# How to Appear in Google AI Overviews

> Source: https://rankxai.com/blog/rank-in-google-ai-overviews · Last updated: 2026-09-08

Appearing in Google AI Overviews requires a page that is indexed and eligible for a snippet, which Google states is the only requirement. Selection is separate. Roughly a third of cited URLs also rank in the top ten, and more than half of top-ten results are never cited at all.

## How do you appear in Google AI Overviews?

Google AI Overviews are the AI-generated summaries that appear at the top of Google Search results, assembled from multiple web pages and shown directly in search results with links back to them. Appearing in one takes two separate things. Eligibility, which Google publishes and you control. And selection, which Google does not publish and you influence only indirectly. Eligibility is a low bar most sites already clear, and almost all the advice on this subject describes the bar.

Four things carry the practical weight and change your chances of appearing, and none of them is a new file or a new markup format.

1. **Stay indexed and snippet-eligible.** A `nosnippet` rule removes the page from AI Overviews outright.
2. **Rank somewhere for the question and its neighbours.** About a third of cited URLs rank in the top ten for the query itself; the rest arrive from elsewhere.
3. **Answer the sub-question in one passage.** AI Overviews fan a query out and quote passages, so a section competes rather than a page.
4. **Cover the neighbouring questions on their own pages.** Two thirds of cited URLs do not rank in the top ten for the query they are cited on.

The retrieval mechanics underneath those four belong to the [GEO guide](/blog/generative-engine-optimization) and to [query fan-out](/blog/query-fan-out), and the comparison across the five assistants belongs to the [platform pillar](/blog/how-to-rank-in-ai-search). This article stays on Google AI Overviews: what Google requires, what the citation data actually shows, which controls change what appears, and what a citation is worth once you have one.

## What does Google say is required to appear in an AI Overview?

Google states one requirement and denies the existence of any other. Its Google Search Central guidance on generative AI features says a page must be [indexed and eligible to be shown in Google Search with a snippet](https://developers.google.com/search/docs/appearance/ai-features), fulfilling the Search technical requirements. The same paragraph then adds that there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimisations necessary.

The same page rules out three things the market sells. Google says you do not need to create new machine-readable files, AI text files or markup to appear in these features, and that there is no special schema.org structured data you need to add. The AI text files sold for this purpose are covered by that sentence, and [llms.txt](/blog/llms-txt) sets out what the format does and does not do.

Eligibility being the only stated requirement leaves the interesting half unstated. Every indexed page that clears the bar is a candidate, which on a competitive query means thousands of pages competing for a handful of citation slots. Google's guidance tells you how to avoid being excluded and says nothing about how to be picked.

Google names the mechanism that finds the candidates, in permissive terms. AI Overviews and AI Mode may use a query fan-out technique, issuing multiple related searches across subtopics and data sources to develop a response. That one sentence explains why the citation data below looks the way it does.

## Do you have to rank in the top 10 to be cited in an AI Overview?

Ranking in Google's top ten is neither required nor sufficient for an AI Overview citation, and both halves of that come out at roughly a third. Ahrefs analysed 863,000 keyword SERPs and 4 million AI Overview URLs, publishing on 2 March 2026, and found that [37.9 per cent of cited URLs also appeared in the first ten blocks](https://ahrefs.com/blog/ai-overview-citations-top-10/) of the search results page.

Ahrefs ran that as two separate tests, and the three numbers are easy to quote mixed together. Counting every block on the page, including ads, featured snippets, People Also Ask and video packs, gives one distribution. Counting standard blue links only gives another. Take the three numbers from one row of this table or the other, never one from each.

| Ahrefs test, published 2 March 2026 | Cited URLs in the top 10 | Ranked 11 to 100 | Not in the top 100 |
| --- | --- | --- | --- |
| All SERP blocks, including features | 37.9 per cent | 31.2 per cent | 31.0 per cent |
| Standard blue links only | 37.1 per cent | 26.2 per cent | 36.7 per cent |

An independent measurement five months later landed close to both. Position Digital ran [92 AI Overview parses across 31 SaaS keywords](https://www.position.digital/blog/ai-overview-citation-source-analysis/) in August 2026, capturing 857 citations across 227 root domains, and found 33.2 per cent of cited URLs also ranking in the organic top ten. Two datasets of very different sizes agreeing on roughly a third is worth more than either figure alone.

