then behind the scenes they just don't let it happen. So it's not it's it's this is an existing like software tactic that companies take. Doesn't mean Spotify is doing it, but it kind of sounds like they might be. And for a playlist, it's I could see them be like, we think this playlist is for some reason very bad for our platform. So, we are going to make it so that it's deranked and then when people follow it, we don't actually follow them. Basically, it's just pushed to purgatory. It's never going to grow again or it's going to grow so slow that they're just going to abandon. That would kind of be fine. Like, it's it's not, but let me give you an instance where it is fine. Let's say someone is actually um doing something super shady with their playlist or you know they're selling slots on it or whatever. Like Spotify could ban them, but then that person's going to know they're banned and then go and take some action. So what if instead of banning them, they shadowban them where the playlist can kind of never recover. Um and in that case it's happening to a bad a bad actor, right? Like let's say they're they're breaking a bunch of rules. It's like, okay, cool, fair enough. But if they're manipulating the stream counts on it, like let's say an actual person does find the playlist and then stream the music to manipulate those numbers would actually just be taking m taking royalties away from rights holder that like if it's an actual human listening to a playlist even if that playlist did take a bad action. It wouldn't be an excuse for for for throttling the streams on it which um I don't know have you run that test? Well, some of that would require you to have like SFA data, for example, or first-party data. And unfortunately, platforms like Luminate, which I use, I'm a subscriber, don't show you playlist level data. They show you track or album level data. But on a playlist level, that's the kind of data that only someone at Spotify, for example, would have. So, that would be really tough to, you know, to determine that. Well, in your case, the some of the playlists that that this has happened to that you have your catalog in them, >> correct? >> You must have s S4A access to these, right? >> I do. I do. I do. And I'm seeing I'm seeing like, you know, pretty pretty harsh decreased consumption. I mean, just for just for context and clarity, you know, as a result of this this uh spam filter system, as Spotify calls it, I'm seeing about, you know, 50% decrease in consumption across my entire catalog at the moment, which is, you know, we're a small modest record label. Uh it's it's it's pretty devastating and uh concerning. So, >> so for for example, like I what I will probably be a great way to test this is that you go to one of your playlist that's that's throttled. Like for example, the one you spent like 250 to 500 a day and you go grab a song in your catalog that fits the playlist but is either very low in the playlist or isn't on there yet. And you slap it in slot two or something and you let it sit there with no ads for a week. So you get a baseline for what it's getting from that playlist for a week and then you ramp up the ads to 250 a day. And then like if nothing happens, you kind of have your answer that that it's not just the followers being throttled, it's the streams. It's not concrete because like someone could always argue like, well, they're not they're not getting that far on the playlist or they're skipping before 30 seconds. And so there I think no matter what, there's a way to to dispute it um from Spotify's perspective because of that. But it would be pretty pretty strong evidence in my opinion that like they're doing something to manipulate the numbers for the play. So I hope they're not doing that, >> right? To be clear. But uh if someone were to prove that that was the case, I think Spotify would have some some answers, some questions to answer. So what's next? So what's what's like what's the path forward here for for you know labels and curators and and artists? what's what's the kind of takeaway? What action should people take? >> Yeah, I I can't speak for others. I think, you know, if you were to ask different uh people that own playlist, for example, independent record labels that use UGC playlist as a distribution model. All of them would probably have a different answer, including me. So, I'm certainly not the the emissary of the playlist space or want to speak on their behalf. But from our perspective, we have hired a law firm named Adams and Reese in Nashville. There's a litigator, well-known litigator in music named Richard Bush that's uh we've hired to make certain inquiries at Spotify at the moment. And we're hoping that Spotify gets back to us soon to answer some very simple questions on this new spam classification system, how it works, whether our songs and playlists are being classified as such, and perhaps ask them nicely to perform a review of our of our accounts, our playlists, and perhaps perhaps, you know, fix fix mistakes where they've been made. So, I'm just my overall point is that um everyone supports cleaning up fraud and spam. We do too. You know, no one likes that. It steals from legitimate artists. Everyone wants to protect the royalty pool. But my perspective is at the moment the system is probably painting with too wide of a brush. It reminds me last year, and this might get me in trouble, Andrew, so I I apologize, but a year and a half ago, a lot of independent artists, either through DIY distributors or inside their Spotify for artist dashboard, were getting these very aggressive artificial streaming notices. And often times these were people that weren't even these were, you know, songs that they had released years ago. They were doing no marketing to them. You know, the song had kind of come and gone. And I during that period talked to many distributors, small ones, big ones, major label ones. A lot of them privately uh expressed um skepticism on some of these notices. In some case, they were privately overturned. Distributors are really cautious about their relationship with with with uh