Gustav Söderström, Spotify

Gustav Söderström's Preparation for Co-CEO Role and Daniel Ek's Leadership Style

Gustav Söderström spent three years as co-president alongside Alex Norstrom, running the daily operations of Spotify. This extensive period allowed them to manage the full P&L and balance sheet, providing thorough preparation for eventually stepping into the co-CEO role and handling the business side.

Daniel Ek's leadership approach is highly unusual and delegating. He trusts his team, even new hires, with significant responsibility, as exemplified by Gustav being tasked with a deep integration with Facebook early in his tenure. This constant delegation creates a dynamic environment where employees get a 'new job almost every year' through evolving responsibilities.

Ek is described as a non-typical 'alpha male founder,' which Gustav believes is key to his success. His style, characterized by being non-threatening, humble, and tenacious, is particularly effective in navigating the dual nature of Spotify as both a technology company and a media company dealing with powerful personalities from labels and publishers.

This unique leadership allows Ek to rally a team to build world-class technology while also uniting diverse, strong personalities in the media business to work towards common goals, even when they might individually prefer not to. His ability to foster collaboration without force is a central tenet of Spotify's organizational success.

I've had both great bosses and not so great bosses. Daniel is the, is the great one, and, he's incredibly, delegating, you know, already from the start.

Spotify's Synchronized Operating Model for Unified Product Experience

Spotify's Gustav Söderström and Alex Norstrom adopted a 'synchronized swimming' operating model to counter Daniel Ek's star-pattern communication style, which often led to disconnects between departments. They recognized that while divide and conquer is efficient, it risks making the company unsynchronized.

Their solution is a weekly, three-hour 'E-team' meeting involving 14-15 SVPs from all internal functions, including product, technology, business, content, marketing, and ads. The goal is to resolve blocking issues in real-time and eliminate the phrase 'let's take it offline' from their vocabulary by having all relevant decision-makers present.

This intensive cross-functional meeting ensures that leaders gain a holistic 'CEO perspective' on everything from machine learning problems to P&Ls and balance sheets. Though conversations may not always seem immediately relevant to every participant, the shared context informs future decisions, such as a personalization expert understanding ad monetization strategy.

Spotify's complex product, aiming to be a 'super app' with music, podcasts, and books, has many dependencies. This synchronized model is an investment that prevents the 'org chart from showing up in the product,' ensuring a unified user experience rather than externalizing internal complexity to the user.

Our biggest risk is to ship the org chart.

Long leadership tenure enables functional organizations to succeed by fostering trust

Gustav Söderström discusses various organizational structures, including functional, matrix, and division-based models, and notes that while a functional organization, like Apple's, can make inter-departmental cooperation challenging, it excels at building complex products that feel seamlessly integrated.

He recounts his experience at Yahoo, where frequent leadership changes led to infighting among executives. Söderström observed that Apple's functional structure, despite theoretical challenges, avoids this pitfall. He realized that Apple's success comes from the exceptionally long tenure of its senior leadership.

This extended tenure fosters deep trust and synchronization among functional leads, allowing different departments, such as hardware or software, to take the lead when appropriate. Without such long-term trust and stability in leadership, functional organizations often devolve into internal politics.

Söderström has applied this principle at Spotify, where many of his direct reports have worked for him for many years, with an average tenure of seven to eight years, which he believes makes their functional structure effective.

Without the tenure is the key.

There is No Single Right Organizational Structure

Gustav Söderström explains that no single organizational structure is inherently perfect; every choice involves trade-offs. He contrasts Larry Ellison's belief in keeping a core product team together for decades with Elon Musk's reportedly opposite view, even though Musk was mentored by Ellison. This illustrates that even industry leaders hold fundamentally different philosophies on optimal organizational design.

