| In JC’s Newsletter, I share the articles, documentaries, and books that I enjoyed the most in the last week, with some comments on how we relate to them at Alan. I do not endorse all the articles I share, they are up for debate. I’m doing it because a) I love reading, it is the way that I get most of my ideas, b) I’m already sharing those ideas with my team, and c) I would love to get your perspective on those. If you are not subscribed yet, it's right here! If you like it, please share about it on social networks! Share 💡JC's Newsletter
👉 Spotify: A Product Story - Episode 2: How do you charge for nothing? (Spotify) Context: After months of negotiations and several white papers on encryption technologies sent back and forth, we landed on a price: 10 bucks a month. That was to keep and listen to 30 000 encrypted tracks offline for up to a month before you had to reconnect and confirm that you were still a paying user. Now we just had to also figure out the tiny little problem of why on earth anyone would actually pay for this service... if they were already getting it for free via their pirated music solution? When we surveyed our users we found that - not surprisingly - that was 10 bucks a month that literally no one said that they would pay.
➡️ User research is great. For understanding what people think they will do. Not what they will actually do. How to use User Research (and not to use): So, yes, do your user research. Look at what the data says. But treat it for what it is: people trying to predict their own future reactions. And if you have a strong hypothesis for why they will react differently in reality - such as them underestimating their own willingness to actually pay for convenience - you should follow that, as the contrarian hypotheses are the most valuable ones!
Blow your user mind! Every product person knows that the hardest thing you can do in product development is to change an existing habit. If you're going to try to tear people out of an existing habit, that already works for them, you need to be 10 times better - especially if that existing habit happens to be free while you're planning to charge $120 per year for it! If this works, this is going to blow people's minds. Definitely.
➡️ How do we blow people’s minds more? Product x Business Model x Distribution: Really great product development is more than just a cool new feature or even a breakthrough new technology. Really great product development almost always combines technological innovation with business model innovation. A huge success factor with Spotify was our licenses, because they were unique. At that time no one had the licenses we had. Again, others had the tech, the streaming. Many product managers forget about the business innovation part, leaving the real product half done. Even some of Spotify’s early competitors essentially seemed content to launch first and figure out the details later.
Can you afford to be wrong? Adapt your product process based on this: Spotify literally had to negotiate every new feature or addition to its business model - with at least 4 major labels - meaning that it took substantially longer to pilot new ideas than for a normal company - like maybe a Facebook, Google or Twitter that could just ship code and A/B test it instead of always debating its benefits, and drawbacks and consequences. Spotify couldn’t afford to be wrong very often because being wrong would be too expensive. Licensing the wrong thing for 4 years could have killed the company. So we learned to be more careful, prioritize brutally, scenario plan, game out and think through second and third order consequences of our decisions to a larger degree than perhaps other companies have had to.
Benchmark on freemium: After a few weeks, we saw the “conversion moment” starting to happen. First a small percentage of our user base, then gradually more and more people ended up paying. The bar for freemium models at the time was Skype, which converted about 7% of their users to pay at least something - maybe a dollar. We hit that in a few months, but it just continued from there -- past 10%, 15%, 20% - and we realized that we had stumbled on what might just be the most effective freemium model the world had ever seen.
👉 Spotify: A Product Story - Episode 3: This party is going to end (Spotify) Never try to fight a macro wind, you will lose! Get the wind at your back instead of in your face. Test big or go home: Test as different as possible. The goal is not to go deep on one idea, it’s to cover as much area as you possibly can. And sometimes, that means you have to test ideas that aren’t necessarily even feasible in the real world, or are even bad ideas, like capping the amount of time people could use the app for free, which was diametrically opposed to the principle of: the more you play, the more you pay. The funny thing here is that, you have to let them run a long time before you know how this behavior works. Because as we said earlier, the free tier is an experience that enables consumers to convince themselves to go “paid” and that music is way more valuable than it used to be. That journey for people takes time. And it takes a different amount of time for different people. So the shapes of the retention curves are very different. We're talking months. Like nine months later.
How to think about conversion:
➡️ Being patient when you think about conversion about a freemium product to see emergent behaviors is really important. First principles and thinking as a member when defining a product: I think that informed us about all these different caps. If the caps made you use the product less and cap your usage, that would somehow be inherently bad for conversion down the line. Kind of the intuition that we had was that even when you get into it as a consumer, if you know that there is a cap, you're not going to invest. You're not going to fully engage with this product because you know you're going to run out of it. So it's no point to even start. So we believe that it really would mess with people's minds to kind of go along on that journey and allow the experience to convince themselves to go paid.
