He Needed Software
to Sell Snowboards.
Every Tool He Tried
Was Wrong.
So He Built the Right One.

The Setup: A German Apprenticeship, a Canadian Girlfriend, and No Work Permit

Tobias Lütke was born in Koblenz, Germany, in 1980. His parents gave him a Schneider CPC computer when he was six years old. He sat down and never really got up again. By eleven or twelve he was not just using software — he was rewriting it. Modifying the code of games he played. Discovering that the machine was not a fixed thing but a malleable one that responded to instruction.

He faced real challenges at school. ADHD and dyslexia made the conventional academic path difficult. He finished tenth grade and stopped. Not because he was not intelligent. Because he had found something more useful to do with his time. He entered the German apprenticeship system — not a consolation prize but a rigorous, experience-based pathway that puts a young person inside a real company doing real work from the first week.

He landed at BOG Koblenz, a Siemens subsidiary. His supervisor was a man named Jürgen — long-haired, rode a BMW motorcycle to work, did not wear a suit like he was supposed to. Lütke has described what followed as the most important thing that ever happened to him professionally.

After the apprenticeship, he met Fiona McKean while snowboarding in Canada. He moved to Ottawa to be with her. Moved into her parents' house while he figured out what came next. He could not get a work permit for a local job. He learned he could start a business instead. He and Scott Lake, a family friend he met through Fiona's parents, decided to build an online snowboard store. They each put in twenty thousand Canadian dollars. They called it Snowdevil.

Lütke started looking for software to run it. He found nothing that worked. So he built it himself.

The Constraint: Every Tool in 2004 Was Built for Someone Who Was Not You

Lütke has described the e-commerce options available in 2004 with clarity. Yahoo Stores was built for companies with web teams. Miva was unstable. osCommerce was so complicated to configure that setting it up correctly required a developer's full attention just to maintain it. None of them assumed that the person running the store was also the person doing everything else — the design, the marketing, the shipping, the customer service, the accounting.

Every hour he spent troubleshooting the software was an hour he was not spending on the actual business. That frustration is the most valuable raw material in software product development. More valuable than market research. More valuable than competitive analysis. Because it produces a product that reflects the real experience of the real problem rather than a product team's theory about what that experience might be like from outside it.

He built the platform in two and a half months using Ruby on Rails — a programming framework that had just been released and that almost nobody outside the developer community had heard of. He chose it because it matched his engineering instincts. Clean. Fast to build with. Minimal configuration required. When the platform was finished he launched Snowdevil. First customer came from Pennsylvania. He remembers it still. Profitable first season.

Then something unexpected happened. Developers in the Ruby on Rails community started asking about the platform. How did you build this. Can I use it. Can I see how it works. By Christmas 2004, Lütke and Lake looked at each other and made the decision that changed everything. The snowboard store was not the business. The software was the business.

The Playbook: Four Decisions That Turned a Snowboard Store Into $292 Billion in Annual Transactions

The first decision: build the tool for the problem you personally cannot solve any other way. This is the most fundamental principle in the Shopify story. Lütke did not study the e-commerce market and identify a gap. He tried to sell snowboards, found that every available tool made it harder than it needed to be, and built the tool that would let him do the actual job he wanted to do. That origin produces a specific quality in the product that market research cannot manufacture. When the builder has the problem, every decision is tested against the real experience rather than against a theory about it.

The second decision: choose the emerging technology before consensus has formed around it. Ruby on Rails in 2004 was not established. It was new, written by one developer at a small company in Chicago, used by almost nobody. Lütke chose it because it matched his instincts. That choice had compound effects he could not have predicted — it put him inside the Rails community at a moment when that community was small, specific, and intensely interested in each other's work. The first people who found Shopify were Rails developers who understood what had been built and why it was well made. Their enthusiasm was credible in a way that a general marketing campaign could never produce.

The third decision: stay in Ottawa when every pressure pointed elsewhere. In 2008, investors in San Francisco invited Lütke to come down. The implication was clear: move to the Bay Area, raise a serious round, join the ecosystem. He went to the meeting. He heard it out. He went back to Ottawa. His reasoning has been consistent across every interview since. Being inside the consensus is a disadvantage for a company trying to build something the consensus has not yet imagined. The conversations optimised for what San Francisco investor circles care about may or may not have anything to do with what a small business owner in Pennsylvania needs from an e-commerce platform. Distance from the narrative helped him build the actual company rather than the story's version of it.

That same year, 2008, the financial crisis hit. Most businesses contracted. Shopify grew. People who lost jobs started businesses. People who had been sitting on ideas decided the downside of trying had just become much smaller relative to the downside of depending on an employer. The number of people who needed to sell things online grew precisely because the employment security that had kept them from trying had evaporated. Shopify was positioned for that moment exactly — affordable, accessible, reliable enough for a newly self-employed person who needed the software to work without daily maintenance. Monthly revenue reached $60,000 — eight times what it had been a year earlier. The recession was their growth engine.

