tech · 2009–2014 · Tiny Speck (later Slack Technologies)
Slack's pivot from Glitch
How a failing game studio noticed it had built the wrong product — and shipped the right one to itself in the process.
9 min read · 5 sources cited
// background
In 2009, Stewart Butterfield — fresh off selling Flickr to Yahoo — co-founded a small studio called Tiny Speck. The plan was to build a massively multiplayer online game called *Glitch*: a quirky, non-violent, hand-illustrated 2D world about exploration, cooperation, and gentle absurdity. The team raised roughly $17M from First Round, Accel, and Andreessen Horowitz across multiple rounds on the strength of Butterfield's track record and the game's visual charm.
Tiny Speck was distributed across several offices and time zones from day one. The team's chosen tools — IRC for chat, plus a homegrown layer they wrote on top of it for searchable history, file sharing, and inline integrations — became central to how they actually worked together. Butterfield later described the internal tool as "what we needed to build the game at all."
By 2012 it was becoming clear *Glitch* wasn't going to find a sustainable audience. The game launched, drew some passionate fans, and quietly leaked players. In November 2012, Tiny Speck announced it would shut Glitch down. Most of the staff was laid off; a small core of about a dozen stayed behind. The studio had roughly $5M of remaining cash. Butterfield's stated plan to his investors was to return that cash. Most boards in his shoes would have agreed.
Instead, the small remaining team noticed something the players hadn't: the internal communication tool the company had built to make Glitch wasn't just useful — it was vastly better than the off-the-shelf options. Butterfield's pitch to the team and the board was simple: ship the tool. What followed is one of the most-cited pivots in modern tech.
// the decisions
1. Shut down Glitch and return the cash, or pivot
Late 2012. Player retention is poor and shows no sign of reversing. The team is staring at $5M of cash and a clear ethical norm: when you can't deliver what you raised the money for, you give it back. The internal chat tool is functional but unrefined; nobody outside Tiny Speck has ever seen it.
options on the table
- A.Wind Glitch down and return the remaining capital to investors.
- B.Continue iterating on Glitch with reduced scope and a smaller team.
- C.Pivot the company onto the internal communication tool with the existing team and remaining capital.
what they actually did
Butterfield called each major investor individually and laid out the pivot honestly: this is the tool, here's why we think it's better than what's out there, here's the plan, you can ask for the money back instead. Every investor agreed to keep the capital deployed against the new direction.
consequence
The team retooled around the messaging product. Slack launched in beta in August 2013, hit 8,000 companies on its first paid day in February 2014, and reached a $1.1B valuation within a year. By the time of the IPO via direct listing in 2019, Slack was used by majority-of-Fortune-100 organisations. Salesforce later acquired it for $27.7B in 2021.
lesson
A pivot is a decision about which hypothesis the company is testing — not just a new feature direction. Butterfield's transparency with investors was as important as the product insight. Pivots that hide the cash question fail; pivots that surface it ask 'do you still believe in us when we're testing a different hypothesis?' That's a question the strongest investors actually want to be asked.
2. How to launch something nobody had asked for
By mid-2013 Slack was usable internally and at a handful of friends-of-the-team companies (Cozy, Rdio, others). Conventional B2B SaaS marketing wisdom said start with a private alpha, then a controlled beta, then a public launch tied to a press push. Slack's product was almost the opposite of legible: it was a chat app in a market full of chat apps.
options on the table
- A.Stealth → enterprise sales motion: lock down the product, build a sales team, sell to CIOs.
- B.Classic developer-first launch: open up to teams that already used IRC and let them migrate.
- C.Preview launch: invite teams who heard of it, expand quietly via word-of-mouth, focus on first-week retention not signups.
what they actually did
Tiny Speck went with the third option. The 'preview release' lasted six months. There was no top-of-funnel marketing. The team measured one thing obsessively: whether teams who tried it kept using it after week one. They iterated on the product based on what made teams stick, then announced general availability when their internal retention curves convinced them the product was actually working.
consequence
On launch day in February 2014, Slack hit 8,000 paid teams. The waitlist had been organic word-of-mouth from companies that had been using it during the preview. Slack publicly noted that they had effectively no marketing spend during the preview — the spending happened on product, not acquisition.
lesson
When you're in a category that already exists, the question isn't 'how do I get users?' — it's 'how do I know the product is working before I scale?' Slack's preview-release model is the canonical answer: optimise for first-week retention before optimising for acquisition. If you can't get the first hundred teams to keep using it, paying to acquire ten thousand more won't save you.
