Google AI Overviews and AI Mode are AI-generated answers inside Google Search that summarise a topic and link to supporting web pages. Google says there are no special technical requirements to appear: a page must be indexed and eligible to show a snippet. Pages get cited when they answer a part of the question clearly, accurately and from a trustworthy source.
For many businesses, AI Overviews changed what a “ranking” looks like. A buyer can now read a summary, see three or four cited sources and decide what to do next without scrolling through ten blue links. This guide explains how the two features work according to Google’s own documentation, what independent research has measured about their effect on clicks, how pages become supporting links, what you can control, and how to measure your presence. It is written for marketers and founders who want a practical, evidence-based plan rather than speculation.
Key takeaways
- No special markup or file is needed. Google’s documentation states there are no additional technical requirements beyond being indexed and snippet-eligible.
- AI features search more widely. Google says AI Overviews and AI Mode may use “query fan-out”, running multiple related searches across subtopics to find supporting pages.
- Clicks behave differently. Pew Research Center found in March 2025 that users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% without one.
- You can now measure it. Google says Search Console’s generative AI performance report, showing AI Overviews and AI Mode impressions, was available to all websites by 31 August 2026.
- The foundations are the same. Clear answers, depth, trustworthy sources and sound technical SEO decide whether you are cited.
What are AI Overviews and AI Mode?
AI Overviews appear at the top of some Google results pages. They give a short AI-generated summary of a topic with links to the pages used to support it. AI Mode is a separate, conversational search experience where a user can ask longer questions and follow-ups and receive a fuller AI-generated response with links to supporting websites.
Both features have grown quickly. Google’s I/O 2025 announcement on 21 May 2025 said AI Overviews had scaled to 1.5 billion monthly users in 200 countries and territories, and that AI Mode was starting to roll out to everyone in the United States. Availability and behaviour continue to change by country, language and query type, so treat any snapshot as temporary.
| Feature | Where it appears | Typical query | What the user sees |
|---|---|---|---|
| Classic web results | Standard results page | Any | Ranked links with titles and snippets |
| AI Overviews | Top of some results pages | Questions where a summary helps | Short summary plus supporting links |
| AI Mode | Separate AI Mode tab or entry point | Longer, multi-part, exploratory questions | Conversational answer, follow-ups, supporting links |
| Featured snippets | Top of some results pages | Direct factual questions | A quoted passage from one page |
How do pages get chosen as supporting links?
Google’s documentation on AI features and your website, last updated in December 2025, is the authoritative source. Three points matter most:
- Eligibility. To be shown as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet. Google states there are no additional technical requirements.
- Query fan-out. Google says both features may issue multiple related searches across subtopics and data sources to find supporting pages while the response is generated. A single user question can therefore pull in pages that rank for related sub-questions, not only the exact query.
- A wider set of links. Google says these features can display a wider and more diverse set of helpful links than classic web search.
The practical implication of query fan-out is important. If a buyer asks “how should a small B2B team choose and set up a CRM?”, the system may look for pages on CRM comparison, setup steps, data migration and cost. A site with one strong page for each of those sub-questions has several chances to be cited. A site with one generic page has fewer.
One question, several searches
The sub-searches shown are illustrative. Google does not publish the exact sub-queries it runs.
What does independent research show about clicks?
The most rigorous public study so far comes from Pew Research Center, published in July 2025. It analysed the browsing of 900 US adults during March 2025, covering 68,879 unique Google searches.
Users clicked results about half as often when an AI summary appeared
Pew also found that 18% of the Google searches in its sample produced an AI summary, and that Wikipedia, YouTube and Reddit were the most cited sources in both AI summaries and standard results. The data is US-only and from one month, so it does not describe every market, but it is clear evidence that AI summaries change click behaviour.
Two conclusions follow. First, for many informational queries, being cited in the summary matters more than ranking fifth below it. Second, traffic alone is no longer a complete measure of search visibility. You need impressions, citations and enquiries in the same report.
What can you control?
Google gives site owners the same controls it uses for regular snippets. According to its AI features documentation:
| Control | Effect | When to use it |
|---|---|---|
noindex |
Page is not shown in Search at all | Pages that should never appear, such as thank-you pages |
nosnippet |
No text snippet is shown from the page | Rarely; it also removes your normal snippet |
data-nosnippet |
Marked text is excluded from snippets | Pricing details, legal text or quotes you do not want reused |
max-snippet |
Limits snippet length | When you want short snippets only |
| Google-Extended | Limits use in some other Google AI systems for training and grounding | A policy decision; Google’s documentation points to it separately from Search |
For most businesses that want visibility, the right setting is to allow snippets and focus on content quality. Restrictive controls reduce the chance of being cited and can also make your normal search snippets worse.
The Bargaon citation readiness model
Because Google publishes eligibility rules but not selection rules, we assess pages against five practical conditions that make a page a strong candidate to support an AI answer.
Five conditions for a citable page
A page that meets all five is a strong candidate. Selection is still Google’s decision.
1. Eligible
Check indexing with URL Inspection, make sure robots.txt is not blocking the page or its resources, and confirm there is no restrictive snippet control. Our technical SEO guide and the SEO audit checklist cover these checks.
2. Answerable
Put a direct, self-contained answer in the first paragraph. Avoid opening with history or a sales pitch. Use the question as a heading where it helps. Our answer engine optimization guide explains how to write answers that stand alone.
3. Complete
Cover the obvious follow-up questions on the same page where they share intent, and on linked pages where they do not. Query fan-out rewards sites that answer the whole journey, as described in our guide to topical authority and content clusters.
4. Verifiable
Name sources and years in the sentence, link to the primary source, and show who wrote the page and why they know the subject. Google’s guidance on E-E-A-T is covered in our E-E-A-T and trust signals guide.
