AI search optimization: how to get your site cited in AI answers

AI search optimization is the work of making your website easy for AI answer engines (Google AI Overviews and AI Mode, ChatGPT search, Perplexity and Microsoft Copilot) to find, trust and cite. In practice it’s mostly good SEO done carefully: let the right crawlers in, answer questions clearly near the top of each page, keep your business details consistent everywhere, and measure the referrals.

Here’s why it matters for a business site. In a Pew Research Center study of 900 US adults’ browsing in March 2025, people clicked a traditional search result in 8% of visits that showed an AI summary, against 15% when there was no summary. When fewer people click, being one of the sources the answer names is how you stay visible.

This guide sticks to what Google, OpenAI, Perplexity, Microsoft and Anthropic say in their own documentation as of November 2026, from crawler access to measurement.

What is AI search optimization (GEO and AEO)?

AI search optimization, also called generative engine optimization (GEO) or answer engine optimization (AEO), means making your content easy for AI systems to retrieve, understand and cite when they write an answer. The three names describe the same job.

A generative engine is a search tool that writes a summary from several web pages instead of only listing links. Classic SEO aims to rank a page in a list. AI search optimization aims to make your page one of the handful of sources an AI answer quotes and links to.

The two overlap far more than GEO vendors suggest. Google’s guide to its generative AI features says optimising for them is “still SEO”, because AI Overviews and AI Mode draw on the same Search index and ranking systems. Bing’s February 2026 guidelines, as summarised by Search Engine Journal, add that GEO can’t guarantee a citation any more than SEO guarantees a ranking.

If your site already has crawlable pages, clear answers and a good reputation, you’re most of the way there. If it doesn’t, that’s the first thing to fix, and you can tell us what you need when you’re ready.

How do AI search engines choose which pages to cite?

Each AI search engine runs a search against an index (its own or a partner’s), picks pages that answer parts of the question, and writes a summary with links to them. The first filter is the same as classic search: if a page isn’t indexed, it can’t be cited.

Google AI Overviews and AI Mode

Google’s AI features documentation says there are “no additional requirements to appear in AI Overviews or AI Mode”. A page must be indexed and eligible to appear in Search with a snippet. Both features use query fan-out: the model runs several related searches on subtopics, then shows a wider set of supporting links than a classic results page.

In Alphabet’s Q2 2026 earnings remarks on 22 July 2026, Sundar Pichai said AI Mode had passed 1 billion monthly active users and that AI features in Search send “billions of clicks to websites every week”.

ChatGPT search

ChatGPT finds pages through OpenAI’s search crawler, OAI-SearchBot. OpenAI’s crawler documentation is blunt: sites that opt out of OAI-SearchBot “will not be shown in ChatGPT search answers”. OpenAI hasn’t published which outside search providers ChatGPT uses today. Microsoft did announce in May 2023 that Bing would be the default search engine in ChatGPT, so being well indexed in Bing is a sensible safeguard.

Perplexity

Perplexity runs its own crawler. Its documentation says PerplexityBot is “designed to surface and link websites in search results on Perplexity” and isn’t used to train AI foundation models.

Microsoft Copilot

Copilot answers are grounded in Bing’s index, which Bingbot crawls. Per Search Engine Journal’s summary of Bing’s 2026 guidelines, a NOARCHIVE meta tag keeps a page out of Copilot answers altogether, so check your templates don’t set it by accident.

Which AI crawlers should you allow in robots.txt?

Allow the search crawlers (Googlebot, Bingbot, OAI-SearchBot, PerplexityBot and Claude-SearchBot) if you want AI tools to cite you. Training crawlers are a separate decision: blocking them keeps your content out of future model training without removing you from AI search answers.

A robots.txt file is a plain text file at the root of your domain that tells crawlers which paths they may fetch. User agent names change, so check each company’s page before editing yours.

Google-Extended trips people up. Google’s crawler documentation says it “does not impact a site’s inclusion in Google Search”, so it isn’t the way to leave AI Overviews. For that, Google offers a Search generative AI control in Search Console (Settings, then Search generative AI), rolled out to all sites by 31 August 2026. It removes a property from AI Overviews, AI Mode and Discover’s AI features while keeping it in normal results.

OpenAI and Perplexity both say robots.txt changes take up to about 24 hours to apply. And robots.txt isn’t the only gate: Google asks owners to allow crawling “through any CDN or hosting infrastructure”, because a firewall rule can block a bot that robots.txt allows.

To opt out of OpenAI training but stay in ChatGPT search, add a group with User-agent: GPTBot and Disallow: /, and leave OAI-SearchBot allowed.

How should you structure pages so AI tools quote them?

Write pages that answer one question each, put the answer in the first one or two sentences under each heading, and back it with specifics. AI engines lift passages rather than whole pages, so each passage should make sense when it’s quoted on its own.

What that looks like on the page:

• Question-style headings that use the words customers type or say, such as “How much does a kitchen refit cost in Leeds?”
• Answer first. One or two sentences that still read correctly if quoted alone, with no “this” or “as mentioned above”.
• Specifics. Numbers with units, dates, named standards and linked sources.
• Tables for comparisons. Microsoft’s AI Performance launch post names clear headings, tables and FAQ sections as formats that help AI systems reference content accurately.
• One topic per page. Bing’s guidelines say single-topic pages are more likely to be chosen for grounding.
• Main content in the HTML. Vercel’s December 2024 analysis found that “none of the major AI crawlers currently render JavaScript” (Google’s Gemini, which uses Googlebot, and Applebot were the exceptions), so text that only appears after scripts run may be invisible to them.

