Small software products built alone

The research was done by AI agents that open web pages. Some sites refuse them; where that happened we say so. The small numbers point to the source list at the end. Where a figure is our own sum or guess, it is marked “our estimate” and the basis is given.

The claim

These are examples, copied from our collection of 52 tweets about this way of making money¹. We call every method a “scheme” in the neutral sense of a plan; the word does not mean that the method or anyone quoted is dishonest. Some of the tweets state a goal or a result, not a promise to others. View counts are as collected on 6 October 2026. Dates are worked out from each tweet’s ID number. The page judges the scheme, not the people who tweeted.

“My vibe coded app grew by +$23,000/month MRR this month. 🤯 It was my first-ever SaaS that I built without knowing how to code.”

@ModestMitkus, 30 October 2025, 62,002 views¹

“Bro vibe coded a $20k MRR app in 14 days.”

@starter_story, 5 December 2025, 126,746 views¹

“Left my job. Left my home. Living on savings. […] Goal: Build a $10k/month SaaS and document the process.”

@robj3d3, 20 January 2025, 1,147,962 views¹

“Want to make $100,000 this month? Build a micro-SaaS with no-code. DM 50 potential customers a day. Get it to $2,000 MRR. List on Micro Acquire. Sell it for $100,000.”

@samuelthompson, 16 June 2021, view count not collected¹

The collection also holds much smaller claims, such as one report of about $250 of monthly recurring revenue from a few hundred users¹. We did not check any single tweet’s figure against the author’s payment records, with one exception described below.

What the scheme is

“SaaS” means software as a service: a program people use through a website or app and pay for by subscription, usually monthly. “Micro-SaaS” is a small one, built and run by one person. “Vibe coding” means building software by describing what you want to an AI tool, which writes the code.

The number in almost every tweet is MRR, monthly recurring revenue. It is the subscription price times the number of people paying this month. It is revenue, not income: it is counted before payment fees, app-store commission, tool bills, advertising and tax.

The scheme has two versions. In the first, you keep the product and collect the subscriptions. In the second, you build it up to a modest MRR and sell (“flip”) the whole product on a marketplace.

The arithmetic

Step 1: what one subscriber is worth after fees

In a dataset of more than 115,000 subscription apps, the most common monthly price is $10. The report gives the median as $8 in one place (“Most common at $10, median risen from $7 to $8”) and $9.99 for North America in another⁶. “Median” means the middle one: half charge more, half less. Business tools on the web often charge more, so the table runs up to $99.

On the web. A US account with Stripe, a common payment processor, pays 2.9% plus 30 cents on each domestic card charge, 1.5% more for international cards, 1% more if currency is converted, and 0.7% for Stripe Billing, the part that runs subscriptions².

On a $10 charge with a domestic card: $0.29 + $0.30 + $0.07 = $0.66. The seller keeps about $9.34 (our sum from²). Stripe showed us its Finnish prices on a direct visit, so the US page was read through r.jina.ai, a service that fetches a page as plain text.

Many solo sellers use a “merchant of record” instead: a company that legally makes the sale and handles sales tax and VAT. Paddle and Lemon Squeezy each charge 5% plus 50 cents per transaction, which leaves $9.00 of a $10 charge (our sum from³).

In an app store. Apple keeps 15% of subscriptions from developers who made up to $1 million in proceeds in the prior year, and charges $99 a year for the developer account⁴. Above $1 million the standard rate applies. Apple gave it as 30% when it announced the programme in 2020; the current programme page does not restate the number⁸³. Google Play charges 10% plus a 5% billing fee on auto-renewing subscriptions in the US, Europe and several other markets⁵. Either way a $10 subscription leaves $8.50 (our sum).

Step 2: how many paying subscribers the claim needs

These are subscribers who are still paying, counted before tool, hosting, advertising and tax costs. All figures in the table are our sums from the fees in² and⁴, rounded up to the next whole subscriber.

These are not large numbers, which is why the claim sounds reachable. The difficulty is in three figures that are absent from the text of most of the tweets: traffic, conversion and churn.

The app figures in Steps 3 to 5 describe consumer mobile apps. Business tools sold on the web charge more and lose customers more slowly¹³, but their buyers are harder to reach.

Step 3: how many visitors that takes

“Conversion” is the share of people who try a product and end up paying.

Mobile apps. RevenueCat’s report gives no single figure for all apps. The median share of downloads that become payers within 35 days is 2.6% in North America and 1.4% in India and South-East Asia, and by category it runs from 1.0% (gaming) to 2.9% (health and fitness)⁶. Apps that ask for money before anything else convert 10.7%; apps with a free version convert 2.1%⁶.

Taking 2% (our estimate from those figures), 118 payers needs about 5,900 downloads, and 1,177 payers needs about 59,000 downloads.

Web products. ChartMogul surveyed 200 business software products in January 2026. Per 1,000 visitors, “Free trial products see 45 free signups and 3.6 paying customers”, and “Credit card required trials see 35 free signups and 10.5 paying customers”¹².

At those rates 108 payers needs about 10,000 to 30,000 visitors, and 1,071 payers needs about 100,000 to 300,000 (our estimate from¹²). The report says a typical respondent has $1 to $10 million of yearly revenue and charges $50 to $249 a month¹². These are established companies, so the rates probably flatter a newcomer.

Step 4: how many of them stay

“Churn” is the share of subscribers who cancel.

On monthly app plans, typically 53% to 61% of subscribers pay a second time, and as few as 42% for social and lifestyle apps. After a year, 6% to 14% are still paying⁶. Yearly plans keep 20% to 40% at the one-year mark⁶. For apps built around AI the one-year figures are lower: 6.1% on monthly plans against 9.5% for other apps⁶.

For web products, the only figure by price that reaches down to cheap products is old. ChartMogul’s data from late 2021 put median monthly customer churn at 6.1% for products earning under $25 a month per account, against 2.2% for those above $500¹³. At 6.1%, a product holding 1,071 subscribers must sign about 65 new payers every month just to stand still (our sum from¹³).

Two newer figures were found on 2026-10-07. Neither covers products as small as a solo beginner’s. A survey of just under 700 founders of software products built without investors, collected in late 2023, put average monthly churn at about 5.5% where a card is required to start a free trial and at almost 11% where there is a free plan. These are founders’ own answers, read in a podcast transcript by the survey’s authors and not in the report itself⁷⁹. ChartMogul’s December 2025 analysis of 3,500 software companies with more than $250,000 of yearly revenue found that products built around AI and priced under $50 a month keep 23% of the revenue from existing customers over a year, before upgrades. That is about 20 points worse than other software at that price. AI products priced above $250 a month keep 70%⁷⁸. This measures revenue kept, not customers kept.

One point in AI apps’ favour: in RevenueCat’s data they earn more per paying customer in the first year, $30.16 against $21.37 for other apps⁶. They then keep fewer of those customers, as above⁶.

So an MRR figure in a tweet is a photograph of one month. For a consumer app, most of the people in it are gone within a year⁶.

Step 5: can you buy the customers with ads?

For mobile apps at median numbers, no. Adapty, a company that sells tools to app makers, analysed Apple’s ad system across more than 8,000 apps in 2025. It reports that “$1,000 buys 65 paying subscribers in Brazil and 15 in the US”¹⁴. That is about $67 per US subscriber (our sum). The same page’s other US figures, a median cost of $2.51 per install and 1.92% of installs paying, give about $130 (our sum from¹⁴).

RevenueCat’s median value received per payer is about $23 on Apple’s store worldwide⁶. Spending $67 to $130 to gain about $23 loses money. The figures come from different samples and cover consumer apps only, so this is a rough guide⁶¹⁴.

Step 6: what it costs to try

Prices on the vendors’ own pages today: the AI code editor Cursor is $20 a month; Claude Pro is $20 a month; the app builder Lovable is $25 a month for 100 credits; the database service Supabase is from $25 a month; the hosting service Vercel is $20 a month; the app builder Replit is $20 or $100 a month¹⁵. Vercel says its free plan “is for personal, non-commercial use”, and Supabase pauses free projects after a week without activity, so a product that charges money cannot rest on the free plans¹⁵.

A paid set of tools runs from $65 a month (Cursor, Supabase, Vercel) to $110 for the first five, or about $190 with Replit’s $100 plan in place of Cursor. That is about $800 to $2,300 a year, plus $99 a year for Apple (our estimate from⁴ and¹⁵). It leaves out advertising and the bill from the AI provider if the product itself uses AI.

The first version of this page said we could not price that bill. A rough figure is now possible. Anthropic’s price list today charges, per million tokens (a token is roughly a short word), $1 for input and $5 for output on its cheapest model, Haiku 4.5, and $4 and $20 on its dearest, Opus 5.5¹⁵. If one subscriber makes 200 requests a month, each with 2,000 tokens in and 500 out, that is about $0.90 a month on the cheapest model and $3.60 on the dearest (our estimate from¹⁵). The usage level is our assumption and is the weak part. Heavy users cost several times more. Set against the $8.50 kept from a $10 app-store subscription, the cost of serving a subscriber is no longer near zero.

