Claimed vs verified: what the revenue numbers of 168 AI-built products really show
We compared founder-reported and payment-provider-verified revenue across the whole MissumAI catalogue. Self-reported figures sit one band higher, and the likeliest reason is who gets interviewed.
- AI-built products analysed
- 168
- median verified monthly revenue band
- $1K–10K
- median self-reported monthly band
- $10K–30K
- $100K+/mo listings that are self-reported
- 6 of 7
Data snapshot: 168 listings, taken . Download the data (CSV) · Method · Limits
Key findings
- 01The typical verified listing (revenue read from a payment provider) earns $1K–10K a month. The typical self-reported one says it earns $10K–30K, one band higher.
- 0235% of self-reported figures are $30K a month or more (26 of 75). Among verified figures it is 17% (8 of 46).
- 03Six of the seven listings at $100K+ a month rest on the founder's word. One is verified.
- 04Small numbers are nearly all verified: 14 of the 21 listings under $1K a month come from a payment provider. Almost nobody brings up a few hundred dollars a month in an interview.
- 05The tool doesn't explain it. Claude Code: verified median $1K–10K, self-reported $10K–30K. Cursor: verified median between $1K–10K and $10K–30K, self-reported $10K–30K. OpenAI Codex: verified median $1K–10K, self-reported $10K–30K.
Why we looked
Every listing on MissumAI says how its revenue figure is known. Some numbers come straight from a payment provider through a public profile such as TrustMRR, which calls itself "the database of verified startup revenues"[1]. Others come from a founder's post, podcast or interview. We label the first kind verified and the second self-reported, and we explain the labels on the About page.
With 168 products in the catalogue, we can stop asking about one number at a time and look at the pattern: do the two kinds of figures tell the same story about how much AI-built products earn? They don't.
The data
Of the 168 listings, 139 carry a traction figure and 29 carry none. 129 figures are revenue we could place in a monthly band; the rest are user or customer counts. Here is how the figures split by proof.
| Proof level | Listings (share) | With a band | Median band |
|---|---|---|---|
| Verified (payment provider) | 46 (27%) | 46 | $1K–10K |
| Dashboard shown by founder | 15 (9%) | 15 | $10K–30K |
| Self-reported | 75 (45%) | 67 | $10K–30K |
| Press | 3 (2%) | 1 | n = 3 |
| No figure | 29 (17%) | 0 | – |
Median band counts every listing in the group, including those without a revenue band (treated as the lowest value). 8 self-reported listings give users or customers instead of revenue. Groups under 5 show their size instead of a median. Shares may not add to 100% because of rounding.
Where the money sits
The chart compares the three proof levels with enough listings, as a share of each group, so the different group sizes don't distort it. The verified bars lean toward the bottom bands; the self-reported bars lean toward the top.
| Band | Verified | Dashboard shown | Self-reported |
|---|---|---|---|
| $100K+ a month |
| ||
| $30K–100K a month |
| ||
| $10K–30K a month |
| ||
| $1K–10K a month |
| ||
| Under $1K a month |
| ||
| No revenue figure |
| ||
Each cell shows the number of listings and its share of that proof level. Bands are monthly revenue, as used in our ranking formula.
Read it from the top. 39 listings sit at $30K a month or more. 8 of them are verified, 5 show a founder's dashboard and 26 are the founder's word. At $10K a month or more, it is 39% of verified listings against 65% of self-reported ones.
The dashboard group sits in between: its median is $10K–30K a month, and 33% of it is $30K a month or more. With only 15 listings, that is a hint, not a result.
Does the tool change the picture?