Say a 15-person physiotherapy practice in Brisbane has one Services page listing twelve treatments in a single block of text. Ask an assistant whether dry needling helps shin splints and nothing on that page answers it. A dry needling page that opens with a two-sentence answer, a table of session lengths and prices, and the physiotherapist’s name gives the engine something to quote. (This is a hypothetical example.)

Don’t overcorrect, though. Google says there’s “no requirement to break your content into tiny pieces”, and that pages made for every variation of a query, mainly to manipulate AI answers, break its scaled content abuse policy.

Does structured data help with AI search optimization?

Structured data helps engines understand what a page describes, but it isn’t a ticket into AI answers. Google says there’s “no special schema.org structured data that you need to add” for AI Overviews or AI Mode, and that any markup must match the visible text on the page.

Structured data is code (normally JSON-LD using the schema.org vocabulary) that labels facts on a page, such as a business name, an author or a product price. It still earns rich results in classic search, and it removes ambiguity, so it’s worth having for the basics:

• Organization or LocalBusiness on your homepage, with name, logo, address, phone and a sameAs list linking your official profiles.
• Article on blog posts, with the author and the date you last updated it.

Google’s Organization markup guide describes sameAs as the URL of “a page on another website with additional information about your organization”. That’s how you tell engines that your LinkedIn page, Google Business Profile and website all belong to the same company.

Why do entity consistency and third-party mentions matter?

AI engines build answers from what several sources say about a business, so your name, address, services and people should read the same everywhere. When details conflict, an engine has less reason to be confident that two pages describe the same company.

An entity is a distinct thing that search systems can recognise and describe, such as a business, a person or a product. Bing’s 2026 guidelines, as summarised by Search Engine Journal, recommend stating facts directly, keeping entity names clear and consistent, and placing essential information near the top of the page.

If you sell in more than one country, keep your Google Business Profile and Bing Places listings in each market matched to your registered name: your state filing in the USA, Companies House in the UK, your ABN record in Australia, and your company or GST registration in India.

Third-party mentions count because they’re evidence an engine didn’t get from you. Real ones come from industry directories, local press, partner pages, conference talks and reviews. Google’s guide says chasing artificial mentions across the web doesn’t help, because its systems focus on quality and filter spam, so skip paid link schemes.

Do you need an llms.txt file?

No major AI search engine has publicly confirmed that it reads llms.txt, so treat the file as optional. Google’s generative AI guide says creating one for other services is fine and “will neither harm nor help your site’s visibility or rankings in Google Search.”

llms.txt is a Markdown file at the root of a site that summarises the site and links to its most useful pages for language models. Jeremy Howard proposed it on llmstxt.org on 3 September 2024. It isn’t a robots.txt replacement and doesn’t control crawling.

It can still be worth a few minutes, since generating one from your sitemap costs almost nothing. Just don’t expect it to change whether ChatGPT or Google cites you, and don’t pay anyone who promises that it will.

How do you measure AI search traffic?

Use three tools together: Search Console’s Generative AI performance report for Google, Bing Webmaster Tools’ AI Performance report for Copilot, and Google Analytics 4 (GA4) for visits from ChatGPT, Perplexity and other assistants. None of them shows everything on its own.

Google Search Console. The Generative AI performance report, rolled out to all websites on 31 August 2026, counts how often links to your site appeared in AI Overviews and AI Mode. It reports impressions only. Clicks from these features sit inside the normal Performance report under the Web search type.

Bing Webmaster Tools. Microsoft launched AI Performance in public preview on 10 February 2026. It shows total citations, average cited pages per day and sample grounding queries across Copilot, AI summaries in Bing and “select partner integrations”.

GA4. The default channel group now includes an AI Assistant channel. Google’s help page names ChatGPT, Gemini, Deepseek, Copilot and Grok as example sources, and it keeps clicks from AI Overviews and AI Mode in Organic Search. To catch anything it misses, such as Perplexity or Claude, create a custom channel group:

1. Go to Admin, then Data display, then Channel groups, and create a new group.
2. Add a channel called AI assistants with the condition Source matches regex, for example chatgpt|perplexity|claude|copilot|gemini.
3. Drag it above Referral, because GA4 assigns a visit to the first channel it matches.
4. Save. Custom channel groups apply to past data too, and standard properties can have two.

Finally, once a month, ask the main assistants the five questions your customers ask most and note which sources they cite. Answers can differ by country, so check from each market you sell in.

Want help setting up the tracking or fixing what it shows? Talk to us about your project.

What to do next

Start with access, then content, then measurement. None of these steps needs a rebuild:

1. Check robots.txt and your firewall against the crawler table above, and decide on training crawlers separately from search crawlers.
2. Verify your site in Google Search Console and Bing Webmaster Tools, and submit your sitemap to both.
3. Rewrite your five most valuable pages with question headings and answer-first paragraphs.
4. Make your business name, address and founder identical across your site, listings and profiles, and add Organization markup with sameAs.
5. Set up the GA4 channel group and review all three reports once a month.

If your site renders content with JavaScript only, mixes topics on every page or can’t take structured data, a rebuild may be cheaper than patching. Our guide to website redesign without losing SEO covers the migration, and how much a website costs in 2026 covers the budget.

Clean, semantic HTML also helps people using screen readers, as our website accessibility guide explains. If you’d like a software development company to plan and build it with you, request a quote.