
Google AI Mode: How It Picks Sources and What to Change on Your Site
AI Overviews summarize a topic at the top of a normal results page. AI Mode is the conversational layer, and Google now lets searchers flow from one into the other. Both answer by fanning a question out into several related searches, which is why fewer citations come straight from the top 10 (about 38% in Ahrefs’ 2026 data). Search Console folds both into your normal Web numbers. The site work that helps is ordinary SEO done well: direct answers under clear headings, accurate structured data, real updates, consistent business details, and pages Google can crawl and quote.

If you run marketing for a contractor, a clinic, or a machine shop, you have probably seen two AI things on Google this year and wondered whether they need two strategies. They don’t. They do behave differently, though, and the differences change which pages get cited, what your Search Console numbers mean, and where your next few hours of site work should go. At the end there’s a five-part checklist you can hand to whoever maintains your site.
What Is the Difference Between AI Overviews and AI Mode?
Google’s own AI features documentation draws the line by the job each one does. AI Overviews help people get the gist of a complicated topic quickly, with links to dig further. AI Mode handles questions that need more exploration, reasoning, or comparison. An AI Overview shows up on a regular results page above the blue links. AI Mode is a conversation: you ask, it answers with links, you ask again, and it keeps the context.
The border between them has gotten thinner. In January 2026 Google made Gemini 3 the default model for AI Overviews and let mobile users ask a follow-up question right from an AI Overview and continue in AI Mode. At I/O in May 2026, Google said that same flow was live on desktop and mobile worldwide. In the July 2026 earnings remarks, Sundar Pichai described it as having brought together AI Overviews and AI Mode into one seamless Search experience.
So think of them as two depths of the same system. The Overview is the quick answer on the results page. AI Mode is where a curious buyer goes when the quick answer raises three more questions. That second group matters more than it did a year ago: Google reported in May 2026 that AI Mode had passed one billion monthly users, with queries more than doubling every quarter since launch.
| AI Overviews | AI Mode | |
|---|---|---|
| Where it appears | Top of a normal results page, when Google judges it helpful | A conversational view, reached directly or by asking a follow-up from an Overview |
| Built for | Getting the gist fast | Comparisons, multi-step questions, follow-ups |
| Answer length | Short summary | Roughly 4x longer, per Ahrefs |
| Brands named per answer | 1.3 on average | 3.3 on average |
| Search Console position | The whole Overview holds one position; every link in it gets that position | Same position method as a regular results page; each follow-up counts as a new query |
How Does Each One Pick and Cite Sources?
Both start from Google’s regular index and ranking systems. Google’s guide to optimizing for generative AI features names two techniques. The first is retrieval-augmented generation, or grounding: Google retrieves relevant pages from its index, reviews the specific information on them, writes a response, and links the pages that support it. The second is query fan-out: the model fires off a set of related searches at the same time to gather more material.
Google’s own example is a search for how to fix a lawn full of weeds. Fan-out queries might include the best lawn herbicides, removing weeds without chemicals, and preventing weeds in the first place. Translate that to a business question like “should we replace or repair a 20-year-old rooftop HVAC unit” and you can guess the sub-searches: typical replacement cost, efficiency gains, refrigerant rules, repair-versus-replace math, financing. Each one is a separate chance for your page to be pulled in.
AI Mode fans out harder. When Google introduced it in 2025, it described AI Mode as breaking a question into subtopics and searching them simultaneously, and its Deep Search option can issue hundreds of searches for a single research request. More searches means more candidate pages, and longer answers mean more room to cite them.

That explains a result that surprises a lot of owners. Ahrefs compared the two features on the same queries using September 2025 US data and found that only 13.7% of citations overlapped, even though the answers scored 86% on semantic similarity. Same conclusions, different sources credited. The same study found that when a brand appears in an AI Overview, there is a 61% chance it also appears in the AI Mode answer, usually alongside more competitors.
The fan-out effect also shows up in rankings. In a March 2026 study of 863,000 keyword results, Ahrefs found that about 38% of URLs cited in AI Overviews also appeared among the first 10 results on the page for the original query. In its July 2025 study the figure was roughly 76%. Ahrefs attributes the drop to heavier use of fan-out after the Gemini 3 switch.
For a small business, that cuts both ways. You no longer need to outrank a national directory for a head term to get cited. You do need a clear, quotable answer to the narrower question that one of those sub-searches will ask.
What Does Search Console Show for AI Mode and AI Overviews?
