AI visibility · 2026-10-28 · 6 min

How is your business found in ChatGPT, Copilot and other AI assistants?

A tourist heading to the coast for a long weekend no longer types "hotel Noordwijk" into Google, but asks ChatGPT: "Which hotel in Noordwijk is within walking distance of the beach and has a good restaurant?" The answer names three or four places, with a link. If your hotel is not among them, that guest books somewhere else. For a restaurant, a bike rental or an office furniture supplier it works the same way. This article is for the owner or marketer of an SME who wants to know where that answer comes from and what they can do about it.

Where an AI assistant gets its answer

An assistant such as ChatGPT, Microsoft Copilot, Perplexity, Claude or Google Gemini draws on two sources. The first is what the language model read during training: a snapshot of the web, months or years old. The second is a search at the moment of the question. For questions about a hotel, a restaurant, a price or something nearby, the assistant almost always searches live, and names the pages it found as sources.

That second part is where you have influence. The assistant uses a search index for it: Bing's, Google's or its own, and sends its own crawler to the pages it wants to read. If you are not in that index, or the crawler cannot read your page, you do not appear in the answer. So how well you rank in Google does not tell the whole story: ChatGPT and Copilot partly rely on Bing.

Step 1: let AI assistants' crawlers in

The robots.txt file on your website states which crawlers may read which pages. The AI companies use separate names for fetching pages to answer a question and for collecting training data. You can allow or refuse them separately:

  • For search and answers: OAI-SearchBot and ChatGPT-User (OpenAI), Claude-SearchBot and Claude-User (Anthropic), PerplexityBot and Perplexity-User, and Bingbot for Copilot.
  • For training: GPTBot (OpenAI), ClaudeBot (Anthropic) and CCBot (Common Crawl).
  • Google-Extended is a separate case. It controls whether Google uses your content for Gemini: for training, and to ground answers in the Gemini app. According to Google it does not affect your place in Google Search.

If you want to be found but not end up in training data, allow the first group and exclude the second. Also check your hosting or CDN: since July 2025 Cloudflare blocks AI crawlers by default for new sites, regardless of what robots.txt says, and other providers may do the same. Ask your web developer or hosting provider how this is set up for you.

Step 2: make sure the text is readable without JavaScript

The crawlers of OpenAI and Anthropic, among others, do not run JavaScript; an analysis by Vercel showed this at the end of 2024. They read the HTML the server sends back, and that is it. If your website only builds the text in the browser, as some website builders and single-page apps do, the crawler sees an empty page. Check it by viewing the page source in your browser (right-click, "View page source"): if your service description is there as plain text, you are fine.

The same goes for text in images, PDFs behind a button and information that only appears after clicking a tab. What a visitor needs to know to choose you is best placed as plain text on its own page with a clear address.

Step 3: write what an assistant wants to quote

An assistant looks for a page that answers the question, and quotes a few sentences from it. Generic text like "we deliver tailor-made solutions with passion" gives it nothing to quote. What does help:

  • A separate page per service or product, with in the first paragraph what it is, for whom, and in which region.
  • Concrete facts: location or service area, opening hours, starting prices if you state them, what you do and do not offer, who you work for.
  • Answers to the questions customers actually ask, in their words. An FAQ with short, factual answers works well.
  • A date on articles and news, so an assistant can see whether the information is current.

Step 4: make your business details the same everywhere

An assistant combines what it finds about you: your website, your Google Business Profile, the Chamber of Commerce register, LinkedIn, industry sites and reviews. If they show different addresses, phone numbers or names, it is not sure they are the same business, and mentions you less readily.

Put name, address and phone number exactly the same in every place. Add structured data to your website (schema.org, as JSON-LD) with your organisation, address, contact details and links to your profiles elsewhere. Search engines read that data directly, and assistants that run on their index benefit from it too.

Step 5: get mentioned on other sites

For a question like "which company is good at…" an assistant leans heavily on what others write about you. Your own website alone is rarely enough. Think of Google reviews, for hotels and restaurants also reviews and listings on booking and review platforms and in local guides, listings on the site of your industry association or a quality mark, a customer case the customer publishes themselves, an article in the regional or trade press, and comparison sites in your industry. This is the same work as good ordinary findability, and it takes time.

The llms.txt file

There is a proposal to put an llms.txt file in the root of your website: a short, factual summary of your business in plain text, meant for language models. None of the large AI companies has confirmed that their assistants read it. It costs little and it forces you to write down what you are in a few sentences, but do not expect miracles from it.

Measuring whether it works

  • Ask ChatGPT, Copilot, Perplexity and Gemini the questions your customers ask, with search switched on. Do it a few times: answers differ each time and per person.
  • Look at which sources the assistant names. If competitors or comparison sites are listed, you know where you need to be mentioned.
  • Check your web statistics for visits from chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com. ChatGPT adds utm_source=chatgpt.com to some of its links; the rest arrives as direct traffic.
  • Register your site with Bing Webmaster Tools if you have not done so. It shows whether Bing knows your pages, and through IndexNow you can submit changes right away.

What we do on our own site

On axonfactory.ai we allow the search and training crawlers of the large AI companies; being found and cited is the goal. Every page is built as HTML in advance, so the text is readable without JavaScript. Name, address and phone number come from one source in the code, appear exactly that way on the site and in the JSON-LD, and we keep them identical to our Google Business Profile and our Chamber of Commerce registration. There is an llms.txt with a factual summary of what we do. And after every publication we submit every page in our sitemap through IndexNow to Bing and other search engines. None of these steps guarantees a place in an answer; together they make it easy for an assistant to find and understand us.

Nobody can guarantee a place in an AI assistant's answer. A promise of a fixed place is about something nobody controls. What you do control: that the assistant can read your site, that it says what you do, and that others write about you.

CONVERSATION
Axon Factory

Want to know how your business shows up in AI assistants, or what your website needs technically? We take a look with you in a short call.

Get in touch →