Google AI Overviews optimization

Google Search Central’s position is that AI Overviews and AI Mode are Search. The optimization guide says generative AI features on Google Search use the core ranking and quality systems, and that optimizing for them is still SEO. There is no separate AEO switch to flip, and Google says indexing is not guaranteed even when a page meets the requirements. CiteScore does not query Google. The product page for what our report infers is AI Overviews visibility.

The two documents to read are AI features and your website and optimizing your website for generative AI features on Google Search. ChatGPT and Perplexity publish different controls, covered on how to get cited by ChatGPT and how to get cited by Perplexity.

What has to be true before an Overview can link you

Google’s AI features guide says a page must be indexed, and eligible to be shown in Google Search with a snippet, to be eligible as a supporting link in AI Overviews or AI Mode. There are no additional technical requirements. The optimization guide adds one site-level condition: the site must be included in Search generative AI features in Search Console.

AI Overviews and AI Mode can use different models, so the links differ. Google says both may use query fan-out: the model issues related searches across subtopics, then attaches links. Their example starts from “how to fix a lawn that’s full of weeds” and fans out to herbicides, non-chemical removal, and prevention. AI Overviews are shown when Google’s systems decide an overview adds something beyond the classic results, and Google says they often do not trigger.

Meeting the technical requirements does not promise a crawl, an index, or a place in the response. Google says that directly.

Four controls, four jobs

Search Console Help says the generative AI control was rolled out to all sites as of 31 August 2026. After you switch a site to exclude, Google says it generally takes a few days, with content excluded within 1–2 days and some URLs taking longer because of caching. To limit training of the models used in those Search features, the same help page points at Google-Extended, not at the Search control.

This group is a Gemini training and grounding opt-out. It is the wrong group if the goal is to leave AI Overviews.

User-agent: Google-Extended
Disallow: /

The token’s write-up is Google’s common crawlers. The Search Console control is Search generative AI control. How the CiteScore audit records the token, which is a different question from Google’s inclusion rule, is on the robots.txt guide.

What Google says is worth doing

The AI features guide lists the same fundamentals as the rest of Search. Crawling allowed in robots.txt and at the CDN. Internal links so pages can be found. A good page experience. Important content available as text. Images and video when they actually support that text. Structured data that matches the visible text. Merchant Center and Business Profile kept up to date when you use them.

The optimization guide is more specific about the writing. Google says content people find unique, compelling, and useful is likely to matter more over time than the other suggestions in that guide: a first-hand review rather than a summary of pages that already exist, organized with headings a person can follow. Their commodity example is “7 Tips for First-Time Homebuyers.” Their non-commodity example is “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line.” If a generative model helped draft the page, Google says the result still has to meet Search Essentials and the spam policies.

Fan-out is not a brief to publish one URL per variation. Google says that doing so to manipulate rankings or AI responses violates the scaled content abuse policy, and that the systems can match a page to a query without an exact phrase match. There is no ideal length, and Google says you do not need to rewrite the page in a special style for AI systems.

What Google says to skip

• Special files. Google says you do not need a new machine-readable file, AI text file, markup, or Markdown file to appear in Google Search, including its generative features. An llms.txt file neither helps nor harms visibility or rankings in Google Search, because Google Search ignores it. Keeping one for some other system is fine.
• Special schema. Structured data is not required for these features, and there is no schema.org type to add for them. Google still says structured data is worth using for rich results, and that any markup should match the text on the page.
• Chunking. Google says there is no requirement to break a page into tiny pieces for a model. The systems can use part of a longer page.
• Inauthentic mentions. Generative features can reflect what other pages say about a product. Google says chasing inauthentic mentions is not the way those systems work, because they depend on the same quality and spam systems as Search.

JavaScript is allowed. Google says it can process content in JavaScript when the scripts are not blocked, and that JavaScript sites are more work to keep eligible. Semantic HTML does not have to be perfect. Google says the web is not valid HTML and that semantic markup mainly helps people, including people using screen readers.

How Google says to measure it

AI Overviews and AI Mode are included in the Search Console Performance report, under the Web search type. There is also a Generative AI performance report for generative AI features on Search and Discover. Google says clicks from results pages that include AI Overviews are higher quality, in the sense that people are more likely to spend more time on the site. The page does not give a percentage. Google also says no third-party tool has its internal ranking or AI metrics.

A SaaS reading of the same rules is the AI search readiness checklist for SaaS. How GEO differs from a rank report is GEO vs SEO.

Free check, the $39 audit, and the template pack

Run the free check first. It reports whether robots.txt allows or blocks Google-Extended, along with GPTBot and PerplexityBot. Google-Extended is the Gemini token in the table above. The free check does not read the Search Console generative AI control, and it does not query AI Overviews. A passing row means that token is not disallowed.

The $39 GEO audit is one public URL, a 0–100 CiteScore, and about ten prioritized fixes, as a PDF. The score comes from on-page signals — definitions, entities, evidence, schema, freshness, crawler access — and, when a language-model key is configured, a review of the HTML we fetched. We do not log into ChatGPT, Perplexity, or Google AI Overviews. The per-engine snapshot is an inference, and the report says so.

The GEO Content Template Pack ($19) is a separate $19 purchase from the check and the audit.

Sources

• AI features and your website, Google Search Central. Eligibility, snippet controls, the SEO fundamentals list, and where AI features show up in Search Console Performance.
• Optimizing your website for generative AI features on Google Search, Google Search Central. The Search Console inclusion requirement, non-commodity content, and the list of tactics Google says to ignore, including llms.txt.
• Google’s common crawlers. What Google-Extended controls, and that it is not a Search inclusion or ranking signal.
• Search generative AI control, Search Console Help. The include and exclude choices, the 31 August 2026 rollout, and the 1–2 day exclusion window.