What is AI visibility, and how do you measure it?

AI visibility is whether AI systems such as ChatGPT, Claude, Perplexity and Google AI Overviews can reach your site, read it, and cite it in their answers. You measure it in two layers: inputs you control (crawler access, structured data, llms.txt, readable content) and outputs you can observe (impressions in Google's generative AI features, and whether assistants name you).

What is AI visibility?

AI visibility is the chance that an AI answer engine can use your pages as a source. We break it into three questions: can the AI systems reach you, can they read you, and will they cite you? Each depends on the one before it.

Reach is technical. An assistant's crawler has to request your page and get the real content back, not a 403 from a firewall or a challenge page. Read is about structure: a clear H1, headings that match questions, passages that answer them directly, and structured data that says what the page is and who wrote it. Cite is the outcome. It is the only part you do not control, which is why we never promise it.

You will see the same idea called AI search visibility, AI brand visibility or LLM visibility. The labels differ in emphasis. AI brand visibility leans towards whether your company is named; LLM visibility leans towards the language models themselves. The mechanics underneath are the same: if the systems cannot reach and read your pages, nothing else you do will register.

How is AI visibility different from SEO?

Less than the new vocabulary suggests. For Google, it is the same discipline; for ChatGPT and Claude, it adds a second set of crawlers you have to let in.

Google's own guide to optimising for generative AI features is blunt about it: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO." It explains that AI Overviews and AI Mode are "rooted in our core Search ranking and quality systems", and that to be eligible, "a page must be indexed and eligible to be shown in Google Search with a snippet." So on Google, AI visibility starts with being indexable.

The same guide adds that you do not need new files to appear: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them." It also says LLMS.txt files "will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them." We still score llms.txt, because other services may read it, but we weight it accordingly and do not sell it as a Google lever.

Outside Google, the picture changes. ChatGPT and Claude run their own crawlers, with their own names, and a site can welcome Googlebot while quietly refusing them. That is where AI visibility and classic SEO part ways.

Which crawlers decide your AI search visibility?

For ChatGPT and Claude, the search crawlers matter most, because they are the ones the vendors tie to appearing in answers. The training crawlers are a separate decision.

OpenAI's crawler overview says "OAI-SearchBot is used to surface websites in search results in ChatGPT's search features" and that "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links." GPTBot is described separately, as crawling content "that may be used in training", and the page notes that each setting is independent of the others.

Anthropic's help page on its crawlers describes three bots. ClaudeBot collects content that could contribute to training. Claude-SearchBot "navigates the web to improve search result quality for users", and disabling it "may reduce your site's visibility and accuracy in user search results." Claude-User fetches pages when someone asks Claude a question.

Robots.txt is only half of reach. A rule in a firewall or CDN can refuse a crawler that robots.txt welcomes, and nothing in the robots file will show it. We cover the full list of user agents in our AI crawlers list.

How do you measure AI visibility?

Measure the inputs with a fixed rubric, and the outputs with whatever reporting each platform gives you. Inputs move when you fix things; outputs lag and are partly outside your control.

For inputs, we score every site on the same AI Visibility rubric, and freeze it at baseline so the after score is comparable with the before:

• AI schema and structured data: 25%
• AI crawler access: 20%
• llms.txt: 15%
• AI content readability: 15%
• AI content gap: 15%
• AI review signals: 10%

For outputs, Google gives you a direct measure. Search Console's generative AI performance report lets you "See how your organic impressions from Search generative AI features changes over time" and see which pages get the highest or lowest impressions. Impressions there are "how many times links to your site were shown to a user in a generative AI feature on Google Search." If the report says your site "hasn't received enough impressions", that is itself a reading.

For ChatGPT, Claude and Perplexity there is no equivalent report we can point you to, so we track a fixed set of questions a buyer would ask and record whether the site is named or linked. That is a sample, not a census, and we treat it that way. We go through the manual method in how to check if your site appears in AI Overviews.

What we have seen: AI visibility from 18.35 to 66.35

One SEO Autopilot engagement, a US healthcare analytics company on WordPress behind a firewall, shows what the inputs look like before and after. Baseline was frozen on 15 July 2026 and re-measured on 18-19 August 2026 by fetching all 39 sitemap URLs.

On that site the firewall returned 403 to OpenAI's and Anthropic's search crawlers on every page. The trigger was a generic bad-bot rule matching the substring "searchbot". Eleven other crawlers passed. The client's monitoring tool reported all sixteen AI crawlers as accessible, because it read robots rules instead of sending requests.

Two lines in that table deserve as much attention as the headline. Performance fell from 48 to 28, because removing the firewall rule took the CDN edge cache with it (median time-to-first-byte 0.42s to 2.39s). And AI review signals did not move at all. A rubric score measures whether you can be reached and read; it does not measure whether anyone has said good things about you. The full write-up is in AI visibility before and after.

How to check your own AI visibility

Start with reach, because a fault there hides everything else. Then work through readability and the outputs.

1. Read your robots.txt and note what it says for OAI-SearchBot, ChatGPT-User, GPTBot, Claude-SearchBot, Claude-User and ClaudeBot. Decide search and training separately.
2. Send real requests with those user agents and record the status code. A 403 or a challenge page means the crawler is refused, whatever robots.txt says.
3. Check that each important page has exactly one H1, headings that match the questions people ask, and an answer in the first paragraph.
4. Check your structured data: who the organisation is, who wrote the page, and what the page is about.
5. Open the generative AI performance report in Search Console and note which pages get impressions.
6. Ask the assistants the questions your buyers ask, and record whether you are named or linked. Repeat with the same questions each month.

Our free AI visibility checker covers the first steps for you. It reads robots.txt, llms.txt and the sitemap, crawls 50 pages, and requests the homepage once with each crawler's user agent for 11 AI crawlers. That homepage request is the checker's own, not the real crawler's, and it covers the homepage only, so treat a pass as a good sign rather than proof that each crawler can reach your site. A firewall rule like the one above is exactly why we send requests instead of only reading files; we explain that case in how a firewall blocked AI crawlers.

Frequently Asked Questions