Tobi Lütke on Learning, Creative Constraints and Leading Shopify

Shopify co-founder and CEO Tobi Lütke tells Harry Stebbings that he sees happiness as temporary and contentment as a different goal. Their 2023 conversation explores how he learns, designs constraints, thinks about remote work and leads a large product company.

Lütke grew up in a small German town and encountered a computer at six. He learned programming from magazine listings before he even knew the name for it. By around twelve he wanted to spend his life programming, and although his work changed, he never stopped.

The first experience of typing a letter and seeing it appear on a screen felt magical. He wanted to understand how the machine worked: he dislikes black boxes and tells his children that things are understandable and often more interesting than they first appear.

Computers gave a child in a rigid school system an unusual sense of agency. He could make something, control what it did and share it. Lütke believes entrepreneurs need frontiers—places to reach for something new and create value.

Asked what he is running from, he names boredom and ignorance, but says he is more drawn toward what he might become. His personal image is of meeting the person he could have been at the end of life and trying to minimize the gap.

For Lütke, learning updates a mental model of the world. He carries that model into decisions and wants it to be close to reality. That is why he wants to understand transformer models and embeddings well enough to reason with his teams instead of treating them as black boxes.

He says he follows curiosity because he has repeatedly found valuable lessons in unexpected subjects. Harry asks how to commit scarce time when the eventual value is uncertain. Lütke weighs the effort and reversibility of an experiment.

If a test is cheap and easy to reverse, he suggests trying it. An afternoon with material already available can produce a preliminary scorecard before committing to deeper research or a new channel. Even an unsuccessful test fills a gap in the model.

On goals, Lütke describes himself as more of a craftsperson than a scientist. Learning for its own sake can be worthwhile, but he values turning knowledge into something other people can use, judge or explore. If he cannot see a path to sharing its value, he tends to lower its priority.

Learning guitar and playing in a band started as a personal hobby after years focused on Shopify. It unexpectedly taught him about teams and creativity. In jazz and blues, musicians choose a tempo and key, then improvise within that shared structure.

The constraints make collaboration possible rather than eliminating creativity. Skilled musicians can bend or change a rule and others can follow; a team similarly learns one another’s strengths while building something. Lütke says good teams actively seek useful boundaries.

At Shopify, he says projects move through proposal, prototype, build and release. Prototype phases have limits on time and people: he describes a maximum of six weeks and three people, while arguing that a week and repeated fresh attempts might be better.

Early on, Lütke and an engineer paired on code under a stricter constraint: they threw away the source code each day except the tests and implemented the feature again. He credits the exercise with sharpening their work and says creating constraints is itself a creative act.

Harry asks how a well-funded company makes artificial constraints real. Lütke points to culture, stories about valued work and a project management system that limits staffing and duration at different stages. Rules can also be challenged or changed when they prove harmful.

Is the best CEO primarily a resource allocator? Lütke considers that an incomplete explanation. In his view, good allocation is often the visible result of a more accurate model of the world, effective communication and trust that lets people build toward it.

He also emphasizes subtraction. Companies add practices because others do them, yet removing a process is often harder than adding one. If nothing is ever removed, the system accumulates unnecessary work.

Harry notes that a CEO can call experts directly; Lütke agrees that access helps but says podcasts, books, papers and online discussion also make knowledge available. Growing up outside a technology hub and living in Ottawa helped him develop a habit of finding and assembling information from a distance.

He listens, takes notes and checks unfamiliar ideas, including on Wikipedia. The deeper distinction for him is between seeking a view accepted by one’s group and seeking what is true. The latter can be lonely and requires returning to primary sources and people willing to challenge assumptions.

On sunk costs, Lütke acknowledges the fallacy: people can keep doing something because changing course appears to condemn earlier choices. He wants to ask instead what the best next move is given current knowledge. New evidence should be an opportunity to update a belief, not a threat to identity.

He stresses that this is an aspiration rather than a claim to perfect rationality. He states his current view strongly partly to invite disagreement; if another person has better information, he wants to learn from it and change his position.

How much evidence is enough? Lütke cites a rough 40–70% information heuristic, then adds a key dimension: the cost of reversing the decision. A cheap, reversible choice can be tested early; a costly or irreversible one warrants more care.

Lütke gives Shopify’s move to remote work as an example of responding to changed conditions. He strongly values physical proximity and recalls learning as an apprentice by sitting near an exceptional mentor, hearing conversations and receiving feedback on code.

His preferred small team is diverse in expertise yet close enough to learn by osmosis. He describes an order of preference from a shared room to a floor, building and time zone. Shopify had designed offices for collaboration around those beliefs.

As Shopify grew, however, it could not recruit everyone it needed in one city and expanded across locations and time zones. Lütke believed a hybrid arrangement risked concentrating influence near headquarters and treating remote colleagues as appendages.

He says Shopify had already experimented by closing its offices temporarily to understand remote employees’ experience and find tooling problems. He thought a fully remote shift might happen within a year or two as the company’s distribution changed.

COVID lockdowns changed an implicit assumption in his model: whether people could leave home. Shopify moved to work from home and then committed to being digital by default. Lütke says he still considers proximity ideal for a small company that can fit on one floor.

