Gagan Biyani had already co-founded Udemy and already watched a second company, Sprig, shut down. When he went looking for a third idea he gave himself an unusual amount of room. On his own blog he describes taking a full three years off, with almost two years of pure exploration before anything resembling company building started. In that period he learned basic Chinese, worked on his Spanish, took up salsa dancing and tried to write a book. While trying to get better at writing he paid for live online courses from David Perell and Tucker Max, and those classes pointed at something specific: a cohort of people learning together beat a library of recorded video.

He did not build software to test that. He tested the class itself. Writing for First Round Review, Biyani lays out the method he uses, which he calls a minimum viable test. His riskiest assumption was economic rather than technical: would people pay roughly ten times the price of a recorded course for a live one? So he ran exactly one course. He asked Sam Parr of The Hustle to co-teach it, which let him borrow an audience instead of building a marketing machine first. The course scored 9 out of 10 with students and took in over $150,000 in its first cohort. It also taught him that he was personally weak at community building and course design, which is why he went after Wes Kao, co-founder of altMBA, as a co-founder. Shreyans Bhansali joined on the engineering side. Five of these tests ran over nine months before version one of the product shipped.

The timeline anchor comes from Biyani's own posts. On 31 March 2021 he wrote on LinkedIn that the team had been "working on this company for exactly 6 months without a name", which puts the start of the company at the beginning of October 2020. He had gone public on 12 November 2020 with a $4.32M seed round led by First Round Capital. Courses started selling on the platform in January 2021, roughly 92 days after that start, and Maven took a percentage of each sale rather than charging instructors up front. So the first paying customers arrived in month four.

The opening went extremely well. In his 2024 thread Biyani says the launch was "wildly successful", that the phrase they coined, cohort based course, ended up in countless pitch decks, and that Maven was on a $4M GMV run rate within three months. In the First Round essay he puts it in dollars: within four months they had done $1 million in sales. Because Maven earns a share of every course sold, that pace put the company's own annual run rate comfortably past $100,000 about six months after the founding date, at roughly day 182. A month later a16z led a $20M Series A.

Then came the part founders rarely publish. Behind the run rate there was almost no product. Maven started as a Notion doc with the team hand holding instructors in the background, and Biyani says instructor revenue retention sat at 73 percent while many instructors told the team they were not critical to their business. The early money came from creators with hundreds of thousands of followers who treated Maven as their third or fourth priority. Good numbers, wrong customer.

Fixing that took about two years and three deliberate reversals, all of which cost revenue on the way. First, the team moved from chasing follower counts to chasing professional expertise, and lost 95 percent of that first $4M in course sales in the process. Second, Maven stopped being a pure software platform and became a marketplace. Biyani had modelled the business on Substack, where the creator brings the audience, and calls that a failure of "business by analogy". Instructors kept asking for a marketplace and students kept asking why they could not find good courses, and he resisted for about eighteen months because he knew from Udemy and Lyft how hard two sided marketplaces are. Third, they let go of the instructor services team, which was half the company, and spent more than two years replacing that work with software.

The reversals worked. By his June 2024 thread, instructor revenue retention had gone from 73 percent to somewhere between 120 and 140 percent, and Maven was growing in a bad market for education companies. In November 2023 Biyani reported that instructors had crossed $20M in lifetime course sales, with 250,000 users, 40 or more instructors past $100k and over 200 past $10k, run by a team of 15 remote employees. In March 2026 he reported 100,000 paid enrollments, weekly course engagement above 80 percent, and a subject mix that had swung from under 5 percent AI at the start to about 70 percent AI. Roughly 35 percent of sales now come through Maven's own marketplace.

One detail from that last post is the tidiest summary of what changed. Biyani writes that audience is overrated when you have expertise, and that many of Maven's top instructors had fewer than 5,000 followers when they started. That is the exact opposite of the customer the company chased on the way to its first $4M.