Söderström emphasizes that companies often switch between organizational models, attempting to fix one problem only to create another. The effective strategy is to consciously identify the company's most crucial priority and design the organization to excel in that specific dimension. This means accepting that the company will be merely average in less important areas, as the worst outcome is being proficient at non-critical tasks while failing at essential ones.

Spotify, for example, deliberately optimized its structure to deliver a single, integrated user experience. This is a complex endeavor because the platform must seamlessly unify disparate business models, such as the royalty-pool-based music business and the ad-supported podcast business, without the end-user perceiving these underlying structural complexities.

You can't win, there is no right org, and companies, they, they sit with one type of org, and then you're gonna be good at something and terrible at something.

Spotify's North Star of Time Well Spent Drives Distribution Strategy

Spotify's core organizing principle, or "north star," is optimizing for "time well spent" by its users. This philosophy deeply influences its strategic decisions, particularly regarding content distribution.

The company identified distribution as the most significant challenge in the market, rather than the existence of good content products. For example, when entering podcasts, numerous quality podcast apps already existed, but none had significant user reach.

Instead of creating separate apps for podcasts or audiobooks, which would simplify internal organization, Spotify opted to integrate these new content types directly into its main application. This decision accepted internal complexity and "pain" to leverage its existing massive user base, prioritizing broad distribution over internal ease.

This approach reflects Spotify's commitment to prioritizing the consumer experience above its own internal organizational convenience or the interests of publishers and labels. The goal is to maximize user access and engagement within a single, adaptable software platform.

We chose to take the pain of doing a single app, with all the complexity that comes with that, for the benefit of reaching what was then already three hundred something million users.

Spotify's 'No Regrets' Strategy Prioritizes User Well-being Over Dark Engagement Patterns

Spotify's core philosophy centers on a deep care for the user, extending beyond mere engagement metrics to how users genuinely feel about their time spent on the platform. The company actively avoids implementing 'dark engagement patterns' that might increase usage or retention but are ultimately detrimental to user well-being.

The platform began with music, a category widely considered an 'unequivocal good' with immense global reach. This deliberate foundation influenced its expansion into podcasts, seen as a counter-trend to fleeting short-form content, and later audiobooks, especially given their popularity in markets like Sweden. These content choices align with a broader 'good for the world' perspective.

Internally, Spotify has long operated under a 'no regrets' strategy, now being communicated externally. This strategy was developed from anonymous third-party surveys that asked users how they felt about time spent across various media services, specifically querying how much time they regretted. Spotify consistently registered the lowest regret among the surveyed platforms, confirming its focus on valuable user experiences.

We were actually the lowest regret.

Spotify announces AI-powered personalized running playlists

Spotify is strategically focusing on becoming a "time well spent" platform, contrasting with survey data showing users regret nearly 60% of time spent on other major platforms. This insight, particularly from Gen Z users, led Spotify to prioritize experiences that make users feel good about their time spent, aligning with their internal saying to become a 'no regrets' area of the internet.

A significant part of this strategy involves deeper investment in fitness, as almost 70% of Spotify users already incorporate the app into their workouts. The company views fitness as an area where people rarely regret their time, making it a natural fit for their "time well spent" philosophy and prompting them to enhance their offerings in this domain.

Upcoming features will include AI-powered personalized running playlists, allowing users to request a playlist tailored to a specific pace, like an eight-minute mile, while matching their musical taste. The system will also be able to beat-match music precisely to the user's steps, either on every downbeat or every other step, and dynamically adjust the music's tempo.

To further enhance the workout experience, the AI can seamlessly mix songs with perfect beat-matched transitions and even overlay spoken commentary. This commentary can guide users through intervals or pace changes, creating a highly customized and engaging fitness journey that helps users achieve a truly "no regrets" workout.

Now you're gonna go into the interval part, and now you're gonna slow down and so forth. So these are the kinds of experience that we're building. This is what I mean with going deeper into fitness. That's one of those no regrets area or time well spent.

