Do statistics, quotes and sources help AI quote my page?
They give a machine something it can check. A claim with a figure, a quoted source or a link to where it came from is easier for an answer engine to repeat and to attribute than a claim that only asserts. The evidence is one controlled study, so we treat it as a useful signal and nothing stronger.
What it actually is
We read the main content of the page, never the menus or the footer, and look for three kinds of checkable material: statistics written out as figures, a source set as a quotation, and links out to more than one outside source. Any one of the three is enough. Pages with too little prose to judge, such as a login screen or a homepage made of short blocks, are left out of this check altogether.
Why it matters
A generative engine has to decide which sources to lean on and how to credit them. In the one controlled study of that, researchers rewrote the same content three ways, adding statistics, adding quotations from relevant sources and adding citations, and each version was used more by the engine they tested than the original. Material a reader can verify is also material a machine can attribute, which is what turns your page into the named source rather than an anonymous paraphrase.
What good looks like
Claims that carry their number written as a number, a quote set as a real quotation with the person or publication named, and links to the original study, standard or report a statement rests on. Nothing stuffed in for the sake of it: one well-placed source beats a page of decorative footnotes.
Being honest about it
This is an indicator and never proof. The study used a simulated engine and one live one, found that the effect differs by subject, and did not measure search rankings; one of its own live results moved the other way. We cannot see whether a figure is true or a source is good, only that they are there. So the check can only pass or warn, and it never fails a page. Because the evidence is that thin, it sits under measured, not scored: it is shown in your report and it does not move your score.
Where most sites go wrong
Prose that makes confident claims with nothing behind them, such as a long article that says things have grown or improved without a single figure or link. The other is evidence that lives only in an image, a chart or a screenshot, which a machine reading the text cannot see.
On the web
In October 2026, 43% of the sites we scanned passed the “Claims come with figures, quotes or sources” check.
| Result | Share of sites |
|---|---|
| Pass | 43% |
| Worth a look | 56% |
| To fix | 0% |
| Couldn't judge | <1% |
Webflow sites pass it most often, at 63%. Next.js sites pass it least often, at 31%.
Goes with
- Sites that pass this are 9 times as likely to pass the “A2A Agent Card” check too.
- Sites that pass this are 5 times as likely to pass the “Auth.md agent registration” check too.
- Sites that pass this are 4.5 times as likely to pass the “MCP Server Card” check too.
Not enough history yet.
Based on the sites Bot Appétit has scanned: ones people chose to scan, plus lists we scan for research. Not a random sample of the web.
Not sure where yours stands?
A free scan checks this and everything else in about 20 seconds — no signup. Or if it's already flagged and you'd rather it were simply handled, we do that too.
Flagged for this one? We handle it. No pitch, just a plain-English chat.
Book a call