The Exchanges

Q Can I ask, can you give me an example of a macro win that you went in the face of and then pivoted and decided that it was better to have it behind your back, just so I get a feeling for it?

A Of course, um, And an example for us, there are two good examples. One is when I joined Spotify, we had what was then a very profitable business model, which was you give the application away for free on desktop computers because people didn't think they wanted to pay for music. So then they started investing in music, playlisting and building libraries. And then they realized that actually wanted to listen to those playlists on the go on their mobile phones. We charge for the mobile application. So simply put, we charge for mobility. And that worked beautifully for several years. And, you know, we loved it. VCs loved it. They invested a lot of money. But then all of a sudden it turned out that this macro wind was blowing called smartphones. And we started seeing people that Not only use their mobile phones more than their desktop or mobile first, but many users that were mobile only, there was no desktop. So we didn't have a free tier at all for these users. So our, our business model broke. Now we were not incentivized to want to see that in the short term. And we could see our user growth actually starting to slow down. But when we accepted this and we took the pain and the risk Of repositioning all of Spotify, figuring out a way to give away the mobile experience for free, even though that was what we were charging for. So we literally changed the entire business model. We …

AI assessment note: “One is when I joined Spotify... this macro wind was blowing called smartphones”

Q Can I ask guys, you know, when we think about the multiple models that we'll use, you mentioned there that we'll have actually kind of, uh, open, closed, very specialized. When we think about, like, cost, the more models we have, the more costly it is also, and margins are impacted. How do we feel about cost and implementation when thinking about model adoption?

A From Spotify's point of view, um, most of the stuff we've built in-house, to Tamara's point, is actually mostly about cost. Right. So if you, if you look at, for example, as we talked about generating a lot of voice, you want to generate like two minutes of voice per day for half a billion people. It's a billion minutes that can ruin you. Like cost is actually already like the biggest factor in if you can deliver products, even though the models are there to generate great voice. The problem is actually doing it cost efficiently. So, so I think that's completely right. A lot of the innovation and technology has to go towards routing often actually for cost purposes. I do think I have a little bit of a different view on there is some chance that per company, I think you will want to embed the entire user history in one model, right? So today, most companies, including Spotify, Many separate systems that are optimized for different purposes, and they sort of collaborate in semi-predictable ways. There's not one system, but if you look at some of the papers coming out of Amazon and so forth, they are starting to look at, you, you re, you literally tokenize all of it. Basically the log, the, how you scroll, how you click, what you listen to, what was in the content that you listened to. And then you just, you just do a prediction, token prediction based on that, based on all of thi…

AI assessment note: “most of the stuff we've built in-house, to Tamara's point, is actually mostly about cost”

Q Why doesn't Spotify create its own models, Gustav?

A Well, partially for that. So we have our own models as well. We, we do both. We have lots of our own models, but we don't have a GPT-IV that we haven't told the world about. I'm sorry to reveal that. But it's also what I said before, like our goals are different. OpenAI's stated goal is AGI. That will drive different incentives. It will not optimize for delivering two minutes of audio to, to have a billion users per day, right? It's not necessarily going to optimize for, for cost efficiency and building pragmatic products. So that's, that's why, you know, there's no point in trying to compete with them for us because we don't even have the same goal. So we're trying to build exactly the things that we don't think they will build. And we would buy the things that they will build or someone else will, but obviously we're working with, with Google and GCP. So that's where we do most of our work.

AI assessment note: “there's no point in trying to compete with them for us because we don't even have the same goal.”

Q Can I ask, can you give me an example of a macro win that you went in the face of and then pivoted and decided that it was better to have it behind your back, just so I get a feeling for it?

A Of course, um, And an example for us, there are two good examples. One is when I joined Spotify, we had what was then a very profitable business model, which was you give the application away for free on desktop computers because people didn't think they wanted to pay for music. So then they started investing in music, playlisting and building libraries. And then they realized that actually wanted to listen to those playlists on the go on their mobile phones. We charge for the mobile application. So simply put, we charge for mobility. And that worked beautifully for several years. And, you know, we loved it. VCs loved it. They invested a lot of money. But then all of a sudden it turned out that this macro wind was blowing called smartphones. And we started seeing people that Not only use their mobile phones more than their desktop or mobile first, but many users that were mobile only, there was no desktop. So we didn't have a free tier at all for these users. So our, our business model broke. Now we were not incentivized to want to see that in the short term. And we could see our user growth actually starting to slow down. But when we accepted this and we took the pain and the risk Of repositioning all of Spotify, figuring out a way to give away the mobile experience for free, even though that was what we were charging for. So we literally changed the entire business model. We …

AI assessment note: “we instead got this wind at our back and we started growing exponentially”

Q listening to it like walking, and I'm kind of looking at it, and I'm in London, and I'm kind of about to get run over, and you know, that would worry Shaq. Um, mother less so, but Shaq more so. Um, you know, how does that change your product paradigm thinking when suddenly you do actually have to worry about like user engagement and video and not just background play?