Spotify Prioritizes User Well-being Over Engagement Metrics

Gustav Söderström reflects on the challenge of truly assessing whether one's work is good for the world, noting that many people convince themselves they are doing good even when they are not. He cites an example of a company pitching a "maximum truth-seeking machine" that was primarily focused on prediction markets for sports game coin flips, highlighting the common disconnect between stated mission and actual practice.

Spotify recently faced a decision regarding video podcasts. While video increases engagement, many users, particularly parents, reported that seeing videos diminished their sense of "time well spent" on the platform. The company recognized this conflict between engagement metrics and user experience.

In response, Spotify made an "anti-engagement decision" by allowing all users to turn off video content if they choose to simply listen. This means potentially taking an engagement hit for some users, but aligns with a commitment to user well-being.

This principled stance is feasible due to Spotify's business model, where nearly 90% of revenue comes from subscriptions. Unlike advertising-based models that incentivize maximizing time spent at any cost, a subscription model encourages optimizing for user value and satisfaction, as users pay monthly for a positive experience. Spotify is now doubling down on this alignment, believing it's the right long-term bet.

We were faced with the decision, like video is good for engagement, but these people feel it is not good for their time, their time well spent. So we made the decision, sort of the anti-engagement decision, to allow people, anyone, to turn off video if they want to.

Spotify Uses Generative AI to Give Users Control Over Algorithms

Spotify is leveraging generative AI, a dual-use technology, to empower users by giving them unprecedented control over recommendation algorithms. Instead of passively receiving suggestions based solely on past listening, users can now actively define their desired audio experience.

This new approach allows users to communicate directly with Spotify in plain English. For example, if the algorithm identifies a user as primarily an EDM listener but they wish to explore classical music or more biographies, they can simply state their preference. The system will then adapt to provide content aligning with their expressed identity, not just their historical data.

The core idea behind this initiative is that generative AI enables computers to understand natural language, making interaction accessible to everyone, not just developers. This transforms user research from small focus groups to a continuous, high-fidelity dialogue with all 761 million users, allowing the Spotify experience to adapt dynamically to individual wants.

Gustav Söderström views this as a contrarian yet essential bet: giving consumers more control over the algorithms that shape their content consumption. This shift aims to create a highly personalized and adaptive experience that truly reflects each user's evolving interests.

The best way I think to describe generative AI is that computers understand English. It used to be a small population of about one million developers on GitHub who could talk to computers. Now, we all can.

The Power Law of Curation: How Active Users Benefit Passive Consumers

User engagement often follows a power law distribution, commonly referred to as the 1-9-90 rule. This means a small percentage of users are highly active as creators, a slightly larger group actively curates, and the vast majority are passive consumers. This principle applies across various platforms, from music streaming to social media, where some listen or post a little, and others do so extensively.

Spotify's success in personalization and recommendations is a prime example. Millions of users created billions of playlists, carefully curating tracks that go well together. This massive dataset, generated by a relatively small fraction of dedicated users, provided invaluable information that allowed Spotify to offer amazing recommendations to the hundreds of millions who were less engaged in music curation.

This power law dynamic is also evident in AI interactions. A few active users will engage extensively with AI systems, providing feedback and guidance to shape its behavior. Their work in refining the AI for themselves will inadvertently benefit a much larger group of passive users who don't have the time or knowledge to do the same, allowing the AI to "do it for them."

that one percent that creates and that nine percent that curates are gonna be incredibly valuable for the other ninety percent that consumes.

Gustav Söderström Details Spotify's Aggressive Early AI Investment

Gustav Söderström's long-standing interest in artificial intelligence was reignited at Spotify around 2009. The company already possessed massive datasets of music and playlists, enabling talented individuals like Erik Bernhardsson to implement machine learning, specifically collaborative filtering, on some of the world's largest Hadoop clusters for recommendations.

