Vibe Coded Websites: Real Revenue, Shipped Products, and Unproven Idea
Vibe coded websites can make real money, but the defensible examples are products with distribution and a narrow paid job, not generic sites assembled…
Vibe coded websites can make real money, but the defensible examples are products with distribution and a narrow paid job, not generic sites assembled because coding became faster. In ProvenStartups’ full matching cohort, 22 of 82 projects publish a clean monthly figure, with a $20K/mo median and a $2K/mo to $500K/mo range; 59 are solo-run. The practical answer is to copy the economics, not the interface.
Contents
The bottom line on vibe coded websites
The best vibe coding website is a small, commercial system: it finds a valuable user, performs one painful task, and charges before its codebase becomes elaborate. We would choose a service-backed tool, focused consumer utility, or workflow SaaS. We would refuse to launch an undifferentiated directory or “app factory” without a distribution edge.
Vibe coding changes the cost and speed of implementation. It does not change demand, trust, acquisition, or retention. A developer who can produce a working site in a weekend has removed only one constraint.
The cleanest small starting point in the named cases is the AEO Service: one client paid a $2,000/mo retainer [F]. That is not proof of a scalable SaaS, but it is proof that a tightly defined outcome can sell before a founder builds a broad platform.
The cohort itself is wider than “website builder” projects. It contains 23 SaaS products, 16 consumer apps, 10 AI services, 6 directory sites, 6 AI websites, 5 platform plugins, and 4 cautionary tales, among other categories. The useful pattern is commercial specificity, not a particular stack or page style.
How strong is the revenue evidence?
Treat a revenue claim as a typed value, not a plain number. ProvenStartups labels third-party verified claims [V], founder-reported claims [F], creator-relayed claims [C], and unverified claims [U]. A target is not revenue, an anonymous client count is not revenue, and a disclosed price is not evidence that anybody paid it.
The grading method matters because impressive numbers otherwise look interchangeable. Across the full 406-project index, the evidence split is 57 [V], 184 [F], 121 [C], and 44 [U]. The matching cohort’s published evidence split records 16 [V] and zero claims in the other three classes.
That is why the Payout class-action discovery app at $20K/mo [V] carries more decision weight than an attractive projection. Payout reached that level in 50 days, but the grade still does not prove another founder can reproduce its distribution.
Read the labels literally:
- ·[V]: strongest basis for benchmarking actual revenue.
- ·[F]: useful operating evidence, but supplied by the founder.
- ·[C]: a creator relayed the claim; verify before modeling around it.
- ·[U]: idea, price, or claim without adequate revenue verification.
The full project index includes failures as first-class records: 38 of 406 ideas are cautionary tales. That prevents the usual survivorship-only view of vibe coding websites.
Which website models actually made money?
The strongest named results come from products attached to a recurring job: discovery, content production, coding, or conversion. The comparison also exposes the non-results. Some entries disclose live revenue, some disclose paying behavior without revenue, and some offer only a target or theoretical upside. Those states should never share one “successful” label.
| Project | Model | Published result | What the result proves |
|---|---|---|---|
| Payout | Consumer claim discovery | $20K/mo [V] | A narrow discovery product can monetize |
| Subscribr | Subscription scriptwriting SaaS | $30K/mo [V] | A repeated creator workflow can support subscriptions |
| Minea / DropMagic | E-commerce research and creative tooling | $750K MRR [F] / $45K MRR [F] | Product plus creator distribution can reach scale |
| Cursor | AI coding product | $500M/yr [V] | The market is real, but this is a scale reference |
| AI App Factory | Long-tail consumer app portfolio | Revenue not disclosed [U] | Paying users were described; earnings cannot be benchmarked |
| AI Directory Site | Search-driven directory | $2K-$10K/mo target [C] | A target, not a shipped revenue result |
| AI Venture Studio | Repeated venture creation | “Uncapped” profitability [C] | Theory, not a measurable outcome |
The Claude Code local SEO service adds a useful middle case: more than $5,000 cumulative [F] from digital products, plus client retainers described only as several thousand dollars monthly [F]. The exact retainer figure was not disclosed, so it should not be manufactured.
