Google AI Overviews are the AI-generated summaries that sit above the traditional ten blue links on a Google search results page. They were announced at Google I/O on May 14, 2024 and powered by Gemini (Google Blog, May 2024). By Google I/O 2025, the feature had scaled to 1.5 billion monthly users across more than 200 countries and over 40 languages (Google Blog, May 2025).

For any business that depends on organic search traffic, this is the most consequential change to Google in a decade. AI Overviews now appear on roughly 13% of queries, and when they appear, organic click-through rates drop by an average of 47% (Seer Interactive, September 2025). The link slot used to be the prize. Now the citation slot inside the AI Overview is the prize.

What an AI Overview actually is

An AI Overview is a Gemini-generated answer that pulls from pages already in Google's search index. It is not a separate AI product crawling the web in real time. It is a synthesis layer that sits on top of the same index Google has always maintained, generated when Google decides a query is better served by a direct answer than a list of links.

The mechanic Google uses internally is called query fan-out. The original query is split into a set of related sub-queries, each sub-query is run against the index, and the pages that appear most consistently across the sub-query results become candidate citations. Gemini then assembles a two to five sentence answer with inline citations back to the source URLs.

How AI Overviews differ from ChatGPT Search and Perplexity

  • Google AI Overviews runs Gemini on the existing Google index. If a page is not indexable by Google, it cannot be cited. Traditional SEO is the prerequisite.
  • ChatGPT Search uses Bing as one input but blends it with OpenAI's own OAI-SearchBot crawler. A site that ranks poorly on Google can still appear in ChatGPT Search. See how to optimize for ChatGPT Search for the full playbook.
  • Perplexity fetches live pages during answer construction and cites four to six sources per response. It rewards freshness and depth more than any other engine.

The signals Google uses to pick AI Overview sources

  1. E-E-A-T. Multiple studies put the share of AI Overview citations coming from sources with strong E-E-A-T signals above 90%. Author bylines, credential pages, and clear publisher identity all contribute.
  2. Content freshness. 2025 citation analysis shows roughly 44% of AI Overview citations come from content published that year, with another 30% from 2024. Old content gets passed over even when it ranks well organically.
  3. Structured data and schema markup. Pages with proper JSON-LD schema show meaningfully higher selection rates than unmarked equivalents. FAQPage, HowTo, Article, and LocalBusiness are the highest impact.
  4. Direct, self-contained answers. Pages that answer the question completely in the first 100 words, structured cleanly, are several times more likely to be cited than pages that bury the answer in marketing copy.
  5. Ranking eligibility, then selection. AIO does not create its candidate pool from nothing. It picks from pages that already rank for the query or its sub-queries. 47% of AIO citations come from pages ranking below position five, which means top-three ranking is not required, but indexable and ranking-eligible content is.

The local business angle

AI Overviews behave differently on local-intent queries. When the query is “best dentist in Naperville” or “personal injury lawyer in Nashville,” Google pulls from a wider source set that includes Google Business Profile data, individual GBP reviews, review aggregators, and local directory listings, not just web pages.

Google began citing individual reviews from Business Profiles inside AI Overviews in 2025. A review that says “Dr. Kim finished my Invisalign in 14 months” is now potentially quoted verbatim in an AI Overview when a prospective patient asks about Invisalign timelines in your city. A review that says “great dentist” is not.

The two-step problem most businesses miss

To appear in a Google AI Overview, you have to clear two gates. First, your page has to be indexable and ranking-eligible for the query or one of its query fan-out sub-queries. Second, your page has to have the structural and quality signals that make Gemini pick it from the candidate pool.

Most businesses skip the first gate because they assume AEO is purely about schema and content format. It is not. Traditional SEO is the prerequisite. The new work sits on top.

Author bylines and credentialed sources

AI Overviews disproportionately cite content with clear authorship. For professional services, that means every substantive page should have a credentialed author byline linking to a real Person schema with jobTitle, alumniOf, memberOf, and sameAs properties. The schema implementation for this is covered in which schemas AI actually reads.

How to test if you appear today

Open Google in a logged-in browser, search for the kind of query a prospective client would type, and look at the top of the results page.

  • “Best [your service] in [your city]”
  • “How much does [your service] cost in [your city]”
  • “[Your service] near [landmark]”

If an AI Overview appears, expand the sources panel. Google shows which URLs were cited. If your domain is not in that list, you have a candidate-pool or selection-quality problem. If your domain is listed but the description is wrong, you have a schema and content problem.

The zero-click tradeoff

AI Overviews suppress traffic. Even being cited inside an AI Overview produces far fewer clicks than ranking #1 used to. The compensation is brand exposure. When Google's AI tells a prospective client “X is a highly reviewed option,” that recommendation carries weight even if the user never clicks through. The slot is more like a verbal recommendation from a trusted source than a clickable link.

Where to start

Run the four query types above for your top services in your top cities. Note which queries trigger an AI Overview, and note whether your domain is cited. From there, the install order that produces the fastest movement: verify Google is not blocked in robots.txt, add LocalBusiness and FAQPage schema to your top service pages, write a direct first-100-words answer on each page, add credentialed Person schema for every named practitioner. The deeper BrightLocal 2026 data deep-dive covers the consumer-side numbers driving the urgency.

To see exactly which queries currently trigger AI Overviews for your practice and where the citation gaps are, run a free AEO Grader scan. We query ChatGPT, Claude, Gemini, and Perplexity with 60 real prompts about your business and return the prioritized fix list in about sixty seconds.