A So back to The contrarian sort of view. Spotify is, you know, the vast majority background application. And so that's what we optimize our music licenses for. First of all, in podcasts, we don't have to license. So we are allowed to play video on demand in the foreground as well. But if you look at what use cases we're going after, Because we're mostly a background application, the use cases we're going after are these long, mostly backgrounded sessions, right? Podcasts. Typically, we're not going after the foreground sort of silent jujitsu videos that make no sense when you background them because it's just a lot of panting, right? We're going after the traditional podcasts. And so what we did starting actually with, with Joe Rogan podcast was we built video into Spotify. We didn't know what the Usage would look like. Maybe it would be that no one actually ever brought up and watched the video, but without saying exact, exact metrics, what happened was exactly what you said. People consume the vast majority of these shows in the, not the vast majority, the majority of these shows in the background, but they do bring it up every now and then to see what they are talking about, who is talking. So we are using video mostly for the Sort of talking heads format, if that makes sense. There are these famous moments of, you know, Elon Musk smoking weed, for example, on Joe Rogan podca…

AI assessment note: “majority of these shows in the background, but they do bring it up every now and then”

Q is very expensive, especially in early stage companies. It's about speed of execution. It's about getting shit done. And, you know, I'm invested in a couple of companies where, oh, we want to talk. We want to debate. Yes. Let's theorize on the future of, you know, privacy. Fuck no. Just ship. Ship, ship, ship. Learn, iterate. Go, go, go. This is your only advantage. Um, why am I wrong?

A Well, I think the way to think about it is there's a difference between being fast and being right. You can, you can get, you can go very fast to nowhere, you know, ship code every day, but if you fundamentally wrong, you're just going to get to nowhere extremely efficiently and fast. And so I've found, especially if you're doing more complex thing, and maybe this is actually because I come from the music industry where building product Had to be licensed as the lowest common denominator between, you know, three, four majors with, you know, the, the impact of being wrong being three, four years. Cause that's a term of one of these licenses. So I, I, maybe we've been trained as a company to be more careful because the cost of being wrong was incredibly high, but I actually think that that cost of being wrong forced us to model the world slightly Better and longer. Um, maybe because we were afraid of mistakes, but I actually think that grew into a skill at Spotify. Now, of course you can, you can debate endlessly. You need ownership. You need to make decisions. And Spotify has certainly been guilty of being too unclear in, in, you know, who owns something and it can slow the organization down. Uh, but fundamentally, I think now we have a really good balance where quite a lot of people actually have a, Fundamental strategic view of what we're doing and why. And, uh, this debate bu…

AI assessment note: “there's a difference between being fast and being right. You can, you can go very fast to nowhere”

Q many cases, often for startups, uh, competition's roadmap drives a lot of their roadmap. Uh, you've said before in, in your show, which is look at the competition and then do something completely different. Uh, always love a contrarian opinion, so this is great. Um, but why do you believe this? And can you explain it to me a little bit on why it's better to do something completely different?

A So I think this is especially important for a smaller company going up against a bigger company. If you, if you assume that the people in the bigger company are smart and competent people, which I think you should assume, Then what they're doing, their strategy probably makes a lot of sense to them. They're doing it because it plays to their strengths. Maybe they're leveraging their distribution, their hardware, their user base, their, um, payments method, something like that. Right? And so if you think about it, should you as a smaller player go up and try to do their strategy with less people, less budget and less assets, you're just going to become a lesser version. I mean, that's a likely outcome. So, especially as a smaller player, I think you should do the opposite. Try to figure out if there is a position where that bigger player doesn't, uh, isn't playing right now, and ideally doesn't even want to play because it's contrary to their business models. You know, there's some sort of collateral damage of what in strategy terms is called a counter position. There is, there are exceptions to this. If you're a bigger player, There, there is a, you know, it's, it's a credible strategy to say, we see some new behavior feature user interface really working. Let's implement it and give it to our user base before they find it in this other user base. And you've seen this play out …

AI assessment note: “should you as a smaller player go up and try to do their strategy”

Q listening to it like walking, and I'm kind of looking at it, and I'm in London, and I'm kind of about to get run over, and you know, that would worry Shaq. Um, mother less so, but Shaq more so. Um, you know, how does that change your product paradigm thinking when suddenly you do actually have to worry about like user engagement and video and not just background play?

