From optician to $28,000/month across three SaaS products: choosing only the battles you can win

Figures are in dollars, with a rough yen equivalent in parentheses converted at 1 USD = 150 JPY.

A business that started from “I can’t code”

Samuel Rondot lived near the Swiss border in France and worked as an optician for three years. He didn’t dislike the job itself, but he felt he couldn’t keep going while physically tied to one place. There was an airport nearby, and he says the sight of it served as a daily reminder pushing him toward change.

He worked on a side business mornings and evenings, and eventually went independent. He now runs three SaaS products in parallel, together totaling roughly $28,000/month (about ¥4.2M/month).

What makes this case interesting is that his very first business wasn’t built with code at all.

The current portfolio

His past track record includes MathPlanner, started in 2017 — an Instagram automation service that reached about $30,000/month (about ¥4.5M).

The technical stack is the same across all three products: frontend on Next.js (Vercel), backend on Node.js (AWS), database on MongoDB. As he puts it, “the tech stack is dead simple, and always the same.” Keeping the configuration unchanged as he adds products is what keeps the load of running them in parallel manageable.

Turning point one — proving demand without writing code

The first turning point was MathPlanner in 2017. It looked like a SaaS product, but there was no code underneath it.

“There wasn’t even any code. Behind a WordPress landing page, real people in India and Bangladesh were doing the automation work by hand.”

Orders came in through the landing page, and the backend was handled manually. This reached $30,000/month and gave him the basis to quit his optician job.

This qualifies as a turning point not because of the amount, but because it factually broke, once, the premise that “I can’t start because I can’t build.” His biggest barrier at the time was, in his own words, simple, “I couldn’t code.” His alternative was a WordPress plugin that “barely worked,” and customizing it was painful. Even so, revenue still materialized.

Before/after: from a salaried optician job to a self-run business grossing $30,000/month. The lever was manually standing in for demand that already existed, not technical skill.

Turning point two — stepping out of a war he couldn’t win

The second turning point was decisive in shaping his current portfolio: rethinking his acquisition channel.

He initially tested paid customer acquisition through Meta ads. What he learned there was the financial reality of his competitors.

“Competitors don’t mind spending $200 to earn $30 in revenue.”

Funded competitors can absorb a loss on a per-unit acquisition cost basis. A self-funded individual who fights on that same ground runs out of money first and disappears. This is where he stepped out of direct competition on paid acquisition.

“Only fight the wars you can win. Most of my competitors are funded, so I need to be more resourceful.”

What he placed there instead was SEO, YouTube videos, and optimization for AI engines. Today, SEO-driven traffic across all products combined is roughly 400 clicks/day. He describes SEO as “slow, but it compounds and becomes the strongest growth channel,” and has made it his long-term main channel. Paid ads (Meta and Google) remain as short-term support, but not the main battlefield.

Why this setup worked

This can be explained across three layers.

On channel choice, he accepted the asymmetry in capital and chose his battlefield accordingly. “Can’t win on ads” reflects a difference in capital structure, not a lack of effort. A funded competitor can pay CAC before recovering LTV. A self-funded individual can’t. Rather than treating this as a matter of grit, he shifted to a channel where the cost is a different kind (time invested, which compounds into an asset. SEO and content). That’s what made the unit economics work. The surface-level tactic reads as “did SEO,” but the underlying structure is “avoided the ground the competitor could buy with money, and stood in the ground that can only be bought with time.”

On process, he places demand validation before building. Before starting anything, he checks search volume, SEO metrics, competitor strength, how competitors acquire customers, and market demand signals. His principle is stated plainly:

“Never build anything before validating demand.”

This connects back to MathPlanner, where demand sold even when handled manually. His approach prioritizes validating fast over building fast. As he says, “I trust my gut but validate with data too. That avoids months of wasted work.”

On structure, the portfolio actually absorbed a loss. UseArtemis dropped to $5,000 from its peak once LinkedIn tightened its automation restrictions. Yet total monthly revenue held at $28,000. When one product gets cut down by an external policy change, the others carry the business. This goes beyond theoretical diversification. It’s a loss that was actually absorbed, in practice.

What isn’t working, and the problems still on the table

The cost of platform dependency is playing out in real time. UseArtemis’s decline is a direct consequence of building a business on top of another company’s platform, LinkedIn. The portfolio protected the whole, but that one product hasn’t recovered.

Churn also remains a structural challenge. Niches like StoryShort inherently have “low LTV, and churn is always an issue,” as he acknowledges. Once a single-purpose use case like faceless video generation is satisfied, users tend to leave. That’s why his growth formula boils down to the obvious three things: new acquisition, reducing churn, and product improvements aimed at raising retention.

And his own view on the $20,000 ceiling is candid. He says reaching $20,000 MRR is “easy,” and the real work begins beyond that. He doesn’t hide the fact that this is where things get difficult, even while noting that StoryShort still has SEO-driven upside remaining.

What can be transplanted, and what can’t

What transfers. The order of checking search volume and competitor acquisition tactics before building anything. The judgment to not fight head-on against well-funded competitors on paid ads, and instead lean into channels where time becomes an asset. Keeping a fixed tech stack even as products multiply, to avoid raising operational load. Holding multiple revenue sources to hedge against a single platform’s policy changes. And the very first move: confirming demand manually, even by hand, before anything else.

What doesn’t transfer. The tailwind of 2017, when Instagram automation was still viable. The regulatory environment on platforms today is entirely different, and the same playbook can’t be run now. Also, the technical skill to self-teach coding and run three products in parallel was built up over years. On AI coding tools, he says: “if I started today, I’d still learn to code, but I’d use AI relentlessly to build the first version faster,” making clear he doesn’t think the learning curve can be skipped.

His next goal is to grow Capacity.so to $100,000/month, on the reasoning that the market for people who want to build tools without writing code is large. Someone who started by hand because he couldn’t code is now building tools for people who don’t want to code, a throughline that summarizes this case with no loose ends.

Related reading

• Plausible — reached $1M ARR through content and search without relying on ads
• ScrapingBee — a small-team SaaS that grew content-first and reached an exit