That study also measured the direction most guides skip. 54.4 per cent of top-ten results were never cited at all, so more than half the pages that won the ranking Google's advice points at got nothing from the AI Overview sitting above them. A top-ten position buys a place in the candidate pool rather than a seat at the answer.

How this figure moved, from 76 per cent on July 2025 data to roughly 38 per cent, and why so many guides still quote the older number, is covered in the [AI search pillar](/blog/how-to-rank-in-ai-search).

### Where YouTube fits

YouTube is the most cited domain in AI Overviews on Ahrefs' own tracking, and a meaningful part of that comes from pages Google does not rank for the query. Among AI Overview citations pointing at pages outside Google's top 100 for the same keyword, 18.2 per cent were YouTube URLs, and those alone accounted for 5.6 per cent of every AI Overview citation in the dataset. Ahrefs reports YouTube citations growing 34 per cent over six months.

Those two percentages have different denominators, so hold them apart: 18.2 per cent is a share of non-ranking citations and 5.6 per cent is a share of all citations. RankX AI's own tracking points the same way on a sample far too small to be a measurement, with youtube.com the most frequent domain among the AI Overview citations captured for its five tracked keywords in the 30 days to 2 September 2026.

## What kind of page does Google cite in an AI Overview?

Google cites explanatory pages far more than commercial ones, and the split runs by query intent. Google describes an AI Overview as a snapshot of key information about a topic or question, with links so you can explore more on the web, and it assembles that from multiple sources rather than one. Position Digital's August 2026 sample says which sources. Educational articles took 70.2 per cent of citations on informational queries, and listicles were the most cited format on commercial ones at 37.6 per cent.

One result in that sample cuts against most of the advice on visibility in AI Overviews. G2, Gartner and 360 Quadrants together accounted for 1.5 per cent of citations across 31 SaaS keywords, so the review platforms teams spend budget getting listed on supplied roughly one citation in sixty-six. YouTube took 14.5 per cent, more than five times Reddit's 2.7, with explainer videos 46 per cent of the YouTube citations and product listicles 29 per cent.

That ratio is a SaaS artefact rather than a platform fact, and the all-topics figure is much closer. Ahrefs tracks more than 3 million US queries across every topic and had YouTube on 22.9 per cent mention share against Reddit's 18.5 when last updated on 2 September 2026, a ratio nearer 1.2 than 5. Mention share there counts a domain's citations against the summed citations of the top 50 sources, so the percentages are not interchangeable with Position Digital's and only the ratios compare.

Vendor pages did get cited, on their own subject. 88.4 per cent of the SaaS vendor citations went to a vendor topically relevant to the query, so being the obvious answer to a narrow question beats being present on a broad one. That is one 31-keyword sample in one sector, and breakdowns of AI Overview citations by content format that publish their method are rare enough to make the caveat worth stating rather than the finding worth discarding.

The page-craft rule with a documented mechanism behind it is narrower than the advice around it. Google's snippet permission is what gates AI Overview input, which is why `nosnippet` removes both at once, so the text Google is allowed to lift is the text it would otherwise show as a snippet. Writing a concise answer in the opening sentences under a heading matching the question is therefore the same work twice, and it is ordinary SEO rather than a new discipline.

## What 118 AI Overview checks on our own site showed

RankX AI ran 118 scheduled AI Overview checks against five tracked keywords in the 30 days to 2 September 2026. An AI Overview was present on 115 of them, 97.5 per cent. The site was cited as a source in none of the 115.

The second half of the measurement is what makes the first half mean anything. On those same checks all five keywords sat outside the top 20 on Google, and Google Search Console put the site's average position across every query at 67.1 for the 28 days to 2 September, on 7,850 impressions and 6 clicks. 115 opportunities, no top-20 position anywhere, and no citations.

One site is a demonstration rather than a study, and the limits are worth stating plainly. Five keywords is a tiny panel: four commercial AI-search queries, where AI Overview coverage is close to universal, plus the brand name itself. So 97.5 per cent says nothing about Google as a whole. The site's current content went live in August 2026, which makes this a runway measurement taken at the start of the runway.

The evidence above rules out the tidy reading of that null. If two thirds of AI Overview citations go to pages outside the top ten, an absent top-20 position cannot on its own account for a zero. What the panel shows is a result consistent with several causes it cannot separate, a site a few weeks old and a few hundred pages among them, and the ranking prerequisite is one candidate rather than the finding.