3. The decision to be opinionated about integrations
Early 2014. Customers are asking for ways to plug Slack into the rest of their stack — GitHub, Trello, Twitter, internal monitoring. Tiny Speck has shipped a handful of inline integrations. The team can either keep building them in-house, build a flexible API and let third parties build their own, or treat integrations as marketing more than product.
options on the table
- A.Stay focused on core messaging; ignore integrations until later.
- B.Build official integrations only; control quality + brand consistency.
- C.Open the platform: ship a public API, encourage third parties to build, accept the messy resulting ecosystem.
what they actually did
Slack opened the platform. They invested in clear documentation, a 'Slack App Directory', and made Incoming Webhooks one of the simplest APIs in B2B SaaS. They explicitly chose breadth over polish: the directory grew quickly, including many low-quality apps, but enterprise customers started saying 'Slack is where our work happens' because every tool they used could pipe into it.
consequence
By 2018, Slack reported that 94% of paid users actively used at least one app. The platform became one of the most-discussed durability moats in B2B SaaS — a lock-in effect that wasn't in the core product but in the ecosystem around it. Microsoft Teams' eventual challenge was forced to compete on platform breadth, not just messaging features.
lesson
Platform decisions look like feature decisions but compound differently. Each integration adds a small immediate utility and a large future switching-cost. If your product can be a platform, the question isn't 'should we build integrations?' but 'should we be an ecosystem or just a product?' The answer determines the ceiling.
// what to take away
- 01A pivot is honest with investors before it's exciting in product reviews. The conversation Butterfield had about returning the cash made everything that followed possible.
- 02Stewart's prior Flickr exit gave him both pattern recognition (Flickr was itself a pivot, originally an internal tool inside another failing game) AND the social capital to do this honestly. Most pivot folklore underrates the founder's existing trust capital.
- 03Slack's launch is the canonical 'optimise retention before acquisition' case study. If your week-1 retention is bad, every new user makes the company worse.
- 04Platform bets compound. Slack's app directory wasn't a feature; it was the moat that pushed competitors into multi-quarter response cycles.
- 05Do not over-romanticise pivots. The vast majority of pivots fail, and the ones that succeed almost always have at least one thing already working before the team changes direction. Slack had a product that worked internally for three years before launch.
// timeline
- 2009Tiny Speck founded by Butterfield, Cal Henderson, Eric Costello, Serguei Mourachov.
- Sep 2011Glitch enters open beta.
- Nov 2011Glitch is taken back to closed beta to address retention issues.
- Sep 2012Glitch relaunches publicly.
- Nov 2012Tiny Speck announces Glitch will shut down. Most of the team is laid off.
- Aug 2013Slack enters preview release with friends-and-family teams.
- Feb 2014Slack launches publicly. 8,000 paid teams sign up on day one.
- Oct 2014Slack reaches a $1.1B valuation in its Series D — "unicorn" status in 8 months.
- Jun 2019Slack lists directly on the NYSE at a ~$23B reference price.
- Jul 2021Salesforce closes acquisition of Slack at $27.7B.
// sources
- We Don't Sell Saddles Here (internal Slack memo) — Stewart Butterfield, Tiny Speck / Slack, 2014
- Slack's Brilliant Strategy: Why Stewart Butterfield's Big Bet on Workplace Chat is Paying Off — First Round Review, 2015
- Glitch Is Closing — Tiny Speck (announcement, archived), 2012
- Slack Form S-1 Registration Statement — U.S. Securities and Exchange Commission, 2019
- Salesforce Completes Acquisition of Slack — Salesforce Press Release, 2021
Practice this kind of decision
The simulator runs scenarios that exercise these same lessons under time pressure. Pick a chapter that exercises discovery + stakeholder.