Loading page…
He bought a GitLab engineer's hobby project for $125K cash, turned on chat support, and built every feature a teacher asked for — $5K/mo to $80K/mo in a year
up from $5K/mo at acquisition · bought February 2024 for $125K all cash at about 2x revenue · ~20,000 monthly searches for the category term
His acquisition filter was not profit. “What would happen if this business is turned off tomorrow? Are there at least 100 people that are going to lose their minds?” — that question is why he paid above ask for a half-built product with traction instead of a polished one without it.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
Chanakya Yerneni — Chan — spent about a decade in corporate product and product marketing roles, most recently as a senior product marketing manager at OpenText. He wanted out, had a spreadsheet full of business ideas, and no idea where to start with any of them. So he skipped the starting part. He registered on Acquire.com, the marketplace formerly called MicroAcquire, and spent roughly 90 days going through listings, eventually looking at 25 to 30 businesses. He had just under $200,000 of savings he was willing to gamble, on the reasoning that losing it would not change his lifestyle. He looked at e-commerce and walked away — the expertise required is supply chain and he did not have it. SaaS felt native because that is where his career had been. Then he found EssayGrader.ai, listed by an engineer who worked at GitLab and had built it as a side project. It graded student essays with AI, for teachers. Chan had been a teacher in a previous life, and he knew exactly what he was looking at: the people who take grading home on weekends, skip dinner, order pizza for the family. He bought it in February 2024 for $125,000, all cash, at about 2x revenue — slightly more than the seller asked — when it was doing $5,000 a month. From seeing the listing to running the business was about three weeks. Then he did one thing: he turned on chat support, roughly 90 to 120 days after taking over, and talked to three or four hundred teachers himself until he nearly burned out. Every feature built since came from one of those conversations. A year later the business was doing about $80,000 a month, roughly a million dollars in recurring revenue. He had privately expected $30,000 or $40,000. He is 37 and says he should have done this at 28.
EssayGrader arrived with customers already attached, and that was the entire point of the deal. A GitLab engineer built it as a hobby project and it was doing $5,000 a month when it reached Acquire.com.
The quick-reference card — acquisition channels, replication playbook, risks & traps, and our verdict — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself. The exact words are quoted below, with a timestamp that jumps to the second each figure is said.
“Imagine acquiring a business, making five grand a month, and then scaling it to $80,000 a month in just one year.”
1:02 · hear it in the video“He was making $5,000 a month up to him making $80,000 a month.”
2:40 · hear it in the videoData credibility: Podcast interview with the buyer-operator on Jaryd Krause's Buying Online Businesses; every number is self-reported and unaudited. The purchase price ($125,000, all cash, about 2x revenue, February 2024), the starting revenue ($5,000/mo), the current revenue (about $80,000/mo, which the host first misstates as $800,000 and corrects on air) and the search volume (about 20,000 monthly searches for the category term) all come from Chan alone. The claim of more than 100,000 educators and ten-plus hours saved per month is marketing copy read out by the host from the company's own positioning, not something Chan substantiates — we have not treated it as a verified figure. He hedges his own recall on the deal timeline (“the details are a bit fuzzy in my head”) and on whether the transition period was 30 or 60 days. No pricing is disclosed anywhere in the interview, so we state none. Nothing here is third-party verified, and the interview format gives him a clear promotional incentive.
The first dollar came with the keys. Chan paid $125,000 in cash for a business already billing $5,000 a month, which means revenue started on day one and the real question was whether the price was right.
The turning point was chat support, switched on around May or June of 2024, three to four months after the acquisition. Chan did it himself and estimates he talked to three or four hundred teachers over the following year, until he was close to burning out on it.
For a buyer, the stretch with no feedback and no growth happens before the purchase, not after.
An AI grading platform aimed at teachers. Submissions go in, grades and feedback come out, and the parts teachers actually argue about are the feedback report and the grammar report — which elements appear in them, in what order, and how defensible the output is when a student contests a mark. The host reads marketing copy describing it as the number one AI grading platform trusted by more than 100,000 educators to save ten-plus hours a month; treat that as a claim from the company's own site, not a verified number. Chan is candid about what he bought: not a finished product. In his words it did not have a fully developed product, but it had traction, which he considered the more important of the two. Our read: this is a thin technical product sitting on a thick workflow problem, and that is the correct shape for a first acquisition. The model does the grading, which anyone can replicate in a weekend. The defensibility is in knowing that a high school English teacher grades on Sunday afternoons, marks differently from a history teacher, and wants the grammar report structured a particular way — knowledge that took three to four hundred chat conversations to accumulate and that a competitor cannot buy off the shelf. The category name it owns in search is the second asset, and it was already in place when he bought it.