The null has now held across two consecutive windows. The 30 days to 21 August recorded an AI Overview on 59 of 60 checks and zero citations, published in the [platform pillar](/blog/how-to-rank-in-ai-search). This window doubled the check count and returned the same answer, while the five assistants named the brand in 20.2 per cent of 218 answers over the earlier window, recorded in [our AI visibility baseline](/blog/our-ai-visibility-baseline).

One measurement trap surfaced in the same export and is worth copying. The brand appeared in the text of 21 AI Overviews, every one of them on the query RankX, the fifth keyword on that panel, whose AI Overview is about the RANKX function in Power BI rather than about this company. A brand-mention count on a term with a homonym measures the homonym. Read what the answer is about before counting it.

## Which queries trigger a Google AI Overview at all?

Which queries are likely to trigger AI Overviews is largely a question of intent: they appear far more often on commercial research queries than on transactional ones, and the gap widened through the first half of 2026. Semrush tracked more than 600,000 US desktop keywords between November 2025 and April 2026 and found the share of commercial SERPs carrying an AI Overview grew 71 per cent across those six months, while transactional SERPs fell 5 per cent.

Query shape predicts presence better than anything about your site. Seer Interactive, tracking 53 brands across 5.47 million queries, recorded an AI Overview on 95.4 per cent of informational comparison queries of the X versus Y kind. A page answering a comparison question is writing into a surface that is almost always there, which changes what that page should look like before it changes anything else.

Treat any single site-wide prevalence figure with suspicion. Published estimates disagree widely between trackers, and the disagreement is mostly the composition of each tracker's keyword panel rather than a dispute about what Google is doing. The presence rate on your own tracked keywords is the number that can inform a decision.

Finance moved most in the Semrush data, with commercial-intent AI Overviews up 231 per cent over the six months. Semrush also found Google Ads and AI Overviews appearing on the same results page roughly twice as often as a year earlier, so the surface is being monetised alongside the answer rather than instead of it.

## How stable are Google AI Overview citations?

Google AI Overview citations change between identical searches, which makes a single check worthless as evidence in either direction. Position Digital ran the same 31 keywords three or four times each in August 2026, 92 runs in total from one account in one session, and Google changed at least one citation between identical repeat searches on 52 per cent of them.

The instability is uneven, and the shape of it is useful. On 25 of 29 keywords, 86 per cent, the top cited URL was the same on every run, so Google holds a firm first choice and rotates the supporting sources beneath it. Being the first citation looks like a durable position. Being the fourth is a coin toss repeated daily.

A far larger measurement of a different kind of instability agrees. Ahrefs watched more than 43,000 keywords, each carrying at least 16 recorded AI Overviews, across a month in November 2025. An AI Overview persisted 2.15 days on average, had a 70 per cent chance of changing between observations, and kept 54.5 per cent of its URLs from one observation to the next, while the answer itself barely moved at 0.95 average cosine similarity.

Read the two together rather than as one number. Position Digital measures non-determinism inside a single session and Ahrefs measures drift across days, and the Ahrefs data predates the Gemini 3 switch in January 2026. The direction they share is the usable part: the sources churn while the answer stays put.

Two things follow for anyone reporting this to a client or a board. A screenshot proves nothing, and a month-on-month citation rate across a fixed panel of keywords is the smallest unit that survives the run-to-run noise. The same argument holds with larger numbers on the assistants, and [measuring AI search visibility](/blog/measure-ai-search-visibility) sets out the method.

## Does structured data get you into Google AI Overviews?

Structured data does not get you into Google AI Overviews, and the AI Overviews arm of the largest test published is the one worth quoting precisely. Ahrefs tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 and matched them against roughly 4,000 control pages. AI Overview citations fell 4.6 per cent, which Ahrefs reports as small but statistically significant at odds of about 1 in 2,500 of arising by chance.

That is the AI Overviews arm alone. The other two engines in the same experiment moved by less and neither result was significant, so summaries reporting one flat null across all three discard the only measured effect it produced.

Read the fall as an absence of benefit rather than as damage, and note the study design before leaning on it. Nobody added the markup for the experiment. Ahrefs found pages that had added it themselves and matched them, and it warns that such pages often change links, content and technical fixes at the same time, so schema cannot be fully separated from those. Both groups were also already declining before the markup arrived.