➡️ Crazy idea: Can we have free unlimited chat with doctors but with ads? Would members want that? My intuition is likely not, but maybe a subset of the markets would like it. This testing was expensive, but it was also invaluable - because it validated our hypothesis that capped usage would drive listeners away, not draw them in. No matter what kind of cap you used: time limits, number of tracks or location, it would also cap how much you could invest in Spotify and therefore how likely you were to convince yourself to later become a paying user.
So now we knew that we needed a free mobile tier that promised the consumer that they could use it 24 hours a day, 7 days a week, anywhere, for the rest of their lives without ever having to pay - just like the desktop client had promised. So it had to be unlimited, but at the same time it still had to be clear why you should pay for the premium version at all. And then I think big learning number 2 was what we came to learn about what people loved about Spotify was the feeling of ownership. So you collect all these songs, you put them in your library and you feel like you're amassing this limitless library that's all owned by you. And we started exploring the idea that “shuffle” might be that perfect connection where you could build playlists. You could have that feeling of ownership, almost celebrating shuffle as a feature rather than a constraint. Just the way that the free tier would work would be that, you know, it's “shuffled only”. That was kind of the hypothesis. Based on those insights, we knew what people loved about Spotify and when we saw what was happening in the market outside, we started to prototype the specific ideas about the shuffle free-tier. At the time, there was no free app in the App Store where you could do that for free, where you could find any music in the world, save it to a playlist, listen to those playlists for free.
➡️ Your free app has to be too good to be true, and you need to create an incentive to convert to premium. It took more than a year of iteration for Spotify to get there. Watch the competition, so that you know exactly what not to do. Look at the competition, and then do something completely different. Your competition is going to do what works best for them and their strategy – what they’re best positioned to do, that leverages their unique assets and skills - that’s why they’re doing it.
Innovation to fuel growth: When to build a new product: Internally, we use this language of climbing mountains, right? So, you know, once - if you use the analogy of a product being a mountain, you climb the mountain and you would reach its peak and the peak is somewhere, right? And so once you hit that peak - or like a maxima in another language - you have to build a new product. And that was what we were doing. But when you jump to a new mountain, there's two questions that you have to ask yourself, right? How high is that mountain? Meaning what's the true potential of this product? And if you actually jump to a position which is lower than where your current mountain is, it's not going to look like a great state, right? Because the product might have much larger potential, but the initial data might actually look like it's slightly worse.
Their new mountain: Machine learning based free tier lets you listen to tracks on demand. These tracks appear on certain, algorithmically generated playlists that are personalized for you, such as Discover Weekly, Daily Mix and Release Radar. It was very incremental. We started to see the inflection and then it started to slowly, slowly grow. And unlike the first product where it was like opening the floodgates, this was more of an incremental improvement, which led to more incremental improvements, which led to more incremental improvements. And today you actually see a quite a big delta. If you look at the retention metrics for the free products, now a few years after, you see the change very clearly. But at that point, you didn't understand if it would happen and what would happen. If you have vision and the data is directionally showing you that you're on the right path, then you just have to believe that the results are going to follow.
👉 Spotify: A Product Story - Episode 4: Human vs Machine (Spotify) Why it took time to start: I’ll be the first to admit that our journey into Machine Learning (or ML, as it’s often called) started with a bit of a misstep. Because for the first few years of Spotify, we didn’t quite see how it would actually fit into our bigger goal of bringing users the perfect listening session. Besides being hard, recommending new music to listeners also seemed secondary. We didn’t discard it, but we didn’t yet see it as core, as our main thing. By 2011, we saw the macro wind that we mentioned in episode 3, the shift from curation-focused to recommendation-focused services, that did a lot of the work for you, starting to really pick up speed. And we realized that recommendations needed to become a part of our core strategy and that we needed to hire for it.
How it started & how they built the team: What happened, of course, was that we didn't really manage to hire anyone because it was impossible to find these machine learning engineers. The internal model would often come up with absolutely amazing and unintuitive suggestions that no human would’ve ever found. But it also made simple mistakes that no human would’ve made. So it still came across as a bit “dumb” to the listener. Throwing Machine Learning at data you don’t fully understand isn’t enough to give you a great product. The problem with collaborative filtering is that it doesn’t actually “listen” to the music or understand it in any way. It just looks at how often it appears together with other songs. For that the Echo Nest (a company acquired by Spotify in 2014 specialized in music intelligence) had this series of crawlers that would go and read the entire Internet, and finding all these blogs, reviews, all sorts of stuff, in order to see how music was being described and then doing natural language processing on top of that.
