AI Search Visibility: How to Get Cited in AI Overviews and ChatGPT

AI search visibility comes from the same foundations as ranking in classic Google Search: your pages need to be crawlable and indexed, and they need to answer questions clearly and credibly. There’s no secret trick, as Google itself says. Below you’ll find how AI Overviews, ChatGPT search, Perplexity and Copilot pick sources, what helps you get cited, and how robots.txt controls who may use your content.

What AI search is, and what GEO means

AI search tools answer with a written response instead of a list of links, with the sources linked alongside. The main ones:

• AI Overviews: the summary Google shows above its regular results
• AI Mode: a separate view in Google Search where people can ask follow-up questions in a conversation
• ChatGPT search: ChatGPT searches the web and links to the pages its answer draws on
• Perplexity and Microsoft Copilot: answer engines that also cite their sources

The work of becoming visible in these answers now often goes by GEO, short for generative engine optimization. Microsoft even uses the term in its own Bing Webmaster Guidelines. The same document says Bing and Copilot rely on the same crawling, indexing and ranking foundation as traditional search. And GEO no more guarantees a citation than SEO guarantees the top spot.

How AI search engines choose their sources

When an AI search tool cites sources, a perfectly ordinary search is running behind the scenes. The model writes its answer from the passages that search turns up. For AI Overviews and AI Mode, Google describes a technique it calls “query fan-out”: a single question triggers several related searches across subtopics, and the answer is built from their results (Google Search Central: AI features and your website).

A good AI chatbot on your own website works the same way: find the relevant sections, then answer from them.

That leads to the most important rule: if a page isn’t in the index, it can’t be cited. In Google’s words, a page must be indexed and eligible to be shown in Google Search with a snippet. There are no further technical requirements. Google also says you don’t need to create new machine-readable files, AI text files or special schema.org markup.

Service	Where the sources come from	Crawler used for search
AI Overviews, AI Mode	Google’s index	Googlebot
ChatGPT search	its own crawler and, according to OpenAI, third-party search providers	OAI-SearchBot
Perplexity	its own crawling of websites	PerplexityBot
Microsoft Copilot	Bing’s index	Bingbot

What helps you get cited in AI answers

None of this is a special trick. It’s what Google and Microsoft recommend anyway.

Put the answer first

Microsoft advises placing essential information near the top of a page and avoiding long introductions. For your writing, that means the first sentence under a heading delivers what the heading promises.

Phrase headings as questions where people really ask that way, such as “How long does a bathroom renovation take?”. Keep each page focused on one main topic. And write every section so it makes sense on its own. A line like “as mentioned above, this applies here too” is no use to anyone who only sees that one section. Microsoft lists exactly this as a criterion: state facts explicitly, and don’t leave key points implied.

First-hand experience, not generic filler

A language model can summarise general knowledge by itself. What gives it a reason to link to your page is content that doesn’t exist elsewhere: how a process actually works at your business, the mistakes you see again and again, your own measurements, your own photos. In its guidance on helpful content, Google asks directly whether a piece offers original information or analysis and demonstrates first-hand expertise.

Show who’s behind the content

The same guidance asks whether it’s self-evident to visitors who wrote a piece, and whether pages carry a byline where one might be expected. Of all the quality factors, Google names trust as the most important. In practice: author names with a line about their experience, an about page, company details and contact information. Microsoft adds that people, organisations, products and places should be named clearly and consistently.

Structured data that matches the page

Structured data such as Organization, LocalBusiness or Article helps search engines understand a page. It isn’t a shortcut into AI answers. Google stresses that it must match the visible text and that there’s no special markup for its AI features. Microsoft says structured data may help but guarantees neither visibility nor traffic from AI answers.

Keep it current

Outdated details are a particular problem, because AI answers can repeat them without context. Microsoft recommends updating content when facts change and removing what’s no longer accurate. A visible “last updated” date helps your readers too.

Clean technical foundations: text in the HTML, fast pages

Google’s list of best practices for AI features starts with the technical side: allow crawling in robots.txt, and also at the level of your CDN and hosting. Some firewalls and bot protection services block crawlers without anyone noticing. Add internal links so every important page can be reached, and keep important content in text form rather than only in images or video.

Make sure your text is already in the HTML the server sends. Google can render JavaScript, but other operators don’t consistently document whether their crawlers do, so content that only appears after JavaScript runs is an avoidable risk.