Google says the same from the other end, with its guidance stating there is no special schema.org markup you need to add. Keep [schema markup](/glossary/schema-markup) for rich results and entity resolution, and see the [GEO guide](/blog/generative-engine-optimization) for the wider list of tactics tested and found not to work.

## How do you stop your pages appearing in Google AI Overviews?

One directive removes a page from Google AI Overviews and it is not the one most advice names. Google's [robots meta documentation](https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag) states that `nosnippet` applies to all forms of search results, listing AI Overviews and AI Mode by name, and will also prevent the content from being used as a direct input for AI Overviews and AI Mode.

Google-Extended does not do it, although a great deal of published advice says it does. Google draws the line itself on the AI features page, sending you to nosnippet, data-nosnippet, max-snippet and noindex to limit what is shown from your pages in Search, and to Google-Extended only to limit training and grounding in some of Google's other systems. The crawler documentation says the same from its side: Google-Extended does not impact a site's inclusion in Google Search and is not a ranking signal there. Blocking [Google-Extended](/glossary/google-extended) leaves AI Overviews untouched.

| Control | Effect on AI Overviews | What it costs elsewhere |
| --- | --- | --- |
| `nosnippet` | Removes the page as a direct input | No text snippet and no featured snippet anywhere in Google |
| `data-nosnippet` | Excludes the marked passage as an input | That passage leaves your ordinary snippet too, the rest of the page stays |
| `max-snippet:[n]` | Limits how much may be used as input | Caps the snippet length in ordinary results too |
| `noindex` | Removes the page from Search entirely | Every Google surface, ordinary results included |
| `Google-Extended` in robots.txt | None | Gemini model training and Gemini app grounding |

The trade in the first row is the whole decision, and it is sharper than most publishers expect. `nosnippet` costs the ordinary text snippet and the [featured snippet](/glossary/featured-snippet) alongside the AI Overview, because Google treats them as one preview permission. No setting keeps a page in the ordinary snippet while refusing it to the AI Overview, which is precisely the setting most publishers want. `data-nosnippet` comes closest and works at passage level in both directions at once: the text you mark leaves the snippet as well.

Anything short of that is a preference rather than a control. Adding a [robots.txt](/glossary/robots-txt) rule for an AI crawler token does not apply here, because AI Overviews are assembled from Google's ordinary Search index and crawled by Googlebot. The last row of that table is the one people misread, and [what Google-Extended does and does not reach](/blog/gemini-seo) is worked through in full alongside the Gemini side of it. Check what you have actually blocked with the [AI Crawler Access Checker](/tools/ai-crawler-access-checker) before assuming a rule does what its name suggests.

## What is a Google AI Overview citation actually worth?

An AI Overview citation is worth less traffic than the same position without one, and the honest figure has been moving. Pew Research analysed 68,879 Google searches from 900 US adults during March 2025 and found users clicked a traditional search result on 8 per cent of visits where one of these AI summaries appeared, against 15 per cent where none did. They clicked a link inside the summary on 1 per cent of visits.

That study is quoted almost everywhere and its browsing data is now eighteen months old. Seer Interactive tracked 53 brands, 5.47 million queries and 2.43 billion organic impressions from January 2025 to February 2026. Organic click-through rate on AI Overview queries fell to 1.3 per cent in December 2025, then recovered to 2.4 per cent by February 2026, against 3.8 per cent where no AI Overview appeared.

A larger study of the same question points the other way. Ahrefs compared 150,000 keywords carrying an AI Overview against 150,000 without, on December 2025 data published in February 2026. The top-ranking page took a 58 per cent lower click-through rate where an AI Overview appeared, against 34.5 per cent in Ahrefs' April 2025 measurement.

They measure different things. Seer follows 53 brands across whatever positions they hold, and Ahrefs isolates position one, which has most to lose to an answer printed above it. Plan for a materially smaller click, and treat the question of whether the loss is still deepening as open.

Seer's dataset answers what this section is really asking. Being cited in the AI Overview delivered 120 per cent more organic clicks per impression than not being cited on the same results, and Seer translates that into clicks per million impressions on informational queries.

| Results page | Organic clicks per million impressions |
| --- | --- |
| No AI Overview | About 33,500 |
| AI Overview present, brand cited in it | About 20,743 |
| AI Overview present, brand not cited | About 9,445 |

A citation roughly doubles what an AI Overview page returns and still leaves it well below a comparable page without one. Worth chasing, and no replacement for the traffic the position used to carry.

