Try it yourself
- Write down the questions your customers ask. Use their words, not yours. If they say “emergency plumber”, do not ask about “heating engineering services”. Good examples: “who fixes boilers in [your town]?” or “how much does a [your service] cost in [your area]?”.
- Start a fresh chat for each question. If you can, log out first, so your own earlier chats about your business cannot play a part.
- Look at the answer. Were you named? Did it link to your website? Which other businesses, and which websites, did it mention?
- Ask about yourself separately. Ask one question with your business name in it, to see what the assistant says about you directly. Keep this apart from the others: being found when someone types your name is not the same as being recommended to someone who has never heard of you.
Why one try is not enough
Answers can differ between sessions: the same question asked twice may return a different list of businesses, in a different order, citing different websites. A single screenshot tells you what happened once.
That is why our audit records it properly. We agree the questions with you for the Full Audit (for the Local Audit we write them from your intake form), ask them by hand in fresh logged-out sessions (in the Full Audit, each question 3 times on each of 2 assistants), and write down every answer and every website cited. You get a count, for example “named in 2 of 3 sessions”, and the websites the assistants rely on, rather than one screenshot.
Check that AI crawlers can reach your website
If your site never appears as a source, the AI companies’ crawlers (the programs that read websites for them) may not be able to reach it. Each company uses named crawlers, and they do different jobs. What the companies say, as of 31 October 2026:
- OpenAI. GPTBot collects content that may be used to train OpenAI’s models. OAI-SearchBot is used for search: sites that opt out of it “will not be shown in ChatGPT search answers”, though they may still appear as navigational links. The two are controlled separately, so you can block training and still allow search. ChatGPT-User makes visits when a user asks, and OpenAI says robots.txt rules “may not apply” to it. (OpenAI crawler documentation)
- Anthropic (Claude). ClaudeBot collects content that may be used for training; Claude-User fetches pages when a user asks Claude a question; Claude-SearchBot indexes content to improve search results. Anthropic says blocking ClaudeBot “signals that the site’s future materials should be excluded from our AI model training datasets”; blocking Claude-User or Claude-SearchBot stops that retrieval or indexing. (Anthropic help centre)
- Google. Google-Extended is a control token you use in robots.txt, not a separate crawler: Google says it “doesn’t have a separate HTTP request user agent string”. It controls whether content may be used to train future Gemini models and for grounding, in Gemini Apps and in Vertex AI. Google says it “does not impact a site’s inclusion in Google Search nor is it used as a ranking signal in Google Search”. (Google’s crawler documentation, updated 30 October 2026)
- Microsoft. We could not confirm on Microsoft’s own pages, on 31 October 2026, how Copilot uses Bing’s index. Being indexed in Bing is worth checking either way.
A one-minute check: open yourwebsite/robots.txt and see whether any of these crawlers are disallowed.
robots.txt is only half the picture. It is a request to crawlers, while some hosting and security settings block them at the server without you knowing. For example, Cloudflare documents a setting that blocks AI crawlers at the server (Cloudflare: Block AI Bots). Testing what your server really returns to each crawler is part of the Full Audit.
What a quick check cannot tell you
- It is a sample, not a measurement of everyone. Other people, in other places, with other settings, may see different answers.
- It does not tell you why. The assistant does not explain how it chose. The cited websites are a clue, not a reason.
- It does not predict next month. Answers may change as the assistants and the web change. Ask the same questions again later to see a trend.
- It is not your Google ranking. Google’s own results and its AI features are separate things to check.
Do not automate it with Google
It is tempting to write a script to do all this. Don’t point one at Google. Google’s spam policies say that “sending automated queries to Google”, including “scraping results for rank-checking purposes”, breaks its policies and terms without express permission (Google spam policies). Check by hand, or use the official tools: Google Search Console and Bing Webmaster Tools show which searches your site appears for.
Want it done properly?
- The free SEO check tells you, in about a minute, whether your robots.txt lets the crawlers behind AI search tools read your site.
- The audit records the answers by hand in the real apps (10 in the Local Audit, 30 in the Full Audit): whether you were named, whether you were linked and every website the assistant cited, with 3 competitors you name compared. Then it tells you what to change, and who should do it.
No one can guarantee you will be named
An audit can show what assistants said on the days it checked; it cannot decide what they will say. Google says of its own results: “No one can guarantee a #1 ranking on Google.” (Google Search Central) Be wary of anyone who promises that an assistant will recommend you.