Spotify
A music, podcast and audiobook streaming app with a freemium model and personalized recommendations.
Spotify shows up as a reference point more than a case study. Its time-based audiobook pricing (set hours, top-ups, single titles) is admired, and Wrapped went viral by turning listening into identity. Copies often fail. Freemium also built it at scale, yet one source's failed freemium test cut subscriber conversion by over 50%. Lesson: find your own shareable moment, and test freemium before assuming it fits.
- 1Users listen passivelySomething they already do without effort
- 2Package it as identityListening becomes something to show off
- 3Release in compressed windowBeautiful format, all at once
- 4Users share itIf people don't share it, it fails
From Podcast findings on Spotify Wrapped
What was said about Spotify
1Counter-positioning against Apple Music: freemium, personalization, ubiquity
Spotify survived Apple by betting on three things Apple would do badly. When Apple launched Apple Music and told its teams internally it would kill Spotify within six months, Spotify picked deliberate opposites: a good ad-supported free tier, deep data and personalization, and availability on every device.
The bets were that Apple would struggle with free, stay weak on personalization and favor its own hardware over Samsung TVs or Android phones. Apple later pulled back its free radio tier and still struggles on non-Apple hardware. Spotify was tiny at the time, so for it the fight was not optional.
New technology wins only when paired with a contrarian business model. Söderström says technology is necessary but not sufficient: change comes when it is married to a new business model, as with Spotify's access over ownership or Kindle bundling mobile data into the hardware price.
Spotify grew most in periods of change. In 2008 streaming over cellular took 20 seconds to start, and Spotify adapted with offline sync and a mobile business model. He sees AI as another such period, so his rule is to adopt change first: in stable periods market share stays put, in change it is won or lost.
Labels preferred Spotify's licensing, which let it overtake an earlier Pandora. Spotify arrived six or seven years after Pandora but with direct licensing the labels liked, while Pandora relied on a statutory license and assumed cloud streaming of all music would leave it alone. Pandora moved to direct deals too late and Spotify took the lead.
Jimmy Iovine notes Daniel Ek pried licenses from a record business that did not understand him, though the terms copied the iTunes 70/30 split. He criticizes streaming for giving artists no way to reach fans and says being early is only 10% of the game; Spotify had three million subscribers when Beats sold to Apple, and Evan Spiegel cites its growth to about 250 million paid subscribers.
2No Regrets: building for time well spent
A regret survey turned a slogan into Spotify's expansion strategy. An anonymous third-party survey asked users how they felt about time spent on Spotify, YouTube, Apple Music, Amazon and TikTok. Gen Z valued almost 90% of their time on Spotify, while on many big platforms young people regretted 60% or more of theirs.
Söderström had assumed heavy usage meant enjoyment. Spotify now expands only into categories people are glad to spend time on: music, podcasts, audiobooks and now fitness, since about 70% of users already work out with it, with AI-built running playlists matched to pace.
Subscription revenue lets Spotify choose user value over captured time. Almost 90% of revenue comes from subscribers, who pay monthly for the value they feel rather than for time spent, which frees Spotify from the push toward dark engagement patterns that ad models create.
Söderström says dark patterns could lift usage and retention today but would not be good for users. He calls the subscription model a luxury: 'most of our revenue come from subscription to not have that pressure.'
Spotify let anyone switch off video podcasts despite the engagement cost. As podcasts moved to video, parents and other listeners felt videos appeared where they did not want them, so Spotify let users turn video off and just listen, accepting lower engagement.
'If you're going to have these principles, you kind of have to live with them, and they're going to have consequences.'
3One super app: podcasts and audiobooks inside Spotify
Spotify built podcasts into its main app because distribution, not product, was missing. At least seven good podcast apps like Overcast existed, yet Apple Podcasts held about 98.5% of usage. Spotify chose the pain of one complex app to reach what was then 300-something million users, and repeated it with audiobooks.
A separate app would have let teams build in parallel. Spotify now holds over a third of global podcast listening, and more than 75% of new shows on it are hosted with Spotify for Podcasters.
A single-product strategy needs a centralized org, or you ship your org chart. When Spotify ran autonomous teams, several teams put their own controls on the now playing view, so it moved to a model closer to Apple's, with recommendation and UI each run by single organizations.
Söderström says parallel teams feel good for about six months, then the experience crumbles as complexity is pushed onto the user. Amazon-style autonomy ships fast but ships the org chart; Apple-style control gives a simple UX but is slower.
4How Spotify's leadership and teams are organized
One weekly meeting of all SVPs replaces 'let's take it offline'. Co-CEOs Söderström and Alex Norström hold a three-hour E Team meeting every Tuesday with about 14 SVPs from every function, so a feature blocked on licensing is solved with licensing across the table.
It costs 14 or 15 people three hours a week, much of it not relevant to each one, but gives everyone the full CEO view on licensing, ads and machine learning. It suits Spotify because licensors, contracts and product are so interdependent.
Autonomy sits at the VP level, not in small squads or at the top. Spotify moved from autonomous squads of about seven people to teams of roughly 14 per manager, and gives VPs freedom to set strategy for areas like podcasting or the experience layer.
