How to Measure Visibility in ChatGPT and Google AI Results

Search behaviour is shifting. Some people now finish a search not by scanning ten blue links but by reading an AI summary in Google or an answer from an assistant such as ChatGPT. Naturally, site owners want to know: are we in those answers, and how do we measure it?

The honest answer is that this area is not yet as mature as classic SEO. Neither Search Console nor any other official source hands you an "AI visibility score". Still, with the official tools available and a disciplined manual process, you can measure AI search visibility in a meaningful way. This guide walks through that process step by step.

What does "AI visibility" mean?

Before measuring, separate what you are measuring. There are three ways to show up in an AI answer:

• Citation: The answer links to your site, next to or inside the response. This is the only form that can send traffic.
• Mention: Your brand or product is named in the answer, but without a link.
• Content use: The answer relies on information from your page, but neither your name nor your link appears. In practice this cannot be measured from the outside.

Your measurement plan should focus on the first two. For the third, it is enough to check whether you have allowed or blocked access.

Google AI Overviews and AI Mode: what Google says

Google's AI features and your website documentation is clear on several points:

• No extra requirements. To be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown with a snippet. There are no additional technical requirements.
• No special files or markup. Google states that you don't need to create new machine-readable files, AI text files or special schema.org markup to appear in these features.
• Measurement lives in Search Console. Sites appearing in AI features are included in overall search traffic in Search Console and reported in the Performance report under the "Web" search type. The documentation does not describe a separate breakdown for this traffic.
• Control uses existing tools. For Search, robots.txt rules for Googlebot apply. To limit what is shown from your pages, use nosnippet, data-nosnippet, max-snippet or noindex. Google's post on AI experiences in Search adds that more restrictive settings will limit how your content is featured in these experiences.

One point often gets confused: Google-Extended. According to Google's crawler documentation, it is a robots.txt product token that manages whether content Google crawls may be used for training future Gemini models and in some other products. Blocking Google-Extended does not affect your site's inclusion in Google Search and is not used as a ranking signal.

The practical upshot: on Google's side, AI visibility is largely an extension of classic SEO. A page that is not indexed, cannot show a snippet or does not rank well for related queries should not be expected to become a source in AI Overviews. That is why Google rank tracking remains the foundation of measurement.

ChatGPT: OpenAI's crawlers and robots.txt

In its crawler documentation, OpenAI explains that it uses different user agents for different purposes. Three matter for visibility:

These settings are independent of each other. If you want to appear in ChatGPT search answers but do not want your content used for model training, you can allow OAI-SearchBot and block GPTBot. OpenAI also notes that it can take about 24 hours for a robots.txt change to be reflected in search results, and it publishes IP ranges for its crawlers, which you can use to confirm that requests in your logs genuinely come from OpenAI.

A robots.txt group might look like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

This is only an example; which crawlers you allow is a decision for your own content and business strategy. Whatever you decide, make sure your robots.txt follows Google's robots.txt guidance and that layers such as your CDN or firewall are not silently blocking crawlers you want to let in.

llms.txt: do you need it?

llms.txt is a proposal for a Markdown file placed at the site root (or under a path) that gives large language models a short summary of the site and links to its key pages. It is a community-driven proposal, not a standard ratified by a standards body, and its heaviest use is in software documentation.

Google's documentation quoted above says AI text files like this are not needed to appear in AI Overviews or AI Mode. So it is best treated not as a ranking trick but as an optional, low-cost addition. If you have documentation, product information or policy pages that you want AI assistants to find accurately, it may be worth adding.

A measurement method you can use today

1. Check access

The first step is making sure there is something to measure. Our free AI bot and llms.txt checker quickly shows what your robots.txt allows AI crawlers to do and whether you have an llms.txt file; the tool is available today. Use the URL Inspection tool in Search Console to confirm Googlebot can reach your pages.

2. Build a fixed question set

Write 20 to 50 questions your audience might ask an assistant. Split them into three groups:

• Brand questions: "Is company X trustworthy?", "How much does X cost?"
• Category questions: "What are the best … in London?", "Which tool should I use for …?"
• Informational questions: The "how to" and "what is" questions your blog answers

Keep the list fixed. If you measure with different questions every month, you cannot compare months.

3. Run it regularly and record the results

Run each question in ChatGPT (with search enabled) and in Google; note whether an AI Overview appears and which sources AI Mode shows. For each run, record:

From this you can calculate your own metrics: for example, in how many of 40 questions your brand was mentioned, how many linked to your site, and your share compared with competitors. These are simple calculations on your own sample, not an industry benchmark.

Keep in mind that AI answers can vary by person, time, location and conversation history. Google itself notes that AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. Do not decide on a single run; repeat each question several times and, where possible, while signed out. The personalisation and location effects described in why Google rankings change apply here too, often even more strongly.

4. Check your server logs

Filter your access logs for the OAI-SearchBot, GPTBot and ChatGPT-User user agents. ChatGPT-User requests in particular point to moments when a user needed your page during a conversation, and show which pages were visited that way. Because user-agent strings can be spoofed, verify important findings against OpenAI's published IP ranges.

5. Watch traffic in Search Console and GA4

• Search Console: Clicks from Google's AI features are part of your Web performance data. Watching impressions and CTR on queries that trigger AI Overviews gives an indirect but official signal.
• Google Analytics 4: Visits from AI assistants typically appear as referral traffic. Filter the session source report by these services' domains to see which pages receive visits and whether those visits convert.

Adding this data as its own section in your monthly report is the easiest way to see change over time. We cover which metrics to include in which metrics belong in an SEO report.

Common mistakes

• Drawing conclusions from one screenshot. Appearing or not appearing once says nothing about a trend.
• Blocking AI crawlers without realising it. Security plugins and CDN bot protection sometimes block search crawlers too.
• Confusing training with search. Blocking GPTBot or Google-Extended is not the same as switching off search visibility.
• Falling for "special AI optimisation" promises. Google's documentation says SEO fundamentals apply to AI features as well. Page experience is one of those fundamentals, and Core Web Vitals and SEO is a good place to start.

AI visibility in SEOLOK

To be clear: SEOLOK's AI visibility tracking is coming soon and does not measure anything yet. The feature is being built, and when it launches it will appear in your dashboard automatically. We share details and status on the AI visibility tracking page.

What you can use today:

• The free AI bot and llms.txt checker to verify your access settings.
• Google rank tracking to see the parts of the results page beyond the blue links for your tracked keywords. The scan also covers sections such as AI summaries and flags them next to the ranking, so you can follow which keywords have an AI summary on the results page.
• The Search Console and GA4 connections, to follow your search traffic (including AI features) and conversions in one dashboard.

Quick checklist

• robots.txt applies the rules I want for Googlebot and OAI-SearchBot
• I made a deliberate decision about GPTBot and Google-Extended
• My CDN and security layer are not blocking crawlers I want to allow
• Important pages are indexed and eligible for snippets
• I have a fixed list of 20 to 50 questions
• I run the questions regularly, several times each, and record the results
• I monitor AI crawler visits in my server logs
• I track referral traffic from AI assistants separately in GA4

There is no perfect tool for measuring AI visibility yet. But a site that sets up access correctly and keeps regular, consistent records will own the most valuable asset as this field matures: its own history.