Sahil Lavingia called Gumroad a failure, then kept paying creators

The story

On 7 February 2019, Sahil Lavingia published an essay whose title is the argument: he had failed to build a billion-dollar company. The product was Gumroad, a way for creators to sell with a simple link instead of a storefront. He wrote that he had left his job as the second employee at Pinterest in 2011, before any stock vested, after a weekend build and a Monday Hacker News launch that, in his telling, drew more than 52,000 people on day one. He had tweeted, the day before he started building, that he had an idea for his first billion-dollar company. Needless to say, he wrote, that did not happen.

Capital arrived almost immediately. He said he raised $1.1 million from angels and firms including Max Levchin, Chris Sacca, Ron Conway, Naval Ravikant, Collaborative Fund, Accel, and First Round, then $7 million more in May 2012 led by Mike Abbott at Kleiner Perkins. At 19, a solo founder, he described more than $8 million in the bank and three employees. Monthly numbers climbed, then they did not. Venture, he later wrote, was a game of double-or-nothing. They were not doubling fast enough to raise the $15 million-plus Series B they wanted. For the company they thought they were building, every month of less than 20 percent growth should have been a red flag. At the time he thought money in the bank and product-market fit would be enough.

He now dates the stall to a peak in November 2014. Creators loved Gumroad, he said, but there were not enough of them who needed that specific offering. In January 2015, after a last hail-mary, the bank balance dipped below 18 months of runway. He told a 20-person team the next nine months would be hard. They deprioritized everything except features that might move monthly processed volume: Small Product Lab, weekly payouts, debit-card payouts, payouts to the United Kingdom, Australia, and Canada, email additions, recommendations and search, consumption analytics, add-to-cart. In those nine months, he wrote, not a single person left. They still did not hit the numbers they needed.

The public version of the failure arrived as a TechCrunch headline about layoffs and a restructure. He had already laid off 75 percent of the company, including close friends, and had hoped the product would keep growing without anyone far from the company noticing. Customers who relied on Gumroad to get paid asked whether they should leave. Some favorite creators did. He does not blame them. Looking back, he is glad they missed the Series B bar. Raising again and appearing in headlines, he wrote, could have produced a more spectacular collapse.

Investors, in his account, were split on what to do next. Some wanted him to shut down, return remaining capital, and try another billion-dollar company with their money. He tended to agree, then ranked his duties: creators, employees, investors, in that order. About $2,500,000 a month was going into creators' pockets for rent, mortgages, and school, and it was still growing. An acquihire would have meant losing control of the product. He chose to become profitable at any cost. The next year, he wrote, was not fun. Headcount went from twenty to five. They struggled to re-let a $25,000-a-month office. Remaining work went into a premium service.

The monthly snapshot he published is the part later copy likes to flatten into a redemption arc. In June 2015, a few months before the layoffs, he listed revenue of $89,000, gross profit of $17,000, operating expenses of $364,000, and net profit of minus $351,000, all for the month. A year later, June 2016: revenue $176,000, gross profit $42,000, operating expenses $32,000, net profit plus $10,000. Those are his books for two months, not audited annual statements. The second set meant creators would keep getting paid and the company would control its destiny. It also meant Gumroad was no longer the venture-funded rocket investors and employees had signed up for.

The skeleton crew then fizzled from five to one. He was alone, without an office, in a city full of friends raising money. He did not tell his mother about the layoffs; she read the article. He left San Francisco for long stretches and felt more lonely. Each day he answered support, fixed what bugs he could, and often asked former engineers, already employed elsewhere, for help. Happiness, he wrote, is an expectation of positive change. 2016 was the first year that felt worse than the last. After the 2016 U.S. election he left San Francisco for good.

The ownership math changed on 27 November 2017, in an email from Kleiner Perkins. The firm wanted to sell its stake back to Gumroad for $1 after Mike Abbott left and the partnership did not want the board headache. Liquidation preferences, he wrote, went from about $16.5 million to $2.5 million. One other investor joined. He later bought a couple more back. Suddenly there was a path to an independent business not beholden to go-big-or-go-home. He still updates remaining investors by brief email. Gumroad would never be a billion-dollar company, and that started to feel acceptable.

He is blunt about impact versus capture. Gumroad, Inc. may be small, he wrote in 2019, while the company had sent $178 million to creators and turned $10 million of investor capital into that flow. He named Austen Allred, who had raised $48 million for Lambda School after selling a book on Gumroad, as an example of value created off-platform. In April 2018 he began publishing monthly financials because they were profitable and a flat month would not kill the story. One tweet he quoted: April GMV $4.2 million, revenue $273,000, both down 2 percent, gross profit $65,000, up 34 percent. In 2018 they donated more than $23,775, eight percent of profits, to causes he listed. Those figures are founder-reported snapshots.

A later first-person essay, dated 7 January 2021, describes the operating system after the venture costume came off. He said Gumroad peaked at 23 full-time employees in 2015 and was a one-person company in 2016. Asked how many people work there, he answers "ten or so," then clarifies: about 25 if you count everyone, and zero full-time employees, including himself. No meetings, no deadlines. Creators, he wrote then, earn over $175 million a year; Gumroad generates $11 million in annualized revenue, growing 85 percent year over year. Memberships, shipped November 2020, were used by hundreds of creators to earn over $1,500,000 a month. Those later lines sit beside the 2019 confession, not on top of it. He failed at the metric he inherited. He kept the faucet on.