6 hours of work, 12K users, $400/mo — and math that won't close
The Case: 12,000 users, $400/month, $0 infrastructure
What he does. Ships a client-side Chrome extension that find-and-replaces any word on any webpage with a word you pick.
Backstory. He wrote it in 6 hours over a weekend as a gag reply to a Reddit thread. The window that makes it work isn't technical — a commenter claims he built the identical thing in 2 hours years ago, and another remembers the old "cloud"→"butt" extensions. The window is that the Chrome Web Store now behaves like a search engine with almost no competition for absurd, specific, one-job queries.
Product. - Free tier: unlimited single-word replacements - Pro tier: multiple words at once, regex support, per-site custom rules - Stated use cases: spoiler-blocking, making work email tolerable, swapping "AI" for "magic," turning crypto jargon into "monopoly money" - No servers, no databases, no accounts to administer — pure client-side JavaScript - He also built a ChatGPT conversation-branching extension, killed when OpenAI shipped the feature natively
Pricing. Free tier, unlimited basic use. $3/month for pro.
Key numbers. - Build time: 6 hours, one weekend - Active users: 12,000 (his figure) - Pro price: $3/month - Stated revenue: $400/month - Stated upgrade rate: ~10%, which he converts to 1,200 paying users - Arithmetic on his own numbers, as four commenters point out: 1,200 × $3 = $3,600/month, not $400 - Maintenance cost: $0 - Launch distribution: 3 subreddits, ~200 upvotes - Thread traction: 519 points, 148 comments - Account karma at time of posting: 28 (per commenter JW_TB)
Stack. Client-side JavaScript only; he separately names Cursor, Gamma, Trupeer, Mailmodo, Superhuman, and SparkToro as his general tooling, not as this product's stack.
Why this works (and what most readers will miss)
1. The cost floor is actually zero, not "low." No servers, no databases, no auth. That means the $400 figure and the $3,600 figure are both 100% margin. A product with no infrastructure has no bad month — it has months where it earns less, which is a completely different animal from months where it costs you money.
2. The Chrome Web Store is a distribution channel disguised as a directory. He posted in 3 subreddits, got ~200 upvotes, and then the store's own search took over. He got one push and the channel compounded on its own. That is rented land, but it's rented land where almost nobody has bothered to plant anything.
3. The non-obvious one: the gag is the positioning. Nobody searches "text substitution utility." They search for the joke. A stupid premise is a memorable premise, and memorable is the only thing that survives a Chrome Web Store listing with 40 words of copy. The absurdity isn't a limitation on the business — it's the entire acquisition mechanism.
The Distilled Read
Ship the joke, then charge for the boring part. The free tier does the funny thing. The paid tier does regex, multi-word rules, and per-site configuration — which is to say the paid tier is a power-user tool wearing a clown nose. That's the trick worth stealing. Comedy acquires; utility retains. Anyone who tries to build the serious version of this from scratch gets zero installs because nobody was searching for it. He entered through the laugh and monetized the workflow underneath it. You're not selling word replacement. You're selling the twenty minutes a spoiler-free timeline gives someone back.
Templatize the shell, not the gag. He has already run this pattern twice — this extension, and a ChatGPT branching tool that died the day OpenAI shipped the feature natively. That's the real asset: a repeatable client-side extension chassis he can drop a new premise into over a weekend. One of two shipped products got killed by a platform update. That's a 50% mortality rate on a portfolio of two, and the correct response is not to build more durable products, it's to build more products. Six hours per attempt means you can afford to be wrong four times out of five and still come out ahead.
Now the economics, and here's my read, not his number. Two figures cannot both be true. Either 1,200 people pay $3/month — in which case the revenue is $3,600 and he's underreporting by nearly 9x — or the revenue is $400, which puts real paying users somewhere near 133 and the true conversion rate near 1.1%, not 10%. My read is the second: $400 is the number a person reports when they're reading a dashboard, and 10% is the number a person reports when they're rounding a vibe. Nothing else in the post is engineered enough to be a scam; the account has 28 karma and he never named the extension, which is what carelessness looks like, not fraud. Take the $400 as real and the 10% as decoration, and the case gets more useful, not less: 1.1% conversion on a free utility is the honest benchmark you should plan against.
Distribution was one post, and that's the fragility. Three subreddits, ~200 upvotes, then the store carried it. He didn't build an audience, capture an email, or own a single channel — he rented Google's shelf space and got lucky with a listing that ranks. If Chrome's search algorithm shifts or a competitor with the same 2-hour build clones the listing, there's no list to email, no community to fall back on, nothing. He owns 12,000 users and can't reach one of them.
The moat nobody in the thread saw: he's sitting on a browser-side behavioral dataset and treating it like a toy. Every pro rule is a person telling him, in plain language, exactly which words on the internet they can't stand. Spoiler terms. Crypto jargon. AI hype. Manager-speak. That's an intent signal no ad platform can buy, generated by 12,000 people who installed the thing voluntarily. Eighteen months from now the extensions that survive the AI-tooling wash-out will be the ones that turned "what the user chose to filter" into a product — a filter layer people pay $9/month for to make an AI-slop-saturated web readable again. He built the sensor and called it beer money. The winners in this category won't be better word-replacers. They'll be the ones who realized that a browser extension is the last piece of software that sits between a person and everything they read, and that position is worth more than $3/month.
Steal-the-Playbook
- Pick one niche, not the category. Not "browser utility." Pick spoiler-blocking for one fandom, or one industry's jargon. One premise, one search term.
- Build client-side only. No servers, no database, no login. Budget 6-8 hours in Cursor and ship this weekend. If it needs a backend, you picked the wrong idea.
- Write the Chrome Web Store listing before the code. The title is your entire SEO surface — put the exact phrase someone would type into it, and make it funny enough to click.
- Launch free with one paid gate: bulk rules plus regex at $3-5/month. Plan for 1% conversion, not 10%. At 12,000 installs that's ~$400/month — which is exactly what this case actually produced.
- From day one, put an email capture in the options page tied to a settings-sync promise. That list is the only thing you'll own when the platform changes its mind. He didn't do this. It's the whole difference between a hobby and an asset.
Stack: Cursor + vanilla JS + a $5 Chrome developer account. ~$5 total, one weekend.
Bottom Line
A six-hour build with zero infrastructure and 12,000 users isn't a small business — it's a cheap sensor pointed at what people can't stand about the internet. He read the dashboard and saw beer money. The position is worth more than the product. Build things that sit between a person and what they read, and the revenue question answers itself later.