➡️ Do the same with online forums? The Echo Nest had what we lacked - algorithms that said everything about the music itself but nothing about how listeners interacted with it - and Spotify had what they lacked: listening data on how people interacted with the music. Lots and lots and lots of listening data. With that, we’ve arrived at product strategy lesson #3: If you don’t have one side of an equation inside the company -- look outside for it. In some cases this could mean an acquisition, where you join forces with an existing team. It may seem like a steep price upfront, but ask yourself: how many years' worth of time and effort will you save compared to building out the same capabilities in-house? And, furthermore, how many years into the future will you jump by combining forces now?
Only about a year after Spotify acquired The Echo Nest, we launched Discover Weekly, our first fully algorithmically generated playlist, individualized for every user. Every Monday morning, you’d open up the app to find a brand new playlist full of gems that you missed from Spotify’s vast back catalogue of millions of tracks, served up for you by the algos.
➡️ “Your Prevention Weekly: a full list of tips from nutrition, to sleep, to check-ups individualized for every member”. We had reached a level of Machine Learning accuracy where we could switch from just giving users even better tools to playlist themselves, to just giving them a weekly playlist and let them save the tracks they really liked. We switched our vision from “even better tools to playlist yourself” to “you should never have to playlist again”.
The job of a PM in a ML-world: The job of a product manager in a machine learning world is to find metrics, objectives and source datasets that can objectively describe what “great” actually looks like - or in our case - sounds like at scale. You might be surprised to hear me say this but in several aspects, the editorial team’s playlists were superior to the algos’ in that they were better at tapping into the zeitgeist and being able to target complicated human emotions and situations, because they were built by complicated humans who understood those situations. But at the same time they were very limited in their ultimate ability to perfectly fit any specific listeners’ taste because they had to be an average of everyone’s taste in that use case. In order to find a big audience for a use case, situation or mood, each playlist needed to be largely right for everyone - but not perfect for anyone. Conversely, the algo playlists could cater to an individual users’ taste, but they lacked depth and understanding of the situation. It turns out that humans are pretty great and still very smart compared to most algorithms. But the thing about humans is they don’t scale so well.
We discovered that when you think of the playlist as a product, and the editor as the product manager - sourcing the “test set” - the pool of tracks that Meg created, ML can deliver a truly personalized listening session. And thus algotorial - algorithmic plus editorial - was born, and 1 + 1 finally equaled 3 for Spotify.
👉 Spotify: A Product Story - Episode 5: When your winning bet becomes your losing bet (Spotify) Lesson #1: Don’t get attached to the status quo. Meanwhile, the rest of the internet had started to catch up. The pay off in terms of latency and reliability that came from running our own data centers diminished every day, as cloud computing got better, faster, and cheaper. The challenge is that when you spent enough time to get into the nitty gritty details of doing something yourself, zooming back out and realizing that this isn't the most important thing that I should be doing right now, that's like a personal challenge and not something that comes natural to humans. I believe that's a fight that you have to take and really be purposeful about. Otherwise, it's so easy to just stay in your little bubble and continue iterating on your little problem that you've learned so much about and become so good at. The reality is, it took us longer than it probably should have to accept that our data centers were becoming such a bottleneck. Partly because we were too attached to the “fantastic technology” that we originally pioneered Lesson #1: Don’t get attached to the status quo. When you’re really good at something, you continue doing it -- because you’re really good at it, not because it’s necessarily the right thing to do!
How to do migrations: So this is what I think many people miss with migrations like this. It sounds logical because you're switching from your own premise to the cloud. And actually, ideally, the cloud would require even less people. But that's not how it works. Like for two years, you're going to have to have twice the people because you need to keep the old system alive until the last service from the old system has migrated completely. So during two years, you're going to behave like you're running two companies at the same time, right? It’s not like you can switch them over. And people don't really get how expensive and hard that is. I remember ordering millions and millions of dollars of servers as we were doing the migration because we had to have this overlap. And I mean, it doesn't feel great doing that, knowing that these are going to be in use for a short amount of time. But that's just the way - that's what you had to do. This brings us to product strategy lesson #2: the question isn’t if you can do it better - it is if you should do it better. As a company, you have limited resources, so your biggest cost isn’t actually going to be the direct cost of the thing you would buy. The biggest cost is actually the opportunity cost of what you would have been able to do with all those smart people, if they didn’t have to build this thing instead - the thing that no other company could build and that you couldn’t buy anywhere - the thing that actually differentiates you.
