Debunking the Myth of Overnight Success in Micro-SaaS

Written with AI assistance. Figures without a traceable source were cut before publishing.

The same piece is posted as a thread on GitHub — that copy has a reply box under it, and this one does not.

A six-hour Chrome extension pays $400 a month. Nothing runs on a server. That sounds like something invented to sell a course. I read the Chinese indie-developer accounts every day, and that is where this one came from. A text-replacement extension, one weekend of work, about six hours of it. 12,000 active users. Roughly 10% of them pay $3 a month for the premium tier. The free version replaces text on any page, as often as you like. The $3 tier adds multi-word replacement, regex matching, and per-site rules. That split only makes sense once you already know which users are the impatient ones. All the logic runs inside the browser. No server bill, no database, nothing that can page you at 3am. Top line and bottom line are the same number.

That is a clean figure attached to a small object, which is why it travels so well in a screenshot. It is also the less useful half of the story.

What Happened Before Line One

The half that does not screenshot well is the month or two of unglamorous selling that comes first. Before writing a contract-comparison tool, one builder handled three to ten comparisons by hand at $29 a document, and only turned the routine into software once the same people kept coming back and paying for it. Pricing follows that same order. The sweet spot for a tool-shaped micro-SaaS in North America sits somewhere between $9.90 and $49 a month, but that range is a result rather than an opening decision. You do not set $19 and wait for buyers. You find buyers, and $19 turns out to be what they will pay. A user who says they love the idea is worth nothing at all. A user who hands over a card is a fact.

If you cannot sell the thing by hand, you do not have a product yet. You have a preference.

My favourite version of this test is also the smallest one I have read. Someone posted screenshots of the steps for setting up an AI tool on a personal account, watched the comments fill up with people who wanted the same setup and could not follow the official instructions, priced a walkthrough at 9.9 RMB to see whether the interest survived contact with a payment screen, and took two orders inside 48 hours. Total revenue: 19.8 RMB, roughly the price of a sandwich. The number is irrelevant and the signal is not. Two strangers paid for a clearer installation path, which is a far better reason to build one than a hundred people saying it sounds useful.

Distribution deserves the same treatment before any code exists. Chasing broad keywords means bidding for attention against companies with budgets, and the acquisition cost eats the margin before the first renewal lands. The cheaper route is a free utility aimed at an oddly specific query, something closer to “how to write SEO product descriptions for a photography portfolio” than to “SEO tool”, with the paid tier sitting one click behind it. Complaints in a narrow subreddit are the other end of that same pipe. They hand you the exact words people use when they are annoyed enough to type.

The Cold Start Nobody Screenshots

The extension did not land in front of 12,000 people on its own. Its author posted it in 3 relevant subreddits, collected about 200 upvotes, and converted those into the first installs. After that the Chrome Web Store took over, because the store runs its own search and the extension surfaced for queries like “text replacement” and “modify web page”. First 200 users by hand, every user after that from search.

Naming is part of the same trick and it costs nothing. A product name and a domain that mirror the phrase your buyer already types into the address bar are search intent bought at zero.

Narrow Beats Clever

Puff Count took one high-frequency, hidden, slightly shameful behaviour, vaping, and turned it into a number the user sees every single day. Then it sold a subscription around quitting, and reached $44,000 in monthly revenue without ever being a general health app. Zigpoll went in the opposite direction. It embedded a boring feature, the survey, into the one moment when an e-commerce buyer is honest, and got to $125K MRR. A third of its new signups arrive through the Shopify App Store, its single largest source, because that store filters out casual browsers and shortens the path to trust. BackPedal is the least software-like of the three. James Dunn paired GPS trackers with a team that physically goes out and recovers stolen bikes, which is precisely the unpleasant part nobody wants to copy, and the thing runs at $55k MRR.

None of those three won on code. Each of them won by being unreasonably specific about who was in pain.

Zigpoll is worth one more sentence, because a survey tool in 2026 is about as unfashionable as software gets, and it still beat most of the AI wrappers launched the same year by sitting inside a workflow that already had money moving through it. Its best accounts turned out to be agencies, the kind of customer who installs the same tool for the next client and the client after that, which is a distribution channel disguised as a user segment. Old lanes with a redefined slice in them are not picked over. They are just quiet.

Where AI Actually Helps

Spinning up twenty AI wrappers a week does not solve distribution, and the speed of code generation was never the bottleneck. Models pay for themselves in the boring, high-volume corners instead. Generating local SEO content at volume gets a small site ranking sooner, and sites like that rent to local businesses for $500 to $3,000 a month each. A designer building restaurant websites reached £30,000 a month by keeping the taste and the client judgement human and handing the repetitive layout work to the machine. Order matters more than tooling here: a two-week minimum demo, then a real customer reaction, then the actual system. Silq’s public record on TrustMRR shows the same staircase, web before app, free before paid, one language before many.

There is a second reason to keep the founder out of the product, and it shows up at the exit rather than at the launch. A content-plus-subscription structure that does not depend on anybody’s personal brand is a thing a stranger can buy and keep running, and at least one such business went from public listing to a completed acquisition in under four months, which is not a timeline available to a business whose audience follows a face.

I have not audited any of these revenue figures myself. They come from the founders’ own accounts and the public records those founders point at, so read them as claims with a name attached rather than as audited books. My guess is that survivorship here is brutal and that most attempts at each of these patterns earn nothing at all.

Three Paying Strangers, Then Code

The six-hour build is real, and it is also the cheapest part of what happened. Those six hours sat on top of a decision to keep every rule client-side, which is what made a $3 subscription at a 10% conversion rate profitable instead of merely busy. Zero servers meant all $400 stayed. The extension worked because its author knew which 200 people to show it to first and where the next 11,800 would come from. So pick something small enough to do by hand, charge three strangers for it this week, and open an editor only after the third one pays.

Also readable on Telegraph.

Read next

All 32 write-ups

The 18 figures in this piece — each with the sentence it came from — are in the figures table, alongside 317 more, as JSON and CSV.

Topics: SaaS Business · Niche Market · Productivity · Artificial Intelligence

Part of llm-api-pricing — field notes on AI coding agents.

Did this save you an afternoon? A star on the repository is the whole ask — it is what puts these in front of the next person looking; the data is CC BY and does not require starring. Want a figure that is not in here yet? Say which metric, which provider, which unit — in one line. One required field, and the page you came from is already filled in. Got a better number? Open an issue — that form knows which write-up you came from too; corrections and counter-data are the point.