A So back to The contrarian sort of view. Spotify is, you know, the vast majority background application. And so that's what we optimize our music licenses for. First of all, in podcasts, we don't have to license. So we are allowed to play video on demand in the foreground as well. But if you look at what use cases we're going after, Because we're mostly a background application, the use cases we're going after are these long, mostly backgrounded sessions, right? Podcasts. Typically, we're not going after the foreground sort of silent jujitsu videos that make no sense when you background them because it's just a lot of panting, right? We're going after the traditional podcasts. And so what we did starting actually with, with Joe Rogan podcast was we built video into Spotify. We didn't know what the Usage would look like. Maybe it would be that no one actually ever brought up and watched the video, but without saying exact, exact metrics, what happened was exactly what you said. People consume the vast majority of these shows in the, not the vast majority, the majority of these shows in the background, but they do bring it up every now and then to see what they are talking about, who is talking. So we are using video mostly for the Sort of talking heads format, if that makes sense. There are these famous moments of, you know, Elon Musk smoking weed, for example, on Joe Rogan podca…

AI assessment note: “we are using video mostly for the Sort of talking heads format”

Q How do you think TikTok did that? Sorry, help me understand that. We're, surprisingly, I'm showing my age here, a bit so down with the goods. We're very big in TikTok, um, but in venture capital startup TikTok, like how did, sorry, how did TikTok kind of reinvent a new medium format of music?

A So it started in 2014 as Musical.ly and there was a lot of trial and error. I've spoken to, to, uh, both the founders and there was a lot of trial and error, um, And they sort of happened on this use case and they didn't start by going and licensing stuff. You know, there's a, there is a beauty in being small, small enough that people don't care. And that's actually a problem with being big. We can't really just try things. We have to license them first, which means we take the cost up front. Uh, so back then they were small enough to try this and they tried to do something different actually, but they found product market fit. You know, they, they kind of saw in the day that what people were actually doing with these clips was recording dance videos and they leaned into that. And then later they started licensing, and then they got acquired by ByteDance and, and got more distribution and so forth.

AI assessment note: “what people were actually doing with these clips was recording dance videos and they leaned into that”

Q Can I ask, was this the case with the introduction of podcasts? Obviously we made the shift purely away from just music to obviously, you know, Podcasts and then also kind of video podcasts as well. Was that the case where actually it took a year to invest to reach feature parity with podcast players? I'm intrigued on that one.

A For sure. Actually podcast was mostly additional, so we didn't go down in metrics. Those, those cases have been when we change a user interface, for example, completely, or you change the different recommendation algorithm, or you change something fundamental, your metrics may go down. Podcast was different because it was mostly additional, but it is true. That it took a long time before Spotify was sort of our, our internal goal was, you know, get the podcast catalog on the service and build a, uh, best in class podcast player, you know, feature parody. That was the first stage. And then obviously the idea was to pull away. But to your point, that took over a year before we had everything in place that people even expected from a podcast player. It was painful to spend that time, but it wasn't the problem of actually, you know, losing altitude in the meantime.

AI assessment note: “to your point, that took over a year before we had everything in place”

Q Yeah, no, I totally get you on that one. I was intrigued. Sorry, before we do like you as a leader, where did you lose altitude? If you think back, where was it like, oh God, we're back in the bottom of this mountain. Fuck.

A So a typical example was, um, there's actually, let's, let's take the same example that we discussed because there's context around it. When we finally launched this mobile free tier that was, you know, contrarian to YouTube and that risked our business model, what happened was Leading up to that, because fewer and fewer people use the desktop free tier, actually from the outside, what happened was our conversion metrics looked better and better. It looked like a larger and larger percentage were becoming premium, but that's not because more were converting. That's because the free user intake was going down, right? So from the industry point of view, the conversion was at an all time high, you know, and what do you think happened when we launched this massive, massive free tier? It just dropped. Everyone downloaded the free app. No one converted for months, right? So the, the MAU metrics, the growth was very inspiring. That was an exponential curve, but premium conversion was an exponential curve in the other direction. And as you can imagine, many people, labels and investors were wondering if that was going to catch up or if this was, if we were down to like single percentage points, premium conversion, instead of like It did catch up. Uh, first, you know, we, we also needed to invest in ads monetization. So basically the, the analogy here is user metrics went up, but moneti…

AI assessment note: “monetization, both free ad monetization and paid conversion, you know, that dropped severely in altitude.”

Q Not on schedule, but I'm intrigued. Do you agree with Mike Mignano's suggestion of the disregard of the social graph and the movement into the media recommendation engine of the future?

