How to rank in Google AI Overviews & Gemini
Google's AI surfaces — AI Overviews and AI Mode in Search, and the Gemini app — sit on top of Google's core index and ranking systems. The strongest lever is being a genuinely authoritative, well-structured page that already ranks. This guide covers how Google's AI selects sources and how to earn a place in its answers.
How Google AI (Gemini & AI Overviews) decides what to cite
Unlike ChatGPT or Perplexity, Google's AI is anchored to Google's own index, ranking systems, and quality frameworks (E-E-A-T, the helpful-content system). AI Overviews are largely assembled from pages that already rank well for the query, so classic Google SEO and AI-Overview visibility are tightly coupled — with structured data and Merchant Center feeds giving an extra edge for products.
Classic Google rankings
AI Overviews are drawn largely from pages that already rank on page one for the query and its sub-questions. Strong traditional SEO is the single biggest input to AI-Overview visibility.
E-E-A-T and helpful content
Google applies its Experience, Expertise, Authoritativeness, and Trust signals to AI answers. First-hand experience, clear authorship, and genuinely helpful content are favoured over thin or templated pages.
Googlebot, not Google-Extended
AI Overviews and AI Mode are grounded in the Search index Googlebot builds. Google-Extended is a separate opt-out for Gemini-app grounding and training; per Google's documentation it does not affect Search, AI Overviews or AI Mode.
Structured data & Merchant Center
Product, FAQ, and Organization schema help Google extract facts, and a healthy Merchant Center feed is what surfaces products in AI shopping experiences.
How to optimize for Google AI (Gemini & AI Overviews), step by step
- 1
Win classic Google rankings first
Because AI Overviews source from top-ranking pages, invest in the fundamentals — intent-matched content, internal links, and Core Web Vitals. Ranking on page one is the strongest path into the Overview.
- 2
Decide on Google-Extended, and never block Googlebot
Confirm your robots.txt handles Google-Extended intentionally (allow to contribute to Gemini-app grounding and training; disallow to opt out). It has no effect on AI Overviews or AI Mode, which run on Googlebot. Check it with the Robots.txt AI-Crawler Checker.
- 3
Demonstrate E-E-A-T
Add named authors with credentials, first-hand experience, cited sources, and clear last-updated dates. Google's AI favours pages that visibly demonstrate expertise and trust.
- 4
Answer the sub-questions
AI Overviews and AI Mode stitch together several sub-answers. Structure pages with clear H2/H3 questions and concise answers so you can be pulled in for multiple facets of a query.
- 5
Add structured data & Merchant Center
Emit Product, FAQ, and Organization JSON-LD (generate and validate with the free tools), and keep a complete Merchant Center feed so products can appear in Google's AI shopping surfaces.
- 6
Track your AI Overview visibility
Monitor which queries trigger an AI Overview that cites you, and how Gemini describes your brand, so you can prioritise the pages worth strengthening.
Free tools for Google AI (Gemini & AI Overviews) optimization
How Gemini, AI Overviews and AI Mode actually retrieve and cite pages
Google runs three AI answer surfaces on one index. Knowing which crawler feeds which surface, and which robots.txt token controls what, removes most of the confusion.
| Surface | How it is grounded | Controlled by |
|---|---|---|
| AI Overviews | A summary at the top of the results page, built from indexed pages for the query and its sub-queries, linking the sources used. | Googlebot and the standard Search controls (noindex, nosnippet, data-nosnippet, max-snippet). |
| AI Mode | Conversational Search that breaks a question into sub-queries (query fan-out), searches the index for each, and writes a cited answer. | Googlebot and the same Search controls. Google-Extended has no effect. |
| Gemini app | A standalone assistant that can ground answers in Google Search and show its sources. | Google-Extended governs use for Gemini-app grounding and training; it does not touch Search. |
Googlebot crawls and indexes your page as it always has. When Google decides to answer a query with AI, it runs that query and related sub-queries against the index, pulls candidate pages from the results, and writes a summary grounded in them, linking the pages it used. There is no second crawl and no separate AI index. A page that is not indexed, or does not rank for the query or its sub-queries, is not a candidate.
