Loading page…
Two pages, one input box, zero social media: an exact-match domain that rewrites AI text into human-sounding text and gets 2.6 million Google visitors a month on the back of 1,700 referring domains
DR 48 · 1,700+ referring domains · ~7,000 backlinks · plans at $19 / $29 / $79 per month · revenue is NOT reported — the $200K-$500K/mo headline is the creator's 1%-conversion arithmetic
The whole site is two pages and the homepage has no call to action — just the tool. It gets 2.6 million visits a month because 1,700 domains link to it, not because anything on it is clever.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
The site launched in February 2024. Within about a year it was doing more than 2.6 million organic visitors a month, according to the Ahrefs data the creator pulls up on screen. It is called Humanize AI Pro, and it does one thing: you paste text written by an AI, and it rewrites it so it reads as though a person wrote it. The timing bet is the whole business. As ChatGPT and Gemini went mainstream, a second-order need appeared — people who had generated text now needed it not to look generated. The site's operators saw that and built the tool for it before the category was crowded, on an exact-match domain. What makes this worth studying is not the product, which is a single language-model call behind a text box. It is the extreme minimalism of everything around it. There are two pages on the entire site: an AI humanizer and, added later, an AI content detector. The homepage carries the tool itself and, as the creator points out with some surprise, no calls to action at all — just a pricing link in the menu and the footer. No blog empire, no social presence, no email funnel visible. The honest caveat, which we are putting up front rather than burying: the revenue figure in the video's title is not a reported number. The creator takes 2.6 million monthly visitors, assumes 1% subscribe, arrives at 26,000 paying users, and concludes the site "could be pulling anywhere between $200,000 to $500,000 every single month." That is arithmetic on an assumption, not a receipt. What is real and third-party-sourced is the traffic and the backlink profile.
The first visitors came from the search term itself, which is the only reason the site exists. As ChatGPT and Gemini went mainstream, "humanize AI text" went from a query nobody typed to one carrying millions of monthly searches, and the site launched in February 2024 on the exact-match domain for it.
The quick-reference card — acquisition channels, replication playbook, risks & traps, and our verdict — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself. The exact words are quoted below, with a timestamp that jumps to the second each figure is said.
“And this is why this micros website at least makes $500,000 a month.”
0:02 · hear it in the videoData credibility: Two very different grades of evidence in one video, and the distinction is the whole point. STRONG: the traffic and link data come from Ahrefs, shown and described on screen — 2.6M organic visits per month, domain rating 48, 1,700+ referring domains, ~7,000 backlinks, and the named top countries (Philippines, US, India, Canada, UK) and head keywords. The February 2024 launch date, the two-page site structure, the .pro exact-match domain, the $19/$29/$79 pricing, the annual discount, and the stack (Next.js on Vercel, React, Hero UI, Cloudflare CDN, PostHog + Google Analytics) are all as stated in the video. NOT VERIFIED: the revenue. The video's title claims $500K/month, but the narration reveals the derivation — 2.6M visitors, an assumed 1% subscription rate, therefore 26,000 paying users, therefore "anywhere between $200,000 to $500,000 every single month." No revenue figure is reported by anyone connected to the business, and a 1% visitor-to-paid rate on cold search traffic to a free tool is optimistic by roughly an order of magnitude against typical freemium benchmarks; at a tenth of that assumption the same traffic implies $20K-$50K/month. We report the traffic as the hard number and treat the revenue as an estimate. Also note the conflict of interest: the creator runs a software development agency (Dev House, self-described as 50+ developers and 93+ startups over five years) and pitches it twice during the video, including a build-this-for-you offer. The 3-4 week build estimate and the 30-day roadmap are his agency's framing, not the reference company's account. No information about the operators' location, team or nationality appears anywhere in the transcript, so region is null; the named countries are the audience's, not the founders'.
Money arrives one way: three subscription tiers at $19, $29 and $79 a month with a meaningful discount for paying yearly, and no ads or affiliate arrangements anywhere.