So the cheap-to-try part of the pitch is true. The larger cost is time.

Step 7: how many products reach the numbers

Subscription apps. In RevenueCat’s 2026 report, 17.3% of newly launched apps reach $1,000 in monthly revenue within two years, and 4.6% reach $10,000⁶. That is about 1 in 6 and about 1 in 22. These are shares of apps, not of people. The report counts only apps that “have active subscription revenue, meet a minimum threshold of installs or revenue”⁶. Apps that never charged are left out, and apps made by companies with staff are included.

Both shares fell in a year: the $1,000 share from 19% to 17%, and the $10,000 share “from 5.3% to 4.6%”⁷.

Among the apps that do reach these levels, the median time is 58 days to $1,000 and 109 days to $10,000⁶. So the fast stories are real, and they describe the winners.

The typical app looks different. RevenueCat puts median revenue one year after launch at about $72 a month; the top quarter is above $429 and the top tenth above $2,574⁶.

The flood. New subscription apps on RevenueCat’s platform rose from about 2,000 a month in January 2022 to more than 14,700 in January 2026, about sevenfold⁶. Some of that is RevenueCat itself gaining customers, so the true rise in competition is smaller and we cannot size it. An outside source confirms that a rise is real. TechCrunch, citing the app-data firm Appfigures, reported that worldwide app releases in the first quarter of 2026 were up 60% on a year earlier across both stores, and up 80% on Apple’s⁸¹. That counts all apps, not only subscription apps, and we read the news report, not Appfigures’ own figures. RevenueCat says apps launched “in 2025 or later (aka, the vibe coding era) account for just 3%” of subscription revenue, while apps launched before 2020 take 69%⁶. Part of that gap is that new apps have had less time.

Web products. TrustMRR is a public list where founders connect their payment account so that revenue is read from the payment company, not from a screenshot¹⁶. Two tweets in our collection cite it as proof¹. Its statistics page today shows this spread of listed startups⁹:

The chart is described as a snapshot of all verified revenue, and the charts beside it are all-time totals. So these bands are most likely total revenue ever taken, not monthly revenue. Read that way, two thirds of listed startups have taken under $1,000 in their whole life. Beside it, the median startup connected through Stripe has taken $258 in total⁹.

Two outside analyses of TrustMRR’s data put the median near $150 a month. One, by a forum user who pulled 5,079 projects in March 2026, found a median of $169 a month, $156 for AI projects, and the top tenth above $10,000 a month¹⁰. The other, by a company that sells a database of software ideas, found “the median indie SaaS earns $145 per month” among startups with any revenue. It puts the top tenth at $5,107 and says 6.1% reach $10,000, “about 1 in 16”¹¹. That second analysis calls the revenue “self-reported by founders to the underlying revenue-tracking network”¹¹. We recomputed neither. Founders choose whether to list, so both samples lean towards products with something to show.

Every one of these samples leaves out people who never finished the product, never added payments, or never listed themselves.

Step 8: the flip

The tweet says: reach $2,000 MRR, sell for $100,000, “this month”¹.

Sale prices are usually quoted as a multiple of yearly profit: at 3.9 times, a business with $10,000 a year of profit sells for about $39,000. Acquire.com, the marketplace the tweet calls by its earlier name, reports that software businesses sold there at a median of 3.9 times annual profit in 2024 and 2025, and that those with under $100,000 of yearly profit averaged 3.7 times⁴⁹. The figures come from “136+ deals that reported their multiples”, all of them completed sales. The report does not say what share of listings never sell⁴⁹.

$2,000 MRR is $24,000 a year. At the 71% average profit margin Acquire reports, that is about $17,000 of profit, and 3.7 times that is about $63,000 (our estimate from⁴⁹). Acquire then takes 8% of sales under $250,000, plus a monthly listing fee of $25 to $100⁴⁸. On $63,000 the 8% is about $5,000 (our sum).

A secondary article, mixing live listings and completed sales, gives 2.5 to 4 times annual revenue, or $60,000 to $96,000 here. It says listings ask about 10.7 times profit on average, far above the 3.7 times Acquire reports for completed small sales⁴⁹⁵⁰.

The sale alone takes 80 to 90 days on average, so “this month” is far faster than a typical sale⁴⁹. A flip is possible at roughly 60% of the tweet’s figure, for a product that is profitable and still holding its subscribers. The tweet is from 2021; we found no sale prices for that year, so this tests it against 2024 and 2025 prices.

One small sale, added on 2026-10-07, fits that estimate. A seller’s post shown on TrustMRR’s marketplace page reads: “Sold for $85K ($75K sale + $10K consultation) I grew SocialKit to $3.3K MRR + ~$800/mo one-time payments”⁶⁹. That is about 1.7 to 2.1 times yearly revenue (our sum). The price is the seller’s statement. The product’s own TrustMRR listing confirms the product, a July 2025 founding date and $15,499 of verified revenue in total⁶⁹. Another seller on the marketplace page wrote: “Just sold my SaaS for $5,000”⁶⁹. An earlier draft of this page said TrustMRR had closed more than 115 sales in six months. We could not find that statement when we re-read the page, so it is removed.

What the tweets supply and omit

This counts the text of the 30 most-viewed tweets in the collection. Attached images, videos and threads were not viewed, so a tweet may supply there what its text does not. None of the texts states a subscription price. One gives a subscriber count. One gives costs. Three, all from one account whose tweets advertise its own paid database of sites, give a traffic figure for someone else’s site. One says the author started with no followers. None gives a churn figure, a conversion rate, ad spend, fees, refunds or hours worked¹.

What people who tried it report

The broad samples

Older data shows the low hit rate is not new. Indie Hackers is a forum for people building small software businesses. A 2022 study of 937 products with Stripe-verified revenue there found that “more than 54% of the products are not making any revenue at all”²². Under a 2021 Indie Hackers post about 1,387 such accounts, commenters calculated a median of $500 MRR and said 49 accounts earned 80% of the revenue. They did not show their working and we did not reproduce it²³.

Age matters more than launch speed. In the 5,079-project analysis, projects in their first year had a median of $148 a month. Five-year-old projects had a median of $2,509¹⁰. These older projects are survivors, so the figure flatters them. TrustMRR’s own statistics show a product’s total revenue to date rising with its age, and revenue rising with its Domain Rating, a score for how many other sites link to it, and only weakly with the founder’s follower count⁹. It gives the strength of each link as a correlation, where 1 is a perfect link and 0 is none: 0.47 for age, 0.43 for Domain Rating and 0.25 for followers on X⁹. The age chart is titled “Older startups trend toward higher total revenue”⁹. Older products have had longer to add up revenue, so part of the age link is automatic. Of listed founders, 54.3% have under 1,000 followers and 31.0% show no count⁹.

One tweet we could check

The most-viewed tweet in our collection is the January 2025 post by @robj3d3 about leaving a job to build a $10,000-a-month product¹. His product SuperX is listed on TrustMRR with $21,866 MRR, $247,682 in total revenue and 518 active subscriptions, verified through Stripe¹⁹. The page is live: an earlier reading the same day showed $21,915, $247,458 and 519, and this page first carried those figures¹⁹. That is about $42 per subscription (our sum); the listed prices are $49 to $199 a month¹⁹. So at least one claim in the set was met, and exceeded.

Three things bear on how repeatable it is. The listing dates SuperX to August 2025, seven months after the tweet, so the result came about 21 months after he started¹⁹. The product is a tool for growing an audience on X, aimed at creators, founders and solo builders¹⁹. And the author had about 65,000 followers when we collected the tweet in October 2026; we do not know how many he had at the start¹. Costs and profit are not shown¹⁹.

Individual accounts

• A solo builder’s four-year tally: 26 projects launched, 8 earned anything, about $115,000 in total, of which about $79,000 came from one product. Self-reported, no payment proof²⁴.
• A thread from August 2026 on Hacker News, a discussion site for programmers, asked who makes $500 a month from side projects. The products people named are mostly several years old and earn roughly $600 to $5,000 a month. The thread’s author wrote: “I myself am at $0/mo from my side projects”²⁵.
• A write-up titled “I Quit My $200K Job to Build a SaaS. 18 Months Later: $147 MRR” reports $87,000 of savings spent. Pseudonymous, no proof²⁷.
• A summary of 423 founders’ public post-mortems (write-ups of why a product failed) says the usual failure is that “products ship, traffic and signups sometimes appear, and almost nobody pays”. This is secondary; we did not open the underlying studies²⁶.

The winners, in their own words

Marc Lou runs TrustMRR and is named in one tweet in our collection¹. A seller’s post shown on TrustMRR’s marketplace page calls it a “great platform by @marclou”⁶⁹. His newsletter says: “I’ve made $3M with my 36 startups. Solo. 85% margin. Bootstrapped. It took 10 years” and “90% of my 36 startups failed. And it’s OK!”¹⁸. “Bootstrapped” means built without investors. One of his products, DataFast, shows $30,251 MRR from 1,422 subscriptions, roughly 27 months after it was founded, verified through Stripe²⁰.