Who to partner with: By 2016, Amazon Web Services was the industry leader, with Google Cloud Platform as the newer entrant - the challenger. Amazon was - in many ways -- the obvious choice. But that doesn’t mean it was the right choice. It seems like we deliberately chose to bet on someone who was a challenger, who was inclined to be more aggressive in helping us figure out solutions, just more hungry, because we would be one of the biggest customers for them instead of just yet another customer. There was also a really strong fit, honestly, in engineering culture, which if you speak to our engineers, they say it was very important.
➡️ It is mostly always better to be the biggest, most important customer of a smaller player than customer n°1000 of a bigger one. So, Nicole leaned even further into that partnership Tyson and Urs had established. Every two weeks the product and technical teams got together with Google’s engineers to discuss the trade-offs and tactical issues that we were facing. It started to feel less like we were two separate companies and more like we were one team, working towards the same goal of getting Spotifiers the data they needed without breaking the bank.
👉 Spotify: A Product Story - Episode 6: Hardware is hard (Spotify) How to think long-term: When we’re facing an important product decision - or just envisioning what the next new feature might look like - we come back to that idea of: how would this experience work if it were perfect? And then, we figure out how to get from here to there. In 2011, we took a gamble on the switch to connected speakers. From a numbers point of view, it didn’t really make any sense yet - at the time, only a tiny slice of users even had connected speakers and they were quite expensive, but we put a substantial percentage of our engineering brainpower towards developing Spotify Connect. It wasn’t until seven years later that we finally saw our return, when voice-activated speakers hit the mainstream. If you have an ambitious vision for a future, you need to allow it to take significant time to materialize. You need some strong conviction in the beginning, and then you need enough management support and cultural support to be able to keep that investment alive for a long time. But looking at it like in financial terms here and now, it's not the kind of investment that makes sense. But I think that's true for a lot of the things that we've done over the years. Today - Spotify Connect is found in more than 2000 different devices from over 200 different brands.
➡️ I love how they were able to make a bet on the future, even if it is not big today. What are the similar trends you believe in? Avoid surveys: We could actually go sit with people on their couch and watch them play in a game and, you know, see how they interact with Spotify in that moment. So we rely on those insights and the user behavior actually in context, which is much better than having somebody fill out a survey or, you know, just asking questions over a hangout call.
➡️ Why I don’t usually believe in surveys to get really good insights. 👉 Spotify: A Product Story - Episode 7: Spotify’s podcast bet (Spotify) Spend time debate on ideas and problems: What we've learned is that a lot of these things can be debated away - like we could debate or wait to understand if something's not going to work. So let's not even do this test. Let's spend time upfront talking because it's pretty cheap to talk, getting the right experiences, the right perspectives and data from different angles on the problem so we that the tests that we do and the bets that we put, have a higher likelihood of success. So this is what we did here. As we then moved to the next phase of the podcast strategy, we tried to figure out why we're doing it. Why should we be doing podcasts? What's the purpose? What can we do for consumers? What can we do for creators? What can we do for advertisers? What can we do for ourselves? And what can others do? What can we uniquely bring to this marketplace?
This is the idea from science, kind of, that you can get surprisingly far by reasoning and people still underestimate how far you can get if you have a diverse set of smart people, you can get really far with reasoning. But this is not what a company normally does. A company tends to push out, divide and conquer responsibilities and budgets and P&Ls and so forth. So it's something surprisingly hard to do because you need to talk about other people's strategies and they need to get into your strategy. It's very messy, but surprisingly effective on a longer time scale.
Bundling & personalization: We saw the success of companies like WeChat in China, which had gone in the opposite direction and bundled everything together instead of creating a bunch of separate apps, and we had a hunch that should be our strategy: bring the podcasts to the people, not the people to the podcasts. Personalization is a subtle feature that by definition gets better the more you use it. It’s not the reason that people try an app for the first time, it’s the reason they stick with it for years.