Autonomy at the leaves produced '100 squads with 100 strategies'; autonomy at the top bottlenecks every decision.
No org model wins, so optimize for what matters and accept average elsewhere. Citing Steven Sinofsky, Söderström says Amazon, Apple and Elon Musk's companies are organized very differently and each became trillion-dollar companies. Spotify optimizes for a single experience even though the backend mixes royalty pools and ad models.
Switching models just swaps one weakness for another. The worst case is being excellent at the unimportant thing and weak at the important one.
Long leadership tenure keeps a functional org out of politics. Many of Söderström's reports have stayed 14 or 15 years, averaging 7 or 8, which builds the trust a functional org needs. He saw Yahoo's EVPs fight and hide information, with four CEOs in about two and a half years.
To offset tenure's downsides, Spotify runs a Rising Stars program of 20 to 30 people as a 12-month internal MBA, and hires or acquires new skills, as with The Echo Nest for machine learning near MIT.
Spend no more than about 10% of a cycle planning. Spotify plans for about two weeks per six-month cycle, the same increment Brian Chesky uses at Airbnb.
Its product phases are think it, build it, ship it, tweak it, with think it kept cheap to reduce the risk of being wrong before spending jumps. The labels are still used about 12 years later.
5The home redesign: recall versus discovery
Flipping home from recall to discovery sent users hunting in search. Spotify's home was about 90% recall and 10% discovery. A redesign flipped it to 10% recall and 90% discovery, and traffic shifted from home to search and library as users could no longer find their playlists.
Recall needs a dense UI with many items; discovery needs a low-density feed with big visuals where a miss costs a swipe. Using a discovery feed for recall feels 'like a slot machine'.
Match the interface to how often the recommendation is right. Background listening needs about nine hits in ten, while breaking out of a taste bubble works at one in ten, so Spotify built swipeable sub feeds for podcasts, playlists and new genres where users go for new music and save songs.
A single big play button only works near 100% accuracy; at one hit in five, show five options plus an easy exit. Midjourney showed four fast low-res images because roughly one in four was good. The fault-tolerant interface idea comes from Chris Dixon.
Use new user cohorts to tell change-aversion from a wrong design. Redesigns are not voluntary, so anger can mean broken habits or a real mistake. Compare new cohorts, who have no habits, against old ones and read shifts like traffic moving between tabs.
Söderström warns teams before launch that a redesign will be painful.
Believe the hypothesis fully, then change execution when data says no. Spotify kept the goal of helping users break out of their taste bubble but tested several home variants, and now tests a home with recall intact and discovery tools available but voluntary.
The MVP for a redesign has to be big: a new UI without the algorithms for a single-item feed cannot tell you whether the idea or the machine learning failed. That internal cost was the main pain of the home redesign. 'You have to believe in things 100% until the data says no.'
6The AI DJ and Spotify's generative AI bets
The AI DJ solved 'I don't know what I want' by doing very little. Spotify's only truly generative product so far pairs a digitized real voice with AI-written commentary, for the case radio once handled and Spotify had failed at for years. It skips weather and news because people came for the music.
It was built bottom-up by the personalization team, with support from big partners. In testing, users wondered how a person could have recorded so many lines about their music; Söderström says fine-tuning and scoping down is what crosses from impressive to magic, and that feeling correlates with going viral.
Treat generation as a new era, not better recommendations. Söderström compares generative AI to the earlier shift from curation to recommendation, which forced rethinks of UX and sometimes the business model, and asks teams for new product shapes.
Language models already improve existing features with bigger vectors, more cultural knowledge and safety classification of unheard podcasts. On AI music, he calls models a more powerful instrument and says artists benefit only once a new business model pays rights holders.
Spotify invested before LLMs, buying Sonantic for cheap voice generation. Söderström read the 2017 Transformer paper early and began investing. Buying voice at market prices for 700 million users listening minutes a day would have bankrupted Spotify, so it bought Sonantic.
'We just bet on intercepting that curve instead of waiting for it.'
AI gives users control of the algorithm instead of making it addictive. Spotify shows users who its models think they are and lets them say 'that's what I do, but not what I want to be', such as an EDM listener asking for classical.
Only about 1% create and 9% curate, but their roughly 10 billion playlists improve recommendations for everyone; he expects users who correct the AI to help similar users who never will. He likens it to user research with all 761 million users.
7Discover Weekly, Wrapped and designing for delight
Discover Weekly works better with some familiar songs mixed in. Discover Weekly was meant to be all new tracks, but a bug inserted songs users already knew. When engineers fixed it, every success metric dropped, so the familiar songs were restored.
Users want a little surprise, not total novelty.
Deep delight comes from meeting a functional and emotional need together. Nesrine Changuel classes Discover Weekly and Jam as deep delight (finding music while feeling seen) and Wrapped as surface delight, and suggests a roadmap of 50% functional, 40% deep delight, 10% surface delight.
Spotify's roadmap mixed search improvements with video podcasts and Canvas looping clips. She segments users by motivation: some search for a track, others want inspiration or to feel less lonely. Spotify sustains this with quarterly squad health checks, monthly hack days and a yearly hack week.