Where the data contradicts the popular claim
The popular claim is that cheap AI coding makes shipping the decisive advantage. ProvenStartups’ data says the opposite: implementation is abundant, while verified demand and distribution remain scarce. A vibe coder website can be technically complete and commercially empty. Faster code increases the supply of products; it does not create buyers.
Across the site, 211 distinct projects mention at least one AI coding or no-code tool. ChatGPT appears in 100 cases, Claude Code in 50, Cursor in 46, Bolt in 40, and Lovable in 19. Tool choice plainly does not separate winners by itself.
The revenue distribution reinforces the point. Of 106 cases with a clean monthly figure, 8 are under $1K/mo, 18 are at $1K-$10K/mo, 54 are at $10K-$100K/mo, and 26 exceed $100K/mo. Yet NoFap reached $6K/mo [V] in its first month and is still filed as a cautionary tale. Revenue does not erase product, platform, or execution risk.
We would therefore reject three common shortcuts:
- ·Counting generated pages as distribution.
- ·Counting a price, target, or waitlist as revenue.
- ·Treating a scale reference such as Cursor as a realistic solo-founder baseline.
Use Claude Code’s documentation to understand the tool. Use customer behavior to decide whether the product deserves to exist.
What we would build, and refuse to build
We would build the smallest paid loop that can be tested manually, then automate the repeated bottleneck. Start with one buyer, one acquisition channel, one measurable outcome, and one reason to return. Refuse multi-product portfolios, broad directories, and infrastructure-heavy concepts until a specific audience repeatedly asks for the underlying result.
- 1.Choose a paid job. Prefer claim discovery, script production, reporting, lead qualification, or another task with an observable finish.
- 2.Sell the manual version. The $2,000/mo [F] AEO retainer is a better starting signal than a polished free dashboard with no buyer.
- 3.Instrument the loop. Track activation, repeat usage, conversion, churn, and support burden from the first customer.
- 4.Automate only the repeated work. Keep human review where wrong output destroys trust.
- 5.Publish the evidence state. Separate collected revenue, run rate, targets, and undisclosed results.
For a vibe code website, difficulty should match the proof available. ProvenStartups’ 266 software/SaaS cases include 12 at difficulty 1, 100 at difficulty 2, 104 at difficulty 3, 40 at difficulty 4, and 10 at difficulty 5. Do not accept difficulty 5 before demand proof when a difficulty 2 test can answer the commercial question.
FAQ
The short answers are simple: vibe coding is an implementation method, not a business model; published revenue must retain its evidence grade; and “launched” is different from “profitable.” For a solo founder, the best next move is a narrow paid test whose failure is cheap and whose result can be measured.
What are vibe coded websites?
They are websites or web products built substantially through natural-language collaboration with AI coding tools. The label describes how code is produced, not what customers buy. A vibe coding website still needs a defined user, a valuable job, distribution, payments, reliability, and maintenance. Generated code removes none of those commercial requirements.
Can a vibe coding website make real money?
Yes, but the evidence varies sharply. Subscribr publishes $30K/mo [V], while the AI Directory Site publishes only a $2K-$10K/mo target [C]. Both can appear in a list of ideas, but only the first is a verified operating result. Compare evidence classes before comparing models, features, or stacks.
Which model is best for a solo vibe coder?
Start with a service-backed tool or narrow recurring workflow where one buyer can validate the job. Fifty-nine of the 82 matching projects are solo-run, so solo operation is common, not exceptional. It still does not make every model sensible. Avoid businesses requiring marketplace liquidity, mass content, or enterprise infrastructure before demand is proven.
Did every named project launch?
No such conclusion is supported. Some cases clearly report operating revenue; the AI App Factory reports paying users but no revenue figure [U]. The AI Directory Site supplies a target [C], and AI Venture Studio supplies theoretical upside [C]. ProvenStartups treats those as unproven results, not as permission to invent launch status or earnings.
Is Cursor the best tool for building a vibe code website?
The data cannot establish that. Cursor is itself a huge business at $500M/yr [V], but that proves market scale, not that its users outperform users of Claude Code, Bolt, Lovable, Replit, or another tool. Choose the environment that lets you inspect, test, and maintain the code. Distribution remains the harder constraint.