TL;DR: Google's official position on AI Overviews is that nothing special is required: an indexed page, eligible for a snippet, standard SEO. The March 2026 data adds the part Google doesn't say. Only 37% of cited pages rank in the top 10, down from 76% in July 2025 (Ahrefs). Citation has come unchained from ranking. Structure decides who gets picked.
Most Google AI Overviews SEO advice falls into one of two camps: panic about lost traffic, or a pitch for special "AI optimization" that Google itself says does not exist. Neither camp reads the primary sources.
This post does. Every external claim below comes from Google's own documentation and announcements, or from large-scale independent studies: Pew Research Center's tracking of 68,879 real user searches, Ahrefs' analyses of 300,000 and 863,000 keywords, and Semrush's 10-million-keyword tracker. Plus one thing none of those can give you: what building AI citation visibility from a zero-authority domain actually looks like from inside, using this blog's own numbers.
What are Google AI Overviews and how do they actually work?
Answer capsule: AI Overviews are AI-generated summaries that appear above traditional results, launched in the US on May 14, 2024 and expanded to more than 100 countries that October (Google). A Gemini model customized for Search retrieves content, generates an answer, and links a small set of supporting pages. Google says more than 1 billion people see them monthly.
Google announced AI Overviews at I/O on May 14, 2024, rolling out to US users first with a stated goal of reaching "over a billion people by the end of the year." The October 28, 2024 expansion took them to "more than 100 countries and territories" and past 1 billion monthly users (Google, The Keyword blog).
Under the hood, Google describes "a new Gemini model customized for Google Search" that combines multi-step reasoning with its existing Search systems. The sequence matters more than the model: the system retrieves pages, generates a synthesized answer, and attaches links to supporting sources. Your page either gets retrieved and credited, or it plays no visible part.
One more thing worth internalizing from Google's documentation: "AI is built into Search and integral to how Search functions." This is not a lab feature bolted onto the results page. It is the results page.
What triggers an AI Overview, and what doesn't?
Answer capsule: Google's stated rule: "AI Overviews are only shown when our systems determine that it is additive to classic Search." In measured practice, Semrush's 10-million-keyword tracker put the trigger rate at 6.49% of queries in January 2025, a 24.61% peak that July, and 15.69% by November 2025. Longer, more specific research queries trigger them most.
The trigger rate is volatile, and anyone quoting a single flat percentage is selling a snapshot. What is stable is the shape of the queries that trigger them: longer, more specific, research-stage.
The composition is also shifting toward money queries. In January 2025, 91.3% of AI Overview keywords were informational. By October 2025 that was down to 57.1%, with commercial, transactional, and navigational queries taking the rest (Semrush). A follow-up Semrush study of 600,000+ US desktop keywords found commercial-intent SERPs with an AI Overview grew 71% between November 2025 and April 2026, while transactional SERPs declined 5%.
The practical read: Google is positioning AI Overviews in the research stage of buying decisions and keeping them away from the checkout moment. If your business wins customers during comparison and research, this is your fight. If you sell on pure transactional queries, you have more time than the panic suggests.
How does Google choose which pages to cite?
Answer capsule: Through retrieval, not ranking alone. Google uses "query fan-out": issuing multiple related searches across subtopics to build the response. As of March 2026, only 37.1% of AI Overview citations ranked in the organic top 10 for the query, and 36.7% ranked outside the top 100 entirely (Ahrefs, 863,000 SERPs).
This is the finding that should reorganize your strategy.
In July 2025, roughly 76% of AI Overview citations came straight from the top of the original SERP. By March 2026, that was down to roughly 38% (Ahrefs, 863,000 keyword SERPs, 4 million cited URLs). A third of citations now come from pages that do not rank in the top 100 for the triggering query at all.
The mechanism is query fan-out. Google's documentation describes it directly: the system issues multiple related searches across subtopics and data sources to develop a response, which lets it "display a wider and more diverse set of helpful links" than classic results. Your page competes to be the best answer to one of the sub-questions Google generates behind the scenes, and only incidentally for the head term.
That inverts the old sequence. Ranking used to be the prerequisite and citation the reward. Now a page that precisely answers a sub-question can get cited while ranking on page four, and a page that ranks #3 with a meandering answer can get skipped.
Do you need special schema or markup to appear in AI Overviews?
Answer capsule: No. Google's documentation is explicit: "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." The only technical requirement is that a page is indexed and eligible to show with a snippet.
An entire mini-industry has formed around "AI Overview optimization" markup, llms.txt files, and proprietary AI-readiness scores. Per Google, there is nothing to buy. "There are no additional technical requirements."
