Kevin Sahin and Pierre de Wulf spent around 30 hours writing a single blog post, then did it again, for years. That was the acquisition strategy. Their web scraping API, ScrapingBee, went from launch in June 2019 to $1 million in annual recurring revenue in November 2021, then to $5 million ARR and an eight-figure, all-cash acquisition by Oxylabs' group announced in June 2025, with a team of four. No paid ads. The playbook is copyable if your customers are people who search for how to solve a technical problem. Here it is as they actually ran it, with the numbers from their own account.
Step 1: fail toward the niche you can teach
ScrapingBee was their third product in 18 months. ShopToList, a price-tracking extension, got about 1,000 users from a Reddit post and no revenue; they sold it to a web agency in May 2018. PricingBot, a B2B price monitor, converted about 1 percent of trials and was abandoned in April 2019. Both products, whatever their faults, ran on web scraping. When they picked the third idea, they picked the layer underneath the failures: the scraping infrastructure itself, sold to developers. Sahin had already written a book on web scraping. They chose the one niche where they could out-teach everyone.
Step 2: launch to people who already know you
The MVP took a month. Their first paying customer arrived 50 minutes after the launch email went out, per their journey post, because two failed products had left them a small list of people who trusted them. (One avoidable mistake taxed the launch: they named it ScrapingNinja without checking trademarks, and a French company's claim forced a rename and legal fees within a month, as Pierre detailed in a January 2020 Failory interview.) By that interview, seven months in, revenue was $2,900 a month. See also our review of where first customers actually come from: an existing list, however small, beats every cold channel on day one.
Step 3: write the definitive answer, not "content"
In August 2019, two months after launch, they published "Web scraping without getting blocked." It brought 20,000 visitors. That post was not a 700-word listicle; their standard was roughly 30 hours of work per article, written by people who scrape for a living, and they repeated the formula across every language their buyers use: Python, JavaScript, Ruby, PHP, C#. The logic is unglamorous: a developer with a scraping problem googles the exact problem, reads the best answer, and the best answer happens to be published by a company selling the shortcut. They also could not afford the alternative: clicks on "web scraping" keywords were too expensive to buy, which is precisely the signal that they are worth ranking for organically.
Step 4: pay users for conversations, in product credits
While content compounded, they ran about 100 customer interviews in three months, paying 10,000 API credits for each 15-minute call. Credits cost them close to nothing and filtered for genuine users. The interviews shaped the roadmap and the docs, which they shipped with code snippets in seven languages. Money spent: zero. This is a generally available trick for any product with a metered unit: your marginal cost and your customer's perceived value are far enough apart that product credit buys research, reviews and referrals cheaper than cash ever could.
Step 5: one channel until it works
They credit the book Traction for the discipline of testing channels and then committing to one. Early on they scattered across Product Hunt, forums, and communities; from late 2019 they went all in on SEO content. The receipts, from their journey post: $10,000 MRR in October 2020, $20,000 in January 2021, a million organic pageviews a month by March 2021, $1 million ARR that November. They joined TinySeed in May 2020, whose founders later estimated the backing saved them about two years, and made their first hire in June 2021, two years after launch.
Step 6: industrialize what worked, ignore the rest
Scaling, described on Startups for the Rest of Us in July 2025, meant hiring developers who could write, plus an editor and an SEO specialist, producing three to four posts a month at around 4,000 words each. The hiring order matters: they industrialized the proven channel before touching anything else, and the writers were developers first, because the content's credibility was the moat. Nothing else about the company grew. Four people took it to $5 million ARR, absorbed a cease-and-desist from a large tech company along the way (resolved on the strength of precedent from a competitor's case), and went through a three-month due diligence to an all-cash exit. Reported figures put the price around $12 million; the founders confirm only "eight figures."
What to copy, and the 2026 caveat
Copy the shape: a niche you can genuinely teach, one definitive article per real problem, 20 to 30 hours each, in every framework your buyer uses, and no second channel until the first one compounds. Budget two years to the first $10,000 a month; that was the cost even for founders with a book and a mailing list. The honest caveat is that informational search traffic is worth less now than in 2019: AI answer engines are absorbing clicks, a shift we measured in our look at AI search click loss. The playbook still works where the query is specific and the answer needs working code, but assume thinner yields and make the content the start of a relationship (docs, tools, email), not the whole funnel. Distribution advantages decay; owning the definitive answer decays slowest. That was always the real asset, and it is why, as we argued in distribution is the moat, the acquirer paid eight figures for a four-person company.
Discussion
Sign in with Google or just a name. No email link, no password to remember.