Google search doesn’t look the way it did two years ago. Type in a question and there’s a good chance the answer is sitting at the top of the page before you’ve clicked a single link.
BrightEdge tracked AI Overview presence daily for twelve months and found it grew from roughly 30% of tracked queries to 48% by February 2026, a 58% increase.
Now flip that number over, because almost nobody does. About 52% of search queries still return no AI Overview. For most searches, organic results are still the entire experience.
Both things are true. Which one applies to your business depends on what people type to find you.
What Are AI Overviews?
AI Overviews are automatically generated summaries that appear at the top of the results page for certain searches. Instead of scrolling through ten blue links, users get a written answer pulling information from multiple sources, with citations linked underneath. They’re Gemini-powered, generated using Google’s own large language models, and they’ve rolled out across more than 200 countries.
This is different from Google’s older Search Generative Experience (SGE), the experimental version tested before AI Overviews went mainstream. AI Overviews are the permanent replacement.
Google also has AI Mode, a conversational search experience that lets users ask follow-up questions the way they would with a chatbot. Different feature, same idea: answer the query directly instead of pointing you somewhere else.
Worth knowing how much screen space this takes. BrightEdge measured the average AI Overview at over 1,200 pixels tall against a standard desktop viewport of about 900. When one appears, the first organic result sits entirely below the fold. That’s before click-through rate enters the conversation. The listing is just gone from view.
How Do Google AI Overviews Work?
When someone runs a search, Google doesn’t summarize the top result. The process behind AI Overviews is called query fan-out: Google breaks the original query into related sub-questions, searches for answers to each, and stitches the findings into one summary.
This is why ranking #1 and getting cited are two different outcomes. BrightEdge found that only about 17% of sources cited in AI Overviews also rank in the organic top 10, and that number has been flat for months. Roughly five out of six citations come from content that isn’t on page one.
That cuts both ways. Strong search rankings don’t automatically get you cited. Not ranking on page one doesn’t lock you out.
It also varies enormously by industry. Healthcare has the highest overlap at 24%, which fits how cautious Google is with medical content. Finance sits at 11%, meaning nearly nine out of ten citations in that space come from somewhere other than page one.
Not sure how AI tools are describing your business right now?
What Types of Questions Trigger AI Overviews?
Query length and phrasing matter more than most people realize.
The 2025 Pew Research Center study tracked the actual browsing behavior of 900 U.S. adults through March 2025 and found:
| Query type | How often an AI summary appeared |
|---|---|
| One or two-word searches | 8% |
| Full-sentence searches (noun and verb) | 36% |
| Ten or more words | 53% |
| Question-based (“who,” “what,” “why”) | 60% |
Short keyword strings are the least likely to trigger AI Overviews. Long, conversational questions are the most likely. If your content is built around the actual types of questions your customers ask, you’re writing directly into the queries where AI Overviews show up.
Whether AI Overviews Matter Depends On What People Search
Here’s where most articles on this topic fall apart. They quote one blended average and let you assume it applies to you.
Local search is the clearest example. A Whitespark study covered in Search Engine Land’s guide to AI and local search found AI Overviews appearing on 68% of local searches overall. Scary number in isolation. The breakdown by intent is what actually matters:
| Query type | Example | AI Overview appears |
|---|---|---|
| Simple local intent | “tacos san francisco” | ~15% |
| Informational local | “how long does an eye exam take near me?” | ~92% |
| Hybrid intent | “average cost of dental implants in phoenix” | ~97% |
That same study found local packs appearing for more than 90% of simple local queries and 39% of queries overall.
So if you’re a plumber ranking for “plumber philadelphia,” your exposure is small, and your budget belongs in the local pack and your service pages. If you’re a dental practice creating content about implant costs, or a healthcare practice answering procedure questions, you’re in the 92% to 97% range, and this is urgent.
Same SERP feature, opposite strategy. Anyone selling you an AI search package without asking which of those you are isn’t paying attention.
The Click Problem And What Cited Sources Actually Get
AI Overviews are cutting into website traffic. That worries brands and marketers alike, and the Pew data is blunt.
Users clicked a traditional result in just 8% of searches with an AI summary, compared with 15% on pages without one. Clicks on the cited sources inside the summary were rarer still, at 1% of visits. And 26% of users ended their browsing session entirely after a page with an AI summary, versus 16% without.
Zero-click searches aren’t new. Pew found roughly two-thirds of all Google searches end with the user browsing elsewhere on Google or leaving without clicking anything, with or without AI. AI Overviews made an existing problem worse; they didn’t invent it.