Here is the honest version of what structure actually does, since we sell AI SEO services and this cuts against a common pitch: schema and content structure do not unlock AI Overview eligibility. Eligibility is indexing plus snippet eligibility, full stop. What structure changes is extraction odds once you are eligible. A page organized as clean question-and-answer units gives a fan-out retrieval system a self-contained passage to select. A 2,000-word wall of prose makes the system work for it, and the system does not work for it.
We use FAQPage schema and answer-first sections on this blog because they create extractable units, and because structured answer formatting is measurably associated with citation on other AI platforms. Not because Google requires them. Anyone telling you otherwise is selling you Google's own words, marked up.
What do AI Overviews actually do to your clicks?
Answer capsule: Two things are true at once. Google reports that clicks from pages with AI Overviews are "higher quality (meaning, users are more likely to spend more time on the site)." Independent measurement shows there are fewer of them: users clicked a traditional result on 8% of visits when an AI summary appeared, versus 15% without one (Pew Research Center, 68,879 real searches, March 2025).
The Pew study is the cleanest real-user data available: 900 US adults, 68,879 actual Google searches during March 2025. When an AI summary appeared, users clicked a traditional result link on 8% of visits, roughly half the 15% rate without a summary. Links inside the AI summary itself were clicked on just 1% of visits. And 26% of visits with an AI summary ended right there, versus 16% without.
Ahrefs measured the same effect from the site side: comparing 300,000 keywords across March 2024 and March 2025, the presence of an AI Overview correlated with a 34.5% lower average clickthrough rate for the top-ranking page.
Google's counter-claims are about the clicks that survive: they arrive from users who already read a summary, so they land deeper in intent and stay longer. The claims measure different things and can be simultaneously true. Only the independent side publishes its methodology. The informational click pool is shrinking, and what remains is better qualified.
The strategic conclusion: informational content now earns its keep through citation and brand impressions rather than clicks, so build for citation, and put your click expectations on mid-funnel queries where a five-line summary cannot finish the job.
How do you structure content so Google's systems select it?
Answer capsule: Structure each section as a standalone answer to one sub-question: a direct answer in the first 100 words, a question-phrased heading, named primary sources with dates, and current information. Query fan-out means Google retrieves passages, not pages. Every section competes on its own.
The discipline that earns selection is the same one that earns citations from ChatGPT and Perplexity, applied with fan-out in mind:
Enumerate the sub-questions before writing. Fan-out means Google decomposes the query. Your section list should look like that decomposition. Pull the People Also Ask boxes, your own Search Console query report, and the follow-up questions a real customer asks, then give each its own section.
Answer first, support second. State the answer in the opening sentences of each section, then earn it. A retrieval system selecting passages for a summary takes the passage that already is a summary.
Make sections self-contained. A 50–150 word passage that names its subject (not "it" or "this approach"), states a fact, and cites a dated source can be lifted whole. That is the unit of AI citation.
Cite primary sources with dates. A passage carrying "Pew Research Center, March 2025" gives the system a verifiable anchor. A passage asserting the same thing bare does not.
Stay current. The trigger-rate volatility above means the query surface reshuffles constantly. Dated posts about a moving target age out of selection.
None of this is exotic. It is the E-E-A-T substance Google has pushed for years, compressed into an extractable shape. The sites losing to AI Overviews are mostly sites that never structured anything. The 9-point AI SEO audit is the fastest way to find out which camp you're in.
What does our own citation data say about starting from zero?
Answer capsule: On this blog, a new domain with minimal backlinks, one structured post generated 932+ Bing AI citations across query variants, peaking at 171 citations per day on May 7, 2026, while Google organic clicks were near zero. AI citation volume arrived weeks before Google authority did.
First-party data, on the record: the Square Issuing V2 evaluation post on this blog was built exactly the way the previous section describes. Question-phrased sections, answer-first passages, dated primary sources, FAQPage schema. It generated 932+ Bing AI citations across its query variants (May 2026 count), climbing from zero to a peak of 171 per day, then settling in a 56–111 per day range. During that entire window, Google organic clicks were near zero.
Named caveat: Bing AI citations are not Google AI Overviews. Different index, different retrieval system. We use Bing's data because it is the only platform that shows citation counts to site owners, which makes it our measurable proxy for extraction-readiness.
The reason we treat the method as transferable is the Ahrefs March 2026 finding above: with a third of AI Overview citations ranking outside the top 100, Google is also selecting for answer precision over accumulated authority. That is the same selection logic our Bing data measures.