Wrapped spread because it turned listening into an identity worth showing. Wrapped converts passive consumption into stats that make users look good, in a beautiful format with built-in sharing, released in a compressed window.
Copies often fail because the shareable moment has to come from a product's own value; a coffee shop's Wrapped-style recap showed the pattern had jumped the shark. Look for milestones users are proud of, like Zero's weekly fasting digest with click-to-tweet.
8Daniel Ek's changing role and the CEO handoff
Ek handed product to Gustav after being told his reviews weren't helping. Gustav told Ek the team was mostly appeasing him in product reviews. Ek's first instinct was to fire him, but he tried three months without reviews; the team was happier and decided better, so he never took them back.
Ek separately recalls a lieutenant telling him 'you really suck at this' and, once his anger cooled, seeing it was true. He says trust grows about 1% per good interaction and one bad one can ruin it.
Ek keeps moving to wherever he adds the most value. After product, Ek moved to understanding creators; when Alex did that better, he moved to the space between creators and consumers. He now mostly shows up when people need him and protects the first seed of new ideas.
He says companies oscillate between zero-to-one, one-to-100 and optimization, and big companies are good at minimizing variance, which is not how the best ideas arise. He used to be 90% internally focused on running the Spotify machine.
Successors ran the business for three years before taking the title. Ek made Söderström and Norström co-presidents running day-to-day Spotify, with the full P&L and balance sheet, while he stayed CEO and gave them more rope.
Only PR, government and being the public face were new when they became co-CEOs. Ek delegated from the start, sending a newly hired Söderström to set up a Facebook integration with Zuckerberg.
Ek weighed acquisition offers only on whether they'd advance the mission. Already having the money he needed, Ek considered offers 'but not for money', asking whether a buyer cared as much as he did. He never found one.
Tobi Lütke cites Ek as a founder who hires for spikes rather than credentials.
9Anchor inside Spotify
Shielding Anchor for a year kept speed but weakened its ties. Ek told Anchor to say no to meetings and keep being Anchor for a year. It moved fast, but a year later relationships were weak, so it dropped the Anchor name internally and collaborated more, accepting some slowdown.
Anchor stayed relevant through dozens of strategy shifts in almost five years by marketing itself internally and constantly asking how it could help other teams.
Anchor kept its 'move fast' value and challenged needless slowdowns. Anchor shipped a feature every two weeks at acquisition and never built anything over three months, while Spotify's 'small' estimate was about a quarter. Ek told them to keep moving fast and teach Spotify to do the same.
Maya Prohovnik also warns acquired founders to expect a post-acquisition dip, a pattern she heard from many founders acquired by Spotify and others.
10What other apps take from Spotify's freemium model
Freemium suits Spotify's huge funnel, but most apps should start with hard paywalls. Phil Carter says freemium fits companies aiming for a very large top of funnel like Spotify, Duolingo and Strava; one client's freemium shift cut subscriber conversion by more than 50% and was pulled after a couple of weeks.
He says hard paywalls convert 5x better. Others point to the cheaper ad-supported tier with locked features as a way to win people who won't pay full price, and to time-based audiobook pricing with listening hours and top-ups.
Spotify's monthly retention is a benchmark few apps reach. Spotify retains about 96 to 97% monthly, which is why it can rely on monthly plans; most apps should lead with yearly plans.
Opal uses Spotify as its retention yardstick, calling retention the only real moat.
Large user bases don't convert to revenue everywhere. Kunal Shah says Spotify, Netflix and Amazon Prime have not converted Indian users into payers as hoped, because ARPU tracks per capita income and paid products compete with endless free content.
Fan monetization sends money straight to artists through merch and access. Spotify's fan monetization team is making merch more prominent, with rewards like exclusive discounts, exclusive designs or early access for top listeners, and ran listening parties last year.
Merch is the most mature offering; Camille Hearst wants to extend listening parties to podcasts.
11Smaller lessons
A leadership podcast became a recruiting and culture tool. Söderström started an employee-gated podcast interviewing Ek's direct reports, then released it publicly as Spotify: A Product Story with no special promotion. Many tens of new hires say it was a key reason they joined.
It explains why decisions were made and the constraints at the time. He also runs one-on-ones as walk-and-talks over AirPods, which he found improved strategizing during the pandemic.
Test new channels only where they overlap with your goals. Adam Grenier calls Clubhouse a green light for Spotify because both are audio and its rooms serve music discovery, while a photo app like Paparazzi would be a yellow light at best.
Ami Vora also cites Spotify as growing faster after cutting products.
The Company Apple Couldn't Kill | Spotify Co-CEO Gustav Söderström
Almost 90% of revenue comes from subscribers, which lets it favor user value over engagement. Its No Regrets strategy came from a survey where Gen Z valued almost 90% of their time on Spotify, and it let users switch video podcasts off despite the engagement cost.
Freemium suits Spotify's huge funnel, but most apps should start with hard paywalls. Phil Carter says freemium fits companies aiming for a very large top of funnel like Spotify, Duolingo and Strava; one client's freemium shift cut subscriber conversion by more than 50% and was pulled after a couple of weeks.
