How to Rank in Google AI Overviews

Google AI Overviews are short, AI-generated answers that sit above the regular results on many Google searches. Each one explains the topic in a few sentences or a short list, and shows links to the web pages it drew from.

Google describes them simply. AI Overviews “summarize complex topics” and link to sites for further reading, and they only appear when Google’s systems judge they add something beyond classic results (Google Search Central).

The feature grew out of Google’s Search Generative Experience test and became the default for all US users on 14 May 2024 (9to5Google). A year later, at Google I/O in May 2025, Google said AI Overviews were available in more than 200 countries and territories and more than 40 languages (Business Today). By July 2025, Sundar Pichai reported 2 billion monthly users (TechCrunch). At I/O in May 2026 that figure had grown to more than 2.5 billion (Neowin).

To make the rest of this guide concrete, we’ll follow one hypothetical business all the way through: a physiotherapy clinic in Bangkok. It wants to be cited when people search questions like “how long does office syndrome take to heal” or “is physiotherapy or massage better for back pain”.

If you want the bigger picture across ChatGPT, Gemini and Perplexity as well, start with our step-by-step AI search optimization guide.

AI Overviews appear on top of a normal results page and give a short summary. AI Mode is a separate, chat-style tab inside Google Search, built for questions that need more exploring, reasoning or comparing, and it returns a longer answer with supporting links.

Google is clear that the two can behave differently. Its documentation says AI Overviews and AI Mode may use different models and techniques, so the answers and the links they show can differ (Google Search Central). A page cited in one isn’t automatically cited in the other.

The line between them is getting thinner, though. Since January 2026, tapping “Show more” on an AI Overview on mobile opens an AI Mode conversation that keeps the context, so people can ask follow-up questions (9to5Google). And AI Mode has grown fast: Google said at I/O in May 2026 that it passed one billion monthly users (Neowin). In October 2025 it also added 35 new languages, Thai among them (Search Engine Roundtable).

So you optimize for both the same way, but track them separately. Our guide on getting your brand into Gemini answers covers the Gemini app, which is a third, separate surface.

AI Overviews appear mostly on long, informational questions, and far less on short, branded, local or shopping searches. That tells you which pages even have a chance.

The biggest study on triggers comes from Ahrefs, which analyzed 146 million desktop results pages in September 2025 (Ahrefs). AI Overviews appeared on 20.5% of them overall. But the rate swung widely by query type.

Questions. 57.9% of question queries showed an AI Overview, compared with 15.5% of non-questions. “Reason” queries (why something happens) had the highest rate of any category, at 59.8%.

Length. The rate climbed from 9.5% for one-word searches to 46.4% for searches of seven words or more. Pew Research Center saw the same pattern in real browsing data: 8% of one- or two-word searches produced a summary, against 53% of searches with ten or more words (Search Engine Roundtable).

Intent. 99.9% of AI Overview keywords in the Ahrefs data were informational. Only 0.9% of navigational keywords got one, and 4.3% of commercial ones.

Topic. Science (43.6%) and health (43.0%) topped the list. Shopping (3.2%) and real estate (5.8%) were near the bottom. Local queries sat at just 7.9%.

Why the studies give different overall numbers

Headline rates quoted online differ because studies measure different keyword sets. Semrush tracked more than 10 million keywords and found AI Overviews on 6.49% of them in January 2025, a peak of 24.61% in July, then 15.69% by November 2025 (Semrush). BrightEdge, which tracks its enterprise clients’ keyword sets, reported AI Overview presence growing from about 30% to 48% of tracked queries between February 2025 and February 2026 (BrightEdge).

So don’t anchor on an industry average. Check your own keywords.

Semrush also spotted a shift worth knowing. Informational queries made up 91.3% of AI Overview triggers in January 2025 but 57.1% by October, as commercial, transactional and navigational queries picked up more AI Overviews. So the feature is spreading beyond pure “how” and “why” questions.

For our Bangkok physio clinic, this means the money pages (“physiotherapy Bangkok price”) will rarely trigger an AI Overview, while the educational questions around them (“why does office syndrome cause headaches”) very often will. That’s where its citations will be won.

Google says there are no additional requirements to appear in AI Overviews or AI Mode. A page has to be indexed and eligible to be shown in Search with a snippet, and it has to meet Google’s normal technical requirements and policies.

In Google’s own words: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” And: “You don’t need to create new machine readable files, AI text files, or markup to appear in these features” (Google Search Central).

That rules out a lot of popular advice: no special schema type, no llms.txt file, no “AI Overview tag”.

What Google does recommend is a list of familiar SEO fundamentals, updated for AI features. It published the list in May 2025, when AI Mode launched in the US (Search Engine Land), and folded it into its documentation.