How to think about conversion: The obvious way to monetize the exclusive content on our platform would be to follow the streaming TV model and simply put it behind a paywall. But remember - our whole goal was differentiation not monetization. And it makes no sense to hide your differentiation behind a paywall where no new user can hear it. Instead we wanted to maximize our differentiation for new users and give away a “too good to be true” listening experience - where anyone can listen to anything for free - as an investment in user acquisition. Somewhat counterintuitively, the more time people spent listening on Spotify for free, the more likely they were to convert to paid users down the line.
👉 Spotify’s Investor Day, Spotify’s Music Aggregation, Podcast Anecdata (Stratechery) Since 2018, the last year before our major podcast investment, our music margins have expanded on average by approximately 75 basis points per year. At our last investor day we told you to expect gross margins in the 30-35% range over the long-term. At the beginning of 2018, we announced the development of our marketplace business We’ve long maintained that our success is not solely tied to renegotiating new headline rates. It’s about our ability to innovate, right along with our partners, to grow a business that benefits both artists and Spotify, and that’s what we’ve done with Marketplace.
👉 Spotify, Netflix, and Aggregation (Stratechery) Spotify isn’t earning money by making margin on its content spend; rather, it is seeking to enable more content than ever, confident that it controls the best means to surface the content users want. Those means can then be sold to the highest bidder, with all of the margin going to Spotify. Spotify calls this promotion — it certainly looks a lot like the old radio model of pay-to-play — but that’s really just another word for advertising.
➡️ Spotify makes artists pay to be discovered, and I think that will be our position with third parties when we reach a big enough scale, so we can keep the price for members/customers as low as possible. Instead of introducing friction in the market, the better to lock-in users, Aggregators want to decrease friction, confident the gravitational pull of their user experience will, all things being equal, draw in more users than their competitors, increasing their attractiveness to not just suppliers but also advertisers (who, in the case of Spotify’s music business, may be the same entities).
➡️ Our aggregator play would be making 0 on insurance (our model of 100% LR), and finding other ways (that protect the data of members) to create value for third parties. Spotify has, for the most part, acted like an Aggregator: the company has fought exclusives in the music business, kept its subscription prices as low as possible, and in the case of podcasts ensured its Anchor platform supports all podcast players.
👉People: Let Opportunities Echo (Spotify) Newly launched Internal Talent Marketplace.
➡️ At what scale does it make sense to build something similar? Echo: So, in addition to new internal job opportunities, Echo has also introduced a Spotify-wide platform to highlight part-time projects and mentorship at scale. Anyone can be a project owner or a mentor now.
➡️ Software for our coaches, coaching structure. ➡️ Should we dig more into what they do with Echo? 👉 Learning & Development Under The Spotlight (Spotify) GreenHouse team to enable, empower, and learn faster than the world is changing. That we do not become bottlenecks in the development journeys of our workforce. We need to master ways of producing videos, creating animations and building learning opportunities in a fast and appetite teasing way. We are not suggesting leaving the L&D remit, we still need to keep our core competency.
➡️ Interesting article about learning and development at Spotify. 👉 Lyra Health gets $235M, soars to $5.58B valuation with new acquisition for global expansion [January 22] (Fierce Healthcare) ➡️ It is interesting that it is only 75 companies. Prior to the acquisition, Lyra covered more than 2.2 million members globally. Lyra offers evidence-based care across a range of mental health concerns, including more serious issues like alcohol use disorder and suicidality, with programs for those conditions announced last September and expected to roll out in early 2022.
👉 HTN Weekly Health Tech Reads 1/30 (Health Tech Nerds) Eucalyptus, an Australian company building a portfolio of D2C digital health brands, raised $60 million. It seems like a sign of the times that we’re starting to see more CPG-like "house of brands" plays emerge in the digital health space. Seems like we're going to see a lot more of this over the coming years. Link.
👉 Amazon Acquires One Medical, One Medical Allure, Reasons for Skepticism (Stratechery) Amazon Acquires One Medical: Amazon will gain access to a practice that operates more than 180 medical offices in 25 U.S. markets and works with more than 8,000 companies to provide health benefits to employees, including with in-person and virtual care. That adds significantly to a smaller service Amazon launched in 2019 and for which it had signed up a limited number of employers in the last year.
It’s already over! Please share JC’s Newsletter with your friends, and subscribe 👇 Let’s talk about this together on LinkedIn or on Twitter. Have a good week! | |