A For sure. I mean, Spotify went through this journey and this is one of the macro wins that we saw quite early and you know, that, that, uh, and Mike and I have discussed this at length. We, Spotify started as a, uh, I wouldn't call it social media. I would call it curation media, different version of the same thing, which means the internet started with you. Taking something offline, friends, books, music, and then asking your users to curate it into graphs, playlists, friend graphs. That was what, that was the pattern, right? And that's where Spotify grew up. And then it shifted from curation to recommendation, where users weren't prepared to put in all of that work to soundtrack themselves. We actually had an in between, which was editorial, where we had editors or professional playlisters that helped curate for you. But the natural extension of that is to start combining the editors with, with machine learning, right? And now we have many pure machine learning set. So we actually made a bet on the shift from curation to recommendation or social media to recommendation media. Four or five years ago and rebuilt the application quite drastically. And today we, I would say we are a, a recommendations based company much more than a curation based company.

AI assessment note: “For sure. I mean, Spotify went through this journey”

Q did you get over the fear of the cannibalization of an existing business model that was also working? You make it free. You could have a lot of the premium users, uh, you know, churn and just be ad supported users, but free. Um, and actually you saw the cannibalization of existing user bases. How did you get over that concern and what did the internal discussions look like there?

A So we actually used a lot of data. It was, uh, Mary Meeker, I know precisely when we realized this, it was a deck from Mary Meeker that showed the projections of smartphone growth overtaking desktop growth, and we looked at that and we said, holy crap, you know, in, in the 2012, it was almost like a date. This is going to blow up. So we discussed that a lot. We kind of found the confidence ourselves to do this, but the big problem we had was actually that we have other stakeholders in the labels and the labels didn't see it. They were not looking at Mary Meeker's slides and they were not so keen to risk the entire business model that they had just figured out after, you know, the whole, you know, uh, piracy era and so forth. So that was actually the bigger problem. And we had to wait all until, all up until we had stopped growing and actually started losing users before they agreed to give us the licenses we wanted. But by then we had, um, prototyped and researched and AB tested a bunch of different concepts. So when we got the licenses, we actually knew what we wanted to do and had already built the experience and we launched it almost overnight when we signed the, the contracts.

AI assessment note: “So we actually used a lot of data. It was, uh, Mary Meeker”

Q is very expensive, especially in early stage companies. It's about speed of execution. It's about getting shit done. And, you know, I'm invested in a couple of companies where, oh, we want to talk. We want to debate. Yes. Let's theorize on the future of, you know, privacy. Fuck no. Just ship. Ship, ship, ship. Learn, iterate. Go, go, go. This is your only advantage. Um, why am I wrong?

A Well, I think the way to think about it is there's a difference between being fast and being right. You can, you can get, you can go very fast to nowhere, you know, ship code every day, but if you fundamentally wrong, you're just going to get to nowhere extremely efficiently and fast. And so I've found, especially if you're doing more complex thing, and maybe this is actually because I come from the music industry where building product Had to be licensed as the lowest common denominator between, you know, three, four majors with, you know, the, the impact of being wrong being three, four years. Cause that's a term of one of these licenses. So I, I, maybe we've been trained as a company to be more careful because the cost of being wrong was incredibly high, but I actually think that that cost of being wrong forced us to model the world slightly Better and longer. Um, maybe because we were afraid of mistakes, but I actually think that grew into a skill at Spotify. Now, of course you can, you can debate endlessly. You need ownership. You need to make decisions. And Spotify has certainly been guilty of being too unclear in, in, you know, who owns something and it can slow the organization down. Uh, but fundamentally, I think now we have a really good balance where quite a lot of people actually have a, Fundamental strategic view of what we're doing and why. And, uh, this debate bu…

AI assessment note: “there's a difference between being fast and being right.”

Q many cases, often for startups, uh, competition's roadmap drives a lot of their roadmap. Uh, you've said before in, in your show, which is look at the competition and then do something completely different. Uh, always love a contrarian opinion, so this is great. Um, but why do you believe this? And can you explain it to me a little bit on why it's better to do something completely different?