The Gemini app is the partial exception. It can ground answers in live Google Search, on the same index, but the Google-Extended token lets you opt out of Gemini-app grounding and training. That is all Google-Extended does: disallowing it will not remove you from AI Overviews or AI Mode, and allowing it will not add you.
What Google's AI favours when it picks a source
Everything that already earns rankings still counts. On top of that, four properties decide which of the ranking pages actually gets quoted.
✓Passage-level structure that maps to sub-queries
Fan-out turns one question into several searches. A page with clearly-headed sections, each answering one sub-question in its first sentence, can be cited for several; one long section answering everything is cited for none.
✓Freshness where the query implies it
For prices, versions, availability and anything with a year, Google prefers current pages, and the summary inherits that. An accurate dateModified and a visible last-updated line signal the page is current.
✓Corroboration and E-E-A-T
Google's quality systems reward first-hand experience, named expertise, and claims other credible sources also make. The summary prefers to attribute a statement to a page that demonstrably knows the subject.
✓Entity clarity
Google builds its understanding of your organisation from your site, your Organization schema, your Business Profile and every consistent third-party description. The clearer the entity, the more confidently the AI names you rather than "some providers".
One section, many sub-questions
<h2>Everything about our warranty</h2> <p>We offer a comprehensive warranty. Coverage varies by product line; accidental damage is handled separately; transfers are possible in some cases...</p>
One section per sub-question
<h2>How long is the Acme warranty?</h2> <p>Two years from purchase on all hardware, in every region.</p> <h2>Does the warranty cover accidental damage?</h2> <p>No. Accidental damage is covered by the optional Acme Care plan.</p>
When AI Mode fans "is the Acme warranty any good" out into duration, coverage and claims sub-queries, the second page is a candidate for each and the first is a weak candidate for one. It is the structure that earns featured snippets.
How to rank in Google AI Overviews and Gemini
Ranking in AI Overviews is ranking in Google plus being the page the summary can quote. Google's own guidance is that no special optimisation is required beyond the fundamentals; the steps below are ordered by how cheaply each finds a problem.
- 1
Confirm the page is indexed and Googlebot is unrestricted
Use URL inspection in Search Console to confirm the page is indexed and the rendered HTML contains the body text. Check robots.txt for anything catching Googlebot or the wildcard, and for stray nosnippet directives.
- 2
Make a deliberate Google-Extended decision
Allow it to let your content ground Gemini-app answers and contribute to training; disallow it to opt out. Give it its own user-agent block. Either choice leaves AI Overviews and AI Mode unchanged.
- 3
Rank for the query and its sub-queries
List the questions your buyers ask and, for each, the two to four sub-questions an AI would need answered. Check which you rank for; the gaps are the content plan.
- 4
Restructure target pages around the sub-questions
One H2 per sub-question, phrased the way it is searched. A complete answer in the first sentence beneath it, then detail, a table for anything comparative, and a link to the primary source for anything non-obvious.
- 5
Show the experience and expertise
A named author with real credentials, first-person detail that could only come from having done the thing, and dates that are true. Google's AI attributes claims to pages that visibly demonstrate E-E-A-T.
- 6
Validate structured data; feed Merchant Center
Organization, Article with an honest dateModified, FAQPage on the question blocks you have written, Product where you sell. Validate it. If you sell physical products, keep the Merchant Center feed complete; Google's AI shopping surfaces draw on it.
- 7
Track AI appearances separately from rankings
A page can hold a ranking and lose its place in the summary, or gain a citation from position six. Run your prompt set against Google's AI surfaces on a schedule and record citations per prompt; Search Console does not break AI-surface traffic out separately at the time of writing.
Opting out of the wrong thing
# "keep us out of Google AI" User-agent: Googlebot Disallow: /blog/ User-agent: Google-Extended Allow: /
Deliberate, separate decisions
User-agent: Googlebot Allow: / # Gemini-app grounding + training # opt-out. Does not affect Search, # AI Overviews or AI Mode. User-agent: Google-Extended Disallow: /
The first file does the opposite of what its comment says: it removes the blog from Search and every AI surface built on it, while leaving Gemini-app grounding and training allowed. The second keeps the blog in Search and opts out of the one thing Google-Extended controls.