He says hard paywalls convert 5x better. Others point to the cheaper ad-supported tier with locked features as a way to win people who won't pay full price, and to time-based audiobook pricing with listening hours and top-ups.
Daniel Ek, Spotify | David Senra
Ek's company, now in its 19th to 20th year, where the first ten years were zero-to-one and the last ten were not. He handed product to Gustav, moved to understanding creators, then to the space between creators and consumers, and turned down early acquisition offers because no buyer cared as much about the mission.
Spotify survived Apple by betting on three things Apple would do badly. When Apple launched Apple Music and told its teams internally it would kill Spotify within six months, Spotify picked deliberate opposites: a good ad-supported free tier, deep data and personalization, and availability on every device.
The bets were that Apple would struggle with free, stay weak on personalization and favor its own hardware over Samsung TVs or Android phones. Apple later pulled back its free radio tier and still struggles on non-Apple hardware. Spotify was tiny at the time, so for it the fight was not optional.
New technology wins only when paired with a contrarian business model. Söderström says technology is necessary but not sufficient: change comes when it is married to a new business model, as with Spotify's access over ownership or Kindle bundling mobile data into the hardware price.
Spotify grew most in periods of change. In 2008 streaming over cellular took 20 seconds to start, and Spotify adapted with offline sync and a mobile business model. He sees AI as another such period, so his rule is to adopt change first: in stable periods market share stays put, in change it is won or lost.
A regret survey turned a slogan into Spotify's expansion strategy. An anonymous third-party survey asked users how they felt about time spent on Spotify, YouTube, Apple Music, Amazon and TikTok. Gen Z valued almost 90% of their time on Spotify, while on many big platforms young people regretted 60% or more of theirs.
Söderström had assumed heavy usage meant enjoyment. Spotify now expands only into categories people are glad to spend time on: music, podcasts, audiobooks and now fitness, since about 70% of users already work out with it, with AI-built running playlists matched to pace.
Subscription revenue lets Spotify choose user value over captured time. Almost 90% of revenue comes from subscribers, who pay monthly for the value they feel rather than for time spent, which frees Spotify from the push toward dark engagement patterns that ad models create.
Söderström says dark patterns could lift usage and retention today but would not be good for users. He calls the subscription model a luxury: 'most of our revenue come from subscription to not have that pressure.'
Spotify let anyone switch off video podcasts despite the engagement cost. As podcasts moved to video, parents and other listeners felt videos appeared where they did not want them, so Spotify let users turn video off and just listen, accepting lower engagement.
'If you're going to have these principles, you kind of have to live with them, and they're going to have consequences.'
Spotify built podcasts into its main app because distribution, not product, was missing. At least seven good podcast apps like Overcast existed, yet Apple Podcasts held about 98.5% of usage. Spotify chose the pain of one complex app to reach what was then 300-something million users, and repeated it with audiobooks.
A separate app would have let teams build in parallel. Spotify now holds over a third of global podcast listening, and more than 75% of new shows on it are hosted with Spotify for Podcasters.
A single-product strategy needs a centralized org, or you ship your org chart. When Spotify ran autonomous teams, several teams put their own controls on the now playing view, so it moved to a model closer to Apple's, with recommendation and UI each run by single organizations.
Söderström says parallel teams feel good for about six months, then the experience crumbles as complexity is pushed onto the user. Amazon-style autonomy ships fast but ships the org chart; Apple-style control gives a simple UX but is slower.
One weekly meeting of all SVPs replaces 'let's take it offline'. Co-CEOs Söderström and Alex Norström hold a three-hour E Team meeting every Tuesday with about 14 SVPs from every function, so a feature blocked on licensing is solved with licensing across the table.
It costs 14 or 15 people three hours a week, much of it not relevant to each one, but gives everyone the full CEO view on licensing, ads and machine learning. It suits Spotify because licensors, contracts and product are so interdependent.
No org model wins, so optimize for what matters and accept average elsewhere. Citing Steven Sinofsky, Söderström says Amazon, Apple and Elon Musk's companies are organized very differently and each became trillion-dollar companies. Spotify optimizes for a single experience even though the backend mixes royalty pools and ad models.
Switching models just swaps one weakness for another. The worst case is being excellent at the unimportant thing and weak at the important one.
Long leadership tenure keeps a functional org out of politics. Many of Söderström's reports have stayed 14 or 15 years, averaging 7 or 8, which builds the trust a functional org needs. He saw Yahoo's EVPs fight and hide information, with four CEOs in about two and a half years.
To offset tenure's downsides, Spotify runs a Rising Stars program of 20 to 30 people as a 12-month internal MBA, and hires or acquires new skills, as with The Echo Nest for machine learning near MIT.
Spotify invested before LLMs, buying Sonantic for cheap voice generation. Söderström read the 2017 Transformer paper early and began investing. Buying voice at market prices for 700 million users listening minutes a day would have bankrupted Spotify, so it bought Sonantic.
'We just bet on intercepting that curve instead of waiting for it.'
AI gives users control of the algorithm instead of making it addictive. Spotify shows users who its models think they are and lets them say 'that's what I do, but not what I want to be', such as an EDM listener asking for classical.
