How Google AI Overviews quietly rewrote the SEO playbook

When Google rolled out AI Overviews, most agencies treated it as a tweak to the SERP. It is not a tweak. It is a new layer that intercepts intent before the user ever scrolls to the blue links — and it has changed which content gets rewarded. The discipline built around winning that layer is Generative Engine Optimization.

What is a Google AI Overview, exactly?

An AI Overview is a generated answer that sits above the organic results for a subset of queries. It is assembled, not retrieved: Google's model reads a handful of pages, extracts the passages most relevant to the question, and stitches them into a few sentences with linked citations. The user often gets the answer without clicking. The pages that supplied it get the citation — and, increasingly, the click that follows it.

Google confirms the mechanism in its own documentation: AI Overviews and AI Mode use a query fan-out technique — issuing multiple related searches across subtopics and data sources — to build a response. That single sentence explains why the page ranking #1 for your keyword is not automatically the page the overview quotes: Google is not summarizing one SERP, it is summarizing several.

Which searches trigger an AI Overview — and which don't?

AI Overviews appear on 20.5% of all SERPs — roughly one in five keywords — but that average hides the only split that matters to you. Ahrefs analyzed 146 million SERPs and found the trigger rate runs to 21.4% on informational queries and 57.9% on question queries, while commercial-intent keywords trigger an overview just 4.3% of the time and transactional keywords only 2.1% (Ahrefs, September 2025 desktop data).

The number that should relax a service-business owner: in the same Ahrefs dataset, only 7.9% of queries classified as local searches returned an AI Overview. Somebody typing 'emergency plumber near me' still sees the Local Pack, the ads, and the blue links, because there is nothing to summarize and Google knows the intent is to act, not to read. Your highest-intent money queries are the least disrupted. Your top-of-funnel educational content is where the overview reshapes the click. Earning a spot inside those overviews is its own discipline — the complete GEO playbook covers it end to end.

How much traffic do AI Overviews actually take?

When an AI summary appears, users click a traditional search result on just 8% of visits — versus 15% when no summary is present. That figure comes from Pew Research Center, which tracked the real browsing behavior of 900 US adults across 68,879 Google searches in March 2025 — not modeled CTR, not scraped SERPs, but what people actually did. Pew also found 18% of those real searches generated an AI summary.

The other half of the Pew finding is the one nobody quotes: users clicked a link inside the AI summary itself on only 1% of visits. So the honest framing is not 'get cited and harvest the citation click.' The citation click is small. What being cited actually buys you is your brand name in the most prominent slot on the page, in front of someone actively researching — and the credibility that carries into the branded search they run later. Treat the click loss as real but concentrated at the top of the funnel, not spread evenly across your site.

What actually changed about SEO?

Ranking in the top 10 no longer gets you cited — and there is now hard data on exactly how decoupled the two have become. Ahrefs studied 863,000 keyword SERPs and 4 million AI Overview URLs and found that only 38% of pages cited in an AI Overview also rank in Google's top 10 for that same query, down from about 76% a year earlier (Ahrefs, March 2026). Ahrefs found the rest split almost evenly between pages ranking 11-100 (31.2%) and pages that do not rank in the top 100 at all (31.0%).

That is the whole story in one number. AI Overviews are extractive: they pull short, self-contained passages and stitch them into an answer. Pages that read like magazine articles — long wind-ups, narrative transitions, the real answer buried in paragraph four — no longer extract well. Pages that read like reference manuals — short paragraphs, a definition-first opening, specific numbers, clear headings — extract beautifully. Structure now competes with authority, and for a small service business that is an opening: you do not need the biggest domain to be the cited source, you need the clearest one.

What did not change?

Everything you were already supposed to be doing. Google's own AI features documentation states plainly that to be shown as a supporting link in AI Overviews or AI Mode, 'a page must be indexed and eligible to be shown in Google Search with a snippet' — and that 'there are no additional technical requirements.' There is no AI Overview schema, no secret markup, no vendor tool that buys you in.

So topical authority still matters. Backlinks still matter. Page speed still matters. Ranking is still mostly about being a trusted, relevant result. The shift is in how that trust gets surfaced — and in the fact that ranking on page one is no longer the same as being seen. If the AI Overview answers the question and you are not in it, you are invisible even at position three. Anyone selling you a separate 'AI Overview optimization' product on top of your SEO retainer is selling you something Google says does not exist.

Which content patterns win AI Overview citations?

• Definition-first openings. Lead every commercial page and FAQ answer with a 40-80 word self-contained answer, then expand — the model can lift it whole.
• Question-shaped headings. Use the literal question a buyer would type as an H2 ("How much does dental SEO cost?"), then answer it immediately underneath.
• Specific, attributable facts. Named entities, concrete ranges, and cited sources extract and survive better than vague claims.
• FAQPage schema on every page with a genuine Q&A section, so the structure is machine-readable, not just visually formatted.

What are the most common AI Overview mistakes?

Two failure modes show up repeatedly: over-optimizing and under-measuring. Over-optimizing looks like stuffing every page with FAQ blocks that repeat the same generic answer — the model treats duplicated copy as low value and the pages cannibalize each other. Under-measuring looks like having no idea whether you appear in AI Overviews at all, because nobody added it to the monthly report. You cannot improve a surface you are not watching — see our breakdown of GEO vs AEO vs SEO for how this surface fits the bigger AI-visibility picture.

A third mistake is panicking about the wrong pages. Given the 4.3% commercial and 7.9% local trigger rates in the Ahrefs data, your service pages and 'near me' queries are largely untouched. If your traffic is down, check whether the loss is concentrated in blog and how-to pages before you rebuild anything commercial. Rebuilding a money page that was never affected is the most expensive way to respond to an AI Overview.

What should you do this quarter?

• Audit your top 25 commercial pages for passage-level extractability. Lead with a 40-80 word definition, follow with structured FAQ, and cut the narrative wind-up.
• Deploy FAQPage schema on every commercial page that has a real Q&A section — and remove duplicated boilerplate answers that dilute relevance.
• Track AI Overview presence per priority query in your monthly reporting. If you cannot measure it, you cannot optimize for it, and you will not know when a competitor displaces you.

AI Overviews are not the death of SEO. They are SEO maturing into something closer to answer design. The discipline is the same — be the most trustworthy, relevant source — but the unit of optimization has shrunk from the page to the passage. The businesses that adapt to that smaller unit first will compound the citation advantage over the next 24 months. If you want that adaptation handled for you, see our AI SEO service.