What Are Google AI Overviews, and How Does a Page Appear in One?

Google’s May 2024 launch post says a custom Gemini model built for Search writes the summary.

The supporting links sit beside or under the text. AI Mode is a separate, conversational tab built on similar technology.

AI Overviews: how they work (core ranking, RAG and query fan-out)

AI Overviews work by searching Google’s own index, reading the best pages and writing a summary from them.

Google’s guide Optimizing your website for generative AI features on Google Search, updated on 10 July 2026, explains the parts.

• Core ranking. The features are “rooted in our core Search ranking and quality systems.” Pages that rank well are the pool AI Overviews draw from.
• Retrieval-augmented generation (RAG). Google defines it as a technique, also known as grounding, that retrieves fresh pages from its index. The model then reviews them to write a more reliable response with links.
• Query fan-out. The model fires off related searches at once. Google’s example: “how to fix a lawn that’s full of weeds” can spawn searches for herbicides, chemical-free removal and prevention.

Fan-out is why an overview can cite a page that never ranked for the original query. The page ranked for one of the sub-questions instead.

In Digital Magma’s view, that is the most useful idea here: answer the follow-up questions well, and the overview has more ways to find you.

Google’s AI features page adds that AI Overviews and AI Mode may use different models and techniques. So the links each one shows will vary.

When AI Overviews appear and when they do not

AI Overviews appear only on some searches, by design.

Google’s AI features page says they were designed for queries where they add benefit beyond what Search already offers.

It also says they are only shown when Google’s systems judge them additive, and “as such, often don’t trigger.”

Google publishes no list of trigger queries. Its examples point to complicated topics and questions where someone wants the gist fast.

Three practical checks help you see the pattern for your own market:

• Search your top 30 queries by hand each month, logged out, and note which show an overview.
• Check the Generative AI performance report for the pages that already earn AI impressions.
• Compare the queries behind those pages with the ones that never trigger.

This is the monthly routine Digital Magma recommends. It shows where overviews may be taking clicks from you.

Search Labs experiments sit outside all of this. Google’s help page says Search Console does not include their data.

AI Overviews for a business: clicks, traffic and lead quality

AI Overviews change how many people click, and the two main sources disagree on what that means.

The independent data

Pew Research Center analysed the March 2025 browsing of 900 US adults across 68,879 Google searches. Its short read, published on 22 July 2025, found a clear drop.

On search pages with an AI summary, people clicked a traditional result in 8% of visits. On pages without one, the rate was 15%. Links inside the summary were clicked in only 1% of visits.

Google’s own claims

In its May 2024 launch post, Google said links inside AI Overviews get more clicks than if the page had been a traditional listing for that query.

Google’s AI features page also says clicks from results with AI Overviews are “higher quality”, with users more likely to spend time on the site. Both are Google’s statements about its own product.

What it means for leads

For a business, the useful metric is enquiries, not raw clicks. Compare conversion rate and time on site for organic landing pages before and after AI Overviews appear on their main queries.

Digital Magma reports enquiries first for this reason. A planned answer covers whether AI Overviews reduce clicks in more depth.

How to appear in AI Overviews: what to do

Appearing in AI Overviews needs the same groundwork as ranking, plus one setting that is new in 2026.

1. Meet the technical requirements. Google says a page must be indexed and eligible to show with a snippet. There are no extra technical requirements beyond that.
2. Stay included in Search Console. The July 2026 guide adds that a site must be included in Search generative AI features in Search Console. Inclusion is the default.
3. Allow crawling end to end. Check robots.txt for Googlebot, then your CDN and hosting rules. Make sure important text is in HTML that Googlebot can render.
4. Write non-commodity content. Google asks for first-hand, expert takes that go beyond common knowledge. A summary of other sites adds little the model does not already have.
5. Support the text with images and video. Google says its AI features can bring in relevant images and video, giving more ways to appear.
6. Keep business data current. Merchant Center feeds and Google Business Profile details can surface in AI responses, according to the guide.

None of this guarantees a link. Google says plainly that indexing and serving are never guaranteed.

Controlling AI Overviews: nosnippet, max-snippet, noindex and Google-Extended

AI Overviews controls are often confused, so this table sets out what each one actually does.

Google’s help page for the Search generative AI control says it reached all websites worldwide on 31 August 2026. It lives under Settings in Search Console.

After an exclusion goes live, content leaves within one to two days, though caching can stretch that. The setting does not affect AI training.

The Google common crawlers page describes Google-Extended separately. It covers training of Gemini models and grounding in Gemini Apps and Vertex AI, and Google says it does not impact inclusion in Search.

Digital Magma’s advice is to change these controls only with a clear business reason. Each one costs visibility somewhere.

How to track AI Overviews in Search Console

Tracking AI Overviews became far easier once Google added a dedicated report.

The Generative AI performance report shows impressions for AI Overviews and AI Mode. An impression counts each time a link to your site was shown in one of them.

• Dimensions: pages, countries, dates and devices.
• Search type: “Web: text-based” for typed queries, and “Web: multimodal” for Lens, Circle to Search and image searches.
• Aggregation: the chart counts two links from one site in one feature as a single impression.
• Missing report: usually too few AI impressions, or a site set to exclude.

The data comes from the Web search type of the main Performance report. Clicks from AI Overviews still sit inside that report’s totals, so pair both views with GA4 engagement data.

AI Overviews myths: schema, llms.txt and chunking

AI Overviews myths spread fast, and Google’s July 2026 guide answers the biggest ones directly. This Digital Magma entry pairs each myth with Google’s reply.

• Schema gets you in. Structured data is not required, and there is no special schema.org markup for AI features. It still helps with rich results.
• You need llms.txt. Google Search ignores llms.txt and similar files. Making one neither helps nor harms in Google.
• Content must be chunked. Google says its systems understand several topics on one page, and there is no ideal page length.
• One page per fan-out query. Building pages mainly to catch fan-out variants breaks the scaled content abuse policy.
• Third-party tools see inside. Google says no outside tool has access to its internal ranking or AI systems.

AI Overviews: frequently asked questions

AI Overviews questions site owners in India ask us most.

AI Overviews: related terms, sources and author

AI Overviews connect closely to these entries:

• Google AI Mode: Google’s conversational search mode
• Query fan-out: the related searches behind each overview
• Google-Extended: the Gemini training and grounding token
• AI citation: the supporting link a page can earn
• Schema markup: structured data, useful but not required

Sources

Google Search Central: the July 2026 generative AI guide, the AI features page and the common crawlers page. Google Search Console help on the generative AI report and control.

Google’s blog posts of May 2024 and October 2024, and its India blog. Pew Research Center, 22 July 2025. Digital Magma checked each one on 21 October 2026.