Learn how to optimize your website for Google AI Overviews. Discover proven SEO, AEO, and GEO strategies to improve your chances of appearing in AI-generated search results.
Quick answer: Google cites AI Overview sources from its normal index: a page must be indexed and eligible for a snippet, with no special schema needed. Because Google fans one query out into many related searches, pages that clearly answer specific sub-questions get picked, even outside the top 10. Fix eligibility, cover sub-questions, keep facts in text and grow brand mentions.
Google AI Overviews now sit above the organic results for a large share of searches, and Google says they reach over 2 billion monthly users across more than 200 countries and territories. Getting cited inside one is not the same job as ranking a blue link, but it is not a mystery either. Google has published fairly specific guidance, and several large studies show which queries trigger AI Overviews and which pages get picked. This guide sets out what Google says, what the data shows, and the tactics that follow from both.
Google picks supporting links from its normal search index, using pages that are indexed and allowed to show a snippet. Its documentation states that to be eligible, a page must be indexed and eligible to be shown in Google Search with a snippet, and that there are no additional technical requirements.
Three details in the same document explain why the selected pages often differ from the top organic results:
There is no AI-specific markup. Google states plainly: There's also no special schema.org structured data that you need to add, and no new machine-readable files or "AI text files" are needed.
Google's answer is to keep doing SEO fundamentals well, with a few points that matter more for AI features. Its documentation lists these as worthwhile:
Google's May 2025 Search Central post adds the content angle: focus on unique, non-commodity content because in AI search experiences users are asking longer and more specific questions, plus follow-ups. Commodity content that restates what fifty other pages say gives the system no reason to pick yours.
Google also documents the controls. To limit what appears, use nosnippet, data-nosnippet, max-snippet, or noindex. Be careful here: a restrictive max-snippet or nosnippet setting added for other reasons can quietly remove a page from AI Overviews, because snippet eligibility is the entry ticket.
Estimates range from about 11% to over 75% of searches, because each study uses a different keyword sample, country, device and month. That spread is the honest answer, and it is why any single "AI Overviews appear on X% of searches" claim needs its source and date.
| Source and date | Sample | AI Overview prevalence |
|---|---|---|
| BrightEdge, May 2025 | BrightEdge-tracked queries | Over 11% of Google queries |
| Pew Research Center, March 2025 data | 68,879 searches by 900 US adults | 18% of searches; 53% of searches with 10+ words; 60% of question queries |
| Semrush, 2025 | 10M+ keywords | 6.49% (Jan), 24.61% peak (Jul), 15.69% (Nov 2025) |
| Ahrefs, Nov 2025 | 108M AI Overview queries by country | US 20.5%, UK 19.1%, India 26.8%, Singapore 14.5% of keywords |
| Ahrefs, Sept 2026 | 232.8M US desktop result crawls | 78.3% of non-branded and 73.3% of branded keywords |
Two patterns hold across these studies. First, longer and question-style queries trigger AI Overviews far more often, as the Pew data shows. Second, AI Overviews have moved well beyond informational searches. In Semrush's data, the informational share of AI Overview keywords fell from 91.3% to 57.1% during 2025, while commercial rose from 8.15% to 18.57%, transactional from 1.98% to 13.94% and navigational from 0.84% to 10.33%. Ahrefs' latest study found AI Overviews on 82.91% of US desktop results for branded keywords as of 29 September 2026, and in its brand-name sample, 48 of 63 AI Overviews appeared in position one, above the brand's own site.
For Asian markets, Ahrefs' country data matters: India's 26.8% coverage was fourth-highest of the 50 countries listed, while Singapore's 14.5% sat 36th. Both figures are from November 2025, and Ahrefs' later US data shows coverage has kept rising since.
Google's own numbers explain why the feature keeps spreading. In July 2025 it said AI Overviews were driving over 10% more queries globally for the types of queries that show them, and that the effect was growing over time. A feature that increases searches is one Google has every reason to show more often, so plan for higher coverage, not lower.
It helps, but it matters less than it did. In Ahrefs' July 2025 study of 1.9 million citations, 76.10% of AI Overview-cited pages ranked in the top 10, and the first cited URL typically ranked around position 2. By March 2026, Ahrefs' larger study of 863,000 keyword results and 4 million cited URLs found that 38% of pages cited in AI Overviews also rank in the top 10, with 31.2% ranking 11 to 100 and 31.0% beyond the top 100.