Tony Dinh reported in November 2024 that his product TypingMind “has reached $1M in revenue in the last 12 months”, 20 months after he launched it. It began as a one-time purchase, not a subscription²⁸. Pieter Levels wrote in March 2026 that his Photo AI was “making 105,000/mo revenue and 80,000/mo profit”²⁹. The first version of this page said the line gives no currency. That is true of the post’s title only: the body of the post gives “$105,000/mo revenue” and “$80,000/mo profit”²⁹. Both are self-reported; we saw no payment verification for either²⁸²⁹. Later figures for both are in “The upside” below.

At the far end, TechCrunch reported that a six-month-old product, Base44, was sold to Wix for $80 million. It had one owner and, by the time of the sale, eight employees. Its founder had earlier founded a company funded by investors, and the product is itself a tool for vibe coders³⁰.

Correction, 2026-10-07: the $80 million headline overstates what was paid at the sale. Wix’s annual report for 2025, filed with the US Securities and Exchange Commission, records total purchase consideration of about $92.2 million. About $18.1 million of that was cash. The other $74.1 million is the estimated value of an earnout: later payments that depend on revenue targets for 2025 to 2028⁶⁴. Separately, the founder and staff can receive about $43.0 million in retention payments if they stay⁶⁴. We prefer the filing to the news report. On staff, an Israeli business paper reported six employees, all hired the month before the sale, where TechCrunch reported eight; both are secondary³⁰⁶⁶.

The small claims in our own collection

Three tweets in the set report small sums: about $250 of monthly recurring revenue from a few hundred users after about $40 of tools; about $2,500 a year from a clone built in two days; and a little over $500 in one month¹. None is verified. But $208 to $250 a month sits close to the medians above ($72 to $169 a month, depending on the dataset)⁶¹⁰. The $10,000-and-up figures in the other tweets sit in roughly the top 5% to 10% of products that got as far as charging: 4.6% of apps, and 6.1% to 10% of listed web products⁶¹⁰¹¹.

What the rules allow now

Building and selling a small software product is legal and allowed on every platform we checked. The risk lies in the shortcuts some tweets recommend.

App stores

Clones and thin apps. Apple’s review rules say: “Don’t simply copy the latest popular app on the App Store, or make some minor changes to another app’s name or UI and pass it off as your own”³². They name categories, “such as dating, flashlight, sound effects, wallpaper, simple timers, and fortune telling”, where Apple “will not accept new submissions unless they offer a meaningfully different or improved experience”, and warn that repeated submissions “may lead to removal from the Apple Developer Program”³². One tweet in our set praises a pay-per-use flashlight app¹.

For 2024, Apple reported reviewing 7.7 million submissions, rejecting more than 1.9 million, and rejecting over 320,000 as copies, spam or misleading. Separately, and for other reasons, it terminated more than 146,000 developer accounts³¹.

The 2025 figures are higher on every line. Apple reviewed 9.1 million submissions and rejected more than 2 million, of which over 1.2 million were new apps. It rejected more than 371,000 as copies, spam or misleading, and terminated 193,000 developer accounts⁸². It also says more than 306,000 new developers joined in 2025⁸².

Other rules. Since November 2025 Apple requires apps to get “explicit permission” before sharing personal data with a third-party AI³³. Inside an iPhone app, subscriptions must use Apple’s own payment system³².

Android timing. Google Play requires new personal developer accounts to “run a closed test for their app with a minimum of 12 testers who have been opted in continuously for at least 14 days” before the app can be published. A closed test is a release to invited people only. Registration costs $25³⁴. So a new developer cannot publish an Android app in the first two weeks, however fast it was built.

Bans. Google does “not allow apps that only have limited functionality and content”³⁵. If an account is terminated, “any related Google Play developer accounts will also be permanently suspended”, and new accounts are terminated too, with no refund of the fee³⁵.

Web payments

Stripe’s agreement says it “may terminate this Agreement or close User’s Stripe Account at any time”³⁶. Its list of restricted businesses includes “no-value-added services, including the sale or resale of a service without added benefit to the buyer” and “reduced price trials with unclear or hidden pricing”³⁶. Ordinary software is allowed. One tweet in our set advises reselling another company’s AI product under your own name through its API, with features added¹. Whether a resold product adds enough is Stripe’s call. We do not say that the tweet’s advice breaks Stripe’s rules; setting the two side by side is our reading, not a statement on Stripe’s page.

Card disputes (customers asking their bank to reverse a charge) carry penalties. Sellers placed in a card network’s monitoring programme “can incur monthly fines and additional fees”, though Stripe calls these programmes comparatively rare³⁷.

Subscription and tax law

US federal law requires a subscription seller to disclose terms clearly, get the buyer’s express consent, and provide “simple mechanisms for a consumer to stop recurring charges”⁴³. A stricter 2024 “click-to-cancel” rule from the Federal Trade Commission (FTC), the US consumer regulator, is not in force. In February 2026 the FTC published a revision of the rule “To Conform These Rules to Federal Court Decisions”, and in March 2026 it asked for public comment on new changes⁴³.

The EU runs a VAT registration scheme for sellers based outside it: “Any taxable person, not established in the EU, who supplies services to non-taxable persons taking place in the EU, can register in the non-Union scheme”⁴⁴. The first version of this page said, from memory, that VAT on digital services is due in the customer’s country from the first sale. Read at source, that needs correcting. The EU’s guidance for businesses confirms that electronic services sold to consumers are taxed in the customer’s country. It also says “A VAT threshold of EUR 10 000 applies” each year, below which a seller established in the EU may charge its home country’s VAT⁸⁴. So “from the first sale” is not right for sellers based in the EU. The page does not say whether the threshold covers sellers outside the EU⁸⁴. App stores and merchants of record handle it for the fees in Step 1.

In the EU, Apple shows a business seller’s address, phone number and email on the app’s store page⁴⁷.

Security

• Veracode, which sells security testing, tested more than 100 AI models and found “45% of code samples failed security tests”³⁸. The test, from July 2025, covers isolated coding tasks, not finished apps.
• CVE-2025-48757, an entry in the public register of software flaws, records that sites generated by Lovable up to 15 April 2025 could let outsiders “read or write to arbitrary database tables”. Lovable disputes the record, saying each customer is responsible for protecting their own app’s data³⁹.
• Escape, which sells security scanning, scanned 5,600 live vibe-coded apps and found more than 2,000 vulnerabilities, over 400 exposed secrets such as passwords for other services, and “175 instances of PII (including medical records, IBANs, phone numbers, and emails)”⁴⁰. PII means personal data; IBANs are bank account numbers.
• The Register reported in February 2026 on one app built with Lovable and shown on its showcase page, whose flaws exposed 18,697 user records. Lovable said the project “includes code not generated by Lovable and the vulnerable database is not hosted by Lovable”, and that the builder had not acted on its security scan⁴¹.
• The Register also reported one case of Replit’s AI coding agent deleting a user’s live database against instructions; the data was later restored⁴².

Both of Lovable’s replies make the same point this page does: the builder, not the tool, holds the users’ data. Someone who cannot read the code cannot check what the tool produced.

Suppliers and selling

A product built on an AI provider depends on that provider’s prices and terms. Anthropic’s commercial terms let it change rates with 30 days’ notice⁴⁵. We could not open OpenAI’s terms.

Flipping is allowed. Fees are in Step 8; Flippa, another marketplace, charges a 10% success fee on small sales⁴⁸. One open question: the US Copyright Office says “copyright does not extend to purely AI-generated material”, and that prompts alone “do not alone provide sufficient control”⁴⁶. The report does not rule on AI-generated code; whether it weakens what a buyer of a fully prompted codebase owns is our question, not its conclusion.

Who makes money from it

What the 52 tweets lead to

By the labels in our collection file, 21 of the 52 tweets show nothing for sale. The other 31 lead somewhere: 7 ask readers to follow and reply to receive a link by private message, 6 promote the author’s own app, 6 sell paid lists or databases of sites said to earn money (5 from one account), 4 sell services, 3 point to a video or podcast channel, 2 sell tools, and one each a newsletter, coaching and a guide¹. We did not re-open the profiles.

Parties paid whether or not the product earns

• Tool sellers. Cursor, Replit, Lovable and the others are paid every month by everyone who tries¹⁵.
• Referral payers. Lovable pays “up to $100 for each first-time subscriber you refer”⁵³. ShipFast, a set of ready-made code to start a product from, pays those who refer buyers 30% to 50% of each sale⁵¹. Shipped, a similar product, advertises 40%⁵⁴. We did not check whether any tweet in the set carries a referral link.
• Platforms. Apple and Google keep 15%, and payment companies keep their fees (Step 1).
• Marketplaces. Acquire.com charges 6% to 8% at closing⁴⁸. TrustMRR charges a 3% fee on sales arranged through it and pays half to whoever referred the buyer⁵².