A So I think this is especially important for a smaller company going up against a bigger company. If you, if you assume that the people in the bigger company are smart and competent people, which I think you should assume, Then what they're doing, their strategy probably makes a lot of sense to them. They're doing it because it plays to their strengths. Maybe they're leveraging their distribution, their hardware, their user base, their, um, payments method, something like that. Right? And so if you think about it, should you as a smaller player go up and try to do their strategy with less people, less budget and less assets, you're just going to become a lesser version. I mean, that's a likely outcome. So, especially as a smaller player, I think you should do the opposite. Try to figure out if there is a position where that bigger player doesn't, uh, isn't playing right now, and ideally doesn't even want to play because it's contrary to their business models. You know, there's some sort of collateral damage of what in strategy terms is called a counter position. There is, there are exceptions to this. If you're a bigger player, There, there is a, you know, it's, it's a credible strategy to say, we see some new behavior feature user interface really working. Let's implement it and give it to our user base before they find it in this other user base. And you've seen this play out …

AI assessment note: “try to do their strategy with less people, less budget and less assets, you're just going to become a lesser version”

Q Can I ask, was this the case with the introduction of podcasts? Obviously we made the shift purely away from just music to obviously, you know, Podcasts and then also kind of video podcasts as well. Was that the case where actually it took a year to invest to reach feature parity with podcast players? I'm intrigued on that one.

A For sure. Actually podcast was mostly additional, so we didn't go down in metrics. Those, those cases have been when we change a user interface, for example, completely, or you change the different recommendation algorithm, or you change something fundamental, your metrics may go down. Podcast was different because it was mostly additional, but it is true. That it took a long time before Spotify was sort of our, our internal goal was, you know, get the podcast catalog on the service and build a, uh, best in class podcast player, you know, feature parody. That was the first stage. And then obviously the idea was to pull away. But to your point, that took over a year before we had everything in place that people even expected from a podcast player. It was painful to spend that time, but it wasn't the problem of actually, you know, losing altitude in the meantime.

AI assessment note: “to your point, that took over a year before we had everything in place”

Q How do you think TikTok did that? Sorry, help me understand that. We're, surprisingly, I'm showing my age here, a bit so down with the goods. We're very big in TikTok, um, but in venture capital startup TikTok, like how did, sorry, how did TikTok kind of reinvent a new medium format of music?

A So it started in 2014 as Musical.ly and there was a lot of trial and error. I've spoken to, to, uh, both the founders and there was a lot of trial and error, um, And they sort of happened on this use case and they didn't start by going and licensing stuff. You know, there's a, there is a beauty in being small, small enough that people don't care. And that's actually a problem with being big. We can't really just try things. We have to license them first, which means we take the cost up front. Uh, so back then they were small enough to try this and they tried to do something different actually, but they found product market fit. You know, they, they kind of saw in the day that what people were actually doing with these clips was recording dance videos and they leaned into that. And then later they started licensing, and then they got acquired by ByteDance and, and got more distribution and so forth.

AI assessment note: “what people were actually doing with these clips was recording dance videos and they leaned into that”

Q Can I ask, are you shitting yourself in this moment? You have your board, like putting pressure on you. You see single digits conversion, your investors are putting pressure on you. Are you nervous as a product leader?

A I think you should always be somewhat nervous and paranoid. And, and yes, I would say this maybe not. So I had a lot of faith that conversion would go up because we had seen so many years that the core correlation was the more you use the free product, the more likely you are to convert to paid. We call it the more you play, the more you pay, which always made sense to me. Like you're going to start paying for a product you use and love. You're not gonna pay for something you don't use. Right. So getting it to use it more. Means you're gonna not want the ads. You're going to want offline. You're going to want speakers. I was pretty certain of that. That was my worry. The worry when we launched this free tier was if it was actually good enough. This contrarian hypothesis, hypothesis that we had, it could have been BS. It turned out to be true, but that we couldn't test at scale really. So I was very nervous about that. I think investors and labels were much more nervous about the conversion metrics and the ads metrics. Um, And also, to be quite honest, I wasn't responsible for the conversion of the ad metrics. So maybe other people were very nervous, but they did a great job and they caught up.

AI assessment note: “I think you should always be somewhat nervous and paranoid. And, and yes”

Q know, as well as I do, humans are lazy, humans are not as disciplined as you are, you know, the most disciplined person on the planet, Gustav. Um, but, um, like, point being, it doesn't lead to a great investment often, and that's the thing that I find so hard as someone who loves product as much as me, Is you have to be more multifaceted. Does that make sense?

A It does. And actually I'm glad you brought that up because that's a really, you asked about like insights sort of through my career. I've always been incredibly passionate about technology, even for the sake of technology and product. And I find so much joy in like, I wonder if we could build this thing, just see if we can build it. And, and at the beginning of my career, that means you undervalue the importance of the business model, but what made it click for me? And, um, and I think this is true for many. So what I talk to my teams about now is, you know, the, the, The greatest innovations are when you find a new technology and you change not just the product and how you use it, but you change the business model. Those are always the most disruptive changes. And, you know, to people, to some people who care, you know, mostly about design or product, it may almost feel like, you know, making money is something bad and, and, you know, something negative, and it's only VCs who care about money. The way I phrase it is, If you want to build the greatest product on earth, it's going to require a lot of engineers. And so if, if, um, if you have a, you know, 20% margin and your comp, your competitor has a 40% margin, you both make a dollar. They're going to hire twice engineers. They're going to build twice as a good product. Like you want to finance that product passion. So you nee…

AI assessment note: “you want to finance that product passion. So you need a good business model.”

