How to rank in Google AI Overviews and get cited: what actually works in 2026

How to rank in Google AI Overviews is, for the most part, how to rank in Google: AI Overviews and AI Mode pull their sources from Google's normal search index using its core ranking systems, so a page has to be crawlable, indexed, eligible for a snippet and genuinely useful before it can be cited. What changes is the unit that wins. AI answers quote passages, so pages whose sections answer specific questions clearly, with something other pages do not have, get picked more often than pages that only rank for the head term. So the answer to how to get cited in AI Overviews is to rank first, then make each section worth quoting. This guide sticks to what Google, OpenAI and other providers have documented, and labels speculation as speculation.

Quick answer: how to rank in Google AI Overviews

• There is no separate AI ranking system to game. Google's AI optimization guide (opens in a new tab) (last updated July 2026) says optimizing for generative AI search "is optimizing for the search experience, and thus still SEO."
• Eligibility is basic: the page must be indexed and eligible to be shown in Google Search with a snippet. No special file, markup or format is required. If a page is stuck outside the index, see crawled, currently not indexed.
• Make every important page crawlable by Googlebot, Bingbot and the search crawlers of ChatGPT, Claude and Perplexity, with the main content in the HTML.
• Write answer-first sections under clear headings, and include specifics only you can provide: your data, your process, your pricing logic, your experience.
• Build real entity signals: consistent business details, Organization and Person markup, a complete Google Business Profile, genuine mentions and reviews.
• llms.txt is optional. Google Search ignores it.
• Measure with Search Console's Generative AI performance report, Bing's AI Performance report and AI referral traffic in GA4.

How AI answers choose sources: what is known and what is speculation

Here is what Google has actually published about AI Overviews and AI Mode, mostly in AI features and your website (opens in a new tab) and the newer optimization guide:

• To be eligible as a supporting link, a page must be "indexed and eligible to be shown in Google Search with a snippet." There are no additional technical requirements.
• Retrieval relies on Google's core Search ranking systems to find relevant, up-to-date pages from the index.
• Both features may use query fan-out: the model issues several related searches across subtopics and then assembles an answer. Because of this, Google says AI features can show "a wider and more diverse set of helpful links" than a classic results page.
• Meeting the requirements does not guarantee crawling, indexing or being shown.

For ChatGPT, OpenAI documents that its search features use the OAI-SearchBot crawler, and that sites which block it (opens in a new tab) "will not be shown in ChatGPT search answers." Anthropic and Perplexity document similar search crawlers.

What is speculation: almost every "AI ranking factor" percentage you will read. Studies by tool vendors sample a set of prompts, record which pages were cited, and look for patterns. They are useful for ideas, but answers vary between runs, users and days, and no provider has published its weighting. Treat any claim of the form "X% of citations come from Y" as a sample, not a law.

Rankings still matter, but citations reach beyond the top 10

Since AI features retrieve from the search index, strong organic visibility remains the foundation. Query fan-out changes the picture in one useful way: when someone asks a complex question, the system also searches the sub-questions behind it. A page that ranks well for one of those sub-questions can be cited even if it is nowhere near the top for the original query.

The practical move is to cover the questions your customers actually ask on the way to a decision, each answered properly on the page where it belongs. Say you sell solar installation: "how long do panels last", "do panels work in monsoon season" and "what happens if I move house" are each fan-out candidates. Google also warns, in the same guide, that mass-producing pages just to target fan-out variations can violate its scaled content abuse policy. Depth on real questions, yes; a page per phrasing, no.

Make pages crawlable and renderable

This is where the most fixable problems hide, because they are invisible until someone looks. The checks:

1. Do not block the crawlers that feed AI answers. Search crawlers and training crawlers are separate, and you can treat them differently:

A robots.txt that stays visible in AI search but opts out of model training looks like this:

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: Google-Extended
Disallow: /

User-agent: *
Allow: /

Sitemap: https://www.example.com/sitemap.xml

Whether to block training is a business decision, not an SEO one. Blocking the search crawlers, on the other hand, removes you from those answers.

2. Check your CDN and firewall. Bot protection at the CDN or host can block crawlers that robots.txt allows. Some providers now offer one-click "block AI bots" settings. Look at what is enabled, and check server logs for 403 responses to the user agents above.

3. Put the main content in the HTML. Google says it can process JavaScript as long as it is not blocked, but do not assume every AI crawler renders pages the way Googlebot does. If your product pages or articles only appear after client-side rendering, server-render them. My Laravel SEO guide covers this for Inertia and SPA front ends.

4. Do not switch off snippets by accident. nosnippet, data-nosnippet and max-snippet limit what Google can show from a page, in AI features as well as classic results. A max-snippet:0 left over from an old experiment quietly removes a page from consideration.

A technical SEO audit catches all four of these in one pass.

Answer-first writing and passage structure

Google is explicit that you do not need to break content into tiny chunks or write "in a specific way" for AI. Its systems understand synonyms and can find the relevant part of a long page. What it asks for instead is "non-commodity content": content people find "unique, compelling, and useful."

So the goal is not AI-flavored writing. It is writing a busy person can use, which also happens to be easy to quote. Answer engine optimization best practices that hold up:

• Use headings that name the question or topic, then answer it in the first sentence below. "How long does a kitchen remodel take?" followed by "Most kitchen remodels take 6 to 12 weeks once work starts, depending on..." is a passage that stands alone.
• Be specific. Numbers, conditions, steps, examples and trade-offs. "It depends" is only useful if you say what it depends on.
• Add what only you know. Your process, your pricing logic, results from your own projects, photos of your own work. A page that restates the top five results gives a model nothing new to cite.
• Show who wrote it. A named author with real experience and a page about them. This matters for people first, and it gives every system a clearer signal of who stands behind the claims.
• Keep it current. Update dates only when you update content, and remove what is no longer true.