Ahrefs suggests the shift is linked to Gemini 3 and a larger role for query fan-out: the system is citing pages that rank for the sub-questions, not just the original query. YouTube stood out, as the most-cited domain overall, with YouTube URLs making up 5.6% of all cited URLs.
Take a hypothetical search such as "office renovation contractor Singapore". Behind that one query, the system may also look for typical costs per square foot, how long a fit-out takes, which permits are needed and how to compare quotes. A contractor page that only says "we do office renovations" matches the first search. A page with a clear section on each of those questions can be cited for several parts of the answer, even if a directory outranks it for the head term. The same logic applies to any service business: the sub-questions are where smaller sites win.
The practical reading is that a strong page ranking position 15 for your head term, but answering a specific sub-question better than anyone, now has a real chance of citation. Ranking still correlates; it no longer gates.
Work through eligibility first, then cover the fan-out questions, then build the signals Google's systems appear to weigh. In order:
Two unglamorous fixes come up on almost every site we audit for AI Overview visibility: snippet hygiene and page speed.
Because eligibility runs through the snippet, we start with titles and descriptions. On our own site, 36 titles were over 60 characters, many because a long " | Blog" suffix was cutting off blog titles, and 48 descriptions were too long or too short. On a Singapore exhibition-display client, we rewrote titles on 72 pages and descriptions on 73 pages to fit Google's limits. None of this guarantees a citation, but it makes each page's topic unambiguous, which is what a fan-out query needs to match.
Page experience is the other one Google names explicitly. Our own homepage, which has a 3D WebGL hero, scored 27 on mobile Lighthouse. The causes were code that re-measured element positions on every animation frame, a third-party LinkedIn feed loading on page load, and the full 3D engine loading on phones. We now show a still image of the 3D element on phones, lazy-load the widget and stopped the per-frame layout reads. Mobile Lighthouse went to about 74, service pages from 66 to 89, and blog posts score about 98.
These are the first checks in our SEO and AI search engagements, because they are cheap and they remove reasons for Google to pass a page over.
Treat AI Overviews as a sub-question game, not a head-term game. The Ahrefs data shows citations drifting away from top-10 rankings as fan-out grows, and Semrush shows AI Overviews spreading into commercial and transactional searches. That favours pages that answer specific buyer questions clearly over pages that try to rank for one broad term. Google's own advice is unusually blunt: no special markup, no AI files, just eligible, useful, non-commodity pages. We would spend the first month on eligibility and snippet hygiene, the next on sub-question coverage of your top service pages, and keep building brand mentions throughout. If you are worried about the traffic side, our analysis of zero-click search covers it, and our guide to getting cited by ChatGPT, Gemini and Perplexity covers the other assistants.
Only partly. Google says robots.txt rules for Googlebot control crawling for Search, including AI features, and Google-Extended does not affect Search. To limit what appears, you can use nosnippet, data-nosnippet or max-snippet, but these also restrict your normal search snippets. There is no separate switch that removes a page from AI Overviews while leaving its regular listing untouched.
Not directly. Google's documentation says there is no special schema.org structured data needed for AI Overviews or AI Mode. Structured data still helps Google understand a page and powers other search features, but it must match the visible text. Clear question headings with direct answers in the visible content matter more than the markup itself.
Usually because their page answers a sub-question better. Google's query fan-out runs several related searches behind one query, and Ahrefs' March 2026 data found 62% of cited pages did not rank in the top 10 for the original query. A competitor with a clear section on cost, comparison or a specific use case can be cited for that part of the answer.
Yes. Ahrefs' November 2025 country analysis found AI Overviews on 26.8% of tracked keywords in India, fourth-highest of 50 countries, and 14.5% in Singapore. Google said in July 2025 that AI Overviews reach more than 200 countries and territories in 40 languages. Rates have risen since in Ahrefs' US data, so expect higher coverage today.
Only after Google recrawls and reprocesses the page. Google notes crawling can take anywhere from several days to several months, depending on how often its systems decide a page needs refreshing. You can request a recrawl in Search Console with URL Inspection, then watch the generative AI performance report for changes in impressions over the following weeks.