How to Appear in Google AI Overviews

To appear in Google AI Overviews, a page has to be indexed by Google and eligible to show a snippet in normal Search results. That is the full eligibility list. Google's documentation on AI features says there are no additional requirements and no special optimisations needed for AI Overviews or AI Mode — no new markup, no special file. Whether your page is actually chosen is then a question of whether it answers the question better than the other candidates.

That is less exciting than the courses selling "AI Overview secrets", but it is useful, because it tells you exactly where to put effort. For how this fits with ChatGPT, Perplexity and the rest, see the AI search optimisation guide.

What AI Overviews and AI Mode are

AI Overviews are the AI-written summaries that appear at the top of some Google results pages, with links to the pages they drew on. AI Mode is a separate, conversational search experience in its own tab, aimed at longer and more exploratory questions; it launched in the UK in July 2025.

Both draw on Google's normal Search index. Google describes a technique called query fan-out: the model breaks a question into related sub-queries, runs them, and builds the answer from the combined results. Google's own example is a question about fixing a lawn full of weeds producing extra searches about herbicides and chemical-free weed removal.

That one detail explains most of what follows. Your page is not just competing for the query the user typed. It is competing for each sub-question the system decided to ask.

How to appear in Google AI Overviews: the eligibility checklist

Work through these before thinking about content.

1. The page is indexed. Check it in Search Console with URL Inspection. If it is not indexed, nothing else on this page applies. How Google crawls and indexes a website covers the usual reasons.
2. The page can show a snippet. A nosnippet robots meta tag or header removes the page from AI features as well as from normal snippets. A max-snippet value of zero does the same. Check your SEO plugin and any headers your host adds.
3. The text you want quoted is not inside data-nosnippet. Some themes and plugins wrap content in this attribute without the owner knowing.
4. The content is in the rendered HTML. If your main text loads only after a click, or through JavaScript that fails for Googlebot, Google may not see it.
5. The site is not excluded in Search Console. There is now a Search generative AI setting under Settings. Its default is to include your site, but check nobody has changed it.
6. The page meets Google's technical requirements and spam policies. Pages caught by spam policies are not eligible for anything.

If all six pass, the page is eligible. Eligibility is necessary, not sufficient.

What decides whether your page is cited

Google does not publish a selection formula, and anyone claiming to know it precisely is guessing. What Google does say, in its AI optimisation guide, is to focus on unique, valuable, people-first content, keep a sound technical structure, and include good images and video where they support the text.

From that and from how retrieval systems generally work, the following are reasonable inferences rather than documented rules:

• Pages that already rank for the topic are in the candidate pool. Because the system searches Google's index, pages that rank for the sub-queries are the ones it sees.
• A clear, self-contained answer is easier to use. If the passage that answers the sub-question is buried in a long story, the system has to work harder to use it than it would with a competitor's direct paragraph.
• Specific beats generic. A summary built from ten identical pages can cite any of them. A page with a detail the others lack gives it a reason to cite you.
• Covering the sub-questions matters. For a query like "how much does a website cost", the fan-out might cover platforms, maintenance, hosting and timescales. A site that answers those across connected pages has more ways in.

A worked example

Say a plumbing firm in Kent wants to appear when someone asks Google why their combi boiler keeps losing pressure.

A typical page on their site reads: "We fix all boiler problems quickly and professionally. Call us today." It is indexed and eligible, but it has nothing to cite.

A page worth citing would:

• Answer the question in the first paragraph: the common causes, in plain terms.
• Have a short section per cause, each with a heading that names it, explaining how to tell which one you have.
• Include a table of what a homeowner can check safely versus what needs a Gas Safe engineer.
• State what the firm actually sees most often, from its own jobs, as an observation rather than a statistic.
• Link to the service page for boiler repair in context.

Nothing in that list is an AI trick. It is a better page. That is the point Google keeps making, and it is the right one. How to structure content for AI answers goes deeper on the structure side.

AI Overviews vs AI Mode: does the approach differ?

Not in what you can control. Google's documentation treats them together: the same eligibility rules, the same snippet controls, the same Search Console report. The difference is in the questions people bring to them.

The practical implication is a reasonable inference rather than anything Google states: AI Mode's longer, multi-step conversations give more weight to sites that cover a topic thoroughly, because the follow-up questions fan out further. A business with one page on a subject has fewer chances to be useful than one with a well-linked set of pages.

Troubleshooting: indexed but never shown

If a page is eligible but never appears in the generative AI report, work through the likely causes in order.

Work top to bottom. Most pages fail on the first two rows.

The controls you have

You may decide some content should not be summarised — paid guides, for instance, or pages where you want the visit rather than the summary.

The Search Console setting is new. Google announced it in June 2026, tested it first with a subset of UK site owners, and says it rolled out to all websites on 31 August 2026. Its help page states the setting is not used as a ranking or inclusion signal for the rest of Search, and that exclusion takes effect within a day or two, sometimes longer.

One common confusion: Google-Extended, the robots.txt token for Gemini training and grounding, does not control AI Overviews. Google's crawler documentation says it does not affect a site's inclusion in Google Search.

How to measure AI Overview visibility

At the time of writing (October 2026), Search Console has a generative AI performance report. It shows:

• Impressions — how many times links to your site were shown in a generative AI feature.
• Breakdowns by page, country, device and date.
• Coverage of AI Overviews and AI Mode, with a separate report for Discover.

It does not show clicks or the queries that triggered the impression. So you can see which pages are being shown in AI features, but not exactly why. Pair it with the normal Performance report: if a page's AI impressions are rising while its ordinary clicks drop, the summary may be answering the question for people. The Google Search Console guide covers the rest of the tool.

Google's AI optimisation guide also warns against third-party tools claiming to have access to internal Google metrics on AI features. None do.

What does not help

These are recommended widely and contradicted by Google's own guidance:

• Special files. Google says it ignores llms.txt and does not need any new machine-readable file for AI features.
• Chunking. Google says you do not need to break content into small pieces for AI; its systems understand pages that cover several sub-topics.
• Rewriting for AI. Google says its systems understand synonyms and meaning. Rewriting good pages to match AI phrasing does nothing.
• FAQ schema as an entry ticket. Google stopped showing FAQ rich results in May 2026, and no special structured data is needed for AI features.
• Manufactured mentions. Google calls pursuing inauthentic mentions across the web ineffective.

Where entities come in

Google's AI answers about businesses draw on what Google knows about the business as an entity — Business Profile, Knowledge Graph, consistent details across the web. For brand and local queries this matters as much as any page on your site. Entity SEO and the Knowledge Graph covers it, and AEO vs SEO explains which parts of the old answer-box playbook still apply.

Next step

Run the six-point eligibility checklist on your five most important pages this week, then open the generative AI report in Search Console to see which pages are already being shown. If you would like that done as part of a wider audit, it is included in the SEO service.