The loop is links buying rankings and rankings paying for more links. With a domain rating of 48 sitting on a weak top-level domain, the site holds head terms only because 1,700 domains point at it, and the revenue from that traffic funds the next round of placements.
A two-page site gives you almost nothing to do while you wait, which is the particular difficulty of this model. There is no content calendar to fill, no funnel to test and no audience to talk to — the only lever is referring domains, and links take months to drag a profile from nothing up to a domain rating of 48.
An input box. You paste AI-written text, it comes back rewritten to sound natural and human. The stated audience is content writers, marketers, students, bloggers, influencers and businesses — which is a polite way of saying anyone who needs generated text to pass as unassisted. A second tool was added later on its own page: an AI content detector, which scores how likely a passage is to be machine-written. The creator's read on why that was smart is purely about acquisition — "by adding new features, they capture new keywords, new traffic and obviously the new users." The detector is not really a second product; it is a second keyword cluster with a working demo attached. The technical surface is thin by design: a prompt, a model, a rate limiter, a paywall. The creator's own estimate is that a small team — one back-end developer, one front-end developer and a graphic designer — could build the first version in three to four weeks. Our read: the product is deliberately not the moat, and pretending otherwise is how people lose money copying this. Two things are the moat. First, mover advantage on an exact-match domain for a term that went from zero to millions of searches in a year. Second, 1,700 referring domains. Neither is a feature, and only one of them is still available to you.
Subscriptions only. The creator is explicit that there are no ads and no affiliate arrangements — "a very clean SaaS sales setup." Three tiers: $19 a month for the entry plan, $29 for standard, $79 for pro, with a meaningful discount for annual payment. That is the entire monetisation. No credits marketplace, no enterprise tier mentioned, no upsell path beyond the three plans. And here is where we have to be careful, because the video's headline is an estimate dressed as a fact. The creator's calculation is: 2.6 million monthly visitors, at a 1% subscription rate, gives 26,000 paying users, which at these prices would produce somewhere between $200,000 and $500,000 a month. Every step after the traffic number is an assumption. A 1% visitor-to-paid rate on cold, top-of-funnel search traffic to a free tool is optimistic by an order of magnitude in most freemium businesses; a tenth of that would be more typical, which would put the site nearer $20,000-$50,000 a month. We are not asserting either figure. What we can say with confidence is that a two-page site receives 2.6 million organic visits a month and sells $19-$79 monthly subscriptions, and that whatever the conversion rate is, the revenue is substantial and the gross margin is whatever is left after inference costs. The honest framing is: verified traffic, unverified revenue.
One channel. Search, and nothing else — the creator says outright that the site is "only focused on search engine optimization." The keyword set is the head of a brand-new category: AI humanizer, humanize AI, AI checker, humanize AI text. These are high-volume terms with millions of searches and, crucially, perfect intent match — somebody typing "humanize AI text" wants exactly the thing the page does, immediately. Top traffic countries are the Philippines, the United States, India, Canada and the UK, which tells you the audience skews toward markets where writing work is outsourced and where academic AI policing is aggressive. The domain is an exact-match domain on a .pro top-level domain. The creator flags this as a genuine handicap — .pro is weaker than .com, .net or .ai — which sharpens the point that the ranking is not coming from domain prestige. It is coming from links. Domain rating 48, more than 1,700 referring domains, and almost 7,000 backlinks, mostly from guest posts, blog placements and mentions across platforms. That is a heavy, deliberate link-building programme for a two-page site, and the creator names it as the reason the site "is still growing like a beast" despite the weak TLD. Their explicit strategic choice was to ignore the usual authority-and-expertise signals and pour everything into backlinks. The second page — the AI detector — is a distribution decision more than a product one: launch one feature, then expand to capture the adjacent keyword cluster.