Q I'm in London. It doesn't work that way either. But no, I get you totally. I'm sorry to interrupt. Can I ask that? Is there a time when actually You sat around the table, and you went, oh shit, we should have copied them. It could be YouTube, it could be any of the other competitors. We're actually doing the different didn't work out.

A Well, I would say this. There are certainly tons of things. It should have been faster. I think lyrics is an example, uh, of that, which, which we now have is very engaging. That was obvious. So we've certainly been super slow for various reasons on these things. In that case, it wasn't really because we had a contrarian view. It was actually, which is often the case because of how the music industry works. So if you work normally at Twitter or Google or Facebook or something, You're, you're as a product person, What you do is you try to figure out what the best possible thing is that you could build, and then you try to build that. What you do when you work in the music industry is you try to figure out what the best possible thing is. Then you go to the labels and you try to license the lowest common denominator of what is licensable. And that often means that you can't build the best possible product. You have to build with much, much narrow constraints. And often actually the cost is a constraint because the job of a music label is to take a piece of IP And try to sell it to you as many times as possible. You know, foreground rights, background rights, video rights, sync rights, all of these rights. So you can see that as, as very annoying as a product person, or you can do what I do and see it as constraints are what makes the world interesting. You know, can you do someth…

AI assessment note: “I think lyrics is an example, uh, of that, which, which we now have”

Q I remember congregating with you in a corner, uh, but, uh, no, I, I totally get you and I, and I appreciate that. Final one on just the leadership. I do have to ask you, you come across as so self-assured, confident with conviction. When you review your leadership today, Gustav, is there any elements where you're like, I could be better here? That is a weakness of my leadership.

A Of course. I think everyone has imposter syndrome. And, and certainly when you work in, when you're fortunate enough to work in a company that grows this much, even if you have the same title, I've been sort of CPO CTO for, for several years. It changes every year because first you're a CPO of a tiny Swedish company, then you're CPO of a Scandinavian company, then of a global company, and then I'm a, you know, listed New York company. It's, the job changes all the time, and every time you feel like there's, there's no way I can be that person. You know, you're, you're an imposter all the way, I think. So I, I certainly feel like that. Um, and I know, I think I've grown to understand my weak spots. Uh, better with age. I already told you one, which is, um, I sort of have this global CEO perspective, which is helpful. I think it's one of the reasons why, um, Danny likes having me around because he feels like I care about the, the, the outcome of the company, but it is also tricky when you get in other people's shit. That's, that's always tricky, and I have to, like, balance, you know, how, how I do that in a diplomatic way. So I have, I have all kinds of, of flaws.

AI assessment note: “it is also tricky when you get in other people's shit.”

Q is also buried like three clicks in, and I didn't know about it until someone told me about it on Twitter. And when I think about the shift from curation to recommendation being so central to the future of content discovery and content platforms, why is it so buried? And is that a product mistake? I would love to say that someone else's question, but it's mine. So forgive me.

A It's a perfect question for me. So one answer is stay tuned. You will be happy. Uh, Another more telling answer is, The way to think about it, we kind of had to, we made choices between what the main paradigm is. The main paradigm of Spotify was playlisting. That's what Spotify was and what we grew up around, you know, user curation. And then, you know, we started playlisting for you and we, we had to decide, should we have the radio channel as the analogy or should we have the personalized playlist as the analogy? And they kind of do The same job, a personalized playlist within a, you know, a certain genre like chill. It is very similar to starting a chill radio. Literally you will get the same songs actually. So we've struggled a little bit with which user paradigm to use. And as you can imagine, older people, they like, uh, radio paradigms and younger people, maybe not so much, not calling you old here, but you know,

AI assessment note: “we've struggled a little bit with which user paradigm to use.”

Q within the org, I, I spoke to Daniel before the show and, uh, the number one thing that he said that I had to ask you was he said, ask him about talk is cheap. And so I want, I want to dive into this. What, what, what is talk is cheap? What do you mean by this? And how does that come out in how you communicate? Bleed. Oh.