Sellers to the builders

Starter Story posted 5 of the 52 tweets, including “$20k MRR app in 14 days” and “$340K/month app in 60 days”¹. It sells a programme called Starter Story Build: “Price is $395, which includes lifetime access to all course materials”²¹. Its page shows customer testimonials but no figures for how buyers do overall. On the two pages we read, we found no statement of whether the revenue figures in its interviews are checked; we did not search the rest of the site²¹. A later reading found its page on method. It says: “Most Starter Story profiles aren’t written by the founder. They’re compiled by our research system from public sources”⁸⁵. By its own account the figures are checked against those public sources and should be treated as best-effort⁸⁵. As we read it, that checks that a figure was stated in a public source, not that the money was received. The page does not cover the video interviews the tweets promote⁸⁵. The About page says Starter Story became part of the software company HubSpot in February 2026 and that “we may receive a commission if you click a link and purchase something that we have recommended”²¹.

Marc Lou’s verified page shows where one well-known builder’s revenue came from. As read on 7 October 2026 it shows 19 listings and $3,212,777 of total verified revenue; a later reading the same day showed $3,220,870¹⁷. Of that, $1.3 million is ShipFast and $833,000 is CodeFast, a coding course¹⁷. Together that is about two thirds (our sum). Both are one-off sales to people who want to build products, and both show $0 MRR¹⁷. Eight of the 19 listings show under $100 in the last 30 days, and twelve show under $300; three of the 19 are duplicate, test or old listings of the same products¹⁷. The revenue is payment-verified. The point is about the method: most of it came from selling to other builders, not from subscriptions to many small apps.

ShipFast is listed at $299 to $349 and was sold at $100 off ($199 to $249) on the day we read it. Its page says: “After you’ve got access to the repo, ShipFast is yours forever, so it can’t be refunded.” One of its own customer quotes reads “I launched a week and a half ago and I’m at $450 MRR”⁵¹.

The proof site is a business too. TrustMRR’s own listing showed $48,238 of revenue in the last 30 days when read on 2026-10-07⁵². An earlier version of this page gave $40,899. It verifies money received, not profit, ad spend or refunds¹⁶.

Others in the set. The site of Shipped gives, in its page details, the X account that posted a “free guide” tweet with 110,833 views as its creator (read 2026-10-07)¹⁵⁴. Shipped is priced at $157 to $207, shown as reduced from $257 to $307⁵⁴. We do not know whether the guide leads to it. Another tweet reports one client paying several thousand dollars a month for a product its author vibe coded¹. The site of CodeSpring, a planning tool for vibe coders priced from £39 to £199 a month, names that tweet’s author as the site’s author in its page details (read 2026-10-07)⁵⁵. We do not know whether CodeSpring is the product in the tweet. One tweet is an income report from a seller of a list of micro-SaaS examples: a little over $500 for the month, most of it from the list itself¹.

Regulators

We found no court or regulator action about micro-SaaS or vibe-coding offers. Web search was not available for this question in the first round of research. A second round with search found one nearer case, described below, and no action about micro-SaaS or vibe coding as such. The earlier nearest cases are the FTC’s September 2024 lawsuits against three sellers of AI-branded online store businesses. Of one, it alleged losses “based on deceptive earnings claims that rarely, if ever, materialize”⁵⁶. These were allegations; the cases were pending when the release was published, and we have not read how they ended. They concerned ready-made online shops, not software.

The nearer case: in August 2025 the FTC sued Air AI Technologies. It alleged deceptive claims about earnings and refunds in the sale of business coaching and of licences to resell the company’s AI software, with losses of up to $250,000 per buyer⁸⁶. These were allegations in a complaint. In March 2026 the FTC announced a settlement under which Air AI is to be banned from marketing business opportunities, and a stipulated order was filed on 24 March 2026⁸⁷. A settlement is not a court finding that the allegations were true. The FTC’s case page, read on 2026-10-07, still lists the case as pending⁸⁷. We have not read Air AI’s answer to the complaint and did not open the order. The case concerns earnings and refund claims made when selling coaching and resale licences. It is not about any tweet in our collection.

These cases concern the companies named in them only. We found no regulator action against any account quoted or named on this page, and we do not suggest any connection.

A US rule requires sellers of a “business opportunity” who make an earnings claim to hand over a written earnings statement⁵⁷. Nothing we read settles whether a course or starter kit counts.

The upside

Success here is real and documented. It is also uncommon. Every figure in this section comes from a sample that leaves out people who never finished a product or never charged for it.

What the top groups earn

The two web analyses draw on the same public list, TrustMRR, and still disagree at the top. We recomputed neither. The first is published by a company that sells a database of software ideas¹¹. The second is by one forum user who was building a paid digest of projects for buyers¹⁰.

The same company’s analysis says 23.6% of web products with any revenue reach $1,000 a month, “about 1 in 4”. It puts the top twentieth above $13,205 a month, says 1.2% reach $50,000 a month, and says 0.58%, 22 businesses, reach $100,000 a month¹¹. It calls $1,000 to $5,000 a month “a genuinely strong, achievable outcome, not a consolation prize”¹¹.

By kind of app. The odds differ by category in RevenueCat’s data. Among gaming apps, 20.0% reach $1,000 a month and 8.9% reach $10,000. Among photo and video apps, 21.4% and 7.3%. Among business apps, 14.7% and 1.6%⁶. One year after launch the top tenth earns above $4,554 a month in gaming and above $822 in travel, and the middle half of all apps earns between $16 and $429 a month⁶.

By age. Revenue grows with years, for the products that last. In the 5,079-project analysis the median was $148 a month for projects in their first year, $334 for one-year-old projects, $656 at two years, $2,399 at four and $2,509 at five¹⁰. In the other analysis, products founded in 2020, 2021 and 2022 had medians of $1,200, $1,091 and $948 a month, and about half cleared $1,000 a month (53.5%, 52.6% and 50.0%). Products founded in 2025 had a median of $168, and 23.5% cleared $1,000¹¹. Both count only products still listed. The second source says so itself: “dead projects stop being tracked”, but “revenue takes years, not months”¹¹.

The gap is widening. Stripe, the payment company, studied thousands of companies with a single founder that were set up through its Atlas service in 2022 and 2023. It gives ratios, not dollar amounts⁵⁸. It also reports on newer solo companies. In 2025 the top tenth of solo founders made 61 times the median’s revenue in the first six months, up from 34 times four years earlier⁵⁸. Median first-six-month revenue fell 23% in a year while the top tenth’s rose 19%⁵⁸. So among solo founders the upside grew for the top tenth and shrank for the middle.

At the very top, working alone is not a ceiling. Among companies without investors, solo founders at the 99th percentile came within 5% of the revenue of companies with several founders after two years. At the top tenth, companies with several founders made 53% more⁵⁸.

Across all Atlas companies, teams and funded companies included, first revenue is arriving faster. In 2025, 56% more companies than in 2024 reached $100,000 of revenue in their first six months, in 108 days against 121. A fifth charged a first customer within 30 days, against 8% in 2020⁵⁹. Across all Atlas companies the middle rose too: the median company set up in 2025 made 39% more in its first six months than the median 2024 company, the 90th percentile (the top tenth’s entry line) 52% more, and the 10th percentile 18% more⁵⁹. The fall of 23% is for solo founders only⁵⁸. Stripe gives the growth in the count, not the share that gets there⁵⁹. Atlas companies paid to set up a US company, so they are more committed than a typical reader of these tweets, and Stripe sells Atlas⁵⁸⁵⁹.

The best documented outcomes

Verified through a payment company. Each of these is read from the founder’s payment account by TrustMRR. It shows money received, not profit¹⁶. Figures are from live pages on 2026-10-07.

We do not know whether Post Bridge’s founder sells a course, whether Conductor, Web3Forms or Podawaa are run by one person, or how many followers any of these founders had at the start⁶⁰⁶¹⁶². An earlier draft of this table also listed LLM Gateway. It is removed; the reason is under “Two cautions” below.

Per subscription, these products take in from about $14 a month (Web3Forms) to about $157 (Conductor); SuperX is about $42, DataFast about $21, Post Bridge about $29 and Podawaa about $107 (our sums: monthly recurring revenue divided by subscriptions)^1920606162. TrustMRR’s listings label Conductor, Podawaa and DataFast as selling to businesses, and SuperX, Post Bridge and Web3Forms as selling to consumers^1920606162.

Post Bridge has taken $524,331 in total and $385,236 in the last 12 months⁶⁰. Conductor has taken $465,715 in total, about $157 per subscription a month (our sum), and is listed for sale at an asking price of $1,250,000. An asking price is not a sale⁶¹. Podawaa has taken $1,196,479 in total⁶².

Conductor is the clearest case we found of a narrow tool sold to businesses at a high price by someone with a small following⁶¹. Post Bridge and SuperX are the opposite case: the founder’s audience is also the product’s market¹⁹⁶⁰.

These are rare. The listing ranked 50th by monthly revenue on TrustMRR shows $41,125, and TrustMRR says it lists more than 15,000 startups¹⁶. So about 50 listings in more than 15,000 show $41,000 a month or more: about 1 in 300 (our sum). The listing ranked first shows $3,569,654 a month¹⁶. In our first reading only ranks 26 to 50 loaded; a later reading the same day loaded all 50. Several of the top 50 are agencies, courses or content services, not small software¹⁶.