Only about 1% create and 9% curate, but their roughly 10 billion playlists improve recommendations for everyone; he expects users who correct the AI to help similar users who never will. He likens it to user research with all 761 million users.
Successors ran the business for three years before taking the title. Ek made Söderström and Norström co-presidents running day-to-day Spotify, with the full P&L and balance sheet, while he stayed CEO and gave them more rope.
Only PR, government and being the public face were new when they became co-CEOs. Ek delegated from the start, sending a newly hired Söderström to set up a Facebook integration with Zuckerberg.
Monetizing passions, scaling marketplaces, and stories from a creator economy vet | Camille Hearst
Camille Hearst heads fan monetization there, connecting artists and fans so money goes directly to artists. Merch is the most mature offering, with rewards like exclusive discounts, designs and early access for top listeners, and listening parties were run last year.
Freemium suits Spotify's huge funnel, but most apps should start with hard paywalls. Phil Carter says freemium fits companies aiming for a very large top of funnel like Spotify, Duolingo and Strava; one client's freemium shift cut subscriber conversion by more than 50% and was pulled after a couple of weeks.
He says hard paywalls convert 5x better. Others point to the cheaper ad-supported tier with locked features as a way to win people who won't pay full price, and to time-based audiobook pricing with listening hours and top-ups.
The science of product, big bets, and how AI is impacting the future of music | Gustav Söderström
Moved away from 7-person squads to teams of about 14 per manager, and puts autonomy at the VP level with a centralized model. Its home redesign flipped from 90% recall and 10% discovery to the reverse, traffic shifted to search and library, and a new test keeps discovery feeds optional. It spends about 2 weeks planning per 6-month cycle. The only truly generative product at Spotify so far, built bottom-up by the personalization team. It solves the 'I don't know what I want' case, does as little as possible (no weather or news), and users wondered how a person could record so many lines about their music.
Spotify's monthly retention is a benchmark few apps reach. Spotify retains about 96 to 97% monthly, which is why it can rely on monthly plans; most apps should lead with yearly plans.
Opal uses Spotify as its retention yardstick, calling retention the only real moat.
How Tolan 3X'd Subscriber Conversion after Shifting to Freemium
Mehta admires its time-based audiobook pricing: subscribers get set listening hours, can buy top-ups, and can dip into single titles. He recalls roughly 30 to 40% of users being paid but isn't sure of the figure.
Freemium suits Spotify's huge funnel, but most apps should start with hard paywalls. Phil Carter says freemium fits companies aiming for a very large top of funnel like Spotify, Duolingo and Strava; one client's freemium shift cut subscriber conversion by more than 50% and was pulled after a couple of weeks.
He says hard paywalls convert 5x better. Others point to the cheaper ad-supported tier with locked features as a way to win people who won't pay full price, and to time-based audiobook pricing with listening hours and top-ups.
7 Ad Monetization Hacks Top Apps Use Right Now
Free users hear video ads and some features stay locked, a standard content-app model where people accept ads for a lower price.
Labels preferred Spotify's licensing, which let it overtake an earlier Pandora. Spotify arrived six or seven years after Pandora but with direct licensing the labels liked, while Pandora relied on a statutory license and assumed cloud streaming of all music would leave it alone. Pandora moved to direct deals too late and Spotify took the lead.
Jimmy Iovine notes Daniel Ek pried licenses from a record business that did not understand him, though the terms copied the iTunes 70/30 split. He criticizes streaming for giving artists no way to reach fans and says being early is only 10% of the game; Spotify had three million subscribers when Beats sold to Apple, and Evan Spiegel cites its growth to about 250 million paid subscribers.
Why Opal Stopped Chasing Revenue to Build a Billion-User App – Kenneth Schlenker
Opal's early retention is said to match or beat Spotify's.
The AI Growth Playbook for Subscription Apps – Phil Carter, Elemental Growth
Built a huge company on freemium, which fills the top of the funnel with far more users.
Wrapped spread because it turned listening into an identity worth showing. Wrapped converts passive consumption into stats that make users look good, in a beautiful format with built-in sharing, released in a compressed window.
Copies often fail because the shareable moment has to come from a product's own value; a coffee shop's Wrapped-style recap showed the pattern had jumped the shark. Look for milestones users are proud of, like Zero's weekly fasting digest with click-to-tweet.
Evan Spiegel, Snapchat: Building a Multi-Billion Dollar Company
Spiegel cites Spotify as the subscription pattern, growing from three million paid subscribers to about 250 million per Jimmy Iovine.
Labels preferred Spotify's licensing, which let it overtake an earlier Pandora. Spotify arrived six or seven years after Pandora but with direct licensing the labels liked, while Pandora relied on a statutory license and assumed cloud streaming of all music would leave it alone. Pandora moved to direct deals too late and Spotify took the lead.
Jimmy Iovine notes Daniel Ek pried licenses from a record business that did not understand him, though the terms copied the iTunes 70/30 split. He criticizes streaming for giving artists no way to reach fans and says being early is only 10% of the game; Spotify had three million subscribers when Beats sold to Apple, and Evan Spiegel cites its growth to about 250 million paid subscribers.
