Ranking and being quoted are two different jobs
A search engine ranks documents. An answer engine writes an answer and then decides which sources deserve a citation beside it. Those are different problems, and a page can be good at one and useless at the other.
The gap shows up constantly. A page ranks on the first page for its target term, but every fact on it arrives via JavaScript — so a crawler that doesn’t execute scripts fetches it and finds an empty shell. Or the page answers the question perfectly, but only in paragraph nine, under a heading that matches nothing anyone would actually ask. The ranking is fine. The citation never happens.
That is why AuditSEO scores AI visibility separately from your SEO score instead of folding both into one number. Averaging them would hide the exact problem you need to see.
What answer engines actually need
Nobody outside OpenAI, Google, Anthropic and Perplexity knows their source-selection rules, and anyone selling you a formula is guessing. But the failure modes aren’t mysterious, because they’re mechanical:
- The content has to be in the HTML. Several AI crawlers don’t run JavaScript, or run it unreliably. Client-rendered content is content an answer engine may never see.
- The answer has to be findable inside the page. A direct, self-contained response sitting under a matching heading is extractable. The same fact buried mid-paragraph, leaning on three earlier paragraphs for context, is not.
- The structure has to carry meaning. Real headings in a sane order, lists that are lists, tables that are tables. Layout built from styled divs looks identical to a human and is invisible as structure to a parser.
- The claims need to be attributable. Specific, checkable statements with dates, figures and named sources get quoted. Vague marketing copy gives a model nothing safe to cite.
- The crawler has to be allowed in. If robots.txt disallows
GPTBotorClaudeBot, none of the above matters.
Most of this overlaps with what good technical SEO already wants. The difference is that SEO tolerates a lot of sloppiness that AI extraction doesn’t.
What the checker measures
Paste a URL and you get two scores out of 100 — on-page SEO, and AI visibility — plus the specific reasons behind each. On the AI side it reads the rendered page and reports on answerability (is there a direct answer near the top?), heading structure and semantic markup, structured data and entity clarity, content depth and specificity, and the freshness and authorship signals that help a model decide you’re worth trusting.
Every result says what to change. The free Chrome extension runs the same two scores against whatever tab you’re on, which is faster when you’re working through a site page by page.
Two things this check honestly can’t tell you
An analyzer running inside a rendered page has hard limits, and they’re worth naming rather than quietly scoring anyway:
- Whether your content survives with JavaScript off. By the time an in-page analyzer runs, the scripts have already executed. Working out what a non-executing crawler would have seen needs a separate fetch of the raw HTML.
- Whether robots.txt lets the AI crawlers in. That lives at your domain root, not in the page, so it takes a second request to read.
The extension flags both and links here instead of guessing. The full audit fetches your site directly and answers both properly.
Check your AI visibility — free
Two scores, the reasons behind them, and what to fix. No signup.
Run the Free Check →How to improve AI visibility
- Server-render the content that matters. The highest-leverage fix on this list, and it helps ordinary SEO too. If a fact exists only in the client bundle, assume it may not count.
- Answer the question in the first two sentences under the heading, then elaborate. This is the easiest structural change available and it makes a page dramatically more quotable.
- Use headings that match real questions. “How much does X cost?” earns citations that “Pricing” never will.
- Be specific. Numbers, dates, named sources, version numbers. Specificity is what makes a sentence safe for a model to quote and attribute.
- Add the structured data that fits — FAQ, HowTo, Article, Organization — with the free schema generator. It disambiguates entities; it doesn’t magically rank you.
- Decide your robots.txt on purpose. Check which AI crawlers you allow and make it a choice rather than something you inherited.
- Keep pages current and dated. Stale, undated pages are risky to cite, and models behave accordingly.
AI visibility FAQ
What is AI visibility?▾
AI visibility is how usable your page is to an answer engine - ChatGPT, Perplexity, Claude, Google's AI Overviews - when it assembles an answer and picks sources to cite. It is related to SEO but not the same thing: ranking decides whether you appear in a list of links, extractability decides whether a model can lift a clean, attributable answer out of your page.
Is AI visibility the same as AEO or GEO?▾
Broadly yes. AEO (answer engine optimization) and GEO (generative engine optimization) are the two names the industry is settling between for the same discipline: making a page easy for an AI system to quote and attribute. We say “AI visibility” because it describes the outcome rather than the acronym.
Can you guarantee ChatGPT will cite my page?▾
No, and be sceptical of anyone who says otherwise. No answer engine publishes its source-selection algorithm, and outputs vary between models and between runs. What a checker can honestly do is measure the things that demonstrably block citation - content that isn't in the HTML, no clear answer near the top, missing structure, crawlers disallowed - and tell you which of them apply to you.
Do AI crawlers run JavaScript?▾
Inconsistently, and several don't. If your main content is injected client-side, there is a real chance an answer engine fetches your page and sees an empty shell. Server-rendered HTML is the single highest-leverage fix for AI visibility, and it costs you nothing in SEO terms either.
Does blocking GPTBot hurt me?▾
It's a genuine trade-off, not a mistake either way. Disallowing GPTBot, ClaudeBot or PerplexityBot in robots.txt keeps your content out of those systems, which is a legitimate choice - but it also removes you from consideration as a cited source. Make it deliberately rather than inheriting it from a copied robots.txt.
Is the checker free?▾
Yes. The on-page audit and the AI visibility score are both free with no signup, in the browser and in the Chrome extension. The extension makes no network requests at all.