Two cautions. The forum analysis names Speel.co as a breakout¹⁰. Its listing today shows $65,794 of monthly recurring revenue but $0.00 of verified revenue in the last 30 days and $180,966 in total, and it is for sale at $500,000⁶³. We could not reconcile the two figures from the listing alone; a lapsed payment connection or yearly billing could explain them. We leave it out of the table for that reason, not because we found anything wrong. LLM Gateway shows $78,764 of monthly recurring revenue, but its listing says the payment connection expired on 7 September 2026 and the revenue was last read then. So the figure is a month old and not currently verified. The listing names two cofounders, so it is not a one-person product. Its last-30-days revenue ($259,539) is far above its monthly recurring revenue ($78,764), which is normal for a product that also charges by usage but means the recurring figure does not describe it well. We leave it out of the table for these reasons, not because we found anything wrong⁶².

The person’s own statement. We saw no payment verification for these.

• Tony Dinh reported $1 million of revenue in 12 months by November 2024²⁸. In October 2025 he wrote that TypingMind “is making ~$130-160k/month” and that recurring revenue from its version for business teams had passed half of the total⁶⁷. He says the launch took in $22,000 in its first seven days and that he has a large following on X²⁸. TypingMind is not a product about making money.
• Pieter Levels reported “$105,000/mo revenue” and “$80,000/mo profit” for Photo AI in March 2026²⁹. He launched it in February 2023 and reported $61,808 a month by July 2023²⁹. In September 2026 he put all his businesses together at “about $200K-$250K/mo”²⁹. He has built products since at least 2014 and sells a book about building startups²⁹. In 2021 he wrote: “Only 4 out of 70+ projects I ever did made money and grew”⁶⁸.
• Marc Lou reports $3 million from 36 startups over 10 years¹⁸. His verified page supports the total; about two thirds of it came from products sold to other builders¹⁷.

From a company filing. Base44 is the largest outcome we found and an extreme outlier. Wix recorded about $92.2 million of purchase consideration for a product about six months old. About $18.1 million of that was cash, and $74.1 million is the estimated value of later payments that depend on revenue targets³⁰⁶⁴. Wix said in August 2025 that it expected Base44’s yearly recurring revenue to grow “from just a few million in June to $40 to $50 million by the end of 2025”. That is the buyer’s forecast, not an audited figure⁶⁵. The founder had started an investor-funded company before, and the product is a tool for vibe coders³⁰. An Israeli business paper reported that he had taken no outside money and said he had put in NIS 30,000 of his own; that is his statement⁶⁶.

A small flip. One seller’s post, shown on TrustMRR’s marketplace page, reports selling a product for $85,000 after about a year, at about $3,300 of monthly recurring revenue⁶⁹. Details are in Step 8.

Contests. Some tool companies pay builders prizes. These are contests, not income. In RevenueCat’s 2025 contest, 812 apps were submitted, and the grand-prize app made $30,017 with 1,750 paying subscribers in the contest period, as RevenueCat reports it⁷⁰. One winner started “with zero mobile development experience” and built an app in three weeks⁷⁰. Bolt’s 2025 contest paid over $1 million in prizes among more than 130,000 registered builders, which is under $8 each on average (our sum)⁷¹.

Two claims that show less than they seem to. Apple says small developers’ earnings on its store “increased by 76 percent between 2021 and 2024”⁷². That is a total for the group, in a study Apple commissioned, so it does not show that a typical small developer earned more. Lovable says 80% of people building with it are in non-technical roles and 8 in 10 of those it surveyed intend to earn from what they build⁷³. A news report on Lovable’s study says some users reached five- and six-figure revenue, “though the company did not disclose how many or provide a distribution of outcomes”⁷⁴.

The far end, for scale. Among 6,525 software companies that use ChartMogul’s analytics, 3.3% reached $1 million of yearly recurring revenue within a year of first charging, 13.4% within three years and 25.1% within five⁷⁷. The author says the sample leans to serious businesses, not hobby projects⁷⁷. A separate ChartMogul report from August 2023 says the best reach that level 9 months after the first paying customer and the median company that gets there takes 2 years and 9 months⁷⁵.

A realistic good result

This is our estimate, from the samples above⁶¹⁰¹¹. It applies to products that get as far as charging.

• A good result about a year in is a top-quarter product: roughly $430 to $900 a month. The app figure is measured one year after launch⁶. The web figures cover listed products of every age, and that source calls its revenue self-reported¹¹.
• A very good result is a top-tenth product: roughly $2,500 to $5,000 a month. One analysis puts that line at $10,000¹⁰.
• About 1 in 4 to 1 in 6 reach $1,000 a month: 23.6% of listed web products with revenue, and 17.3% of subscription apps within two years⁶¹¹.
• The verified cases at about $20,000 to $65,000 a month took from about 21 months to nearly six years. Some had an audience that matched the product; others sold a tool to businesses^1920606162.

These bands describe people who got past the first hurdle. They count products, not people; they include products made by companies; and they are revenue before fees and costs. The typical result is lower and is set out at the end of the next section.

What it takes to compete

What separates the top tenth from the middle

Stripe’s study is the only one we found that measures this from payment data⁵⁸. It compares the top tenth of solo founders with the middle.

• They more often sell to businesses. Top solo founders were nearly 30% more likely than the middle to sell to businesses. That is a difference in likelihood, not a statement that most do. By month 24, “revenue for the median solo B2B founder was more than four times that of the median solo B2C founder”⁵⁸. B2B means selling to businesses and B2C means selling to consumers. This held among companies without investors too⁵⁸. The analysis of 3,787 listed products agrees in direction: a median of $198 a month for business products against $99 for consumer products¹¹. Mobile app data points the other way: in RevenueCat’s data 14.7% of business apps reached $1,000 a month and 1.6% reached $10,000, against 20.0% and 8.9% of gaming apps⁶.
• Their customers come back. “Nearly 30% of customers at top-decile solo startups returned the following month, compared with 8% at middle-decile startups”⁵⁸. Top founders were also 20 to 26 percentage points more likely to bill on a recurring basis⁵⁸.
• They sell abroad from the start. Top founders sold into 10 countries in the first month against 3, and took 51% of revenue from outside their home market against 2%. Stripe notes that part of this is founders outside the US selling into the US early⁵⁸.
• More of them build around AI. Top solo founders were about twice as likely to build products with AI at the core, and such products made almost twice the revenue of other solo companies by year two⁵⁸. The TrustMRR samples point the other way: AI is the most crowded category there, with a median of $156 a month among 1,245 projects in one analysis and $203 among 941 in the other¹⁰¹¹. The samples differ. Stripe’s is companies formed in 2022 and 2023; TrustMRR’s is projects that chose to list.

The caveats are the same as before. These are people who paid to set up a US company, the groupings are Stripe’s own, and Stripe sells the service⁵⁸.

The conditions

It works for a minority. The verified record points to these conditions:

1. You can reach buyers without paying for each one. That means a field you already work in, an audience, skill at ranking in search, or willingness to contact people directly. For mobile apps on Apple’s ad system, at median numbers, ads cost more than a subscriber returns⁶¹⁴. Tony Dinh wrote of his own product: “I tried to run paid ads and sponsorships, but the results were terrible, so I stopped”²⁸. A following helps but is not what the data singles out. On TrustMRR, a product’s total revenue to date is linked more strongly to its age and to how many sites link to it than to the founder’s follower count. Older products have had longer to add up revenue, so part of the age link is automatic. Of listed founders, 54.3% have under 1,000 followers⁹.
2. Your price fits your buyer. An earlier draft said the verified web winners charge $20 to $200 a month and need hundreds of subscribers, not thousands. The listings do not support that, so it is corrected. The verified web products on this page take in from about $14 to about $157 per subscription a month (our sums) and hold from 264 to 3,019 subscriptions^1920606162. Three of the six hold more than 1,400²⁰⁶⁰⁶². The two that take in the most per subscription, Conductor (about $157) and Podawaa (about $107), are both labelled as selling to businesses and hold 264 and 611 subscriptions⁶¹⁶². So a high price to businesses needs few buyers, and a low price needs thousands (Step 2). On mobile, RevenueCat finds high-priced apps return a median of $62.19 per payer in the first year against $10.69 for low-priced apps⁶.
3. You pick a field that is not the most crowded. In the 5,079-project analysis, marketplaces had a median of $1,170 a month, sales tools $711 and tools for online shops $599, against $156 for AI projects¹⁰. In the other analysis, sales tools had a median of $640 and 42.1% cleared $1,000 a month, while developer tools had a median of $68¹¹.
4. You expect most attempts to earn nothing. One winner reports nine failures in ten; another reports 4 that grew out of more than 70; another builder had 8 earners in 26¹⁸²⁴⁶⁸.
5. You have time. First-year projects have a median of $148 a month; five-year-old survivors $2,509¹⁰. Among established business software companies, the median takes about two years to reach 1,000 subscribers and the best take 11 months⁷⁶.
6. You can look after software, not only generate it. Billing, security, support and store rules continue after launch³²³⁸⁴⁰.
7. You keep your income while you try. The tools cost about $65 to $190 a month (our estimate from¹⁵). Quitting a job is what makes the attempt expensive. One survey points the other way in part: among 469 founders already running a software product, companies with full-time founders grew 30% faster than those with part-time founders⁸⁰. That is a link in a survey, not proof that quitting causes growth.