Head of Claude Code: What happens after coding is solved | Boris Cherny
Reported that its best developers have not written a line of code since December.
Ek weighed acquisition offers only on whether they'd advance the mission. Already having the money he needed, Ek considered offers 'but not for money', asking whether a buyer cared as much as he did. He never found one.
Tobi Lütke cites Ek as a founder who hires for spikes rather than credentials.
How ElevenLabs Ships So Fast With Small Teams
Wrapped went viral because it turned passive listening into an identity people wanted to show, in a beautiful format arriving in a compressed window. Copies often fail because the shareable moment must come from a product's own value.
Ek handed product to Gustav after being told his reviews weren't helping. Gustav told Ek the team was mostly appeasing him in product reviews. Ek's first instinct was to fire him, but he tried three months without reviews; the team was happier and decided better, so he never took them back.
Ek separately recalls a lieutenant telling him 'you really suck at this' and, once his anger cooled, seeing it was true. He says trust grows about 1% per good interaction and one bad one can ruin it.
Jimmy Iovine: Building Interscope Records & Beats by Dre
Daniel Ek pried licenses from a record business that did not understand him, and the terms copied the iTunes 70/30 split. Spotify is now worth about $150 billion, roughly double the whole record industry.
A leadership podcast became a recruiting and culture tool. Söderström started an employee-gated podcast interviewing Ek's direct reports, then released it publicly as Spotify: A Product Story with no special promotion. Many tens of new hires say it was a key reason they joined.
It explains why decisions were made and the constraints at the time. He also runs one-on-ones as walk-and-talks over AirPods, which he found improved strategizing during the pandemic.
Tobi Lütke: 21 Years of Building Shopify
Lütke cites its founder as someone who, like Ramp, hires for spikes instead of credentials.
Discover Weekly works better with some familiar songs mixed in. Discover Weekly was meant to be all new tracks, but a bug inserted songs users already knew. When engineers fixed it, every success metric dropped, so the familiar songs were restored.
Users want a little surprise, not total novelty.
Deep delight comes from meeting a functional and emotional need together. Nesrine Changuel classes Discover Weekly and Jam as deep delight (finding music while feeling seen) and Wrapped as surface delight, and suggests a roadmap of 50% functional, 40% deep delight, 10% surface delight.
Spotify's roadmap mixed search improvements with video podcasts and Canvas looping clips. She segments users by motivation: some search for a track, others want inspiration or to feel less lonely. Spotify sustains this with quarterly squad health checks, monthly hack days and a yearly hack week.
Wrapped spread because it turned listening into an identity worth showing. Wrapped converts passive consumption into stats that make users look good, in a beautiful format with built-in sharing, released in a compressed window.
Copies often fail because the shareable moment has to come from a product's own value; a coffee shop's Wrapped-style recap showed the pattern had jumped the shark. Look for milestones users are proud of, like Zero's weekly fasting digest with click-to-tweet.
How Jensen Works
Founder Daniel Ek was once told by a lieutenant 'you really suck at this' and, after his anger cooled, saw it was true.
Ek handed product to Gustav after being told his reviews weren't helping. Gustav told Ek the team was mostly appeasing him in product reviews. Ek's first instinct was to fire him, but he tried three months without reviews; the team was happier and decided better, so he never took them back.
Ek separately recalls a lieutenant telling him 'you really suck at this' and, once his anger cooled, seeing it was true. He says trust grows about 1% per good interaction and one bad one can ruin it.
Ek keeps moving to wherever he adds the most value. After product, Ek moved to understanding creators; when Alex did that better, he moved to the space between creators and consumers. He now mostly shows up when people need him and protects the first seed of new ideas.
He says companies oscillate between zero-to-one, one-to-100 and optimization, and big companies are good at minimizing variance, which is not how the best ideas arise. He used to be 90% internally focused on running the Spotify machine.
Ek weighed acquisition offers only on whether they'd advance the mission. Already having the money he needed, Ek considered offers 'but not for money', asking whether a buyer cared as much as he did. He never found one.
Tobi Lütke cites Ek as a founder who hires for spikes rather than credentials.
Michael Dell, Dell Technologies | David Senra
It made a private podcast series on its own history, later public as Spotify: A Product Story. It explains why decisions were made and the constraints at the time.
Test new channels only where they overlap with your goals. Adam Grenier calls Clubhouse a green light for Spotify because both are audio and its rooms serve music discovery, while a photo app like Paparazzi would be a yellow light at best.
Ami Vora also cites Spotify as growing faster after cutting products.
A 4-step framework for building delightful products | Nesrine Changuel (Spotify, Google, Skype)
Wrapped is pure surface delight and Discover Weekly and Jam are deep delight. A bug that inserted songs users already knew made Discover Weekly better, and fixing it dropped every success metric, so the familiar songs were restored. Its roadmap mixes search improvements with video podcasts and Canvas, and it runs quarterly squad health checks, monthly hack days and a yearly hack week.