Harry asks whether starting remote would avoid a painful transition later. Lütke says the trade-off includes the talent pool: even if remote collaboration were less productive, access to candidates in many places might outweigh that cost. His 10–20% productivity discount is a hypothetical assumption for the calculation, not a measured Shopify result.

Lütke believes people are generally more productive together, but recognizes that the answer depends on the work. Focused tasks can benefit from remote conditions; open-ended brainstorming and urgent coordination can be harder over video. Distributed teams can also bring people together in person when needed.

Harry asks how to sustain morale when falling stock prices affect employees. Lütke says he tells Shopify staff that the share price is not the company: he distinguishes the market’s changing assessment from the underlying products, people and capabilities. His assessment of those assets is his own, not a market valuation.

Asked about short sellers, he says short selling can serve a legitimate market function and distinguishes it from his criticism of one particular short seller. He alleges that the person made false accusations while holding a short position; the interview presents Lütke’s account and not a finding that those accusations were false.

On personal relationships, Lütke credits his wife Fiona with encouraging him to start the company and helping him think through challenges, even though she has not been on Shopify’s payroll. They sometimes divide reading and share what each has learned.

He describes their relationship as a mix of shared projects and separate interests, sustained by curiosity about one another. He also acknowledges luck rather than presenting his marriage as a formula others can reproduce.

Lütke says his mind rarely switches off and prefers “work-life harmony” to a strict separation. Shopify gives him a reason to explore new technologies and consider how they might help small businesses—an outlook tied to his role and commercial interests at the company.

Harry describes concern about inequality, climate change and the state of the world. Lütke expresses optimism about long-run progress and says he is concerned about climate change but believes technology can solve it. These are his views, not conclusions established by the interview.

Lütke suggests people often feel their own lives are going reasonably well while assuming others are doing badly. He cites approximate figures from memory without a source and acknowledges that real problems and excessive wealth inequality exist. Harry challenges him, saying many people are unhappy with their own living standards.

Lütke agrees that people want change and that real problems exist, but argues that political opponents may share more goals than their disputes suggest. He concedes he is not an expert on these subjects. The exchange leaves their disagreement about public conditions unresolved.

Harry says that even after achieving ambitions he still wants more and asks whether Lütke is happy. Lütke distinguishes fleeting happiness from contentment and says social comparison can undermine the latter, particularly when social media makes the comparison group global.

He also sees the desire for more as a force that can motivate entrepreneurs and researchers. The same drive can produce dissatisfaction, so he does not want it to be the sole measure of a life. He describes this as a personal philosophy rather than a prescription.

Lütke acknowledges that he has been extraordinarily lucky, including financially, and that Shopify created substantial wealth for him. He enjoys well-made surroundings and estimates that such purchases make his days a little better, but considers conversations, creating things, family and health far larger factors.

For him, an appealing luxury can be a good keyboard: he enjoys assembling mechanical keyboards and using tools he appreciates every day. Harry offers a nice car as an example of something that improves his own mood.

In the quick-fire round, Lütke says a trait that can be both useful and difficult for others is his willingness to change his mind. He regrets many small decisions where he wanted a conclusion to be true even though contrary information was available.

He says he once accepted the idea that micromanagement is always bad and now rejects that blanket rule. Harry pushes back: leaders can undermine autonomy, lack the best information and overload themselves if they intervene in everything.

Lütke distinguishes constructive intervention from a bad manager changing things arbitrarily. Harry agrees leaders must intervene when a mistake would be fatal, but argues that many decisions are reversible and people need room to learn. Lütke says leaders should work alongside experts, build trust and then delegate.

He argues that a leader can share responsibility but cannot simply abandon it. In his view, close involvement should bring useful perspective and aim for the best outcome, not merely prove the leader right. He calls this a form of detailed leadership.

A large product should feel coherent, like a book with one voice, even when thousands contribute. Lütke points to Shopify’s Polaris design system as a shared constraint that encodes user-interface knowledge and raises the quality floor while leaving room to create.

He participated in shaping Polaris and reviewing details because the system expressed principles he cared about. He did not do all the work. After building trust in the team and its principles, he says he no longer needs to inspect every dropdown, though he expects to be informed about major changes.

Lütke sees product development as collaboration across organizational layers. Leaders should support the people building, help coordinate the work and make the goal clear. He points to examples from entertainment and technology as inspiration, while cautioning that no one person has every answer.

He thinks company building still has much room to improve and uses a speculative scoring analogy: if current practice is only a few points out of ten, modest improvement can be powerful. He expects future generations to understand the craft much better, but this is his belief, not a measurable forecast.

For coordination lessons, he suggests studying World of Warcraft raiding guilds as well as companies. His broader point is to find useful ideas wherever they occur, keep asking how systems work and avoid leaving them as black boxes.

Lütke returns to Shopify’s mission as he describes it: helping entrepreneurs participate in global commerce and bringing new technology to small businesses. He finds the cycle of exploring what is possible, deciding what matters to merchants and building it especially motivating.

Harry thanks Lütke for the conversation, and Lütke returns the compliment and thanks him.