5. Connected
Link related pages with descriptive anchors so both users and crawlers can see the full body of knowledge. See site architecture and internal linking.
Which kinds of pages tend to be cited?
Google has not published a list, so this is based on observation and the logic of the features rather than a ranking rule. Pages that are often useful as supporting sources share a few traits:
- Definitions and explanations that state the answer plainly before adding detail.
- Step-by-step guides with numbered steps and clear headings.
- Comparisons with a table of criteria rather than vague pros and cons.
- Original data, such as a survey, benchmark or analysis, that other sources do not have.
- Specific, local or niche answers where general sources are thin, such as rules for a particular country or industry.
Thin listicles, pages that repeat what everyone else says and pages with no clear author are weaker candidates, both for AI features and for regular search.
How do you measure your AI Overviews presence?
Measurement has improved. Google states that pages appearing in AI features are included in the overall search traffic in Search Console, under the “Web” search type in the Performance report. In addition, Google says its generative AI performance report in Search Console, which shows impressions from AI Overviews and AI Mode, was rolled out to all websites by 31 August 2026.
| Question | Where to look | Review |
|---|---|---|
| Which pages appear in AI features? | Search Console, generative AI performance report | Monthly |
| Is overall organic visibility rising? | Search Console, Performance (Web) | Monthly |
| Are clicks falling where impressions rise? | Search Console, compare CTR by page | Quarterly |
| Are AI-exposed topics still producing enquiries? | CRM lead source and landing page | Quarterly |
| How is the brand described? | Manual checks of priority questions | Quarterly |
For a wider view across ChatGPT, Perplexity, Gemini and Copilot, read our guide to tracking AI search visibility.
Worked example: improving a service cluster for AI Overviews
Imagine a fictional agency whose “website migration” page ranks on page one but is never cited in AI Overviews. A review using the readiness model finds:
- The first paragraph is a sales pitch; the answer to “how do you migrate a website without losing traffic?” is buried halfway down.
- There is no redirect map explanation, no checklist and no source for any claim.
- The page has no links to related guides on Core Web Vitals, structure or audits.
The team rewrites the opening as a direct answer, adds a redirect-map table and a launch checklist, cites Google’s site move documentation, and links to its website migration SEO guide and related cluster pages. A month later, the page begins to show impressions in the generative AI report. This example is illustrative; results depend on competition and are never guaranteed.
What are the most common mistakes?
- Buying “AI Overview optimisation” packages. There is no special markup or file that guarantees inclusion.
- Blocking snippets to “protect” content. This removes eligibility for AI features and weakens normal results.
- Writing for one exact keyword. Query fan-out favours sites that cover the whole topic.
- Judging success by clicks alone. A cited page may influence a buyer who never clicks. Track impressions and enquiries too.
- Publishing thin AI-generated content at scale. Google’s spam policies treat scaled, low-value content as abuse, however it is produced.
Should informational and commercial pages be treated differently?
Yes. A summary is most useful when someone wants to understand a topic, while a buyer ready to hire usually still needs to evaluate a provider directly. The two page types therefore need different goals:
- Informational guides: aim to be cited. Lead with a direct answer, cover sub-questions and include original evidence. Measure AI feature impressions and branded search, not only clicks.
- Service and comparison pages: aim to be chosen. Make scope, process, fit and next steps clear, so a buyer who arrives from a summary can act. Measure enquiries.
Link the two: every guide that might be cited should offer a clear, honest route to the related service page, so a reader who does click has somewhere useful to go next.
Who should own AI Overviews work?
Give one person responsibility for AI features, usually the SEO lead. They review the generative AI report each month, keep the list of priority questions, and agree with content owners which pages to improve next. Developers handle indexing and rendering fixes; subject experts add the experience and evidence that make pages worth citing. Without a single owner, AI Overviews work often turns into occasional checks with no follow-through.
What this guide cannot promise
Nobody outside Google knows exactly how sources are selected for each answer, and the features change often. The steps here make your pages strong, eligible candidates; they cannot guarantee a citation. Treat AI Overviews as one surface in a wider plan that also includes classic results, other AI assistants and channels you own. For the wider strategy, see our guides to generative engine optimization and how ChatGPT, Perplexity and Gemini choose sources.
Frequently asked questions
Do I need special schema to appear in AI Overviews?
No. Google’s documentation states there are no additional technical requirements. Structured data can help Google understand a page, but it is not required for AI features.
Can I opt out of AI Overviews but stay in normal search?
Google’s controls work at the snippet level. Using nosnippet or data-nosnippet limits what can be shown, but also affects normal snippets. There is no separate switch that only removes AI Overviews.
Why does my page rank well but never appear in AI Overviews?
Ranking for the main query is not the same as answering the sub-questions the system looks for. Check whether the page gives a direct answer early, covers follow-ups and cites trustworthy sources.
Are AI Overviews reducing my traffic?
Possibly, for informational queries. Pew’s 2025 study found users clicked results less often when an AI summary appeared. Compare impressions and CTR by page in Search Console before drawing conclusions.
Does AI Mode use the same rules as AI Overviews?
Google’s documentation covers both features together with the same eligibility rules and the same controls.
How often should I review AI Overviews performance?
Monthly in Search Console for the main pages, and quarterly for a fuller review that includes enquiries and manual checks of priority questions.
Sources and further reading
- Google Search Central: AI features and your website (updated December 2025).
- Pew Research Center: Google users are less likely to click on links when an AI summary appears (July 2025).
- Google: 100 things we announced at I/O 2025 (May 2025).
- Search Console Help: Generative AI performance report.
- Google Search Central: Spam policies.
Want to know where you stand? Bargaon’s answer engine optimization service reviews your priority pages against the readiness model and shows where AI Overviews already cite you, and where they could.