Söderström's belief in the computational nature of human intelligence deepened with DeepMind's work on reinforcement learning, particularly the Deep Q network that started beating Atari games. This demonstrated the scaling potential of AI to handle increasingly complex problems, inspiring him to personally delve into coding recurrent neural networks.

A pivotal moment arrived in June 2017 with the release of the Transformer paper, "Attention Is All You Need." Söderström was profoundly impressed and became an early evangelist for the technology, leading Spotify to aggressively invest in this new AI paradigm.

Spotify began preparing for a future where computers could reason and converse. They acquired Sonantic, a company capable of producing voice very cheaply, even before Large Language Models (LLMs) were sophisticated enough to generate the scripts. This strategic move was a bet on intercepting the exponential curve of AI development rather than waiting for it to mature.

When the, the, what is called the transformer paper came out in, I think, June of twenty seventeen, there was a paper called "Attention is all you need." I think I read it like in the first few days, and I was completely blown away by it, and I started talking to anyone who wanted to listen.

Gustav Söderström's Philosophical Realization: Identity as Information Processing

Gustav Söderström shares a profound philosophical realization from his youth: human identity is not tied to physical atoms. He learned that 99% of the atoms in a human body are replaced every seven years, prompting him to question what truly constitutes a person if their physical components are constantly changing. This led him to conclude that one is not their atoms, but rather the structure or information that those atoms form.

Further refining this idea, Söderström realized identity is not even static DNA, as the brain's physical structure adapts based on what one thinks or practices. He cites examples like pianists developing musical centers, language learners enhancing language centers, and London cabbies who, after memorizing the city, show overdeveloped 3D spatial thinking centers in brain scans. This dynamic re-patterning of atoms, even as they are replaced, indicates that one's thoughts actively shape their physical being.

Ultimately, Söderström concluded that a person literally *is* their thoughts—the constant, processing information pattern. This perspective shifted his understanding of self, viewing humans as computational beings defined by information processing rather than static matter. This core realization profoundly influenced his deep interest in artificial intelligence.

This philosophical foundation developed in the late 1990s, well before current AI capabilities. At the time, computers were not powerful enough to instantiate these ideas, making early neural network experiments unimpressive. This long-standing conceptual framework underscores his deep, personal connection to the field of AI.

You literally are your thoughts. Literally, not as an analogy. It's the only thing that is constant, it's the processing information pattern, right?

Steering Technology to Enhance Humanity and User Well-being

Gustav Söderström emphasizes the importance of steering AI development towards outcomes that benefit the world, such as giving users more control. He criticizes companies that prioritize engagement metrics despite knowing their products negatively impact users' mental health, calling such decisions "insane" for pushing features that degrade well-being.

David Senra echoes this sentiment, expressing a deep desire for technology that enhances humanity rather than replacing it or causing detrimental effects like making people "dumber, fatter, or just depressed." He contrasts this with the current trend of short-form, attention-grabbing content that he believes is destroying users' brains, noting that many technologies currently contribute to negative user states.

In contrast to harmful technologies, David highlights the Amazon Kindle as an example of thoughtful product design that genuinely enhances humanity. He praises not only its passive screen technology but also its innovative business model, specifically mentioning WhisperSync and the inclusion of mobile data that silently updates content in the background, making it a truly user-centric product.

I want the people building the technology that we all use on the scale that we all use it at to build tools that enhance humanity instead of trying to replace it or to make you dumber or fatter or just depressed, which I feel is a lot of what's going on now.

Kindle's Business Model Innovation with WhisperSync

Amazon's Kindle demonstrated a significant business model innovation with its WhisperSync feature. Rather than making users purchase a separate monthly mobile subscription for book updates, Amazon negotiated fixed-cost data deals directly with carriers, calculating based on average book size. This allowed books to update magically and freely over mobile data, which was a remarkable technological and business achievement at the time.

This illustrates a core principle: technology is a necessary ingredient for change, but not sufficient on its own. True disruption and lasting change occur when new technology is combined with a new, often contrarian, business model. Spotify's access-versus-ownership model for music is another prime example of this principle in action.