Gemini and AI Overviews SEO checklist
The short version, for the page you are working on right now.
✓Indexed, and Googlebot unrestricted
URL inspection shows indexed; no robots.txt or snippet directive catches it.
✓Google-Extended decided on purpose
Its own user-agent block; Gemini-app grounding and training only.
✓Ranks for the sub-queries, not only the head term
Each sub-question mapped to a section that answers it in one sentence.
✓Visible E-E-A-T
Named author, real credentials, first-hand detail, honest dates.
✓Schema validated; Merchant Center complete
Organization, Article, FAQPage, Product as they apply.
✓Consistent entity off-site
The same one-line description everywhere Google might read it.
Measure it: track AI Overview and Gemini citations
Search Console tells you the page ranks. It does not tell you whether the summary above the results quoted you, a competitor, or nobody.
AI search monitoring runs a fixed prompt set against Google's AI surfaces and the Gemini app on a schedule and records, per prompt, whether an AI answer appeared, whether you were mentioned or cited, and which competitors and third-party domains were cited instead. Because the prompt set is fixed, a restructured page shows up as a change in citation share rather than one screenshot.
Pair it with rankings. The useful view is queries where you rank and are cited, where you rank and are not, and where you are cited from a page that does not rank on page one. The second group is where restructuring pays off first; the third shows which pages Google's AI already trusts.
Frequently asked questions
How do I appear in Google AI Overviews?+
Rank well in classic Google search for the query and its sub-questions, demonstrate E-E-A-T, structure content as clear question-and-answer blocks, and add structured data. AI Overviews are largely assembled from pages that already rank on page one.
What is Google-Extended?+
Google-Extended is a robots.txt token that controls whether your content is used for Gemini-app grounding and to train Google's models. By Google's documentation it does not affect whether you are indexed, rank, or appear in AI Overviews or AI Mode — those run on Googlebot, which you should never block.
Is ranking in AI Overviews different from normal SEO?+
They are tightly coupled. Because AI Overviews source from top-ranking pages, strong classic SEO is the biggest lever, with structured data, E-E-A-T, and clear sub-question answers giving an additional edge.
How do my products show in Google's AI shopping?+
Maintain a complete, accurate Merchant Center feed and emit Product structured data. Google's AI shopping experiences draw on Merchant Center data and structured product information.
Does Gemini use the same sources as AI Overviews?+
Both are Google systems anchored to Google's understanding of the web, so the fundamentals overlap: authoritative, well-structured, well-ranked content with clear structured data performs across both. Only the Gemini app can be opted out of via Google-Extended.
What is Gemini SEO?+
Gemini SEO is making your pages the ones Google's AI surfaces — AI Overviews, AI Mode and the Gemini app — cite when they answer a question in your category. All three are grounded in the Search index, so it is classic Google SEO plus passage-level structure, visible E-E-A-T and validated schema. There is no separate AI index to submit to.
Does blocking Google-Extended remove me from AI Overviews?+
No. Google states that Google-Extended does not affect inclusion or ranking in Search, including AI Overviews and AI Mode. It only controls use of your content for Gemini-app grounding and training. To keep a passage out of AI summaries while the page stays in results, use nosnippet or data-nosnippet.
How does Google AI Mode choose which pages to cite?+
AI Mode breaks a question into related sub-queries, runs each against the Search index, and synthesises an answer from the pages those searches return, citing the ones it used. Pages that rank for the sub-queries and answer them in one extractable passage get cited, so one section per sub-question makes a page a candidate for several at once.
How do I track AI Overview citations?+
Search Console folds AI-surface impressions and clicks into its normal Web totals, at the time of writing, so it cannot tell you whether the summary cited you. Run a fixed set of buyer questions against Google's AI surfaces on a schedule and record, per query, whether you were cited and who was cited instead.
See how Google AI (Gemini & AI Overviews) cites your brand today
Run a free scan to baseline your visibility, then fix the gaps.
Run a free scan →