RemoteOK: $140K/month to $10K to $41K. What a remote job board's wild swings teach about "market dependency"

Note: yen conversions in this article use an approximate rate of ¥150/$1, for a rough sense of scale.

The full picture in numbers

Why this rollercoaster is worth documenting

From $140K a month to $10K, a 93% decline, this is the record of a business that survived without shutting down, and later climbed back to $41K. Pieter Levels, who runs the remote-work job board RemoteOK, has kept publicly disclosing his portfolio’s revenue for years, so this swing isn’t a boast — it’s data left on the record. Let’s break down, from the published figures, the structural market dependence inherent to the “job board” business model, and the organizational design that let it survive anyway.

The trigger for the decline was outside the business itself

RemoteOK has long captured search demand around “remote jobs” and earns from listing fees paid by companies plus advertising. The $140K/month peak was a number lifted by the remote-hiring boom during the pandemic. The source (PPC Land) describes the fall to the bottom as coming “after the Fed stopped printing money.” The chain runs: monetary tightening → tech companies struggling to raise funding → hiring freezes → job listings evaporating, and the 93% drop from $140K to $10K happened regardless of how well or poorly the business itself was run. A job board’s revenue is a dependent variable of the employment market. “What you earn from” is the flip side of “what you’re dependent on,” and a listing-fee model is tied to a variable you have zero control over: other companies’ hiring budgets.

The “zero fixed cost” structure that survived a 93% drop

A drop from $140K to $10K a month would kill an ordinary company. RemoteOK survived because everything from listing to payment to expiration was fully automated, and there were no employees, no office, and no fixed costs beyond server fees. Even with 90% of revenue gone, expenses stayed essentially at zero. A business that never goes into the red can “wait out” any recession.

The source also documents the technical side underpinning that automation. The stack was a deliberately unglamorous PHP + jQuery + SQLite setup, with zero outside funding. Levels developed at a pace of 37,000 commits in 12 months, with deployments taking 1–2 seconds and, in one recorded case, a bug fix shipped to production in two minutes. Choosing technology optimized not for modernity but for “one person being able to touch everything” is the precondition for running with zero employees. By contrast, companies that hired staff in this same market-dependent business had to disband their teams during the winter and re-enter as newcomers when conditions recovered.

Not shutting down is what made the recovery to $41K possible. Because maintenance cost was near zero, he could keep the shop open through the winter and simply receive the market recovery as it came. As with what Hitode and others say about the volatility of search algorithm updates, for a stock asset, low maintenance cost is itself the survival strategy.

RemoteOK within the portfolio

Viewed on its own, this is a wildly volatile bet, but within Levels’s overall picture, it’s just one holding among several. Per the source’s published figures, PhotoAI was doing $132K/month (May 2025), InteriorAI $40K/month, and other projects each ran $15K–$22K. And Levels himself has said his batting average is stark: “I’ve built over 70, and only 4 made money. More than 95% of what I’ve done has failed” (paraphrased).

Laying PhotoAI’s timeline alongside RemoteOK’s makes the contrast clear: experimentation began in October 2022, official launch came in February 2023, it reached $100K/month in about 18 months (September 2024), an appearance on the Lex Fridman podcast in January 2025 gave growth another push, and it hit $132K by May of that year, a curve driven not by tailwinds in the market but by product and Levels’s own visibility. While the job board was sinking under bad market conditions, a revenue stream with a different character was quietly growing.

RemoteOK’s 93% drop gets smoothed out by being combined with community revenue like Nomad List and AI products whose fortunes barely correlate with the job market. A single business’s swings are only tolerable within a portfolio. That’s the conclusion drawn from a decade of published data. Conversely, if his entire livelihood had rested on RemoteOK alone, the $10K winter would likely not have been “a recession you can wait out,” but pressure to shut down.

Discounting the recovery

The current $41K is still only about 30% of peak. And this recovery can’t really be separated into “operational improvements, pricing, features” versus “the employment market simply recovering on its own.” Given the decline is explained by market conditions, it’s natural that the recovery is too, and it’s reasonable to expect another drop the next time monetary policy tightens. A “downstream” business (a job board, an ad-supported media property, an M&A broker, anything whose revenue is really someone else’s economic sentiment) is dangerous to enter based on boom-time numbers alone. This case yields three structural lessons: the moment you build fixed costs around peak-era revenue, you become fragile. Only those who move boom-time surplus into assets not tied to the same market get real diversification. And the level of automation that lets you “hibernate” rather than shut down is the actual condition for survival. RemoteOK satisfied all three, which is why the $41K figure exists today.

Generalizability for Japanese readers

What generalizes is the structural side: a stock asset that’s captured long-term search demand, a near-zero-fixed-cost automated operation, and diversification across multiple businesses. These work the same way regardless of scale, in Japan too. The wild swings from search-algorithm updates hitting domestic blogs (Hitode’s annual income going from ¥30M to ¥100M and back to ¥30M) is the domestic version of the same underlying mechanic.

But there are conditions here that don’t transplant: the sheer scale of the English-language remote-hiring market, and Levels’s own personal reputation, built over a decade of public output. RemoteOK’s traffic is driven partly by search and partly by Levels’s own social media following, and that flow of traffic belongs to the person, not the business. Building the same kind of job board in Japanese from scratch today, both the market size and the accumulated personal following would be off by an order of magnitude. What’s worth referencing isn’t “the job-board business” itself, but the design philosophy of “running a downstream business with zero employees and zero fixed costs.”

Related cases

• Nomad List — the same operator’s “stable side” asset
• Hitode’s income going from ¥30M to ¥100M and back to ¥30M — the domestic-blog version of the same swings