AI Overviews is not a separate product with its own rules but a layer on top of Google's core ranking. Here is how sources get in, and what independent measurement says about traffic.
You see the AI block above the results and cannot tell whether it is taking your clicks.
It is not obvious which queries turn AI Overviews on at all, or whether you can influence that.
Vendors push "AIO optimization" with no clarity about what is real and what is invented.
AI Overviews use "a customized Gemini model, which works in tandem with our existing Search systems – like our quality and ranking systems and the Google Knowledge Graph". This is not a standalone chatbot: the block is "designed to carry out traditional 'search' tasks, like identifying relevant, high-quality results from our index to corroborate the information presented in the overview" [65, 66, 85].
The key selection mechanism is corroboration. Google writes that "AI Overviews are built to only surface information that is backed up by top web results, and include links to web content that supports the information presented in the overview". For YMYL queries — health, finance — "we have an even higher bar for showing supporting information from reliable and trustworthy sources" [65].
Triggering is deliberately limited. Google says the block was "designed to show up on queries where they can add additional benefit beyond what people might already get on Search today", while the company "aim[s] to not show AI Overviews for hard news topics, where freshness and factuality are particularly important", and restricts election-related queries [65].
Nothing special is required. Google: to be eligible to appear as a supporting link in AI Overviews or AI Mode, a page needs to be indexed and eligible to appear in Google Search with a snippet [66].
No special markup is needed. Verbatim: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add" [66]. The AI optimization guide adds: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add", and there is no requirement to chunk your content into tiny pieces for AI to understand it better [85].
Display is controlled through the standard directives: nosnippet, data-nosnippet, max-snippet, noindex, robots.txt, and Google-Extended for limiting AI training [66].
Google's official line is optimistic: "people who use AI Overviews actually use Search more and are more satisfied with their results", and clicks from the block are "higher quality". The independent preprint by Xu, Iqbal and Montgomery gives a measurable and harsher picture, drawn from 55,393 trending queries across 19 topical categories over a 40-day window (13 March – 21 April 2026) [87].
Overall AI Overviews activation runs at 13.7% of queries, rising to 64.7% on question-form queries; nearly 30% of AIO sources do not appear in the first page of organic results [87].
13.7% of queries
64.7%
AIO domains are on average more authoritative
nearly 30%
98,020
11.0% (omission is the dominant failure mode)
well over half
Two findings sit in tension with Google's position. First, about 30% of cited sources are not on the first page of results: the AI Overviews source set overlaps the classic results without being identical to them, and independent measurement points to partial divergence rather than a contradiction of being rooted in core ranking. Second, 11% of claims are not supported by the page cited — which contrasts with Google's statement that AI Overviews "generally don't 'hallucinate' in the ways that other LLM experiences might" [87].
official PDF — corroboration by top results; a customized Gemini plus core ranking; a higher bar for YMYL
official documentation — eligibility = indexed plus snippet-eligible; no special markup; display controls
official documentation — structured data not required; no chunking requirement
preprint arXiv:2605.14021 — 13.7% activation (64.7% on questions); ~30% of sources outside the top results; 11% unsupported claims
By independent measurement (preprint arXiv:2605.14021), roughly 13.7% of queries overall and up to 64.7% of question-form queries. Google deliberately withholds the block for hard news and sensitive election topics.
No. Google requires only that the page be indexed and eligible to appear with a snippet. No special schema.org markup or AI files are needed.
Partly. Preview and indexing controls (nosnippet, data-nosnippet, noindex) limit snippet display and eligibility. Google-Extended governs the use of content for training and grounding in Gemini Apps and Vertex AI, but it does NOT govern whether a page appears in AI Overviews.
No. Google states that they are corroborated by top results, but the independent preprint found 11.0% of atomic claims unsupported by the cited page, most often through omitted context.
How AI Overviews work. https://www.google.com/search/howsearchworks/google-about-AI-overviews.pdf — official PDF (Google's own explanation).
AI features (Google Search Central). https://developers.google.com/search/docs/appearance/ai-features — official documentation.
AI optimization guide (Google Search Central). https://developers.google.com/search/docs/fundamentals/ai-optimization-guide — official documentation.
An independent measurement of AI Overviews activation and citation. https://arxiv.org/abs/2605.14021 — arXiv preprint.
The Enigma editorial team
18 May 2026
18 May 2026
Peer-reviewed work, preprints and industry research — the type is stated next to each claim
Google quotations come from the original July 2024 PDF and current Search Central pages; the independent figures from the preprint abstract
Preprints and industry reports are cited with their methodological limits; verify the conclusions on your own project
The full list of sources and their reliability levels lives in the research catalogue.