Wrapped spread because it turned listening into an identity worth showing. Wrapped converts passive consumption into stats that make users look good, in a beautiful format with built-in sharing, released in a compressed window.
Copies often fail because the shareable moment has to come from a product's own value; a coffee shop's Wrapped-style recap showed the pattern had jumped the shark. Look for milestones users are proud of, like Zero's weekly fasting digest with click-to-tweet.
Simplify To Win | Lenny & Friends Summit 2024
Spotify grew faster after cutting products, cited as an example of simplicity paying off.
Large user bases don't convert to revenue everywhere. Kunal Shah says Spotify, Netflix and Amazon Prime have not converted Indian users into payers as hoped, because ARPU tracks per capita income and paid products compete with endless free content.
How to Transform ANY Business into a Profitable APP Business #appbusiness #appdevelopers
Its year-end review turns user data into a shareable moment with built-in social sharing and virality, and many apps now try to copy it.
Labels preferred Spotify's licensing, which let it overtake an earlier Pandora. Spotify arrived six or seven years after Pandora but with direct licensing the labels liked, while Pandora relied on a statutory license and assumed cloud streaming of all music would leave it alone. Pandora moved to direct deals too late and Spotify took the lead.
Jimmy Iovine notes Daniel Ek pried licenses from a record business that did not understand him, though the terms copied the iTunes 70/30 split. He criticizes streaming for giving artists no way to reach fans and says being early is only 10% of the game; Spotify had three million subscribers when Beats sold to Apple, and Evan Spiegel cites its growth to about 250 million paid subscribers.
Kunal Shah on winning in India, second-order thinking, the philosophy of startups, and more
Has not converted Indian users into payers, since paid products compete with endless free content.
Spotify built podcasts into its main app because distribution, not product, was missing. At least seven good podcast apps like Overcast existed, yet Apple Podcasts held about 98.5% of usage. Spotify chose the pain of one complex app to reach what was then 300-something million users, and repeated it with audiobooks.
A separate app would have let teams build in parallel. Spotify now holds over a third of global podcast listening, and more than 75% of new shows on it are hosted with Spotify for Podcasters.
Shielding Anchor for a year kept speed but weakened its ties. Ek told Anchor to say no to meetings and keep being Anchor for a year. It moved fast, but a year later relationships were weak, so it dropped the Anchor name internally and collaborated more, accepting some slowdown.
Anchor stayed relevant through dozens of strategy shifts in almost five years by marketing itself internally and constantly asking how it could help other teams.
Anchor kept its 'move fast' value and challenged needless slowdowns. Anchor shipped a feature every two weeks at acquisition and never built anything over three months, while Spotify's 'small' estimate was about a quarter. Ek told them to keep moving fast and teach Spotify to do the same.
Maya Prohovnik also warns acquired founders to expect a post-acquisition dip, a pattern she heard from many founders acquired by Spotify and others.
Billion dollar failures, and billion dollar success | Tom Conrad (Quibi, Pandora, Pets.com, Zero)
Arrived six or seven years after Pandora to far more investor optimism, and labels preferred its licensing structure. It took the lead while Pandora moved to direct deals too late.
Fan monetization sends money straight to artists through merch and access. Spotify's fan monetization team is making merch more prominent, with rewards like exclusive discounts, exclusive designs or early access for top listeners, and ran listening parties last year.
Merch is the most mature offering; Camille Hearst wants to extend listening parties to podcasts.
Building Anchor, selling to Spotify, and lessons learned | Maya Prohovnik (Head of Podcast Product)
Spotify holds over a third of global podcast listening, and more than 75% of new shows on it are hosted with Spotify for Podcasters. It moves fast on strategy, though its 'small' roadmap estimate is about a quarter, versus Anchor never building anything over three months.
New technology wins only when paired with a contrarian business model. Söderström says technology is necessary but not sufficient: change comes when it is married to a new business model, as with Spotify's access over ownership or Kindle bundling mobile data into the hardware price.
Spotify grew most in periods of change. In 2008 streaming over cellular took 20 seconds to start, and Spotify adapted with offline sync and a mobile business model. He sees AI as another such period, so his rule is to adopt change first: in stable periods market share stays put, in change it is won or lost.
A single-product strategy needs a centralized org, or you ship your org chart. When Spotify ran autonomous teams, several teams put their own controls on the now playing view, so it moved to a model closer to Apple's, with recommendation and UI each run by single organizations.
Söderström says parallel teams feel good for about six months, then the experience crumbles as complexity is pushed onto the user. Amazon-style autonomy ships fast but ships the org chart; Apple-style control gives a simple UX but is slower.
Autonomy sits at the VP level, not in small squads or at the top. Spotify moved from autonomous squads of about seven people to teams of roughly 14 per manager, and gives VPs freedom to set strategy for areas like podcasting or the experience layer.
Autonomy at the leaves produced '100 squads with 100 strategies'; autonomy at the top bottlenecks every decision.
Spend no more than about 10% of a cycle planning. Spotify plans for about two weeks per six-month cycle, the same increment Brian Chesky uses at Airbnb.
Its product phases are think it, build it, ship it, tweak it, with think it kept cheap to reduce the risk of being wrong before spending jumps. The labels are still used about 12 years later.