No source we found gives the winners’ starting position as a rate: coding skill, savings, hours a week or audience at launch.

What it costs

• Tools: about $800 to $2,300 a year, plus $99 a year for Apple or a $25 registration for Google Play (our estimate from⁴¹⁵³⁴).
• AI running costs, if the product uses AI: roughly $0.90 to $3.60 per subscriber a month at light use (our estimate, Step 6)¹⁵.
• Selling to businesses can add costs. Tony Dinh says he spent more than $20,000 on compliance, meaning the security and legal checks business buyers ask for²⁸.
• Time is the large cost: from about 21 months to nearly six years in the verified cases above¹⁹⁶².

The field is getting more crowded

• New subscription apps on RevenueCat’s platform rose about sevenfold in four years⁶. App releases across both stores were up 60% in a year, by a news report of outside data⁸¹.
• Solo founders were 63% of the C corporations (one kind of US company) formed through Stripe Atlas so far in the second quarter of 2026, which Stripe calls an all-time high⁵⁸.
• Among existing apps, the top tenth grew monthly revenue by 306% or more in a year, the median by 5.3%, and the bottom tenth shrank by more than 60%⁶.
• Among software companies with under $1 million of yearly recurring revenue in ChartMogul’s data, new-customer revenue was falling by a median of 24% a year in early 2025⁷⁶. These companies are larger than a typical solo product.

How to tell early

Signs you are on the right side, early:

• Strangers pay before or soon after launch. The commonest failure reported is that people sign up and do not pay²⁶.
• First-month customers come back. In Stripe’s data nearly 30% did so at top-tenth solo companies and 8% in the middle⁵⁸.
• Mobile app: more than about 6 in 10 monthly subscribers pay a second time⁶. Web product: more than about 9 in 10 do; median monthly churn for cheap web products was about 6% in 2021¹³.
• Trial users pay in the first week. In ChartMogul’s data conversions peak around day 7 and fall to about 1% by day 14⁷⁶.
• New payers each month outnumber cancellations.

Signs you are on the wrong side:

• Sign-ups with no payments after several weeks.
• The product is a copy of an existing app, or another company’s product with a new name and nothing added³²³⁶.
• The plan depends on selling the product for a price like 50 times its monthly revenue⁴⁹.

The base rate, restated

Our estimate of a realistic result for a capable newcomer: a first product most likely earns between nothing and about $100 a month after a year. Of subscription apps that got as far as charging and passed RevenueCat’s minimum-size cut, about 1 in 6 reached $1,000 of monthly revenue within two years and about 1 in 22 reached $10,000⁶. That is per app, it counts apps made by companies, and it is revenue in one month before fees, not income that lasts. We have no figure for a solo beginner, or for a person who launches several products. For a beginner’s single product we expect it to be lower (our estimate; the samples include companies and experienced builders).

What we did not verify

• The tweets’ own figures. Except for SuperX, no claim in the collection was checked against a payment-verified page. That includes the ”+$23,000/month MRR” tweet, the Starter Story figures, and the “$77,000 in a single month” post, whose subject we did not identify. Tweets were read from our collected file, not re-opened on X.
• TrustMRR. How many startups its revenue chart covers. One researcher saw 11,263 listed startups; neighbouring charts cover 5,739 to 7,079; TrustMRR’s own statement is “over 15,000”⁹¹⁶. We did not test how it verifies revenue. Its pages are live and change daily. The counts on Marc Lou’s page and the figures in the table of verified products are as read on 2026-10-07 and will have moved since¹⁷. For the verified products added on 2026-10-07 we do not know team size, the founder’s following at the start, costs or profit, or whether the founder sells a course⁶⁰⁶¹⁶². Only DataFast’s listing gives a team size²⁰. LLM Gateway’s payment connection had expired, so its figures are not counted⁶². The labels “selling to businesses” and “selling to consumers” on listings were not checked against the products; we do not know who sets them. Its table of medians by payment provider had columns without headings in the text we received, so we used none of it beyond the $258 Stripe median⁹.
• RevenueCat. Its data covers mobile apps, not web products. Its report gives the median monthly price differently in different places (Step 1). Its 2025 report gave figures for the top 5% and bottom quarter of new apps that we could not interpret, so they are left out⁸.
• The two analyses of TrustMRR data were not recomputed, and they disagree on the top tenth and the top hundredth¹⁰¹¹.
• Stripe’s study. It gives ratios and percentages only. The dollar levels are in chart images we could not read. What counts as selling to businesses or as built around AI is Stripe’s own grouping⁵⁸⁵⁹.
• Beginners as a group. We found no source that gives outcomes for people with no earlier software or audience, none that gives results per person for people who launch several products, and none that gives the winners’ starting skill, savings, hours or audience as a rate. We found no data on what share of vibe-coded projects ever launch or charge; Lovable published no spread of outcomes⁷³⁷⁴.
• Who promotes what. Whether CodeSpring is the product in the tweet that reports one high-paying client. Whether the “free guide” leads to Shipped. Starter Story’s research page describes how its written profiles are compiled (see “Sellers to the builders”); nothing we read says the figures in its video interviews are checked against payment records⁸⁵.
• Quotes. Many pages were first read through a tool that summarises them. The quotes from Apple’s review rules and developer news, Google Play’s help pages, Vercel’s price page, Adapty, Lovable’s affiliate page, Starter Story and ShipFast were checked word for word against the page text on 2026-10-07, as were the Paddle and Lemon Squeezy fees and the Base44 figures in Wix’s filing. The seller posts quoted from TrustMRR’s marketplace page did not load when we fetched the page again that day, so they stand as first read⁶⁹. Stripe’s US prices, RevenueCat, Lovable’s prices, BigIdeasDB and the Dev Genius article were read through r.jina.ai. OpenAI’s terms and one security researcher’s write-up refused us. The sources added on 2026-10-07 were quoted from page text fetched directly or through r.jina.ai, not from a summarising tool.
• No data found on web advertising costs, or on how many people start and never launch or never charge. For web advertising only marketing-agency pages turned up, and we did not open them.
• Churn on small web products. The two newer figures in Step 4 come from a founders’ survey read in a podcast transcript, and from companies with more than $250,000 of yearly revenue⁷⁸⁷⁹. We still have no measured churn for products the size of a solo beginner’s. The survey report itself was not opened, and the tiles on the organiser’s page were partly garbled in the text we received⁸⁰.
• AI running cost. The figure in Step 6 uses one provider’s list prices and a usage level we assumed. OpenAI’s price page returned nothing to us¹⁵.
• Crowding. The app-release figures are a news report of Appfigures data; we could not open Appfigures’ own page⁸¹. A figure that the top 1% of app publishers take 92% of in-app revenue was seen only on a site that collects statistics from elsewhere, so it is not used. Totals for Lovable’s projects were seen only in a trade article and are not used.
• Flip outcomes. The share of listings that sell, prices for very small products, and prices in 2021. Acquire.com’s figures come from the marketplace itself. The $85,000 sale in Step 8 is the seller’s own post; no contract or buyer confirms the price⁶⁹. How many sales TrustMRR’s marketplace has closed: a figure in an earlier draft could not be found on re-reading and was removed.
• Contests. The 2025 contest figures are as RevenueCat reports them, and we did not check the contest dates. A 2026 prize pool of more than $740,000 was seen only on a third-party site and is not used⁷⁰.
• Self-reported incomes (Tony Dinh, Pieter Levels, the forum posts) were not checked against payment records. We found no TrustMRR listing for TypingMind or Photo AI.
• Base44. How many people worked there at the sale: one news report says eight, another six³⁰⁶⁶. Whether the earnout targets were met.
• Dates. The publication date of Lovable’s blog post was not in the text we received⁷³. ChartMogul’s growth report dates from August 2023, so its figures are three years old⁷⁵.
• Law. Whether the EUR 10,000 VAT threshold applies to sellers outside the EU (our recollection, from memory, is that it does not). Apple’s 30% standard rate rests on its 2020 announcement⁸³. Apple’s 2026 EU fee changes, US rules on linking to outside payment, EU data-protection law, state subscription laws, and US sales and income tax were not read. The FTC’s February 2026 notice was read by its title only. Whether the court has entered the stipulated order in the FTC’s Air AI case, and what the order says beyond the FTC’s summary⁸⁷. How the FTC’s three September 2024 cases ended⁵⁶. We did not search separately for court or regulator actions under the name of each account quoted or named here.
• Search limits. The shared web-search budget ran out during the first round of research. The survey of sellers and the rules research were done only by opening known pages, so both are thinner than our method asks for. A second round on 2026-10-07, with search, covered the upside, what separates the people who succeed, and regulator actions.