As of August 2026, your standard Performance report already includes both. Google’s AI features page says traffic from these features is counted in the Performance report under the Web search type. There is no filter in that report that separates AI Overview or AI Mode rows from classic results.
The counting rules in the Search Console help center are worth knowing before you read your own data:
- AI Overviews: a click on a link to your page counts as a click. The Overview occupies one position, and every link inside it gets that same position.
- AI Mode: clicks and impressions follow the standard rules, and position uses the same method as a normal results page.
- Follow-ups: a follow-up question in AI Mode counts as a new query. Impressions, positions, and clicks in that response belong to the follow-up, not the first question.
That last rule matters for long-tail reporting. A buyer’s third follow-up might be a query that nobody types into a search box, yet it can land in your query list.
There is one newer piece. On June 3, 2026, Google announced Search Generative AI performance reports in Search Console, with dedicated views for Search and Discover. They show impressions from AI features by page, country, device, and date. Google is rolling them out to a subset of sites first, so many small business properties won’t see them yet. The reports show impressions only, no clicks, and the same data stays in the overall Performance report.
Expect your numbers to look odd for a while. Pew Research tracked 900 US adults’ real browsing in March 2025 and found that people clicked a traditional result on 8% of visits when an AI summary appeared, versus 15% without one. Only 1% of those visits included a click on a link inside the summary. If your impressions climb while clicks stay flat, your site is probably fine; that is what the whole market looks like. Google’s own advice is to look at the full value of visits, such as calls, form fills, and quote requests, instead of clicks alone. Set those up as conversions in GA4 if you haven’t.
The Site Checklist: What to Change
Google says there are no extra requirements or special optimizations for AI Overviews or AI Mode. The five areas below are standard SEO, pointed at how fan-out and grounding actually pick pages.
1. Write passage-level answers
Put the answer in the first sentence or two under each heading, then give the detail. A heading like “How long does a commercial roof replacement take?” followed by “Most 20,000 square foot jobs take one to two weeks, depending on tear-off and weather” is easy for a grounding system to lift and credit. A heading followed by three sentences of warm-up is not.
Cover the sub-questions a real buyer asks on the page where they belong: cost ranges, timelines, permits, what changes the price, what you won’t do. Google’s guide warns against the opposite move, spinning up a separate page for every fan-out variant, and calls it a violation of the scaled content abuse policy when done to manipulate AI answers. It also says you don’t need to “chunk” content into tiny pieces. Write normal sections with honest headings.
The bigger lever is what Google calls non-commodity content. A generic “5 tips for choosing a contractor” page adds nothing the model doesn’t already know. Your actual lead times, your job photos, the code issue you see on half your inspections, a price range from your own invoices: that is material only you can publish, and it is what Google’s systems look for when they choose between similar sources. Our piece on E-E-A-T for small businesses goes deeper on showing that experience.
2. Keep structured data accurate, not exotic
No special schema unlocks AI Mode. Google’s guide lists “overfocusing on structured data” among the myths. Keep the markup that earns rich results and describes your business: Organization or LocalBusiness with your real name, address, phone, and logo; Product and Review where they apply; Article on blog posts; FAQPage only where the questions are visible on the page. The rule that matters most, from Google’s guidance on AI search, is that markup must match visible content. Hours in your schema that disagree with your contact page are worse than no schema.
3. Update for accuracy, not for the date stamp
Freshness matters less for AI Overviews than people assume. An Ahrefs study of nearly 17 million cited URLs found that Google’s AI Overview citations were no fresher than regular organic results, averaging about 3.9 years old, while tools like ChatGPT leaned newer. Changing the date on an unchanged page buys nothing.
Update when the facts move: prices, regulations, product lines, service areas, staff. Show a visible “last updated” date where timing matters. A pricing page still quoting 2023 numbers is a page an AI answer might quote back to your prospect.
4. Strengthen entity and brand signals
AI Mode answers name about three brands on average, so being a recognizable entity helps. Google’s guide points local businesses and sellers to Google Business Profile and Merchant Center, because AI responses can pull in local business and product information directly. Keep your name, address, phone, categories, and hours identical across your site, your Business Profile, and the directories that matter in your trade. Our local SEO playbook walks through that cleanup.
Give your site a real About page with named people and credentials, and link your Organization markup to your official profiles. Earn mentions the slow way: trade association listings, supplier case studies, local press, and reviews from actual customers. Google’s guide says chasing inauthentic mentions doesn’t help, because its AI features depend on the same spam systems as the rest of Search.