Flipping home from recall to discovery sent users hunting in search. Spotify's home was about 90% recall and 10% discovery. A redesign flipped it to 10% recall and 90% discovery, and traffic shifted from home to search and library as users could no longer find their playlists.
Recall needs a dense UI with many items; discovery needs a low-density feed with big visuals where a miss costs a swipe. Using a discovery feed for recall feels 'like a slot machine'.
Match the interface to how often the recommendation is right. Background listening needs about nine hits in ten, while breaking out of a taste bubble works at one in ten, so Spotify built swipeable sub feeds for podcasts, playlists and new genres where users go for new music and save songs.
A single big play button only works near 100% accuracy; at one hit in five, show five options plus an easy exit. Midjourney showed four fast low-res images because roughly one in four was good. The fault-tolerant interface idea comes from Chris Dixon.
Use new user cohorts to tell change-aversion from a wrong design. Redesigns are not voluntary, so anger can mean broken habits or a real mistake. Compare new cohorts, who have no habits, against old ones and read shifts like traffic moving between tabs.
Söderström warns teams before launch that a redesign will be painful.
Believe the hypothesis fully, then change execution when data says no. Spotify kept the goal of helping users break out of their taste bubble but tested several home variants, and now tests a home with recall intact and discovery tools available but voluntary.
The MVP for a redesign has to be big: a new UI without the algorithms for a single-item feed cannot tell you whether the idea or the machine learning failed. That internal cost was the main pain of the home redesign. 'You have to believe in things 100% until the data says no.'
The AI DJ solved 'I don't know what I want' by doing very little. Spotify's only truly generative product so far pairs a digitized real voice with AI-written commentary, for the case radio once handled and Spotify had failed at for years. It skips weather and news because people came for the music.
It was built bottom-up by the personalization team, with support from big partners. In testing, users wondered how a person could have recorded so many lines about their music; Söderström says fine-tuning and scoping down is what crosses from impressive to magic, and that feeling correlates with going viral.
Treat generation as a new era, not better recommendations. Söderström compares generative AI to the earlier shift from curation to recommendation, which forced rethinks of UX and sometimes the business model, and asks teams for new product shapes.
Language models already improve existing features with bigger vectors, more cultural knowledge and safety classification of unheard podcasts. On AI music, he calls models a more powerful instrument and says artists benefit only once a new business model pays rights holders.
A leadership podcast became a recruiting and culture tool. Söderström started an employee-gated podcast interviewing Ek's direct reports, then released it publicly as Spotify: A Product Story with no special promotion. Many tens of new hires say it was a key reason they joined.
It explains why decisions were made and the constraints at the time. He also runs one-on-ones as walk-and-talks over AirPods, which he found improved strategizing during the pandemic.
When to invest in new acquisition channels | Adam Grenier (Uber, MasterClass)
Its goals of new listeners and deeper engagement overlap with Clubhouse's audio and discovery strengths, making Clubhouse a green light. A photo app like Paparazzi would be a yellow light at best.
Test new channels only where they overlap with your goals. Adam Grenier calls Clubhouse a green light for Spotify because both are audio and its rooms serve music discovery, while a photo app like Paparazzi would be a yellow light at best.
Ami Vora also cites Spotify as growing faster after cutting products.
The nature of product | Marty Cagan, Silicon Valley Product Group
Cagan says Spotify works like the best product companies and has held its own against Apple and Amazon for years.
Product Market Fit: The Simple Way to Find It
Cited as an example of fit because it meets clear needs: discovering new music, sorting it into playlists, sharing it and listening offline.
How to Build a UA strategy for High-LTV Subscribers
Retains about 96 to 97% monthly, a benchmark few apps reach, so most should lead with yearly plans.
Spotify's monthly retention is a benchmark few apps reach. Spotify retains about 96 to 97% monthly, which is why it can rely on monthly plans; most apps should lead with yearly plans.
Opal uses Spotify as its retention yardstick, calling retention the only real moat.
App Store Screenshots: How to Write Killer Sales Copy to Increase Downloads
A Sleep Cycle reviewer asked for Spotify integration, so a competitor could lead a screenshot with 'integrated with Spotify' or 'Apple Music and Spotify'.
More in Music & Audio
Music & AudioSuno
An AI music platform that turns text prompts into full songs with lyrics, vocals and instrumentation.
8 sources · 24 tips
Music & AudioAnchor
A podcast creation and hosting app, later renamed Spotify for Podcasters.
1 source · 10 tips
Music & AudioBeats
A headphone and audio brand co-founded by Jimmy Iovine and Dr. Dre.
2 sources · 6 tipsBeats Music
A streaming service built from the MOG purchase that became Apple Music.
1 source · 2 tipsPhonograph
Edison's device for recording and replaying the human voice.
1 source · 2 tipsFirst Roots Media
A startup letting independent musicians sell or share music and track how far fans spread it.
1 source · 1 tipsimeem
A music discovery and social media startup founded by Dalton Caldwell.
1 source · 1 tipsRain Sounds
A sleep and ambient sounds app made by the guest.
1 source · 1 tips