In December 2012, Stewart Butterfield shut down a game called Glitch, wrote severance checks for a team of roughly 45 people, and told his investors he would wire back whatever cash was left. Accel and Andreessen Horowitz had put more than $15 million into a whimsical, non-combat online world set inside the brains of eleven giants. Almost nobody played it.
Nine years later, Salesforce paid $27.7 billion for what survived the shutdown: a scrappy internal messaging tool The team had cobbled together so engineers in Vancouver, San Francisco, and New York could talk to each other. It was called Linefeed. Today, you know it as Slack.
The pivot story gets retold constantly. The decisions underneath it get skipped, and those are the parts worth stealing.
The asset was hiding in the overhead
Tiny Speck built Linefeed because it had a distributed team before distributed teams were even normal. Email was too slow, and threads got lost. The tool was a pure cost center, the kind of thing that never appears in a pitch deck.
When Glitch died, Butterfield ran a cold inventory of what the company actually owned. The IP was worthless. The Flash-based engine had no path to mobile. The one thing with a pulse was the internal tool nobody had ever thought of as a product.
Worth noting how he framed it to investors. He did not pitch a new company. He offered the money back and said he had one idea left, which is why Accel's Andrew Braccia stayed in. Braccia's line was that he had invested in the team, not the game. Ben Horowitz has since admitted publicly that he thought workplace chat sounded like a terrible idea and that he could not have been more wrong.
Run the same inventory on your startup this week. List every internal tool, dataset, process, or script your team built to survive its operations. One of them probably solves a problem other teams pay to fix.
The word "beta" never appeared
Slack opened its doors in August 2013 with something Butterfield called a preview release. The label was deliberate. Beta signals are broken, and users forgive nothing they did not want in the first place. Preview signals early access to something good.
Day one brought 8,000 companies requesting invites. Two weeks later it was 15,000, and the waitlist got so long that Butterfield joked it would take four years to clear.
The controlled rollout mattered more than the number. Before launch, the team hand-recruited six to ten companies through personal connections, starting with a small design shop and then pulling in Rdio, which had about 120 employees. That single onboarding exposed everything wrong with the product at scale. Butterfield described what they saw as pretty gnarly.
Sequencing discipline is the lesson here. They broke the product on purpose, with a friendly customer, before strangers could break it in public.
Butterfield sold horseback riding, not saddles
Two weeks before the preview release, Butterfield sent his team a memo now used as required reading inside product orgs. His argument: a saddle company selling in a world where nobody rides horses should market riding, not leather.
Slack had built something people wanted, and almost none of them knew they wanted it. Group chat was not a budget line item anyone was searching for. So the go-to-market never mentioned features. It named a villain everyone already hated: email. That framing gave journalists a story and offered buyers a reason to care.
The memo also carried a line about tolerance. When customers desperately want something, they forgive flaws. When they do not yet know they want it, they quickly lose interest. That is the real cost of vague positioning.
Freemium with a slow-burning fuse
Slack's free tier was the full product, capped at a searchable archive of 10,000 messages. The economics are clever once you sit with them. Every message a team sends builds a knowledge base and pushes it closer to the cap. Value compounds, and the switching cost compounds with it, so usage itself generates the upgrade pressure.
Bottom-up growth did the rest. The buying unit was a team, not a person, which meant every signup dragged in colleagues by default. Slack hit a $1.1 billion valuation on a $120 million Series D in October 2014, before hiring a CMO, and did not build an outbound sales team until 2016.
Then it lost the war it was winning
Microsoft launched Teams in 2017 and bundled it into Office 365 at effectively no marginal cost. Slack kept winning on product love but lost on procurement. Teams now reports north of 300 million daily users. Slack, by third-party estimates, sits in the 40 million range and holds roughly 18% of the business communications market.
The 2021 sale to Salesforce was a distribution decision. Slack needed an enterprise sales machine it could not build fast enough on its own.
What to take from this
If you are a startup founder or an entrepreneur, here is what to take away from Slack’s win:
1. Audit your cost centers for products.
Internal tools survive because they solve a real, expensive problem your team feels daily. That is validation most founders pay to get.
2. Choose launch language on purpose.
Preview, early access, and beta set wildly different expectations for the exact same code.
3. Stress-test with a friendly whale.
Onboard one customer bigger than your product can comfortably handle while they still have goodwill toward you.
4. Sell the outcome, name the enemy.
Nobody budgets for a chat app. Everyone budgets for less email.
5. Price your free tier against accumulated value.
Let the archive, the history, or the data do the upselling for you.
6. Product love does not beat bundling.
If a platform can give away your category for free, plan your distribution moat before it does.
Forward this message to a founder who is sitting on an internal tool they have never thought of as a business.