The fourth decision: align the revenue model with merchant success before designing any other feature. Shopify's take rate is a percentage of what merchants make. When merchants make more money, Shopify makes more money. When merchants fail, Shopify earns nothing. That structural alignment is not accidental — it is the most important design decision in the business model, made early and maintained consistently. It produces loyalty that does not respond to a competitor offering a lower monthly subscription fee. You do not leave the platform whose success depends structurally on your success.

The Platform Decision: The Move That Turned a Product Into an Ecosystem

In June 2009, Shopify launched its API and App Store. The decision was straightforward in principle and transformative in effect. By allowing external developers to build applications that integrated with Shopify, Lütke turned a single product into a platform. Every application built by an outside developer made Shopify more capable without costing the company a single engineering hour.

By 2024 the Shopify App Store had over 10,000 apps built by external developers. Integrations with specific shipping providers, specific marketing tools, specific inventory management systems, specific accounting packages. Features Shopify could not have built internally — specific tools for specific merchant categories — available to every merchant on the platform because someone outside Shopify had a reason to build them.

This is the philosophical core of what Lütke calls arming the rebels. Amazon is a marketplace. Merchants who sell on Amazon are building Amazon's brand recognition, contributing to Amazon's data advantage, operating inside Amazon's rules. When a merchant succeeds on Amazon, Amazon gets stronger in ways that do not necessarily make the merchant stronger. Shopify is infrastructure. Merchants who sell through Shopify are building their own brand, owning their own customer relationships, controlling their own data. When they succeed they become more independent. The structural difference between these two models is not marginal. It is the reason merchants build loyalty to Shopify that price cuts cannot disrupt.

The Traction Framework: How a Developer Forum Became the Distribution Engine for a $150 Billion Company

Weinberg and Mares in Traction identify 19 channels through which businesses get customers. Their most important observation: most businesses get zero channels to work. One working channel is enough. Here is how Shopify ran the Bullseye across its first decade — and why the channels it found were only available because of specific early product decisions.

The Weinberg and Mares point about channel-product fit applies precisely here. The merchant solutions Shopify added over time — Shopify Payments in 2013, Shopify Capital, Shopify Markets — each deepened the platform's value while simultaneously deepening the switching cost. Each new integrated service made a competing product harder to switch to. The channel and the product were the same thing. Distribution was built into the product architecture, not bolted onto it through marketing spend.

In 2026, the same principle is available to any builder with a product that other developers might want to build on. Build something worth building on. Open the API early, before the product is large enough that it feels like a risk. The developers who find the API first and build for your platform will bring their audiences with them. That is not a marketing channel. It is a structural advantage that compounds for as long as the platform keeps working.

The Mistake: Building Permanent Infrastructure Around Temporary Growth Is the Most Expensive Error a Platform Can Make.

Lütke has been more transparent about the pandemic-era expansion than most public company CEOs manage when things go wrong. During the pandemic, Shopify went on a significant hiring and investment spree. The logic was defensible in the moment. E-commerce was growing faster than anyone had modelled. Building fulfilment infrastructure and logistics capability seemed like exactly the right bet. Headcount expanded dramatically.

Then the pandemic ended. E-commerce growth normalised. The logistics infrastructure bets did not produce the returns they required. Shopify sold its logistics business to Flexport in 2023 at a significant loss. Layoffs followed. Lütke has addressed it without excessive self-flagellation and without deflection. They bet on a trend continuing at pandemic pace. It did not. They corrected.

The recovery was real. In 2024, Shopify posted $8.88 billion in revenue, $1.6 billion in free cash flow, and processed $292 billion in GMV. The pandemic error was costly and the correction was painful. The underlying business was sound enough to absorb both. That soundness was the result of every decision made in the twelve years before the pandemic arrived. The mistake was correctable precisely because the foundation was real.

What You Can Apply: Five Principles From a Snowboard Store That No Longer Exists

Your Move This Week

Lütke did not move to San Francisco. He did not raise money before he had a product worth funding. He did not build a pitch deck before he built the thing. He sat in his girlfriend's parents' house in Ottawa and built the platform he needed to sell snowboards. The snowboards were never the point. The platform was always the point. He just had to build the snowboard store first to find out.

The workflow frustration you right now is your version of that snowboard store. It is not the final product. It is the first question. The answer to that question — the tool you build to eliminate the frustration — might be the thing you were actually trying to build all along.

In 2026, with AI compressing the distance from frustrated recognition to working prototype to days, the only thing that stands between the frustration and the platform is the decision to start with the problem you already have rather than the one you wish you had.

Next issue: The validation framework I use to test any idea in an evening — before writing a single line of code, taking a single meeting, or spending a single dollar on infrastructure.