Recurring subscription revenue, sold by the licence. Chan never states a price point anywhere in the conversation, so do not assume one — the only pricing unit he discloses is the licence, and the only concrete expansion example is a single user turning into twenty licences when they brought their department across. The headline numbers are $5,000 a month at acquisition in February 2024 and about $80,000 a month a year later, which he and the host both round to a million dollars in annual recurring revenue. The host briefly says $800,000 a month and corrects himself on air, so the figure to hold is $80,000. Our read: the revenue shape here is seat expansion inside institutions rather than new-logo acquisition, and that is why the chat-support investment paid so disproportionately. A consumer SaaS would have got a support ticket resolved; an education SaaS got a department head who then pulled in twenty colleagues. It also explains why he was willing to spend a full week of engineering on a single user's request. In a seat-expansion business, the correct unit of analysis is not the user in front of you but the building they work in. Sixteen times growth in twelve months with no described paid acquisition is consistent with that mechanic, though he also freely credits luck and timing on the category.
Chan's process is reproducible, and he narrates it in enough detail to copy. Register on Acquire.com and spend real time — he gave it 90 days before he understood his own criteria, and reviewed 25 to 30 businesses. Set a budget you can lose; his was just under $200,000 and he framed it as money that would not change his life if it vanished. Buy in a domain you can operate, not just one you find interesting: he skipped e-commerce because the core skill is supply chain and his was software. Screen for traction, not product maturity. His clearest articulation is that a first-time buyer with a small budget should not be looking for profitability but for product-market fit — would a hundred people lose their minds if this shut down tomorrow. Verify traction with tooling rather than the seller's deck: he names Ahrefs and Semrush for traffic, keyword position and, now, visibility in AI answers. Red flags he names: revenue concentration (that $700,000 business with 70% from one client, priced at a six-month multiple, which he reads as the seller knowing something), a mature product with no traction and a founder who cannot say who it was built for, a stagnant category like generic invoicing, and — counterintuitively — no competitors at all, which usually means no market rather than a discovery. The host adds the inverse framing that low multiples signal high risk, and that software commands high multiples precisely because recurring revenue and switching costs make it less risky. Chan paid cash and slightly above ask, took a 30 to 60 day transition, and kept the seller on friendly terms afterwards.
Copy the sequencing above all. He bought revenue and then bought information, in that order, and the information came from a chat widget rather than a research budget. If you already have a product with users and no growth, the copyable move takes a week: turn on chat, answer it yourself, and commit to building what the first hundred conversations ask for. Copy the screening criteria, which are unusually well formed — traction over polish, no single-client concentration, a category with competitors in it, and third-party traffic verification before you trust a data room. Copy the willingness to pay slightly over ask when conviction is high; the difference between a 2x and a 2.2x multiple on a $60,000-a-year business is small change against a year of wrong-asset regret. Change two things. First, the capital: $125,000 in cash is the single hardest part of this story to replicate, and the same playbook runs at $10,000 to $30,000 on smaller listings, where the seller pool is larger and the competition thinner. Second, plan the support handoff earlier than he did — he describes exhausting himself in chat, which is a founder bottleneck dressed up as a growth strategy. Our read: the most transferable insight is not about acquisitions at all. It is that a product bought for its distribution can be rebuilt entirely around what its existing users say, and that rebuild is cheaper and faster than acquiring new users would have been.
The category Chan lucked into is now the category everyone wants. He says so himself: competitors have arrived and at least one team went through Y Combinator with the same idea and funding behind it. The 20,000 monthly searches that were a moat in early 2024 are an auction now. Underneath that, the discovery channel itself is shifting — teachers asking an AI assistant which grading tool to use is a different game from ranking for a keyword, and the inherited SEO asset depreciates in that world. Platform risk is real in both directions: the learning management systems that teachers already live in, and the foundation model providers whose output the product resells, are both a product decision away from making a standalone grader unnecessary. Education adds its own layer. Student work is student data, so district procurement, privacy review and parental consent become gating items the moment deals move from individual teachers to institutions — exactly the direction the seat-expansion model pushes. Policy risk sits alongside it: a school board deciding AI may not grade student work removes a whole district at once, regardless of product quality. Operationally the business is one person's attention. The growth engine was Chan personally in chat, and he describes nearly burning out doing it. Our read: the thing most likely to cap this is not a competitor but seasonality plus procurement — grading demand is term-shaped, and moving upmarket to smooth it means selling to districts, which is a different company than the one he bought.
The four founder playbooks — first customers, first dollar, the flywheel, the silent stretch — are in the Unicorn Club, on this idea and on all 1,435 others. The story, the numbers, the sources and every later section stay open to everyone.
Join the Unicorn Club — $5/moCancel anytime, in one click