Fully disclosed and unremarkable, which is the point. Next.js hosted on Vercel, React on the front end, Hero UI for the component layer, Cloudflare CDN for speed, and PostHog plus Google Analytics for tracking. The creator calls it "very common setup in modern SaaS web development" — simple, fast, scalable. Build cost is estimated rather than reported: a back-end developer, a front-end developer and a graphic designer, three to four weeks for the first version. The creator, who runs a development agency and pitches it twice in the video, would know roughly what that costs. The cost that scales is inference. Every humanisation request is a model call on a text payload, and at millions of monthly visitors the free tier is not free to serve. The transcript gives no figures, so we won't invent margin numbers — but note that the abuse surface here is severe: a free text-rewriting box on a page with 2.6 million monthly visitors is an open invitation to scripted bulk use, and rate limiting is a first-order engineering problem rather than a nice-to-have. The creator also offers a 30-day roadmap for building a copy: week one, an MVP that is a simple input box which rewrites AI text; week two, deploy with analytics and tracking; week three, SEO content around the main keywords plus link building; week four, a small launch on Product Hunt, Reddit, LinkedIn and Twitter. That last week is worth noting for what it isn't — the reference site's growth came from none of those.
Copy the shape, not the niche. The transferable pattern is: find a search term that has gone from nothing to millions of queries because a new technology created a new anxiety, register the exact-match domain, put the working tool on the homepage with nothing else on the page, and spend everything on links. Two pages. No blog, no funnel, no social. That is a genuinely repeatable playbook and most people over-build the site instead of over-building the link profile. Copy the tool-as-homepage decision. There is no call to action because the tool is the call to action. A visitor who arrived from a search for exactly this thing should be using it within one second, and the paywall should appear only after they have seen it work. Copy the second-page expansion logic: add features to capture adjacent keyword clusters, not to satisfy existing users. What to change, and this matters more than anything above: do not copy this specific niche. AI humanisers are now one of the most saturated categories on the internet, the head terms are contested by dozens of well-linked sites, and 1,700 referring domains is a year of paid link building you would be starting from zero. Worse, the demand is structurally unstable — AI detection is increasingly regarded as unreliable by the institutions that were driving the anxiety, and if detectors lose legitimacy the market for evading them shrinks with it. Also weigh the ethics honestly. The stated audience includes students, and the value delivered is helping generated text pass as unassisted work. That is a real reputational and policy exposure, and it constrains who will partner with you, advertise for you, or process your payments.
Total Google dependency, which the creator names first. One channel, no email list, no social presence, no brand searches to speak of. A core-update ranking drop takes revenue with it the same week, and there is no second channel to cushion it. This is the single largest structural risk in the business. AI answer engines are the sharper version of the same problem. A tool page survives AI summarisation better than an informational page — you cannot rewrite someone's essay inside a chat answer — but the discovery path still runs through search results that are increasingly being replaced by generated answers. Competitive erasure from above. The creator names Quillbot and Writesonic, and then the real threat: OpenAI or any model provider could ship "rewrite this to sound more human" as a free feature inside the product where the text was generated in the first place. That is a one-line prompt away, and it removes the reason to visit a separate site at all. Demand instability. The entire market exists because AI detection is treated as authoritative. Detection tools have well-documented false-positive problems, and as institutions lose confidence in them the incentive to buy an evasion tool weakens. This is a window, not a permanent category. Abuse and unit economics. A free rewriting box on a page with millions of monthly visits invites scripted bulk usage, and every request costs inference. Link-profile fragility. Nearly 7,000 backlinks concentrated in guest posts and paid placements on a two-page .pro domain is exactly the footprint search engines periodically devalue. The strategy that built the site is also the one most exposed to a policy change. And the evidence problem. The traffic and link figures come from Ahrefs and are the strong part of the case. The revenue is the creator's own multiplication of an assumed 1% conversion rate, presented in the video title as if it were reported. The video is also, in effect, an advertisement — the creator pitches his development agency twice — which is a reason to read the upside framing sceptically.
The four founder playbooks — first customers, first dollar, the flywheel, the silent stretch — are in the Unicorn Club, on this idea and on all 1,435 others. The story, the numbers, the sources and every later section stay open to everyone.
Join the Unicorn Club — $5/moCancel anytime, in one click