1. Focus on unique, valuable content for people.
2. Provide a great page experience.
3. Make sure Google can access your content (robots.txt, CDN and hosting settings).
4. Manage visibility with preview controls such as nosnippet and max-snippet.
5. Make sure structured data matches the visible content.
6. Go beyond text: use good images and video, and keep Merchant Center and Business Profile up to date.
7. Understand the full value of your visits, not just the click count.
8. Evolve with your users.

The documentation also adds two technical habits: use internal links so your content is easy to find, and keep important content in text form rather than locked in images.

So the honest summary is this. Google is telling you that the work is SEO done well. That’s true as far as it goes. What Google doesn’t spell out is how much the writing style, the depth of topic coverage and your reputation elsewhere affect which eligible page gets picked. That’s what the rest of this guide focuses on.

Start by confirming the page is indexed, crawlable and allowed to show a snippet. If any of these fails, nothing else in this guide matters, because the page is simply not eligible.

When our team audits websites, access problems are among the most common things we find, and the owner rarely knows. Here are the checks we run, in order.

Is the page indexed? Use the URL Inspection tool in Search Console. If the page isn’t indexed, find out why: a noindex tag, a canonical pointing elsewhere, thin or duplicate content, or a page Google hasn’t discovered yet.

Can Googlebot reach it? Google says AI features are built into Search, so Googlebot’s robots.txt rules are your control over crawling. Google also asks you to allow crawling “in any CDN or hosting layer” (Google Search Central). Bot-protection settings on a CDN or firewall can quietly block or slow crawlers. You can run any URL through our free AI crawler checker to see which crawlers your setup lets through.

Is a snippet allowed? The preview controls that limit regular snippets also limit AI features. A nosnippet robots meta tag removes the page from snippets and AI Overviews. A max-snippet value that’s very low limits how much text Google can show or use. A data-nosnippet attribute hides a specific part of the page. Useful when you mean them, costly when a plugin set them years ago.

Is the AI opt-out switched on? Since 2026, Search Console has a property-level control that removes a site from AI Overviews, AI Mode and AI Overviews in Discover, both as a link and as grounding. Google first tested it with some UK sites in June 2026 (9to5Google) and said it had rolled it out to all websites worldwide as of 31 August 2026 (Search Engine Roundtable). If someone flipped it, no content work will get you cited.

Is the key content in text? Price lists, opening hours and treatment details saved as images or PDFs are hard for any system to quote. Put the facts in HTML text.

Pro tip

After you remove a nosnippet tag or change robots rules, Google has to recrawl the page before anything changes. Google notes this can take anywhere from days to months. Use URL Inspection to request indexing for your most important pages to speed it up.

Plan each page around the main question plus the sub-questions Google is likely to search behind the scenes. Each sub-question you answer well is one more chance to be pulled into the answer.

Here’s how to guess the sub-questions. Search the main question and read the AI Overview carefully. Its structure usually shows you the parts Google thinks matter: definitions, causes, options, steps, timelines, costs, risks. People Also Ask and AI Mode’s follow-up suggestions are often close to the fan-out queries too.

For “how long does office syndrome take to heal”, the likely sub-questions are what office syndrome is, what affects recovery time, what a typical treatment plan looks like, what you can do at home, and when the pain needs a doctor. Our clinic’s page should answer each of these under its own question-style heading.

Go beyond the single page, too. Google’s documentation recommends internal links so content is easy to find. A small cluster of linked pages (office syndrome, neck pain, lower back pain, posture at work) tells Google you cover the topic in depth, and gives the fan-out more of your pages to find.

Write each section so its first one or two sentences answer the heading directly, with a specific fact, in a way that still makes sense if it’s lifted out of the page. That’s the kind of passage an AI system can use.

Compare two ways our clinic could answer one question.

The first version contains nothing to repeat. The second gives a number, a condition, an exception and a safety note, all in two sentences. (These figures are illustrative for our hypothetical clinic. On a real page, use your own real treatment data.)

Use the question as the heading. “Recovery” tells Google little. “How long does office syndrome take to heal?” matches both the search and the likely fan-out query.

Lead with the answer, then explain. Put the direct answer first, then the reasons, the detail and the exceptions. Readers skim the same way.

Swap adjectives for facts. Numbers, time frames, prices, locations, who something is for and who it isn’t for. These are the details AI answers need and most pages leave out.

Make each section stand on its own. Avoid “as mentioned above” or “this” without a clear subject. A passage that depends on the previous section is harder to use.

Show first-hand experience. Google’s first recommendation for AI features is unique, people-first content. “What we see in our clinic” is something a generic health site can’t write.

Our free AEO Content Checker scores a page for answer-first structure in your browser. For the full method with more examples, read our guide to writing answer-first content.