with Spotify because it's their most important delivery relationship. But it was it was it was very clear to everyone inside the music business that Spotify's know a IML fraud models were making mistakes. Uh and my my concern is that uh they're making mistakes right now with this new spam filter in terms of how they're classifying certain user playlists. And that's why I was so motivated to come on and finally speak. I'm not a public person. Uh doing interviews for me is usually very uncomfortable because I'm usually the one on YouTube watching your channel, watching smart people talk about music. But uh I was compelled to come on and I'm grateful to you that you'd let me come on and talk about this issue and raise awareness about it. >> Yeah, absolutely man. It's it's rare that um someone wants to talk about stuff like this because you know generally people have a lot more to lose than gain by explaining what's been happening to their in the playlist world like everything's very kind of like hush hush about how things >> operate even from a label world because like you you everyone has their own like kind of trade secrets for how they market and so by giving giving giving it away you're opening it up to more competition potentially or whatever. Um, so I appreciate you coming on with regards to what you just said about the um the fraud thing from last year. Like I even got roped up in one of those. I made a video about it. I think like going through a Spotify artificial streaming report because like I I have a I have a distributor like I own a distribution platform and like so I get the actual report straight from Spotify about it. So, in the video I went through it, but like what happened was an artist account that I had that I primarily just used for testing a couple years ago, hadn't released music on it, hadn't promoted it. Um, I log on to the S4 account, I see just like random big spike in streams one day. Um, nothing that I initiated. I'm like, "Oh, great. I'm definitely going to get served from Spotify here." And inevitably, 3 months later after that spike, I get an artificial streaming report and they find me $30, $10 per song. that was that was affected and I'm just like cool that's great. So I asked this the the part distribution partner um like what can I do about this? Like I didn't do anything but like how do you prove that you didn't do something right? I'm dealing with this right now for like I have an Apple card which I overall love. We had a a random airline bill us from Chile and so we dispute it and they're like well send us proof. like how do I prove that I didn't do something right? [laughter] Like how do I do it? And so this is the same thing and and like the distributor is like there's really not anything we can do here. Like if we if we dispute it, they're going to they're going to tell us that they have evidence and they're not even going to show us like the evidence and they they are kind of judge, jury, executioner in that way. And I don't know what I mean, you're pursuing the legal right now cuz I I believe in like in if if in Europe there's a lot of rules around like what data you can request if you're a user, but I don't know how that works if you're if it's not a user thing, if it's more of like a licensing thing, right? Like you can get all your data from Spotify, every data point they have about you as a user, but can you audit their internal findings as a rights holder? Well, it's it's called Discovery. So >> So yes, but you have to hire a hire a law firm essentially, >> which is which it really isn't for I mean Spotify is such a big company. I mean, it's sad that you'd have to do that to get a straight answer from them, but that's really at this point I've reached out to people like Jennifer Massid, for example, who's the head of all indie labels at Spotify. I've spoken to my distributor, ADA, just just for clarity. Uh, I've I've reached out to a lot of people that would be in a position to give me some straight answers and at the moment I'm getting none and many other independent record labels like me are in the same position. So that's why I've hired Adams and Ray. It's not to, you know, pound my fist on the table and and uh yeah, >> you know, to to not be an adult about this. It's just that if you're implementing this system, which you've admitted now in writing exists and you admit also that it perhaps needs some tweaks, I'm simply asking you to perhaps uh write some wrongs here. >> So, I've heard people doing that with Meta before where like account gets unfairly restricted and it like gets severely harming their business and so they hire a law firm to contact Meta and then miraculously, oh, sorry, we were wrong. Here's your ad account back. And it's a shame because like >> you can afford to do that. I could probably afford to do that, but like most people can't really afford to do that. If you're an indie artist or an indie label, like you're running on a shoestring budget. You're probably running at a loss. >> Yes. >> So, it's a shame that it would come to that. Um, and I don't know what from Spotify's perspective, what's best answer here because like they can't afford to have infinite reviewers either, >> right? It's a it's a you know, again, it reminds me of the the issues that people saw around artificial streaming last year. And from my perspective, it went on for a few months and eventually they it seems like they turn the dials down or the models improved, for example. So, I'm hoping that's the case here in time perhaps solves this problem and shakes this out. But, you know, these changes that we're seeing on the 31st are immediate. They're perhaps devastating for some people that have been doing this for a very long time and it's causing people like real economic damage. So, >> well, that being said, Austin, thanks for coming on. Appreciate Sandra is a pleasure.

Grow your music with a proven ad system

Learn the Meta ads and streaming growth playbooks Andrew uses across artists, labels, and agency campaigns.

Explore the course

Keep exploring

Continue in another format

Explore the topicPlaylists