Should my headings be phrased as questions for AI?
Sometimes, and the evidence is mixed. A heading that asks what its section answers sits closest to the question a person types, which is how retrieval systems match. But large studies disagree about whether question-style headings are cited more, and the search engines say no special writing is needed. So we measure it and show it, without scoring it.
What it actually is
On a page with at least three second or third level headings, we count how many are phrased as a question: they end in a question mark, in any language, or open the way a search query opens, such as what is, how do or why does. Statements like what we do or how it works are not counted. The first level heading is left to its own check, and nothing here changes how headings are judged elsewhere.
Why it matters
Retrieval systems split a page into sections and compare each one with the question being asked, and the heading is the label that comparison leans on. One large study found that matching the question in titles and paragraphs helped pages within a site, while features like FAQ blocks did not. Another found question marks twice as common in the text that gets cited. A third found question-style titles cited less. All three are correlations, not experiments.
What good looks like
Section headings that name the question the section answers, with the answer given straight after the heading. This suits pages that really are answering questions: guides, explainers and support content. A page that is not built that way should not be rewritten to fit.
Being honest about it
Because the evidence conflicts, this sits under measured, not scored: it never touches your score. A warning means few of your section headings are phrased as questions, which describes plenty of well-read pages and is not a defect. We only look at how a heading is worded, not whether the section below it actually answers anything.
Where most sites go wrong
Turning every heading into a question, including ones that are really labels, which reads as forced to people and gives a machine nothing extra. The other is a question heading followed by a long lead-in, so the answer arrives late instead of straight after the question.
On the web
In October 2026, 29% of the sites we scanned passed the “Section headings are phrased as questions” check.
| Result | Share of sites |
|---|---|
| Pass | 29% |
| Worth a look | 71% |
| To fix | 0% |
| Couldn't judge | <1% |
Next.js sites pass it most often, at 29%. Webflow sites pass it least often, at 25%.
Goes with
- Sites that pass this are 7.5 times as likely to pass the “A2A Agent Card” check too.
- Sites that pass this are 3.5 times as likely to pass the “AI catalog” 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