Not much. A listing can name more than one tool, so the rows below overlap. Where a tool has at least 10 verified and 10 self-reported revenue figures (Claude Code, Cursor and OpenAI Codex), the verified median is a full band lower for Claude Code and OpenAI Codex and between bands for Cursor, still below its self-reported median. The other tools have too few verified listings to say.
| Tool | Listings | Median, verified (n) | Median, self-reported (n) |
|---|---|---|---|
| Claude Code | 82 | $1K–10K (27) | $10K–30K (29) |
| Cursor | 47 | $1K–10K / $10K–30K (14) | $10K–30K (26) |
| OpenAI Codex | 30 | $1K–10K (11) | $10K–30K (11) |
| Claude | 24 | Under $1K / $1K–10K (6) | $30K–100K (11) |
| Lovable | 18 | n = 3 | $30K–100K (9) |
| Replit | 10 | n = 1 | n = 3 |
| ChatGPT | 9 | n = 1 | $10K–30K (5) |
| Rork | 7 | n = 3 | n = 4 |
Tools named by at least 7 listings. Medians only for groups of 5 or more; smaller groups show their size. Two values mean the middle of an even group falls between two bands. Tool names as written in each listing's sources.
Where verified numbers come from
44 of the 46 verified figures come from TrustMRR profiles and 2 from Indie Hackers revenue pages, both of which read revenue from the payment provider the founder connects. By provider: Stripe 32, RevenueCat 8, Dodo Payments 2, Lemon Squeezy 2, Polar 1 and Superwall 1.
What we think it means
This is not evidence that founders exaggerate. A founder's number can be exactly right. The likelier story is selection. Self-reported figures reach us through interviews, podcasts and launch threads, and those formats go looking for wins: a $40K month makes an episode, a $400 month doesn't. Verified profiles work the other way: they show whatever the connected payment account shows, so they also catch the quiet, small middle.
Three practical reads follow from that.
If you're building: benchmark yourself against the verified column, not the podcast circuit. On our data, a verified AI-built product most often makes $1K–10K a month. Being there is normal, not failure. The full list is on vibe-coded apps making money, which only shows verified and dashboard figures.
If you're reading revenue claims: treat a big unverified number as a picture of the top of the range, not the typical case. It may be true; it just isn't representative.
If you have proof: show it. A small verified number is more useful to other builders than a big round one, and on MissumAI a verified figure scores more than a self-reported one (see how ranking works).
Method
- Sample
- All 168 listings in the MissumAI catalogue on 2026-11-16 (catalogue commit ac27589), including 1 marked shut down, at its last sourced figure.
- Revenue band
- Each listing's revenue tier from our ranking formula: monthly revenue under $1K, $1K–10K, $10K–30K, $30K–100K or $100K+. Annual figures are divided by 12; one-off totals use the implied monthly average. Tiers are set by hand from the sourced figure.
- Proof level
- Verified = read from a payment provider through a public third-party profile. Dashboard shown = the founder showed a revenue dashboard. Self-reported = the founder or the tool maker said it. Press = a named publication reported it.
- Median band
- The middle listing of a group when sorted by band. Where the middle of an even group falls between two bands, we show both.
- Data
- One row per listing, frozen on 2026-11-16: download the CSV. Each listing's page on MissumAI links the source of its figure.
Limits
It isn't a random sample. MissumAI lists products with a public account of being built with AI, so it over-represents founders who talk about their build. That is the same selection the findings are about, and it cuts both ways.
Bands are coarse. Five bands hide a lot: $1,100 and $9,900 a month sit in the same one. The medians are bands, not dollar amounts.
Figures differ in kind and age. Some are MRR, some the last 30 days, some annual run rates; verified figures in the catalogue are refreshed almost daily, self-reported ones are as of the post that stated them. A self-reported figure can be months old.
Small groups. The dashboard and press groups and most per-tool cells are small. We don't draw conclusions from them.
The newsletter
New research, by email
Occasional email with new projects on the leaderboard and new build stories. Every number sourced, like the site.
Sources
- [1]TrustMRR: the database of verified startup revenuesTrustMRR · checked Nov 16, 2026Also a marketplace where startups are listed for sale.
Catalogue figures link to their own sources on each project's MissumAI page. Something wrong? Tell us and we'll fix it.