What the transition looks like from inside, per our July 2026 Search Console review: impressions up 5x since April while average position still sat near 60 — visibility arriving well before rankings. And on the Bing side, citations diversified from 2–4 distinct cited pages per day to 9–15 as newer posts started earning their own citations. A zero-authority domain does not have to wait for domain authority to be selectable. It has to be extractable.
Can you keep your content out of AI Overviews?
Answer capsule: Yes, with the same controls that govern snippets: "nosnippet," "data-nosnippet," "max-snippet," or "noindex." There is no AI-only switch. Google's documentation ties the two surfaces together because "AI is built into Search and integral to how Search functions."
The controls are blunt. "nosnippet" and its variants keep content out of AI features and out of your regular snippets. "noindex" removes the page from Search entirely. You cannot stay fully visible in classic results while withholding content from the AI layer.
For most small businesses, the math favors staying in. The informational click decline hits whether you participate or not; opting out just removes you from the surface where the remaining attention goes. The genuine exception is paywalled or subscription content, where a summary can substitute for the product itself. If that is your model, "data-nosnippet" on the premium passages is the scalpel, and it is a business decision worth making deliberately.
Frequently Asked Questions
How does AI Overview traffic show up in Google Search Console?
It doesn't get its own report. Google's documentation states that sites appearing in AI features "are included in the overall search traffic in Search Console," reported in the Performance report under the "Web" search type. You cannot isolate AI Overview clicks or impressions. The workable proxy is divergence: impressions climbing while average position stays flat suggests AI-surface visibility your position data can't explain.
What is the difference between AI Overviews and AI Mode?
AI Overviews are summaries added to the classic results page when Google's systems judge them additive. AI Mode is a separate, fully AI-driven search experience. Google notes they run on different models, so "the set of responses and links they show will vary." Eligibility is identical for both: indexed, snippet-eligible, no extra requirements. Content selected for one is not guaranteed to appear in the other.
Do Google AI Overviews pull from the same sources as ChatGPT?
No. ChatGPT retrieves through Bing's index; Google's AI features retrieve through Google's own fan-out queries. A page can be a fixture in one system and invisible in the other, which is why we track them separately. The structural signals transfer even though the source pools don't — the same extractable section can win in both pipelines. The ChatGPT and Perplexity mechanics are covered in their own post, and Perplexity's citation system gets its own deep dive.
How quickly can a new page become eligible for an AI Overview?
As soon as it is indexed and snippet-eligible. There is no separate AI indexing queue or waiting period in Google's documentation, and no seasoning requirement. Combined with the Ahrefs finding that 36.7% of cited pages rank outside the top 100, this means a well-structured page can theoretically be cited before it has meaningful rankings. Indexing speed, not authority, is the gate for new content.
Does linking out to primary sources help you get cited?
Nothing in Google's documentation says outbound links affect AI Overview selection. What we can report as operators: every post on this blog that earns AI citations carries dated primary-source links, and the passages that get extracted are consistently the ones anchored to a named source. A retrieval system assembling a factual summary has an easier job selecting a passage whose facts arrive pre-verified. We treat sourcing as extraction insurance, not a ranking trick.
Google published the eligibility rules. They fit in a sentence: be indexed, be snippet-eligible. The selection game is where the work is, and the March 2026 data says that game now favors precise, structured, sourced answers over accumulated authority.
That is a genuinely unusual opening for small sites. It is the same opening we took with this blog: 932+ AI citations from a new domain before Google authority arrived, using the method this post just walked through.
If you want that method run on your business, that is the service.
Kaleb Dickhaut — Founder, ClickWerxs. Kaleb built ClickWerxs from the ground up — from payment processing ISO to the Command Center platform to the AI SEO methodology the blog runs on. He has onboarded hundreds of small businesses onto payment and CRM systems. linkedin.com/in/kaleb-dickhaut
Sources
- StatCounter Global Stats, search engine market share worldwide — Google 91.25% as of June 2026. gs.statcounter.com
- ClickWerxs blog, first-party AI citation data, May 2026 — one post generated 932+ Bing AI citations across query variants from a new domain; daily citations rose from 0 to a peak of 171 on 7 May 2026, settling at 56–111 per day, with Google organic clicks near zero over the same period. Reported as a past result, not a promise of performance. Operator data.
ClickWerxs sells SEO and AI visibility services and earns revenue from those engagements. First-party figures are past results for this blog and are not a promise of future performance. This is operator opinion, not professional advice.