5. Make sure Google can crawl and quote you
To appear in either feature, a page must be indexed and eligible to show in Google Search with a snippet. That is the whole technical bar, and small business sites fail it more often than you’d think. Check these:
- Googlebot isn’t blocked in robots.txt, and key pages return a 200 status with real text in the HTML or rendered page.
- No stray
nosnippet,max-snippet:0, ornoindexleft over from a staging site or an SEO plugin setting. These restrict how Google can show your content in AI features. - Content loaded by JavaScript renders in Search Console’s URL Inspection tool.
- Pages load fast and work on phones.
Two points of confusion come up often. Blocking Google-Extended in robots.txt limits Gemini training and grounding in Gemini apps, but Google states it does not affect inclusion in Google Search. And Google’s guide, last updated in July 2026, also mentions a Search Console setting for inclusion in Search generative AI features. If your property has it, confirm it is set to include.
| Area | What to do | How to check |
|---|---|---|
| Passage answers | Lead each section with a one- or two-sentence answer; cover real buyer sub-questions on the page | Read only the first line under each H2. Does it answer the heading? |
| Structured data | Organization/LocalBusiness, Product, Article; all values match the visible page | Rich Results Test; compare schema hours and phone to the contact page |
| Freshness | Update pricing, rules, and services when they change; show a real updated date | List pages that quote prices or regulations; note the last real edit |
| Entity signals | Consistent name, address, phone; complete Business Profile; named team on About page | Search your business name and compare every listing on page one |
| Technical access | Indexable, snippet-eligible, crawlable, rendered | URL Inspection in Search Console; scan page source for nosnippet and noindex |
What Can You Skip?
Google’s guide spends a full section on things that don’t help in Google Search: llms.txt files and other AI-specific markup (Google Search ignores them, though other tools may read them), rewriting pages in a special style for AI, stuffing in every long-tail variation, and buying mentions. If a vendor’s AI search package leans on those, ask what it does that ordinary SEO doesn’t. Our AEO guide covers the other AI assistants, where the rules differ.
Keep investing in the site itself. The people who click out of an AI answer tend to be close to buying, a point we made in Do You Still Need a Website in 2026?
Frequently Asked Questions
What is the difference between Google AI Mode and AI Overviews?
AI Overviews are short AI summaries that appear at the top of a normal Google results page. AI Mode is a conversational experience built for questions that need reasoning, comparisons, and follow-ups. Since 2026, Google lets people ask a follow-up right from an AI Overview and continue the conversation in AI Mode.
Do I need special schema or an llms.txt file to show up in AI Mode?
No. Google says there are no additional requirements and no special schema.org markup needed for AI Overviews or AI Mode, and that Google Search ignores llms.txt files. Standard structured data is still worth keeping for rich result eligibility, as long as it matches what is visible on the page.
Can I see AI Mode traffic separately in Search Console?
Not in the standard Performance report. Clicks, impressions, and positions from AI Overviews and AI Mode are counted inside the Web search type with everything else. In June 2026 Google announced a separate Generative AI performance report, rolling out to a subset of sites first, which shows impressions only.
Why does my page get cited in AI answers for searches it does not rank for?
Query fan-out. Google’s AI features run several related searches behind the scenes and pull sources from those results too. Ahrefs found only about 38% of AI Overview citations appeared in the top 10 results for the original query, down from roughly 76% in its July 2025 study.
Does blocking Google-Extended remove my site from AI Overviews?
No. Google states that Google-Extended does not affect a site’s inclusion in Google Search and is not a ranking signal. To limit how your content appears in AI Overviews and AI Mode, Google points to snippet controls such as nosnippet, data-nosnippet, max-snippet, and noindex.
Key Takeaways
- AI Overviews give the quick answer; AI Mode handles follow-ups. Google now connects the two.
- Fan-out means citations often come from sub-searches: 13.7% citation overlap between the features, about 38% of Overview citations from the original top 10.
- Search Console counts both inside Web search. The June 2026 Generative AI report shows impressions only and is reaching a subset of sites first.
- Expect fewer clicks per impression (Pew: 8% vs 15%). Judge Search by calls, forms, and quote requests.
- The fix list is careful SEO: direct answers, accurate schema, real updates, consistent business details, crawlable pages.
Not sure how your pages hold up in AI answers? Talk to our team. We can check what Google can crawl and quote on your site and point out the few fixes worth making first.
Written by The WinTech Desk, WinTechnology Inc. | https://www.wintechnology.ai