Seer Interactive’s numbers line up. They tracked 3,119 informational search terms across 42 organizations and found organic click-through rate on queries with an AI Overview fell 61%, from 1.76% down to 0.61%.
Their larger study of 53 brands across 5.47 million queries found brands cited within the AI Overview earn roughly 120% more organic clicks per impression than brands on the same page that don’t get cited.
Now, the part that gets left out of the headline. That same study found cited brands still underperform a no-AI-Overview result by 38%. Getting cited is an advantage over everyone else on that page. It is not a return to what you had. Seer also says plainly that they can’t prove citations cause the lift, since brands with stronger authority and better baseline click-through rates are more likely to get cited in the first place.
Two distinctions worth being precise about, because the industry blurs them constantly:
A mention is not a citation. Being named in the summary text is different from being linked as a source. A mention with no link means the customer has to find you, which is friction you didn’t have before.
Citation position matters. Being the source the summary leans on is not the same as being the seventeenth link in the source tray. Being citation #17 is functionally the same as not being cited.
There’s also no accepted attribution model for any of this. Someone discovers you in an AI Overview and converts three weeks later through branded search. Your analytics credits direct. Anyone promising you clean AI search attribution is selling something.
Before You Panic: The Hype Is Ahead Of The Behavior
Worth keeping in view while everyone yells that search is dead.
SparkToro’s August 2025 research, also covered in that Search Engine Land guide, found that only 20% of Americans use AI tools 10 or more times per month, while 95% still engage with traditional search engines monthly. Semrush found that when people adopt ChatGPT, their Google usage goes up, not down.
Rand Fishkin’s read on his own data: traditional search isn’t going anywhere even among heavy AI adopters, and the AI-versus-search narrative is largely manufactured by media and influencers chasing attention.
Plan for AI-driven search. Don’t abandon the channel still doing 95% of the work.
Want to know whether AI Overviews are actually costing you traffic, or whether your query mix is barely affected?
What Google Says About How To Optimize For AI Overviews
There’s a lot of noise claiming you need a separate strategy to win at AI search. You don’t, and Google has said so directly.
Gary Illyes, speaking at Google’s Search Central Deep Dive event, said that to appear in AI Overviews, you use normal SEO practices and don’t need GEO, LLMO, or anything else. John Mueller made the same point on the terminology fight: what you call it doesn’t matter, but AI isn’t going away, so thinking about how your site’s value works in a world where AI exists is worth the time. Danny Sullivan’s version is shorter. Good SEO is good GEO.
Generative Engine Optimization (GEO) is SEO applied with an eye toward how AI systems pull and summarize information. It isn’t a separate discipline with its own ranking system. Google builds AI Overviews from the same index, crawl, and quality standards that power organic results.
| SEO Fundamental | Why It Also Helps With AI Overviews |
|---|---|
| Content that answers a specific question directly | Matches how query fan-out pulls sub-answers |
| Crawlable, indexable pages returning HTTP 200 | Google can’t summarize what it can’t fetch |
| Backlinks and mentions from reputable sources | Feeds the sources AI-generated answers cite most |
| On-page SEO basics like clear titles and meta descriptions | Improves how your page reads on the search results page |
| Fast, clean technical SEO across desktop and mobile devices | Baseline quality signals Google checks first |
One caveat. That 17% top-10 overlap means organic ranking isn’t a gate you have to pass through to get cited. But the broader overlap, sources ranking anywhere in the top 100, sits around 53% and has been climbing. Being in Google’s index with real topical depth is still the entry ticket. It just isn’t the same race as fighting for position one.
What Site Owners Can Skip When Optimizing For AI
Short list of things being sold as AI search essentials that the data says do nothing.
llms.txt. Ahrefs analyzed 137,000 domains and found 28% publish an llms.txt file, and 97% of those files received zero requests in May 2026. Not zero AI traffic. Zero traffic. Nothing fetched them at all.
Of the 3% that did get read, AI retrieval bots (the ones that actually feed AI search answers) accounted for 1.1% of requests. Slackbot’s link-preview crawler fetched llms.txt files more often than PerplexityBot did. And when Ahrefs looked at 404s for llms.txt files that don’t exist, the AI bot share was zero. Nothing goes looking for one.
Google’s own AI optimization guide has a section titled “mythbusting” stating you don’t need machine-readable files, AI text files, markup, or Markdown to appear in generative AI search. Mueller called llms.txt a temporary crutch for AI coding tools parsing developer docs, and said flatly it’s “not done for search.”
If your customers are developers using coding agents, there’s a real use case. If you’re a dentist, skip it.
Publishing Markdown copies of your pages for bots. Martin Splitt has pointed out that Markdown strips the link structure and heading relationships that language models use to understand context. You’d be removing signal.