A So, It's sort of play on words. Uh, I believe deeply in this notion of Socratic debate, and I love debating with smart people, and I fundamentally believe that if we pay these high salaries that we pay in the tech industry for what are supposedly the best brains in the world, we should really maximize the value of that sort of rented brain power, right? And yet I find that most, specifically most US companies, they're quite hierarchical and very smart people often don't really understand What they're doing or why, and they're underutilized. And if you think about it, that means they're actually overpaid for what they're doing, not because they're not smart enough, but because they're not leveraged enough. So in the tech world, there's always been, for a long time, there's been this notion of talk is cheap and that code decides arguments. So what I decided in order to sort of poke fun at that is to say talk is cheap. So we should do much more of it because In fact, just talking and debating with really smart people who are not afraid of you importantly, which I think is also sort of a Swedish societal trait that you're not so afraid of authority is way cheaper than writing code. Writing code is one of the most expensive things you can do. So this thing of code decides argument, I never really bought that you should only do when you're quite sure, you know, writing and shipping c…

AI assessment note: “to say talk is cheap. So we should do much more of it”

Q Final one for the quick fire. You mentioned the constraints that labels bring and that licensing brings. If labels and licensing We're entirely Spotify's, entirely integrated, just in a hypothetical world. What would you do from a product perspective with the removal of constraints? What would you do in an ideal world?

A So I think what I would do is I would try to make the music experience much more interactive. I think this is what you see with TikTok as well. I think users want to participate in the music. They want to use the music today. They can use it to make dance videos, but I think, you know, there are so many talented musicians out there who sit at home and play their piano or their guitar and they sing and they create music. But the licensing world means that You know, they can't use a beat from, from the song that they love. If you think about coding as an analogy, when you want to learn to code, what you do is you go to GitHub. And then you have this vast library of, of open source code that you can use and you can look at and you can start prototyping from and building and learn. There isn't anything like that in music. There is no GitHub for music. That's probably what I would change. I would try to, and I think that's the reason why music has actually had very little innovation. The track is sort of still three minutes and, you know, an intro and two courses a hundred years later, almost. So I, that's probably what I would try to do. Just innovate on the formats. That's not true. Other media formats have changed. Drastically over the years, but music hasn't so much.

AI assessment note: “I would try to make the music experience much more interactive.”

Q How radical do you think you get with UIs? Everyone here wants to be at the forefront of design, of, kind of, interfaces with consumers. How radical can one be in how far you push it quickly?

A I think I want to build on something Scott said there, because I think it's the definition of what is UI changes. So you took the example of the, of the AI DJ we did, and you're exactly right. It was the design team that user tested. What is this? Is this a utility like Google? Is there an AI person like Alexa or, or do we actually, uh, is the UI that we digitize the real person with the real personality that exists? And we chose the latter. So that is, that is design and, and it is brand. So I think designers need to need to think holistically about the experience in that sense. And that is still design. It is the user experience. And I also think something that, that designers, um, need to get good at in this world is to understand the, the capabilities they have. They need to understand basically GPT-IV as well as they understand a user, what, what, what, uh, what it needs. And how it works. So an example that I think is a good example is Mid Journey. So Mid Journey, the designers there clearly understood the performance of the model when they built the Mid Journey experience, right? It took like two minutes to generate an image. It was wrong one out of four times. They could have built a horrible experience. We waited for two minutes and got disappointed 75% of the time. Instead, because the designers there and product people understood the capabilities, they built a fault-…

AI assessment note: “I think it's the definition of what is UI changes.”

Q is also buried like three clicks in, and I didn't know about it until someone told me about it on Twitter. And when I think about the shift from curation to recommendation being so central to the future of content discovery and content platforms, why is it so buried? And is that a product mistake? I would love to say that someone else's question, but it's mine. So forgive me.

A It's a perfect question for me. So one answer is stay tuned. You will be happy. Uh, Another more telling answer is, The way to think about it, we kind of had to, we made choices between what the main paradigm is. The main paradigm of Spotify was playlisting. That's what Spotify was and what we grew up around, you know, user curation. And then, you know, we started playlisting for you and we, we had to decide, should we have the radio channel as the analogy or should we have the personalized playlist as the analogy? And they kind of do The same job, a personalized playlist within a, you know, a certain genre like chill. It is very similar to starting a chill radio. Literally you will get the same songs actually. So we've struggled a little bit with which user paradigm to use. And as you can imagine, older people, they like, uh, radio paradigms and younger people, maybe not so much, not calling you old here, but you know,

AI assessment note: “we struggled a little bit with which user paradigm to use.”

Q In terms of synchronizing leadership, I'm too interested. What have you done that has worked and what has not worked that you've disregarded?

