Oh Dear: $1M ARR, two founders, zero employees

As of September 2025, Oh Dear is above $83K a month and past $1M ARR. Eight years in, there are no employees, no outside capital, and the same two people running it. Converted to yen in this archive’s frontmatter, that is 1.7 times the median of the 97 SaaS cases here that disclose monthly revenue, and 5.6 times the median across all 251 disclosing cases.

For a developer with twenty years of audience behind him, finding buyers ought to be a solved problem. Asked in a 2026 interview what the hardest part of the business has been, Mattias Geniar answered “distribution, hands down,” and added that “even at $1M ARR, distribution is our biggest challenge.” In the same interview, his single biggest regret was every dollar they put into paid advertising.

They sold to a 500-person list without quitting their jobs

Geniar was a sysadmin at a managed hosting provider and Van der Herten a partner at a web agency. A client’s SSL certificate expires and you hear about it from the customer rather than your monitoring, or a deployment breaks 200 internal links and nobody notices for weeks. Those two failures were the starting point. Existing tools, Geniar says, “just ping your homepage and tell you it’s up. That’s maybe 10% of what can actually go wrong with a website.” The differentiator in version one was a crawler that follows every link on a site the way a search engine would.

He describes the growth curve himself: “Slow. Very slow for years, then it picked up.” The only checkpoint with independent backing is the 2019 doubling, and nothing between that and the $83K of 2025 has been published.

No free tier, and every feature in every plan

Pricing has ten steps decided by one variable, the number of sites monitored. Solo covers 2 sites at $17 a month, Scale covers 200 at $439, and anything above that is quoted individually. No feature is held back for higher tiers, users are unlimited, and the only way in is a 10-day trial. Annual billing takes two months off.

The reason he gives is cost of goods. “Free users cost money and rarely convert. Every monitor we run uses server resources, bandwidth, and storage. If someone isn’t willing to pay after trying the product, they probably weren’t going to pay at month six either.” Flat features, he argues, also help conversion “because people don’t have to figure out which plan they need.”

Here is how that design should be read. Dropping the free tier was an accounting decision rather than a philosophical one. Variable cost in monitoring scales with checks executed, not with headcount, so a free user starts accruing cost linearly the moment they sign up. From the side running over a million jobs a day, a free tier is not a place where people accept fewer features, it is a place where people use the same servers without an invoice. Tally, also two people in Belgium, gives away 99% of its features and reached $150K MRR in four years, but form building carries a cost per submission that is orders of magnitude lighter.

Three channels that failed, and one that gets the credit

The paid channels have no disclosed spend, but the verdicts are unambiguous.

Asked about regrets, he answered: “The money we spent on paid advertising. Every dollar we put into Google Ads and outsourced content would have been better spent on literally anything else. Conference sponsorships, design improvements, even taking a week off.”

So what did work? His answer is “building in public. Full stop,” and the SaaSRat piece closes on that note. Our reading differs. What brought in those 20 customers and $1,500 on day one was not eight years of building Oh Dear in public. It was the balance accumulated before any of that. He says as much in the same answer: “Freek and I both had around 20 years of online presence before Oh Dear existed. We’d been blogging, tweeting, speaking at conferences, and contributing to open source for decades.”

What paid off was not eight years of building in public but the twenty years of balance behind it.

The distinction matters in practice. Anyone can start working in the open today, but whether that turns into 20 paid signups on launch day depends on how much trust was banked beforehand. That is why his advice to founders without an audience is simply to start building one now. Speculating from here: the real reason paid ads failed was not market saturation either. Measured against an entry route made of recommendations from people who already knew them, ad traffic arrives judging the same product by completely different criteria.

At $1M ARR, distribution is still the hard part

The split between them is clean. Geniar handles support, admin, monitoring-related development and proofs of concept; Van der Herten does core PHP work, refactoring and the crawler. Hiring has been considered for eight years and never done. “Adding employees changes everything. Payroll, management, communication overhead, legal obligations.”

The stack is deliberately dull. Laravel, PHP, MySQL, Redis, ClickHouse for analytics, bare metal for dispatch and about 50 VMs across DigitalOcean, Vultr and AWS for global coverage. “When your entire product is about reliability and uptime, the last thing you want is a bleeding-edge stack that introduces instability.”

The heaviest line lands in the quick-fire round. Asked for his best business decision, Geniar said: “Not going solo. Having a co-founder kept us accountable through the slow years. When I was burned out, Freek pushed forward. When he was burned out, I did. A solo founder in our position would have quit by year three.” The first thing a founder of a $1M ARR business names as the cause of that outcome is neither a feature, a price nor a channel, but the fact that neither of them quit. That is worth recording.

The undisclosed numbers, and where the sources stand

Customer count, churn, profit, actual ad spend and the trajectory from 2019 to 2025 are all unpublished. The $1M ARR figure is self-reported and unaudited. A scale number exists in the form of a million monitoring jobs a day, but how many companies or sites that corresponds to does not.

The sources deserve labels too. The most detailed one, SaaSRat, is a software comparison site publishing on its own blog, and a plug for that site sits inside the interview. The Indie Hackers article is paywalled, so only the figures visible in its free opening are used here. The 2019 doubling is the one data point that comes firsthand from a cofounder’s own blog, and the rest of the path rests on secondary reporting.

How far this carries

Running a developer-facing SaaS with a tiny team covers a wide range in this archive. Instatus reached $48K MRR on status pages alone, Oh Dear sits at $83K a month, and Plausible hit $1M ARR with four people and zero ad spend. All three stayed almost entirely off paid advertising and circulated by name inside developer communities.

• Removing the free tier works in your favour only when variable cost tracks processing volume rather than signups, and the product proves its value inside the trial window
• Building in public converts into day-one revenue only when several years of publishing in that field came first, since the act of publishing is a withdrawal against a balance
• A two-person, zero-employee structure survives eight years only when the domains split without overlap, so that either person can absorb the stretch where the other stalls

The limits deserve stating. Oh Dear stands on conditions that are hard to reproduce. Laravel and PHP form a sharply defined community, one cofounder is a widely known open source developer inside that ecosystem, and monitoring is the kind of tool nobody removes once it is installed. Starting in 2017, before price competition in monitoring reached today’s level, is not a neutral detail either. And with churn undisclosed, there is no way from outside to tell how much of that $83K is being refilled by new business every month.