ScrapingBee: Two Failures, $5M ARR, an 8-Figure All-Cash Exit — the Complete “By-the-Book” Journey

The numbers across the full journey

Prehistory: two failed products

Pierre de Wulf and Kevin Sahin were high-school classmates in the south of France. Their first product, ShopToList (December 2017 to May 2018), was a browser extension that alerted users to price drops. The Product Hunt launch went well, but they could not build monetization or an acquisition channel that scaled, and they handed it off to a web agency. The second, PricingBot (July 2018 to April 2019), was a price-monitoring tool for e-commerce merchants. They shipped it after a two-month beta and user interviews, but with no feel for the e-commerce industry, the two could not even tell where their customers gathered.

The turning point came while building PricingBot. They noticed that the tools around the scrapers they wrote every day were “hard to use, slow, and unstable,” and in April 2019 they pivoted from a borrowed problem to one they had themselves. Their own summary: “We built a solution first and then looked for a problem that fit it.” On the third attempt, for the first time, the builders and the customers were the same kind of people.

The business in brief

ScrapingBee is a web scraping API that handles proxy management and browser rendering for you. Founders Pierre de Wulf and Kevin Sahin grew it via the classic content → developers → API billing route: writing a large volume of technical blog posts for developers (scraping how-tos) to build search traffic, then converting those readers directly into API customers. The customer base exceeds 2,500 companies worldwide.

Our editorial take

The placement of the interim goal, “$10K MRR = previous salary + profitability”, was masterful. Rather than shooting for $1M from the start, they set “enough for two people to live on” as the first goal. Once reached, the business could continue indefinitely, and all growth beyond that point became a bonus. This is best practice in goal-setting for indie developers.

The all-cash, eight-figure sale is the polar opposite of earnout hell (Tweet Hunter). Precisely because it was a profitable company with no reason to sell in a hurry, they could negotiate without compromising on terms. The strongest negotiating chip is “being in a position where you don’t have to sell.”

“Two failures → finishing the race on the third try” is a replay of a recurring pattern. Bannerbear’s seven straight failures, Irie’s 29 straight failures, the same story, differing only in the number of failures.

Implementing “content → developers → API billing”

ScrapingBee’s acquisition stood almost entirely on one leg: relentlessly writing technical articles on web scraping (Python scraping how-tos, anti-bot countermeasures, using headless browsers, and so on). People searching in this domain are “developers who are writing scraping code right now and stuck”, and if, just beyond the article’s solution, there sits “an API that takes all this hassle off your hands,” they convert into customers with zero ad spend.

The important point is that the two founders chose this strategy not because it was cheap but because they could write it themselves. Technical blogging can’t be outsourced without losing quality, so for founders who can write, it becomes a barrier to entry. Conversely, the moment a small team centers its acquisition on a channel its founders cannot execute themselves, it gets dragged into a capital fight.

What it meant that the buyer was Oxylabs

The acquirer, Oxylabs, is the largest group in proxy and scraping infrastructure. ScrapingBee’s customers (developers and SMBs) complement Oxylabs’ main battlefield (enterprise), making this an acquisition that broadens the customer base while keeping the brand intact. An infrastructure company buying an API company follows the same “adjacent slot in the same wallet” logic as Semrush buying Prowly.

The timing is telling too: in the generative AI era, demand for “pulling data from the web” is structurally growing, and ScrapingBee sold mid-growth, to a well-matched buyer, not at the peak of demand, which is why they could command such strong terms: eight figures, all cash.

The cease-and-desist arrived just before the exit

What the earlier sections do not cover is when the legal risk surfaced. In a gray-zone business like scraping, it did not show up in ordinary times. It showed up right after the founders hired an M&A advisor, the most expensive moment possible. ScrapingBee held on because of two things: a court win by a competitor in a similar case, which was outside their control, and a fallback they had measured in advance, namely that stopping scraping of that one site would cost only 10 to 15% of revenue. Only the second was something they could prepare themselves.

The growth side also carries an expiry date. Pierre admits the SEO worked best in “the first two years,” so crediting the whole road to $5M ARR to the technical blog does not match the founders’ own account. Our reading is that the all-cash, eight-figure price paid less for the flagship articles than for the years the company kept running profitably after its acquisition playbook had saturated.

Related cases

• Bannerbear — the same “one win after the failures”
• Tweet Hunter — a lesson in contrasting sale terms