AI search can reach almost every site we audited. Most give it little worth quoting

Every month we publish totals from the websites people audit on SEO AI Audits, a free SEO audit tool our team builds. In October 2026, 106 websites were audited, 24,184 pages in all. This post looks at one question: when ChatGPT, Perplexity or Google's AI answers come to a page, can they reach it, read it, and quote it?

The short answer: reaching and reading are mostly fine. Quoting is where sites fall short.

One caveat before the numbers. These are sites people chose to audit with us, not a random sample of the web. Each site counts once, and no site is named.

Access is mostly solved

• 1 site in 106 blocked AI search tools in its robots.txt.
• 4% blocked AI training crawlers. That is a different choice: blocking training doesn't keep a site out of AI search answers.
• 22% had text that only appears after JavaScript runs. Many AI crawlers fetch the raw HTML and never run scripts, so that text may never be seen.

The median AI-readiness score was 88 out of 100, and 34% of sites scored 90 or more. So the doors are open. The problem is what's behind them.

Being quotable is the gap

These were the most common AI-search problems, as the share of sites with the problem on at least one page:

An AI answer tool is looking for a sentence it can lift: a clear claim, ideally with a number, from a source it can name. A page that asks "How long does a website redesign take?" in a heading and then spends three paragraphs on background gives it nothing to lift. A page that says "Most redesigns take 6 to 10 weeks" right under that heading does.

Five fixes you can make this week

1. Answer first, explain second. Put the answer in the first sentence after a question heading, then add the detail. We wrote up how to fix questions without direct answers.

2. Put your real numbers on the page. Prices, timings, counts, results. "Most clients see a change in 3 to 6 weeks" is quotable; "results come quickly" isn't. More in adding statistics to thin pages.

3. Show who wrote it. A name and one line on why they know the subject, on every long page.

4. Tell machines who you are. A small block of Organization schema on the homepage, with links to your profiles elsewhere:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Example Studio",
  "url": "https://example.com",
  "logo": "https://example.com/logo.png",
  "sameAs": [
    "https://www.linkedin.com/company/example-studio",
    "https://github.com/example-studio"
  ]
}
</script>

5. Check what exists without JavaScript. Open view-source on your key pages and search for your main heading and first paragraph. If they aren't there, crawlers that don't run scripts won't see them either.

The old problems haven't gone away

Across every check, not only the AI ones, the most common issues were images without width and height (80% of sites), no Permissions-Policy header (78%), links with no link text (76%) and titles too long for Google's results (75%).

Check your own site

The full October 2026 study has all 19 AI-search problems and the score breakdown. You can run the same audit on your own site for free, with no signup. We publish these numbers on the 2nd of each month, so November's come out in early December.