The story of the company that almost died is told two ways. There is the version the founder tells on a stage years later, where the near-death moment is a forge, the dark night that made everything after it possible, narrated with the calm of someone who knows it worked out. Then there is the version written down at the time, or close to it, by the people who did not yet know. The second version is the useful one, because the calm is gone and the fear is still legible, and what you can read in the fear is the actual decision that saved the company rather than the myth assembled afterward to explain it. There are enough of these first-hand accounts now to read them against each other, and they share a structure the stage version omits.
The structure is not “they believed and pushed through.” Belief is what every dead startup also had. The survivors did three specific things, and the cases that document the near-death moment honestly all show the same three.
Five companies that came close
The cleanest near-death account in startup writing is Slack’s, and it is clean because the company that almost died was a different company. Tiny Speck, the studio behind the online game Glitch, ran out of road in late 2012. In a first-hand account from inside the company, an early employee recounts the morning founder Stewart Butterfield walked him along the Vancouver seawall and said, “We have to shut down the game.” Butterfield’s own arithmetic, as Ben Horowitz later retold it, was brutal: after raising more than $15 million, Glitch had $6 million left, no clear path to raise more, and would cost more than $6 million to finish. Glitch was Flash-based, which Steve Jobs had effectively killed for the iPhone, and players “would finish the game in two days,” which destroyed retention. The company that emerged from that shutdown, built on the internal chat tool Tiny Speck had glued together for itself, became Slack, which Salesforce acquired in July 2021 for $27.7 billion.
Airbnb’s near-death is the one that became folklore, and the folklore is mostly accurate. In 2008, per a contemporaneous account, Brian Chesky, Joe Gebbia, and Nathan Blecharczyk had maxed out their credit cards and were carrying debt while the core rental business made almost nothing. To survive, they repackaged cereal into limited-edition Obama O’s and Cap’n McCain’s boxes timed to the 2008 election, building a thousand boxes by hand with a hot glue gun. The cereal netted roughly $20,000 to $30,000, several times what Airbnb itself was earning, and the story got them into Y Combinator after their actual pitch failed to impress Paul Graham, who funded them because he was “looking for cockroach startups.” The near-death move was not the cereal. It was that the founders treated running out of money as a problem to out-hustle rather than a verdict.
Buffer’s near-death came from the opposite direction: success metabolized too fast. In June 2016, the social-media tool laid off 10 people, 11 percent of its team. Per The Next Web, CEO Joel Gascoigne disclosed in a long, public post that the company had over-hired in an effort to accelerate growth and was “rapidly trending” toward running out of cash. The layoffs saved about $585,000, and Gascoigne and co-founder Leo Widrich took a 40 percent personal salary cut through the end of the year, with the combined measures saving roughly $1.5 million. The near-death here was self-inflicted, the result of spending against a growth curve that the revenue did not support, and the survival move was to name the mistake publicly and cut fast.
Groove, the customer-support software company, documented its near-death in real time. Founder Alex Turnbull’s account opens with him telling a mentor, “We’re dying. If we don’t turn things around fast, there won’t be a Groove.” The company had spent more than six months building a product nobody wanted, “the product that we assumed people wanted to use, rather than the one that our market told us they wanted,” then burned most of its remaining cash and its email list rebuilding it. The relaunch landed to “crickets.” What pulled Groove out was a wholesale move into content marketing and, underneath it, a return to customer-development calls with anyone who would talk.
ConvertKit’s near-death, covered in its own case study, is the slow-motion version. Twenty-two months after launch the product had not just stalled but reversed, bottoming at $1,207 in monthly recurring revenue in October 2014, with advisor Hiten Shah telling founder Nathan Barry to shut it down. Barry instead put in $50,000 of his own savings and killed his profitable $250,000-a-year course business to force his full attention onto the failing one.
The three moves of survivors
Across these accounts the same three behaviors recur, and they are the opposite of the “believe harder” myth.
The first is that survivors looked at reality faster than their pride wanted them to. Butterfield did not pray for rain on Glitch, the option he explicitly named and rejected; he counted the $6 million, concluded the game could not be finished for it, and acted while there was still money to act with. Gascoigne did not quietly hope Buffer’s revenue would catch up to its headcount; he named the over-hire and cut. The dead version of each story is the one where the founder waits one more quarter for the numbers to turn, and runs the cash to zero before admitting what the data already said.
The second is that they cut to a core rather than adding to a sprawl. Tiny Speck’s survival came from killing the entire game and keeping one internal tool. Groove’s came from abandoning a product built on assumption and rebuilding narrowly on what customers actually asked for. ConvertKit’s turnaround began when Barry killed his own profitable side business to remove every distraction from the one that was failing. The instinct under pressure is to add features, add markets, add reasons to hope. The survivors subtracted.
The third is that they paid for survival with founder money or founder pain before asking anyone else to. Airbnb’s founders ate their own debt and built cereal boxes by hand. Buffer’s founders took a 40 percent salary cut. Barry wrote a $50,000 personal check into a company his own advisor had told him to close. In each case the founders absorbed the cost personally before the company could ask the market for relief, which is also what made the eventual recovery legible to everyone watching: the people in charge had skin in the survival, not just the upside.
The dead version of each story is the one where the founder waits one more quarter for the numbers to turn, and runs the cash to zero before admitting what the data already said.
The “fail forward” myth against the record
The genre has a slogan, “fail forward,” that the records do not support in the form it is usually meant. The slogan implies that failure is generative on its own, that almost dying makes you stronger. The accounts show something narrower and less inspiring: almost dying is generative only when it is paired with a specific, painful change of direction made in time. Glitch’s failure did not produce Slack. Butterfield’s decision to shut the game and ship the tool produced Slack, and he made it with $6 million still in the bank, not zero. The failure was not the asset. The clear-eyed reaction to it was.
Horowitz draws the sharper version of the lesson from Slack’s case. The reason Butterfield had a viable product to pivot to, he argues, is that “because he was pulling his hair out” trying to ship Glitch, “he learned where software development was suboptimal” and built tools to fix it. “You have to do something really hard if you want to learn something about the world that nobody else knows or is acting on.” The near-death moment was not the teacher. The years of hard work that preceded it were, and the crisis only forced him to notice what the work had already taught him.
A pre-launch checklist, drawn from the cases
The cases invert cleanly into things to check before the launch that might be the one. Know your zero date, the day you run out of cash at current burn, precisely enough to act months ahead of it rather than weeks, because every survivor moved early and every cautionary tale moved late. Build what the market asked for, not what you assumed, which is the specific mistake Groove names and the specific discipline that saved it. Keep a core you can retreat to, a single thing that works that you could ship if the main bet fails, the way Tiny Speck had its internal tool. And decide in advance how much founder money and founder pain you will personally put in before you ask anyone else, because the founders who survived all paid that price first.
The thing the honest accounts share is that the near-death moment did not feel like a forge while it was happening. It felt like the end. Turnbull hated Fridays because Friday was when the metrics arrived to confirm the death spiral. Butterfield looked, by his employee’s account, exhausted and resolved on the seawall. The stage version sands all of that down into a triumphant arc. The version written at the time records a founder doing arithmetic they did not want to do, and acting on it while there was still money left to act.