Turning every heading into a question and chopping paragraphs into “chunks.” Nobody has demonstrated this works. Readable beats over-engineered.
Schema markup as an AI play. Schema is fine. It’s table stakes for rich results and other SERP features, and it helps Google parse your entities. Nobody has proven it changes whether an AI engine cites you.
Skipping these isn’t laziness. It’s spending the same hours on things with evidence behind them.
5 Ways To Improve Your Chances Of Appearing In AI Overviews
Answer questions directly. Get to the point instead of burying the answer three paragraphs down. Google shows a summary when it thinks it can help users faster than a list of links can.
Build topical depth, not volume. A few strong pieces covering a subject properly signal expertise. Mass publishing thin content does not, and it can damage the rankings you already have.
Earn backlinks and mentions from sources people trust. AI systems repeat what the web says about you. Third-party mentions on review sites, industry publications, and community platforms feed directly into AI Overviews. If nobody credible is saying it, the AI won’t either.
Keep technical SEO clean. Google’s guidance for website owners is specific: don’t block Googlebot, make sure pages return HTTP 200, publish indexable content in supported formats. Make sure your site loads quickly on mobile devices. Add strong images and video, and keep Merchant Center listings current if you sell products.
Change what you measure. Google says clicks arriving from pages with AI Overviews tend to be higher quality, with users spending more time on site. Take that with appropriate skepticism given who’s saying it, but the underlying advice holds: track metrics like leads, calls, form fills, and revenue, not just sessions and snippet impressions.
One caveat on all of it. Showing up in Google AI Overviews does not mean you’ll show up in ChatGPT, Perplexity, or Claude: different systems, different sources, different behavior. If AI visibility matters to your business, you need to check it platform by platform.
Hallucinations Are A Real Business Risk
Separate from the traffic conversation, and underrated.
AI Overviews misrepresent businesses regularly. They generate claims about brands that no cited source actually supports. GatherUp found that 67% of consumers aren’t rigorously fact-checking AI sources before choosing a local business.
So if an AI tool says your practice doesn’t take a certain insurance, or gets your service area wrong, or invents a problem you don’t have, two-thirds of the people reading it take it at face value. You lose the customer and never know it happened.
This is a reputation management problem wearing a search hat, and it’s the single most common reason clients come to us for an AI Visibility Analysis.
Where This Fits Into Your Digital Marketing Strategy
If you’re leaning on paid ads to generate leads right now, that’s not a mistake. Ads get you visibility today. They’re the short-term catalyst. SEO, and by extension GEO, is the long-term engine that keeps producing after the ad budget stops.
What’s changed is what showing up means. AI Overviews currently appear on roughly half of tracked queries, and on those queries, being cited is worth roughly twice as many clicks as not being cited. It’s the only version of position one that still exists there. On the other half, where no summary appears, nothing has changed.
That’s the gap BCC Interactive was built to close. We don’t treat SEO and AI visibility as two products bolted together. Same technical foundation, same content strategy, applied with an understanding of how AI systems pull and cite information. And we’ll tell you when your query mix means AI Overviews aren’t your problem, because plenty of the time they aren’t.
Want a clear read on where you actually stand in Google search and in AI results?
FAQs
AI Overviews are AI-generated summaries at the top of the search results page that give users quick answers pulled from multiple websites. They appeared on about 48% of BrightEdge’s tracked queries as of February 2026. How often you’ll see them depends heavily on query intent and length.
Long, conversational, question-based searches. Pew Research found that question-format queries produced a summary 60% of the time and searches of ten or more words 53% of the time, compared with just 8% for one- or two-word searches.
No. About 52% of tracked queries return no AI Overview. When one does appear, it pushes organic listings below the fold, but the traditional SERP still exists underneath, and rankings still matter.
Not automatically. Only about 17% of sources cited in AI Overviews also rank in the organic top 10. Ranking well helps, but they’re separate outcomes, and roughly five out of six citations come from content that isn’t on page one.
Rarely. Pew Research found that only 1% of visits to pages with an AI summary resulted in a click on a cited source, and users can click the traditional results below instead, though they do so far less often when a summary is present.
Not for Google. Google’s documentation states it doesn’t use machine-readable files, AI text files, markup, or Markdown for Search, including its generative AI features. Ahrefs found 97% of published llms.txt files got zero requests in a full month. The one real use case is if your customers use AI coding agents.
Compare click-through rate on your informational queries against your commercial and branded queries in Search Console over the last twelve months. If informational CTR dropped sharply while impressions held or grew, that’s your answer.