A So I think if we started with what has not worked, the naive, the, the, the mistake you can do in this, not having swim lanes is, it's very unclear who's responsible for, for anything. Everyone is responsible for everything and no one makes decisions. Um, Because they're often in Spotify because they're too nice. That's actually a bigger problem than being, than people being, um, too aggressive. In other companies, you may have the opposite, that if it's unclear, people are gonna power grab. I've seen the opposite. It's like people don't want to step on other people's toes so they don't make decisions. So that's a problem with this model. So you still need to make it clear where the ultimate decision or responsibility and accountability lies. And that's a balance. If you go, if you go too far, um, you're going to create a swim lane and this person is going to run away and do something, uh, and think that everyone else in the company is a moron and everyone else is going to think everyone else is a moron. So I, the, the tricky thing with this is finding the balance of, of, um, accountability. And, uh, and ownership. So you feel agency, um, but still having a single strategy and having alignment. So I'm kind of forcing, I'm saying that even if you own something, Uh, let's say you own what the end user experience looks like. You actually have to take your strategy, bring it up in …

AI assessment note: “if we started with what has not worked, the naive, the, the, the mistake”

Q and likewise, final one on, and actually just you. And then we're going to do a near death experience and then wrap up with a quick fire. Do you have any other, I've seen you in your, your wife and it's this beautiful marriage and relationship. Do you have any like genuine relationship advice for me on what it takes to have such a strong and loving marriage and relationship?

A Hmm. Tricky question. I mean, I actually met my wife when I was. 20, 21. She was 19. And we didn't marry for, you know, many years. So we had a very long sort of trial period. It's like a freemium model, right? So when we married, we were pretty sure. Uh, so I don't know what a good, what good advice might be there. We, We, so we worked together for a long time before we married. Um, I was working quite hard. I was, um, traveling the world before I started working very hard. Uh, she studied and traveled. We lived in different places, so we had a lot of freedom and had known each other for quite a long time before we actually started living together. So I think that was important. Neither one of us felt the need to like, you know, I need to go explore and feel like I'm missing out. We had done many of those things, even though we met earlier. I would also say that You know, my wife and I were certainly different in many ways, but we are very similar in, in values. That helps quite a, quite a lot. We never, we almost never have disagreement on sort of values or what we, what we think about things. Then, then we are very different in terms of our skills, completely different. And I think that's a benefit. You know, she's fantastic at the things I'm really bad at. So we compliment each other really, really well.

AI assessment note: “we are very similar in, in values. That helps quite a, quite a lot.”

Q Final, final one, I promise. I, I read somewhere about Olivia Rodrigo and how she has changed the structure of her, like actual songs to accommodate the changing algorithms of music, whereby choruses are much more brought forward. Um, is that true? And is this a new create like music creation process to fit technology?

A I think that's absolutely true. We've seen small versions of that as well. So when you move from the download world to the streaming world, There is no, in a, in a financial sense, no marginal cost to try another track, right? Which means you're going to start exploring more, which we think is great. So it gives more artists an opportunity. You're also going to start to listen to, to a lot of new music, like sleep music and focus music. Whereas if you paid, you know, a dollar per three minutes, you probably wouldn't soundtrack your sleep so that, you know, so entire industries popped up because the business model changed and they adapted the music. But you also see it because people now have shorter attention spans. You can see the chorus moving closer and closer to the beginning of the song, right? Um, everyone is trying to optimize for the first few seconds, and I don't know if I think that's Good or bad, but it is what it is. Like incentives and systems drives behavior for sure.

AI assessment note: “I think that's absolutely true. We've seen small versions of that as well.”

Q Completely unfair again, but you know, joys of it being my show. Um, which competitor do you most respect and why?

A I have a great deal of respect for, for all of these sort of big tech companies, the usual suspects, the, the Google and YouTube, Apple, Amazon, Facebook, and so forth. What, but what I think is interesting specifically and impressive about this set of companies that I think is different from The previous set of companies is that there, I think there's a reason we call them technology companies. And some people say that makes no sense. Like car companies were also technology companies at the time. Everything was a technology company until it became old. But I do think these companies are a bit different because I do think they think of Technology itself as the strategy. What do I mean by that? I mean that people try to label them. They try to label Amazon as a book company and then as a book and diapers company and as an everything store, but then they started doing AWS and like people can't like pin these companies down. Same with Google. They were a page on the internet search page. Then they became the start page of the browser. And then they became the start page of a browser of an OS and, and they did email and then they did Photos and so forth. Same with Apple. You know, computer company, music company, phone company. These companies, they, they refuse to be categorized as what they are, and they keep changing all the time. And I think it's because they think of technolog…

AI assessment note: “I have a great deal of respect for, for all of these sort of big tech companies”