The contrast between Spotify and Apple Music further highlights this dynamic. While Apple was a massive company, Spotify, a smaller entity, had an existential need to succeed. This drive, combined with its innovative business model, allowed Spotify to overcome larger competitors despite initially having fewer subscribers.

I believe that technology is necessary ingredient for change, but not sufficient. I think you can cause havoc with technology, just like piracy. But when things really change is when you take a new technology and marry it with a new, often contrarian business model.

Spotify's Existential Battle Against Apple Music

When Apple launched Apple Music, the company's internal goal was explicitly to 'kill Spotify' within six months. This aggressive move followed Apple's acquisition of Beats Music and the recruitment of music industry figures like Dr. Dre and Jimmy Iovine, signaling a formidable challenge.

At the time, Spotify was a relatively small Swedish startup. Facing the world's largest and most respected product company, led by Steve Jobs—a figure deeply admired by Spotify's founders—created an overwhelming sense of pressure and an existential threat for the company.

Despite the immense challenge posed by Apple's direct competitive attack, Spotify's team worked tirelessly through this period. The company ultimately persevered and survived, navigating what was one of its most difficult and pivotal moments.

We're gonna kill you. That's what they literally said in their internal meetings. They gave us six months.

Spotify's Three Strategic Bets to Counter Apple Music

Spotify crafted a counter-strategy against Apple, focusing on three key areas where they predicted Apple would struggle. This deliberate approach was critical for Spotify, as it was a fight for survival, a time when they felt "back against the wall".

One major bet was on a superior freemium model. Spotify believed Apple would have difficulty implementing an effective free tier with advertising due to their high desire for control, as seen with their iAd product. Consequently, Spotify heavily invested in building a robust and appealing free music experience.

Their second bet centered on advanced personalization. Spotify observed Apple's past reluctance with data and predicted this would hinder Apple's ability to excel in tailored recommendations. Spotify viewed strong personalization as essential for the future of music streaming.

The final strategic pillar was ubiquity. Spotify anticipated that Apple would prioritize its own hardware ecosystem, such as iPhones, and thus wouldn't effectively support non-Apple devices like Android phones or Samsung TVs. Spotify committed to being available across all platforms.

Gustav Söderström uses a personal AI agent for deeply customized audio updates

Gustav Söderström utilizes a personal AI agent that combines his private documents and Spotify listening history to generate highly personalized audio updates. This system analyzes his interests across various services and local files to understand his preferences.

The AI agent then sifts through global news and social media discussions, filtering out general noise and presenting only the information most relevant to Gustav. It focuses on content being discussed by people he follows, ensuring that the updates are not only personalized but also contextually significant within his professional and personal networks.

Unexpectedly, this personalized algorithm has become a significant source for discovering new music and podcasts. By identifying what is being talked about among his trusted connections, the agent surfaces media that aligns with his interests and the discussions happening in his important circles.

For instance, the agent repeatedly highlighted an episode featuring Evan Spiegel, emphasizing its product wisdom and encouraging Gustav to listen. This demonstrates the agent's capability to deliver specific, actionable insights from a vast array of information.

it's kind of like a personal algorithm that filters out a lot of Of the noise and the, the crazy that I don't wanna hear, and just gives me the stuff that I wanna hear.

Using AI Agents for Premeditated Media Consumption to Improve Well-being

Gustav Söderström actively employs an AI agent to filter the media he consumes, specifically targeting content like "rage bait," "clickbait," and political arguments. This strategy aims to eliminate noise and focus on information that doesn't induce negative emotional responses or elevate cortisol levels, thereby improving his overall well-being.

He refers to this proactive approach as "premeditated media," where individuals decide their future information consumption ahead of time. The goal is to prevent getting "captured in the moment" by sensational or emotionally charged content, recognizing that people often gravitate towards things they know are bad for them, much like choosing unhealthy candy.

