Evidence and data sources
Every claim we make on this site, where its number comes from, and how strong the evidence is. If a figure isn't independently checkable, we say so.
Counted from our own data or linked to a public source.
Our own measurement or user reports, not independently audited.
Example content, clearly labelled, not real data.
Claims register
20,843 validation reports scored with the current method (median 65)
Counted directly from stored reports. Distribution, formula and limits are published.
10,000+ entrepreneurs
Conservative floor. On 16 October 2026 the platform had 44,063 registered accounts.
1,028+ documented startup failures
Counted from the failure dataset at build time. Company status (closed, operating, acquired) carries a source link for 90 companies.
1,798+ structured startup ideas and 1,059+ free tools
Counted from the datasets every time the site is built.
Competitor prices on 40 comparison pages
Each price links to the vendor's own page (65 sources). The site build fails if a price is unsourced or not rechecked within 120 days.
Free tier and credits
One free idea validation, no credit card, optional one-time credit purchases. Credits never expire, as stated in the Terms (v2.4).
Validation in about 2 minutes
Target time for a single validation (120 seconds). The full 6-step journey takes about 30 minutes.
About 89% internal match rate
IdeaProof reports an internal match rate of about 89% between its scores and outcomes users later reported. It is an internal, uncertified metric, not an independent accuracy study — treat scores as decision support, not a prediction.
$2.3M+ raised and 156+ businesses launched by users
Self-reported by users. Not independently verified, and no public source is available yet.
Reddit threads in the homepage demo report
They are labelled "Illustrative, not real posts" and do not link anywhere. Real reports link to the actual threads.
Where report data comes from
- Live web research at the time of each validation, with source links shown in the report.
- Our documented failure database (1,028+ cases), used to flag patterns that sank similar companies.
- 13+ curated research sources used as reference data.
- AI models from several frontier providers, cross-checked. The score itself comes from a fixed published formula.
How we keep claims honest
- Site counts are recalculated from the datasets on every build, not typed by hand.
- An automatic check blocks stronger accuracy wording, outdated credit numbers and invented competitor scores before the site is published.
- Comparison pages show the date each competitor's price was checked, plus the vendor's own link.
- A company is labelled "failed" only when it has actually closed and that is backed by a source.
Spotted something wrong? Email us via the About page and we'll correct it.