Finally, Google mentions a good page experience. Our article on Core Web Vitals explains how to measure and improve load times and responsiveness.

robots.txt: search crawlers vs training crawlers

Many providers now separate crawlers that collect pages for search from those that gather training material for AI models. So you can stay visible in search and still opt out of training.

• OpenAI: OAI-SearchBot crawls pages for ChatGPT’s search features. GPTBot collects content that may be used to train OpenAI’s models. OpenAI says each setting is independent of the other, and robots.txt changes take around 24 hours to apply (OpenAI: Overview of OpenAI Crawlers). ChatGPT-User only visits pages when someone triggers an action in ChatGPT, and according to OpenAI robots.txt rules may not apply to it.
• Google: the Google-Extended token controls whether your content may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. Google states explicitly that Google-Extended doesn’t affect inclusion in Google Search and isn’t a ranking signal (Google Search Central: Google’s common crawlers). There’s no separate crawler for it; Google uses its existing user agents.
• Perplexity: PerplexityBot collects pages for Perplexity’s search results and, according to Perplexity, isn’t used to train AI models. Perplexity-User acts on behalf of individual users and generally ignores robots.txt.

One point worth stressing for Google: AI Overviews and AI Mode are part of Google Search, and Google-Extended doesn’t change that. If you want to limit what Google shows from your pages in Search, Google points to nosnippet, data-nosnippet, max-snippet or noindex.

A robots.txt that allows search but opts out of training could look like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

User-agent: Google-Extended
Disallow: /

What blocking a crawler in robots.txt does

• Googlebot blockedNo snippets in Google Search, so no source link in AI Overviews either
• Google-Extended blockedNo Gemini training or grounding; Google Search unaffected
• OAI-SearchBot blockedNo content in ChatGPT search answers, at most a link with a title
• GPTBot blockedNo training of OpenAI models; ChatGPT search unaffected
• PerplexityBot blockedPerplexity no longer collects the pages for its search results

What about llms.txt?

llms.txt is a proposed file for your site’s root directory that gives language models an overview of the site in Markdown. It isn’t an official standard. Google says you don’t need AI text files to appear in its AI search features. In the documentation from OpenAI, Microsoft and Perplexity that we checked for this article, llms.txt doesn’t appear as a requirement either. The file does no harm, but your time is better spent on your content.

Measuring results: what you can and can’t see

Google: pages that appear in AI Overviews or AI Mode are included in your overall search traffic in Search Console, within the Performance report under the “Web” search type. Google doesn’t mention a separate filter for them, so you’ll see the effect in your regular impressions and clicks.

ChatGPT: according to OpenAI, ChatGPT automatically adds utm_source=chatgpt.com to links from its search results. That makes ChatGPT show up as its own source in Google Analytics or a privacy-friendly alternative.

Other services: visits from Perplexity or Copilot usually appear under the referring site, if the browser passes it on.

Microsoft’s guidelines make a point that holds for every AI search tool: content can be visible as a citation without anyone clicking. Fewer clicks don’t automatically mean less visibility. It’s also worth typing your customers’ key questions into these tools now and then and noting which sources come up. Answers vary, but you get a rough picture.

Local businesses: your Business Profile and reviews

For local questions such as “plumber near me”, the basics of local search apply. Google’s best practices for AI features specifically mention keeping your Business Profile up to date. OpenAI also says ChatGPT can use location data to find local results.

That means a complete profile with the right category, genuine reviews, and the same name, address and phone number everywhere (NAP). Our guide to local SEO walks through it step by step.

Your checklist

1. Check robots.txt, CDN and firewall: Googlebot, Bingbot, OAI-SearchBot and PerplexityBot must be allowed to crawl
2. Decide whether to opt out of GPTBot and Google-Extended
3. Confirm in Search Console that your key pages are indexed
4. Rework key pages: answer in the first sentences, one topic per page
5. Add your own experience, processes and facts
6. Make authors, company details and contact information visible
7. Use structured data only where it matches the visible text
8. Serve content directly in the HTML and check load times
9. Keep your Business Profile up to date
10. Review the Performance report and utm_source=chatgpt.com traffic regularly

Much of this is classic SEO: clean structure, structured data, fast pages and content that answers your customers’ questions. That’s what we take care of as part of SEO and ongoing care, with one fixed point of contact and no ranking promises. Tell us through the contact form which site you’d like us to look at.