Söderström provides an example from his own life, noting his tendency to fall into "dark holes" watching road rage videos, despite knowing the negative impact. By setting a filter to "Don't show me road rage," he maintains user control over his information diet. This allows individuals to establish boundaries for their media intake, consciously avoiding content known to cause sadness or anger.

Like premeditated media.

Honest Feedback and Trust Built Through Long Tenure

Effective leaders require candid, critical feedback to avoid blind spots and make informed decisions. Many leaders struggle to receive such honesty due to a surrounding 'cloud of sycophants' or a corporate culture where challenging superiors can lead to negative consequences.

Gustav Söderström highlights that the key to receiving true feedback is having long-tenured team members. Over time, these individuals develop the trust and comfort to voice strong disagreements without fear of reprisal, even when a leader reacts passionately or dismissively initially.

He recounts an instance where he told Daniel that his performance in running product was lacking. Daniel's initial reaction was to fire him, but upon reflection, he realized Gustav was right and had provided a valuable service. This illustrates how deep trust, fostered through years of working together, enables crucial developmental honesty.

This environment allows for direct challenges to ideas, preventing bad decisions and strengthening leadership. The absence of consequences for disagreeing, built over a long working relationship, is essential for cultivating this level of psychological safety and fostering a truly honest internal feedback loop.

He went home and he told his wife, 'Oh my god, I have to fire this guy.' And then he thought about it, 'Oh, he's right. It's like he just did me a service.'

Spotify's Strategy for Balancing Tenure and New Talent

While tenure brings efficiency and deep organizational knowledge, it also presents risks like a lack of fresh perspectives and a slow career ladder for younger talent. Companies need to actively control for the downsides of long-serving employees to avoid becoming a group of 'old people' resistant to new ideas.

Spotify addresses the challenge of cultivating new talent through its 'rising stars' program. This initiative identifies and accelerates promising junior employees, sending them to different global markets and connecting them with senior leaders like Alex and Gustav. This internal 'MBA-like' program builds cross-organizational relationships and understanding, allowing participants to rise faster than the organic career path.

For completely new skill sets that the existing organization might lack, Spotify employs strategic acquisitions. For example, when deep learning emerged, Spotify acquired The Echo Nest to gain expertise and establish a presence near MIT. This approach allows the company to quickly integrate new, senior-level capabilities rather than waiting for internal development.

The overall strategy emphasizes that companies optimized for tenure must proactively mitigate its drawbacks to ensure continuous innovation. This involves not only fostering internal junior talent but also being prepared to acquire external expertise for rapidly evolving fields to avoid falling behind, a challenge potentially faced by companies like Apple with AI.

if you're optimized for tenure, you have to mitigate for the risk that you don't get fresh blood, that your career ladder is slow

Spotify's Gustav Söderström on AI as a Macro Wave and the 'Always Be First' Strategy

Gustav Söderström views AI as a macro wave that will fundamentally change media habits, similar to past shifts caused by the ubiquity of cheap broadband and personal computers, or the advent of the smartphone. He explains that these periods of new technology and emerging business models often create significant disruption, altering how people consume content.

He illustrates that while stable periods allow for straightforward extrapolation of trends (like 2015-2025), disruptive eras (such as 2005-2015 with the smartphone's rise) invalidate such predictions, causing companies to miss entirely new markets like Uber or Airbnb. Söderström believes the current AI landscape is one of these transformative periods where everything is in flux, and the exact outcome is still unknown.

Despite the inherent risks associated with such profound change, Söderström emphasizes that these are also the greatest opportunities for market share shifts. He notes that Spotify experienced its most significant growth during periods of instability. The company's guiding principle is to "always be first" to adopt new technologies and business models, viewing the current AI wave as a time of immense opportunity to innovate.

My principle is just always be first, be first and adopt it first.