A quick test: read each section on its own, as if it were the only thing on screen. If it still makes sense and answers something, it is citable. A content brief that plans these questions up front makes this much easier than fixing it afterwards.

Entity signals: schema, sameAs and consistent profiles

AI answers name businesses, and to name you correctly a system has to be sure which entity you are. Google says structured data is not required for its AI features and there is no special schema for them, so markup is not a shortcut. It still helps every system read your facts without guessing.

• Add Organization (or the right LocalBusiness subtype) on your home page, with logo, contact details and sameAs links to your real profiles.
• Add Person markup on author pages, with sameAs to LinkedIn or other professional profiles.
• Use the same business name, address, phone and description everywhere: site, Google Business Profile, directories, social profiles.
• Keep markup consistent with the visible page. Google lists this among its best practices for AI features.

I keep ready-made templates in local business schema examples, and my schema markup service sets this up across a site.

Brand mentions and reviews

Being mentioned and recommended elsewhere is plausibly part of why some businesses appear in AI answers more than others, since models and retrieval systems both see the wider web. But be careful here. Google's guide says that "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem," and paid placements or self-written "best of" lists are exactly that.

The mentions worth earning are the ones that would be worth having without AI: a genuine customer story in an industry publication, a useful answer in a community where your buyers ask questions, a listing in a real comparison, a partner page that links to you because you work together. Ask happy customers for reviews on Google and the platforms your industry uses, and reply to them.

Local businesses: Google Business Profile is an AI signal

For local queries, Google says directly that Merchant Center and Google Business Profiles "can help your products and services to be visible in both AI responses and other Google Search results." In practice that means a complete profile: correct categories, services with descriptions, accurate hours (including holidays), real photos, and reviews with replies. If you do one thing for local AI visibility, do this. My local SEO work starts here for the same reason.

llms.txt: what it is, and an honest verdict

llms.txt is a Markdown file at /llms.txt, proposed by Jeremy Howard in September 2024, that gives language models a short summary of a site and a list of links to its most useful pages. The idea is reasonable. The evidence that it affects AI citations is thin:

• Google Search does not use it. Google's AI optimization guide names llms.txt as one of the unnecessary "AI text files" and says Google Search does not use them. John Mueller of Google has compared it to the old keywords meta tag.
• Other AI search providers have not documented using it to choose what to cite, as of October 2026. Some crawlers do request the file, which proves it is fetched, not that it changes answers.
• It costs little. For a documentation-heavy product, where developers point coding assistants at your docs, a well-kept llms.txt can be genuinely handy.

My verdict: skip it for a local business or a typical company site, and spend the hour on your Business Profile or one strong page instead. If you do publish one, llms.txt best practices are simple: an H1 with the site name, a one-paragraph blockquote summary, H2 sections listing your key pages as Markdown links with a short note each, nothing that contradicts your site, and a review whenever those pages change.

Measuring AI visibility

You cannot improve what you cannot see, and in 2026 there are finally first-party numbers:

• Search Console's Generative AI performance report. Google rolled it out to all websites by August 31, 2026. It shows impressions of links to your site in AI Overviews and AI Mode, by page, country, date and device. It appears only once a site has enough impressions in these features.
• Bing Webmaster Tools AI Performance report. Shows how often your pages are cited in Microsoft Copilot and Bing's AI answers, along with the queries used to retrieve them. It is in public preview.
• AI referral traffic in GA4. Build an exploration or a channel group for sessions from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. Traffic from AI Overviews themselves is counted as Google organic.
• A fixed prompt panel. Write 20 to 30 questions your buyers really ask, run them monthly in Google AI Mode, ChatGPT and Perplexity, and record whether you are cited or named. Keep the wording identical each month. It is crude, but it shows direction.

Set up the basics first: GA4 and Search Console configured correctly, with the site verified in both Google and Bing.

A 30-day action plan

Then repeat the measurement monthly. AI answers shift faster than classic rankings, so a single snapshot tells you little.

Frequently asked questions

How do I get my website cited in Google AI Overviews?

Make sure the page is indexed and allowed to show a snippet, then make it the clearest, most specific answer to a question people ask. Google uses its normal ranking systems to find sources for AI Overviews, so the usual SEO basics apply, with extra weight on passages that answer a sub-question directly. Nobody can guarantee a citation for a given query.

Is generative engine optimization different from SEO?

Mostly no. Google's own guide calls optimizing for generative AI search "still SEO." Generative engine optimization strategies that work, like crawlable pages, answer-first structure, original content and consistent entity data, are good SEO practices with more attention on passages and on platforms beyond Google.

How do I optimize for ChatGPT search?

Allow OAI-SearchBot in robots.txt and your CDN, make sure your content is in the HTML, and get indexed in Bing as well as Google. Then apply the same content and entity work as for Google. SEO for ChatGPT search has no separate markup or file.

Does schema markup help with AI Overviews?

Not directly. Google says structured data is not required for its AI features and there is no special markup for them. Accurate Organization, LocalBusiness and Person markup still helps systems identify your business correctly, so keep it, but do not expect it to earn citations on its own.

How long does it take to show up in AI answers?

Fixing crawl access and snippet controls can take effect within days of a recrawl; OpenAI, for example, says robots.txt changes take about 24 hours to reflect in its search. Content and reputation work takes months, as with organic rankings. Measure monthly and judge trends over a quarter, not a week.

If you want a second pair of eyes on your site's AI visibility, my AI search optimization service starts with exactly the audit and plan described above.