Is your website invisible to AI search?

A client asked me last month why their site never appeared when they asked ChatGPT for recommendations in their category. Their Google rankings were fine. Their site looked good. And to an AI crawler it was a blank page.

This is common, it is invisible from inside your own browser, and it has three causes worth knowing about.

Why does my site not show up in ChatGPT or AI Overviews?

Because most AI crawlers do not run JavaScript.

A modern website is often assembled in the browser: the server sends a near-empty HTML document and a bundle of JavaScript, and the JavaScript builds the page once it arrives. Google renders JavaScript and sees the finished page. GPTBot, ClaudeBot and PerplexityBot largely do not. They read the raw HTML, find an empty container, and move on.

You cannot see this by visiting your own site, because your browser runs the JavaScript. You can see it by viewing source, or by disabling JavaScript and reloading.

This site had exactly that problem earlier this month. The fix was prerendering: generating a static HTML copy of every page at build time, so the words are in the document before any script runs. The homepage went from zero readable words to about 1,300.

Am I accidentally blocking the AI crawlers?

Quite possibly, and it is a one-line mistake.

Many sites added blanket rules to robots.txt when AI training became a topic, or inherited them from a template or a plugin. A rule like this:

User-agent: GPTBot
Disallow: /

means ChatGPT cannot read your site at all. Not for training, and not for answering a question someone asked about your industry.

That may be a deliberate choice. It is worth making sure it is a choice rather than something a plugin decided for you. The crawlers worth checking are GPTBot and OAI-SearchBot for ChatGPT, ClaudeBot for Claude, PerplexityBot for Perplexity, and Google-Extended for AI Overviews.

What is the difference between SEO, AEO and GEO?

They overlap, and the distinction is useful.

SEO is being found: titles, structure, speed, links, the things that get a page ranked in a list of results.

AEO, answer engine optimisation, is being quotable. An assistant answering a question needs to identify which sentences on your page answer it. Schema markup, question-shaped headings and direct answers underneath them all make that easier.

GEO, generative engine optimisation, is being readable at all. This is the prerendering problem above, plus crawler access. A page that cannot be read cannot be quoted, however well written it is.

Most sites that invest in SEO have done nothing about the other two, because until recently there was no reason to.

What actually gets a page cited by an assistant?

From watching which pages get picked up, four things matter more than the rest.

Say something specific. Assistants cite pages that contain a fact they can attribute. A page of general advice about web design is interchangeable with a thousand others. A page with a number, a date and a named source is not.

Answer the question in the paragraph after the heading. Not three paragraphs later, after the context. If the heading is a question, the sentence below it should be the answer.

Mark it up. FAQPage and Article schema tell a machine what it is looking at instead of leaving it to infer.

Be attributable. Person and Organization schema, a real author, and sameAs links to profiles that corroborate it. Assistants weight sources they can identify.

How do I check my own site?

There is a free audit on this site that runs 25 checks across SEO, AEO and GEO, including whether your content survives without JavaScript and whether you are blocking any of the AI crawlers. It takes about twenty seconds and needs nothing but your web address.

If you would rather check by hand, the fastest single test is this: open your site, view source, and search the raw HTML for a sentence you can see on the page. If it is not in there, neither is your site as far as most AI crawlers are concerned.