How Slack went from a failed game to a $19.5 billion direct listing

Executive overview

Stewart Butterfield built Flickr, failed twice at an online game, and accidentally invented the fastest-growing enterprise software product ever — by applying gaming analytics and consumer freemium logic to a market that had never seen either. The core tool was Slack's own internal chat layer, built to coordinate a remote game studio.

Freemium gated on searchable message history — not headcount — was the insight that unlocked whole-company adoption before a dollar was paid.

Stewart Butterfield and the path to Slack

• Born Dharma Butterfield on a no-electricity commune in Lund, BC; legally changed name to Stewart at 12
• Philosophy degree at Cambridge; left the PhD track in 1998 for Vancouver's Web 1.0 boom
• Sold his first company, Gradfinder.com, within a year; met Katarina Fake through blogging and married her
• Built Flickr in under 60 days as a side project to fund an online game (Ludicorp's Game Never Ending); sold to Yahoo for ~$23M in late 2004
• Stayed the full three-year Yahoo earn-out; left in 2008 writing a deliberately absurdist resignation memo

Glitch: failure that created Slack

• Regrouped the Ludicorp co-founders in 2009 as TinySpec; raised $18M+ from Andreessen Horowitz and Accel for a nonviolent browser MMO called Glitch
• Scaled to ~40 remote engineers across Vancouver, SF, and New York — and built an internal IRC wrapper called Line Feed to keep them coordinated
• Glitch launched September 2011; pulled back to beta within two months; shut down October 2012 after Butterfield lost faith and told the board directly
• Before doing anything else, the team found jobs for every laid-off employee via a site called Hire a Genius
• Within a week of shutdown, they decided Line Feed was the product; named it Slack — Searchable Log of All Conversation and Knowledge — a name that was quietly dropped from all marketing
• Bought slack.com from a Wisconsin engineer hosting cat photos; brought in Victoria-based design firm Metal Lab to rebuild the UI

Freemium model and go-to-market

• HipChat and every prior enterprise chat tool gated free use on number of users (free up to five); Slack let any size team use the full product free
• The pay gate was searchable message history — invisible until a team needed to find something from months ago or onboard a new hire
• This removed the barrier that would have stopped non-engineering teams from ever trying it
• Growth came from internal word-of-mouth; Slack amplified user tweets to build external demand without cross-company network effects
• Preview launch August 2013: 8,000 teams on the wait list day one; 15,000 within two weeks
• Weeks after launching paid: $1M ARR; by February 2014, 10,000 new users per day and 135,000 paying users

Scale and competitive position

• Revenue trajectory: $105M → $220M → $400M across three fiscal years; 82% YoY growth at listing
• $800M cash on balance sheet; not yet profitable by design — two-to-three-year payback period baked in
• 90%+ first-year paid customer retention; ~80% at five years; CAC:LTV ratio ~13x; cost of sales ~$600 per company vs. $8,000 total acquisition cost (the gap is the cost of running the free tier)
• Magic threshold: teams that sent 2,000 messages retained at 93% — so the entire onboarding funnel was engineered to hit that number as fast as possible
• July 2018: Atlassian sold HipChat/Stride users to Slack in exchange for a small equity stake — Accel held major positions in both companies
• Microsoft Teams launched bundled with Office 365; Slack ran a full-page New York Times ad in response
• Zoom comparison: Slack is internally viral; Zoom is externally viral — Zoom's sales efficiency ran at ~180% vs. Slack's ~111%

The direct listing (DPO)

• Chose a DPO over a traditional IPO: no new shares created, no dilution, no employee lockup, price set by open market supply and demand
• Pre-conditions: no need to raise cash, consumer-level brand awareness, institutional investors already fully positioned
• Investment bankers shifted from underwriting to advisory: surveying existing shareholders on sell intentions and setting a reference price
• Reference price $26; first day of trading June 20, 2019; closed at $38.62 — up ~50% — for a $19.5B market cap
• Cap table at listing: Accel ~24% (atypically large), Andreessen Horowitz ~13%, Social Capital ~10%, SoftBank ~7%, Butterfield ~7%