Sources

1. Does It Pay collection of 52 tweets on this scheme, collected 2026-10-06. Read on 2026-10-07. Counts by “sells” and “kind” use the collector’s labels.
2. Stripe, US pricing. https://stripe.com/us/pricing . Undated live page. Read at source on 2026-10-07 through r.jina.ai (a direct visit returned the Finnish price page). Per-subscriber sums are our estimate.
3. Paddle pricing, https://www.paddle.com/pricing , and Lemon Squeezy pricing, https://www.lemonsqueezy.com/pricing . Undated live pages. Read at source on 2026-10-07.
4. Apple, App Store Small Business Program, https://developer.apple.com/app-store/small-business-program/ ; membership details, https://developer.apple.com/programs/whats-included/ and https://developer.apple.com/programs/enroll/ . Undated pages. Read at source on 2026-10-07.
5. Google Play Console Help, service fees. https://support.google.com/googleplay/android-developer/answer/112622 . Undated page; states changes effective 2026-06-30 and 2026-09-30. Read at source on 2026-10-07.
6. RevenueCat, “State of Subscription Apps 2026”. https://www.revenuecat.com/state-of-subscription-apps . March 2026 (2025 data). Read at source on 2026-10-07 through r.jina.ai. The 2% conversion rate, download counts and the ad comparison are our estimate.
7. RevenueCat, insights from the 2026 report. https://revenuecat.com/sosa-26-insights/ . 2026, exact day not shown. Read at source on 2026-10-07.
8. RevenueCat, “State of Subscription Apps 2025”. https://www.revenuecat.com/state-of-subscription-apps-2025/ . 2025. Read at source on 2026-10-07; publication month not confirmed.
9. TrustMRR, statistics. https://trustmrr.com/stats . States “Last updated: Oct 7, 2026”. Read at source on 2026-10-07.
10. Indie Hackers post, “I analyzed 5,079 Stripe-verified startups”. https://www.indiehackers.com/post/i-analyzed-5-079-stripe-verified-startups-f0f6bd053f . 2026-03-12. Read at source on 2026-10-07; one forum user’s analysis, not reproduced.
11. BigIdeasDB, “State of indie SaaS revenue 2026”. https://bigideasdb.com/state-of-indie-saas-revenue-2026 . 2026-07-03. Read at source on 2026-10-07 through r.jina.ai (direct visit refused); not reproduced.
12. ChartMogul, SaaS conversion report. https://chartmogul.com/reports/saas-conversion-report/ . 2026-02-04. Read at source on 2026-10-07. Visitor counts are our estimate.
13. ChartMogul, “What is a good customer churn rate?”. https://chartmogul.com/blog/good-customer-churn-rate/ . 2022-02-23. Read at source on 2026-10-07.
14. Adapty, Apple Ads for subscription apps. https://adapty.io/apple-ads-for-subscription-apps/ . 2026 (2025 data). Read at source on 2026-10-07. Costs per subscriber are our estimate.
15. Tool price pages: https://cursor.com/pricing , https://claude.com/pricing , https://lovable.dev/pricing (through r.jina.ai), https://docs.lovable.dev/introduction/plans-and-credits , https://supabase.com/pricing , https://vercel.com/pricing , https://replit.com/pricing . Undated live pages. Read at source on 2026-10-07. The monthly totals are our estimate.
16. TrustMRR, home page and directory. https://trustmrr.com/ and https://trustmrr.com/startups . Live pages. Read at source on 2026-10-07.
17. TrustMRR, founder page for Marc Lou. https://trustmrr.com/founder/marclou . Live page. Read at source on 2026-10-07, directly and through r.jina.ai.
18. Marc Lou newsletter, “I’ve made $3M with my 36 startups”. https://newsletter.marclou.com/p/i-ve-made-3m-with-my-36-startups . 2026-09-08. Read at source on 2026-10-07.
19. TrustMRR, SuperX. https://trustmrr.com/startup/superx and its text version, https://trustmrr.com/startup/superx.md . Live pages. Read at source on 2026-10-07.
20. TrustMRR, DataFast. https://trustmrr.com/startup/datafast and its text version, https://trustmrr.com/startup/datafast.md . Live pages. Read at source on 2026-10-07.
21. Starter Story Build, https://build.starterstory.com/ , and About page, https://www.starterstory.com/about . Undated pages. Read at source on 2026-10-07.
22. Scraping Fish, analysis of Indie Hackers revenue. https://scrapingfish.com/blog/indie-hackers-revenue . 2022-07-16. Read at source on 2026-10-07.
23. Indie Hackers, “Indie Hackers are making $60 million in Stripe-verified ARR”. https://www.indiehackers.com/post/indie-hackers-are-making-60-million-in-stripe-verified-arr-bac07f782d . 2021-09-22. Read at source on 2026-10-07; the figures used are from reader comments under the post, so secondary.
24. Indie Hackers, “4 years, 26 projects, $115k”. https://www.indiehackers.com/post/4-years-26-projects-115k-lessons-from-an-indie-hacker-7ab46733da . 2024-09-20. Read at source on 2026-10-07; self-reported.
25. Hacker News thread on side projects making $500 a month. https://news.ycombinator.com/item?id=49417766 . 2026-08-24. Read at source on 2026-10-07 through the Hacker News search service; self-reported.
26. Collab365, “Indie SaaS post-mortems point to no buyers, not bad code”. https://spaces.collab365.com/posts/indie-saas-post-mortems-point-to-no-buyers-not-bad-code-utt3f3m2 . 2026-09-09. Secondary.
27. Dev Genius, “I Quit My $200K Job to Build a SaaS. 18 Months Later: $147 MRR”. https://blog.devgenius.io/i-quit-my-200k-job-to-build-a-saas-18-months-later-147-mrr-33470638c21c . 2026-03-05. Read at source on 2026-10-07 through r.jina.ai (direct visit refused); pseudonymous, no proof.
28. Tony Dinh newsletter, “My first million”. https://news.tonydinh.com/p/nov-2024-my-first-million . 2024-11-22. Read at source on 2026-10-07; self-reported.
29. Pieter Levels’s website. https://levels.io/ . Posts dated 2026-03-06 and 2026-09-23. Read at source on 2026-10-07; self-reported.
30. TechCrunch, on the sale of Base44 to Wix. https://techcrunch.com/2025/06/18/6-month-old-solo-owned-vibe-coder-base44-sells-to-wix-for-80m-cash/ . 2025-06-18. Secondary.
31. Apple Newsroom, App Store review figures for 2024. https://www.apple.com/newsroom/2025/05/the-app-store-prevented-more-than-9-billion-usd-in-fraudulent-transactions/ . 2025-05-27. Read at source on 2026-10-07.
32. Apple, App Review Guidelines (3.1.1, 4.1, 4.3). https://developer.apple.com/app-store/review/guidelines/ . Undated page. Read at source on 2026-10-07.
33. Apple Developer News, guideline update. https://developer.apple.com/news/?id=ey6d8onl . 2025-11-13. Read at source on 2026-10-07.
34. Google Play Console Help, testing requirement for new personal accounts, https://support.google.com/googleplay/android-developer/answer/14151465 , and registration, https://support.google.com/googleplay/android-developer/answer/6112435 . Undated pages. Read at source on 2026-10-07.
35. Google Play, enforcement process, https://support.google.com/googleplay/android-developer/answer/9899234 ; minimum functionality policy, https://support.google.com/googleplay/android-developer/answer/9898783 . Undated pages. Read at source on 2026-10-07.
36. Stripe, restricted businesses, https://stripe.com/legal/restricted-businesses (updated 2026-09-22), and services agreement, https://stripe.com/legal/ssa (modified 2026-09-28). Read at source on 2026-10-07.
37. Stripe Docs, dispute monitoring programmes. https://docs.stripe.com/disputes/monitoring-programs . Undated page. Read at source on 2026-10-07.
38. Veracode, GenAI code security report. https://www.veracode.com/blog/genai-code-security-report/ . 2025-07-30. Read at source on 2026-10-07.
39. CVE-2025-48757, as held in the CIRCL database. https://cve.circl.lu/api/cve/CVE-2025-48757 . 2025-05-30. Read at source on 2026-10-07 (the two official CVE sites returned empty pages).
40. Escape, method note on vulnerabilities in vibe-coded apps. https://escape.tech/blog/methodology-how-we-discovered-vulnerabilities-apps-built-with-vibe-coding/ . 2025-10-29. Read at source on 2026-10-07.
41. The Register, on vulnerabilities in an app built with Lovable. https://www.theregister.com/2026/02/27/lovable_app_vulnerabilities . 2026-02-27. Secondary (news report); article read on 2026-10-07.
42. The Register, on an AI agent deleting a database. https://www.theregister.com/2025/07/21/replit_saastr_vibe_coding_incident/ . 2025-07-21. Secondary (news report); article read on 2026-10-07.
43. US Code, title 15, section 8403, https://www.law.cornell.edu/uscode/text/15/8403 ; FTC negative option rule page, https://www.ftc.gov/legal-library/browse/rules/negative-option-rule (notices dated 2026-02-12 and 2026-03-13). Read at source on 2026-10-07.
44. European Commission, VAT One Stop Shop registration. https://vat-one-stop-shop.ec.europa.eu/one-stop-shop/register-oss_en . Undated page. Read at source on 2026-10-07.
45. Anthropic, commercial terms. https://www.anthropic.com/legal/commercial-terms . Effective 2025-06-17. Read at source on 2026-10-07.
