He set out to build an online game twice, failed twice — and both times the internal tool his team built to survive became worth billions.
Every entry is researched against 30+ structured fields
Stewart Butterfield lived until age three in a log cabin without running water in British Columbia, taught himself to code young, and then did the least commercially obvious thing available: studied philosophy, taking a master's at Cambridge.
His first company, Ludicorp, was founded in Vancouver to build a massively multiplayer online game called Game Neverending. It never launched. But the team had built tools to support it, including a way to share photos inside the game, and that fragment became Flickr — one of the defining products of the early social web. Yahoo acquired it in 2005; reported prices vary between roughly $22 and $35 million.
He stayed at Yahoo, then left around 2008.
In 2009 he started Tiny Speck with roughly $17 million in funding and did the same thing again: built an elaborate multiplayer game, this one called Glitch, where players could milk butterflies and grow eggs on trees.
It did not work, and the reason was measurable rather than mysterious. Butterfield has described it as a leaky bucket — about 97% of people who signed up were gone within five minutes. The small minority who stayed loved it intensely and played for twenty hours a week, which is exactly the trap: passionate users can disguise a broken funnel for years. He has said you could put the numbers in a spreadsheet and see in five minutes that it would never work.
After roughly three and a half years, he shut Glitch down in 2012. The tipping point came during that shutdown: looking at what his team could not stop using while everything else collapsed, he noticed the internal chat tool they had built for their own survival. It had been stress-tested by a company in crisis.
Slack launched in 2013. The critical difference from Glitch was that this product had already been used in anger, daily, by a real team under pressure — which is a stronger validation than any amount of user research.
Growth came without a sales or marketing team. Daily users went from about 10,000 to 90,000 in five months on word of mouth alone. The company was hiring frantically to keep up, going from eight people to twenty to thirty-four within six months.
The defining product decision was the channel — conversation organized by topic rather than by person, which is what made it a system of record for a team rather than a messaging app.
Slack became the fastest-spreading workplace software of its era, went public in 2019, and was acquired by Salesforce in 2021 for $27.7 billion.
Butterfield's own account of why the second attempt at a game company was worth trying is unglamorous: the team was more experienced, capital was easy to raise, and infrastructure had gotten dramatically cheaper. There was no grand insight that Glitch would work — the conditions were simply better.
What makes the record extraordinary is the repetition. Two companies founded to build games. Two games that failed. Two internal tools that became multi-billion-dollar businesses. Once is an accident; twice is a method.
The method is paying attention to what your team actually uses versus what you are trying to sell. Both Flickr and Slack existed as infrastructure before they existed as products, and in both cases the company was too busy pursuing the official plan to notice until the official plan died.
The second lesson is about honest metrics, and it is the harder one. Glitch had genuinely passionate users — the kind founders quote in decks. Butterfield's discipline was looking past them at the 97% who left in five minutes, and refusing to rationalize. Enthusiasm from a tiny minority is the most seductive form of false signal available.
Compare this directly to DHH's entry: same internal-tool-to-product path, opposite capital structures, both successful. The pattern generalizes; the funding model does not.
The framework: when the plan fails, audit what your team built to survive it — that byproduct is often the actual business.
Move 1 — List every internal tool, script, spreadsheet, or process your team relies on daily that was never meant to be a product. Ask which one you would fight to keep if the main project vanished tomorrow.
Move 2 — Put your retention numbers in a spreadsheet and look at the percentage who leave in the first five minutes, not the enthusiasm of the people who stay. A passionate minority on top of a leaky bucket is the most convincing wrong signal in business.
Move 3 — When the numbers say stop, stop, and salvage rather than iterate. He shut Glitch down after three and a half years instead of spending another year on features; that decision is what created Slack.
Budget line: this specific path required venture funding to have a team building internal tools at all. The transferable half — noticing your own byproducts — costs nothing.
Flickr sale price reported inconsistently across sources, roughly $22-35M.
Last verified 2026-09-01
Same structure every time — the tipping point, the first ninety days, and a playbook with real numbers attached.