NinjaPear: After LinkedIn Sued His $10M ARR Company, the Founder Rebuilt in Two Weeks and Reached $15,000 a Month in Six Months

Proxycurl, the predecessor, was a company with roughly $10M in annual revenue, an 80% profit margin, and a rejected $16M acquisition offer in its past. NinjaPear, the B2B data API its founder Steven Goh relaunched alone in January 2026, was bringing in $15,000 a month six months later. In the yen-normalized frontmatter this site uses, that lands exactly on the median of the 245 cases that disclose monthly revenue. The natural assumption is that a founder who ran a $10M business would restart at a completely different speed. The numbers say otherwise. What remains instead is the record of a man who lost his company to a lawsuit and wrote, “I was always trying to exit the company for fear of a lawsuit, until I actually got sued,” plus two assets he kept in the sale negotiation that are now carrying the next business. Goh disclosed the offer amounts, his post-lawsuit decisions and the relaunch user counts, with dates, in an Indie Hackers interview and on his company blog.

Timeline: from turning down $16M to a sale whose price cannot be spoken

At its peak Proxycurl did about $10M a year, and by the founder’s account “80% was pure profit.” About half of that revenue came from scraping LinkedIn. That single fact sets every price that follows.

The buyers had already discounted before the complaint arrived

The 2023 offer of $16M valued a $3M ARR company at more than five times revenue. He turned it down because revenue was doubling every year, so the same multiple would produce twice the number twelve months later. The 2024 offer of $12M shrank the cash portion to $3.9M and stacked the rest in stock and earnouts. The buyers said outright that LinkedIn dependence was “a very potent risk” and warned that “30% or less of the transaction value” might be paid up front.

This is the first structure to understand in this case. The market had priced LinkedIn risk in long before LinkedIn ever acted. The low cash ratio was not stinginess. It was a design that left the loss with the seller if a lawsuit ever came. The founder kept looking for an exit through all of it, kept rejecting the terms, and in January 2025 the complaint arrived and the options were gone.

He gives two reasons for not fighting. Under the American Rule, legal fees are not recoverable even if you win, and the other side was a Microsoft subsidiary with effectively unlimited funds. He settled, announced the shutdown on July 4, 2025, and sold the business to a long-time competitor for a price he says the NDA prevents him from stating. A year later his one-line summary was “I’ll never sell my company again,” citing regrets over how intellectual property was handled after the sale. Our reading is that this sale was priced not against $10M in revenue at an 80% margin, but against a seller who had just finished a lawsuit and wanted out.

Less the two-week rebuild, more the two things he kept

The restart was fast. While recovering from a fractured metatarsal, he prototyped NinjaPear in two weeks with Claude Code and shipped it. But as readers pointed out in the Indie Hackers comments, the code is not what is doing the work. In the sale negotiation he secured the right to keep a newsletter with 20,000 subscribers and the original domain. When a reader wrote that “the years-old email list is doing the work,” he did not disagree.

The NinjaPear name had already been used once, in 2024, for an AI live chat product that ended with 163 registered users and zero integrations. What relaunched on January 30, 2026 is an API that returns company details, employee counts, company updates, and a list of the customers a company does business with. It takes nothing from behind login walls and does not touch LinkedIn. It assembles results from LLMs and deep research of the public web. Billing is by credits: company details cost 2 credits, employee count 2, a customer listing 1 credit plus 2 per result, and the free trial gives 10 credits for 3 days. In March 2026 he wrote that free tiers would be limited to 2 requests a minute, with higher limits split by subscription plan.

Fifty-four days after launch: 1,036 registered users and $66,000 ARR. By June, revenue for the previous month had passed $15,000. That is roughly 2.7x in annualized revenue over three months, and he describes the business as “now profitable and growing exponentially.”

The subject line: “Your competitor just signed up”

He published the tactic that took registrations from 300 to 3,163 in five months. Each time a new user signs up, an agent uses NinjaPear’s own API to find people in similar roles at competing companies, discovers their work emails, and sends a templated cold email with the subject line “Your competitor just signed up.” He describes it as built on three levers: name-dropping the competitor, triggering curiosity, and fear of missing out.

It is dogfooding in its purest form, running the company’s own sales on the company’s own product, and no cost figures are given. The metric it moved is registrations. Paid conversion is not disclosed. We treat the missing exchange rate between 10x registrations and $15,000 a month as exactly that, a missing number. Since he elsewhere calls SEO “the most effective distribution channel,” the safe reading is that this agent drives sign-ups rather than revenue.

What did not work, and the conditions only this founder has

A Reddit post went nowhere. The main churn driver, in his telling, is speed. Proxycurl returned cached LinkedIn data instantly, and NinjaPear runs real-time LLM research on the web, so users compare the two and feel the wait. Legality was bought at the price of latency, charged on every request, and he admits he is still wrestling with it. Asked for the single biggest growth driver, he answered, “I’m still figuring it out. But there isn’t ONE thing. It’s really a constant iteration of the product, marketing, product messaging, ICP, etc.”

Before any talk of replication, the ground this case stands on needs stating. The founder has more than 20 years as a software engineer, two prior exits, and five companies behind him. He made about $1 million on cryptocurrency and has since sold it, and he says that cushion is what supports his “fail 10x” posture. Six months without rent anxiety and six months with it are not the same length.

What can be carried out of this, and what cannot

In the same “platform rules or litigation broke the premise” category sit GummySearch, which closed while still profitable at $35K MRR because it could not get a commercial Reddit API license, and Kleo, whose free extension LinkedIn shut off and which rebuilt in four weeks and was back to $62,000 MRR within three months. The difference between the one that closed and the ones that rebuilt comes down to whether a distribution asset survived.

• Relaunching in the same market after a lawsuit or rules change works only when the sale or shutdown negotiation leaves distribution assets such as a customer list or domain in the founder’s hands
• “Nothing from behind login walls” becomes a selling point only when customers have their own reasons to avoid legal risk and the use case tolerates latency
• A sales agent that multiplies sign-ups tenfold works only in an industry where the product itself can identify prospects, and only if the founder accepts the resentment of the recipients

The limits need stating too. The $15,000 is gross revenue for one month, with no profit or churn figures. The sale price remains unknown under NDA, and the predecessor’s $10M ARR is self-reported. Above all, this restart happened on top of the money, network and 20,000-person list of someone who had run a $10M company, and writing the code in two weeks is the most reproducible part of it and the smallest contributor.