46. US Copyright Office, “Copyright and Artificial Intelligence, Part 2: Copyrightability”. https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf . January 2025. Read at source on 2026-10-07.
47. Apple, EU trader requirements. https://developer.apple.com/help/app-store-connect/manage-compliance-information/manage-european-union-digital-services-act-trader-requirements/ . Undated page. Read at source on 2026-10-07.
48. Acquire.com seller pricing, https://acquire.com/seller-pricing/ , and Flippa pricing, https://flippa.com/pricing . Undated pages. Read at source on 2026-10-07.
49. Acquire.com, acquisition multiples report, https://blog.acquire.com/acquire-com-biannual-acquisition-multiples-report-jan-2026/ (2026-02-11), and webinar recap, https://blog.acquire.com/acquisition-multiples-report-2025-findings-webinar-recap/ (2026-03-20). Read at source on 2026-10-07. The sale-price sum is our estimate.
50. Beancount.io, on valuation multiples for small software businesses. https://beancount.io/blog/2026/07/11/bootstrapped-saas-valuation-multiples-2026-acquire-com-indie-founders-guide . 2026-07-11 (from the address). Secondary.
51. ShipFast, https://shipfa.st/ and https://shipfa.st/affiliates ; CodeFast, https://codefa.st/ . Undated pages. Read at source on 2026-10-07.
52. TrustMRR, affiliate page, https://trustmrr.com/affiliate , and its own listing, https://trustmrr.com/startup/trustmrr . Live pages. Read at source on 2026-10-07.
53. Lovable, affiliates. https://lovable.dev/affiliates . Undated page. Read at source on 2026-10-07.
54. Shipped. https://shipped.club/ . Undated page. Read at source on 2026-10-07.
55. CodeSpring. https://codespring.app/ . Undated page. Read at source on 2026-10-07.
56. Federal Trade Commission, press release on deceptive AI claims and schemes. https://www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes . 2024-09-25. Read at source on 2026-10-07.
57. Federal Trade Commission, business guidance on the Business Opportunity Rule. https://www.ftc.gov/business-guidance/resources/selling-work-home-or-other-business-opportunity-revised-rule-may-apply-you . November 2011. Read at source on 2026-10-07.
58. Stripe, blog post on the traits of top solo founders. https://stripe.com/blog/top-solo-founder-traits . 2026-05-28. Read at source on 2026-10-07 through r.jina.ai. Stripe sells Atlas, the service the data comes from.
59. Stripe, “Stripe Atlas startups in 2025: year in review”. https://stripe.com/blog/stripe-atlas-startups-in-2025-year-in-review . 2025-12-18. Read at source on 2026-10-07 through r.jina.ai.
60. TrustMRR, Post Bridge. https://trustmrr.com/startup/post-bridge.md . Live page. Read at source on 2026-10-07.
61. TrustMRR, Conductor. https://trustmrr.com/startup/conductor.md . Live page. Read at source on 2026-10-07. The per-subscription figure is our sum.
62. TrustMRR, Web3Forms, https://trustmrr.com/startup/web3forms.md ; Podawaa, https://trustmrr.com/startup/podawaa.md ; LLM Gateway, https://trustmrr.com/startup/llm-gateway.md . Live pages. Read at source on 2026-10-07. The LLM Gateway listing states its revenue was last synced on 2026-09-07.
63. TrustMRR, Speel.co. https://trustmrr.com/startup/speel-co.md . Live page. Read at source on 2026-10-07.
64. Wix.com Ltd, annual report on Form 20-F for 2025, filed with the US Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1576789/000162828026015222/wix-20251231.htm . 2026-03-05. Read at source on 2026-10-07.
65. Wix.com Ltd, second-quarter 2025 results release, filed with the US Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1576789/000162828025038117/secondquarter2025.htm . 2025-08-06. Read at source on 2026-10-07; a company forecast.
66. Globes, on Wix buying Base44. https://en.globes.co.il/en/article-wix-acquires-israeli-vibe-coding-co-base44-1001513267 . 2025-06-18. Secondary.
67. Tony Dinh newsletter, October 2025 update, and the newsletter’s archive. https://news.tonydinh.com/p/oct-2025-updates-code-money-and-travel . 2025-10-09. Read at source on 2026-10-07 through r.jina.ai; self-reported.
68. Pieter Levels, “Only 4 out of 70+ projects I ever did made money and grew”. https://levels.io/only-4-out-of-70-projects-made-money . 2021-11-07. Read at source on 2026-10-07 through r.jina.ai; self-reported.
69. TrustMRR, marketplace page, https://trustmrr.com/acquire , and the SocialKit listing, https://trustmrr.com/startup/socialkit.md . Live pages; the seller’s post shown there is dated 2026-08-16. Read at source on 2026-10-07. The sale price is the seller’s statement; the multiple is our sum.
70. RevenueCat, winners of its 2025 Shipaton contest. https://www.revenuecat.com/blog/company/shipaton-2025-winners . 2025-10-13. Read at source on 2026-10-07.
71. Bolt, winners of its 2025 hackathon. https://bolt.new/blog/2025-bolt-hackathon-winners . 2025-08-15. Read at source on 2026-10-07 through r.jina.ai. The per-person average is our sum.
72. Apple Newsroom, on developer billings and sales in the US in 2024. https://www.apple.com/newsroom/2025/05/app-store-in-the-us-facilitated-406-billion-usd-in-developer-billings-and-sales-in-2024/ . May 2025. Read at source on 2026-10-07.
73. Lovable, “A first look at the build economy”. https://lovable.dev/blog/a-first-look-at-the-build-economy . 2026, date not shown in the text received (survey dated May 2026). Read at source on 2026-10-07 through r.jina.ai.
74. The Next Web, on Lovable’s report. https://thenextweb.com/news/lovable-build-economy-500m-arr-vibe-coding . 2026-06-09. Secondary.
75. ChartMogul, SaaS growth report. https://chartmogul.com/reports/saas-growth-report/ . 2023-08-30. Read at source on 2026-10-07 through r.jina.ai.
76. ChartMogul, SaaS go-to-market report. https://chartmogul.com/reports/saas-go-to-market-report/ . 2025-06-11. Read at source on 2026-10-07 through r.jina.ai.
77. ChartMogul, “SaaS growth: the odds of making it”. https://chartmogul.com/reports/saas-growth-the-odds-of-making-it/ . 2025-10-15. Read at source on 2026-10-07 through r.jina.ai.
78. ChartMogul, “SaaS retention: the AI churn wave”. https://chartmogul.com/reports/saas-retention-the-ai-churn-wave/ . 2025-12-10. Read at source on 2026-10-07 through r.jina.ai.
79. Startups for the Rest of Us, episode 721, on the 2024 State of Independent SaaS report. https://www.startupsfortherestofus.com/episodes/episode-721-7-key-takeaways-from-the-2024-state-of-independent-saas-report . 2024-07-09. Secondary (the report’s authors describing it; the report itself was not opened).
80. MicroConf, page for founders at an early stage. https://microconf.com/founders/0-10k-arr . Undated page; data collected late 2023. Read at source on 2026-10-07.
81. TechCrunch, on app releases, citing Appfigures. https://techcrunch.com/2026/04/18/the-app-store-is-booming-again-and-ai-may-be-why/ . 2026-04-18. Secondary.
82. Apple Newsroom, App Store review figures for 2025. https://www.apple.com/newsroom/2026/05/the-app-store-stopped-over-2-point-2-billion-usd-in-fraudulent-transactions-in-2025/ . 2026-05-20. Read at source on 2026-10-07.
83. Apple Newsroom, announcement of the App Store Small Business Program. https://www.apple.com/newsroom/2020/11/apple-announces-app-store-small-business-program/ . 2020-11-18. Read at source on 2026-10-07.
84. European Union, Your Europe, cross-border VAT. https://europa.eu/youreurope/business/taxation/vat/cross-border-vat/index_en.htm . Undated page. Read at source on 2026-10-07 through r.jina.ai.
85. Starter Story, “How we research”. https://www.starterstory.com/how-we-research . 2026-06-15. Read at source on 2026-10-07.
86. Federal Trade Commission, press release on its case against Air AI. https://www.ftc.gov/news-events/news/press-releases/2025/08/ftc-sues-stop-air-ai-using-deceptive-claims-about-business-growth-earnings-potential-refund . 2025-08-25. Read at source on 2026-10-07.
87. Federal Trade Commission, case page for Air AI. https://www.ftc.gov/legal-library/browse/cases-proceedings/airai . Last updated 2026-03-24. Read at source on 2026-10-07.

Corrections

If you are quoted or named on this page and think something is wrong, want your reply shown beside it, or want to be removed, write to [email protected]. We aim to reply within 14 days, and always within one month. Factual errors are corrected with a dated note here. A sentence that is seriously disputed comes down while we check it.

Change log:

• 2026-10-07: wording about named accounts and companies reviewed; one small account is no longer named, the Air AI case now carries its March 2026 settlement, TrustMRR’s own revenue figure is corrected to $48,238, and remarks we could not source were removed or moved to “What we did not verify”.