The figure in circulation is Seer's older one. Its September 2025 update, on 3,119 keywords, put cited brands 35 per cent ahead on organic click-through and 91 per cent ahead on paid. Part of the distance to 120 per cent is the narrower query type rather than the passage of time.

One caution on the word citation. Semrush logged 3,981 domain appearances across 115 prompts, 14 countries and four engines including AI Overviews, and 61.7 per cent were citations where the brand name never appeared in the answer text.

A [zero-click search](/glossary/zero-click-search) that links your site still puts your page in front of the reader as Google's chosen source, and [how AI systems choose what to cite](/blog/ai-citations) covers what that is worth and how to count it.

## How to tell whether you appear in Google AI Overviews

Google Search Console tells you less than half of it, and its own documentation is behind its own product. The crawler documentation still says sites appearing in AI features, AI Overviews and AI Mode included, are reported inside overall search traffic under the Web search type. That is still true of clicks, queries, click-through rate and average position, which have no AI breakdown at all.

What changed is impressions. Google announced dedicated Search Generative AI performance reports on 3 June 2026, rolling out to a subset of sites, and they report how often your URLs appeared inside AI Overviews and AI Mode, broken down by page, country, device and date. Impressions only, and reported for both surfaces as one group, so it cannot tell you which of them showed you: [the two cite almost entirely different sources](/blog/google-ai-mode-vs-ai-overviews). A report that answers whether you were cited and refuses to say what it earned you is worth opening and will not close the question.

That leaves the impression report, two honest methods and one habit to drop. Run a fixed panel of keywords repeatedly, recording whether an AI Overview appeared, whether you were featured in AI Overviews for that keyword, and which domains it cited instead, because that is the only way to separate a real change from the run-to-run variation above. Watch google.com referrals in Google Analytics against your own baseline rather than against a benchmark. And stop reading single searches as evidence.

RankX AI builds the first of those, and the disclosure belongs on a page making this argument. [AI Overviews tracking](/features/ai-overviews) checks each tracked keyword on a schedule and records whether an overview appeared, whether it cited the site, and which domains it cited instead. The [AI Overview Checker](/tools/ai-overview-checker) answers one keyword free with no account, and the [AI Readiness Score](/tools/ai-readiness-score) grades a single page against the extraction rules in this article.

## How long does it take to appear in a Google AI Overview?

There is no published figure and any specific number you are quoted is invented. The bound is indexing rather than ranking: a page cannot be a candidate until Google has crawled and indexed it and it is eligible for a snippet. After that, appearing depends on selection, which changes between identical searches. Measure it as a rate across a fixed panel of keywords over a month rather than as a date, because Position Digital found Google changed at least one citation between repeat searches on 52 per cent of the keywords it tested.

## Can you pay to appear in Google AI Overviews?

No. There is no paid placement inside the AI Overview itself and no submission process. Google Ads do appear on the same results page, and Semrush found ads and AI Overviews now appearing together roughly twice as often as a year earlier, but that is the advertising slot rather than the answer. Anyone selling guaranteed AI Overview inclusion is selling something Google does not offer.

## Does blocking Google-Extended remove you from AI Overviews?

No, and this is the most common mistake in the field. Google-Extended controls whether your content may be used to train future Gemini models and to ground answers in Gemini apps. Google's own crawler documentation states that Google-Extended does not impact a site's inclusion in Google Search, and AI Overviews are a Google Search feature. The directive that does remove a page from AI Overviews is nosnippet, which also removes your ordinary text snippet and your featured snippet.

## Do AI Overviews only cite pages that rank on page one?

No. On Ahrefs' analysis of 863,000 keyword SERPs published in March 2026, 37.9 per cent of URLs cited in AI Overviews also appeared in the first ten blocks of the results page, leaving roughly two thirds coming from further down or from outside the top 100 entirely. Position Digital measured 33.2 per cent independently in August 2026. Ranking helps and it is not a gate, and it is not sufficient either: 54.4 per cent of top-ten results in that August sample were never cited.

## Why did my page disappear from an AI Overview?

Most likely nothing changed on your page. AI Overview citations rotate between identical searches: Position Digital ran 31 keywords three or four times each in August 2026 and Google changed at least one citation on 52 per cent of them, although the single top-cited URL held on 86 per cent. Check whether the AI Overview still appears for the query at all before investigating your page, then compare a month of checks rather than two.