Schema markup won’t get you into AI Overviews by itself. Google says you don’t need special schema types for AI features. But structured data that matches your visible content still helps Google understand your pages, and Google lists it among its recommendations.

Google’s exact advice is to make sure your structured data matches the visible content of the page (Google Search Central). That’s a warning as much as a tip. Markup that claims things the page doesn’t show (reviews nobody can see, an FAQ that doesn’t appear on the page, prices that differ from the text) is a liability.

So we use schema to confirm facts the page already states, with a small set of types.

Organization or LocalBusiness on the home or contact page, with your name, address, phone, opening hours and links to your official profiles.

Article on guides and blog posts, with a real author and the date it was last meaningfully updated.

Product or Service details where you sell something, with prices that match the page.

FAQPage where a visible FAQ really exists on the page, word for word.

You can build clean JSON-LD for these types with our free schema generator. Our article on which schema types actually help AI search covers the details and the common mistakes.

Make it obvious who wrote the page, why they’re qualified, where your facts come from, and that your business is real and consistent everywhere it appears. AI Overviews lean on sources that are safe to repeat.

Google’s first recommendation for AI features is unique, valuable, people-first content, so the E-E-A-T signals (experience, expertise, authoritativeness and trust) carry straight over.

This matters even more in health, finance and legal topics, which SEOs call “your money or your life” (YMYL) topics. These are also the topics where AI Overviews appear most: Ahrefs measured AI Overviews on 34.3% of YMYL queries and 44.1% of medical YMYL queries (Ahrefs). Our physio clinic is squarely in this zone, so trust signals aren’t optional.

A named author with real credentials. For the clinic, a licensed physiotherapist with their license type, years in practice and a short bio page. Not “Admin”.

Reviewed-by details and honest dates. If a marketer drafted a health page, say which clinician reviewed it and when. Show the last-updated date, and only change it when you change the content.

Consistent business facts. Your name, address, phone, prices and services should match across your site, your Google Business Profile, Facebook, LINE Official Account and any directory listings. Conflicting facts make a source riskier to cite.

Proof of real experience. Original photos, your own data, real cases described without identifying patients, and the limits of what you offer.

It’s what a careful reader would want before trusting a page about their health. Our guide on what E-E-A-T means and how to improve it walks through each signal.

Track AI Overview visibility with four things together: Search Console’s generative AI report for impressions, the Performance report for clicks, a fixed question list you check regularly, and analytics for what visitors do next.

Search Console’s Search generative AI report

On 3 June 2026, Google announced a new Search Console report for generative AI features in Search (AI Overviews and AI Mode) and in Discover (Google Search Central Blog). It shows how often your URLs appeared in these features, which pages appeared, and breakdowns by country, device and date. Google’s note on the post says it reached all websites worldwide as of 31 August 2026.

It has two big gaps. It shows no clicks and no queries (Search Engine Roundtable). You’ll know a page appeared, but not for which search or whether anyone clicked.

The standard Performance report

Clicks and impressions from AI features already count toward your normal totals, under the “Web” search type (Google Search Central). For AI Mode, a click on an external link counts as a click, and each follow-up question is treated as a new query (Search Engine Roundtable).

A fixed question list

Because Search Console hides queries for AI features, keep your own list of target questions from Step 2 and check them on a schedule, by hand or with a tool. Record whether an AI Overview appeared, whether you were cited or named, and who else was cited.

For AI Mode, Gemini and ChatGPT, our own tool GrixTrack runs each prompt three times and records mentions and citations, next to your Google rankings by country, city and language. It has a free plan with 10 keywords and 5 prompts. For AI Overviews specifically, pair Search Console’s report with your manual checks or a rank tracker that records AI Overview citations.

Analytics

Google says clicks from pages with AI Overviews are “higher quality” and recommends measuring conversions. So in GA4, track what organic visitors to your AI-cited pages do: bookings, sign-ups, calls and LINE adds.

The full measurement method, including templates, is in our guide on how to track AI citations.

AI Overviews usually reduce clicks to the regular results under them, sometimes sharply. That’s why being cited inside the AI Overview matters more than ranking under it.

The clearest numbers come from Ahrefs. Comparing 300,000 keywords in December 2023 and December 2025, it estimated that an AI Overview means a 58% lower average click-through rate for the page in position 1 (Ahrefs). The effect shrinks lower down: about 51% at position 2, 46% at position 3 and 19% at position 10. Ahrefs’ earlier study, in April 2025, had put the drop at 34.5%.

Pew Research Center found the same direction in real browsing data from 900 US adults in March 2025. People who saw an AI summary clicked a regular result on 8% of visits, against 15% when there was none, and clicked a link inside the summary on just 1% of visits (Search Engine Roundtable). Google disputed Pew’s method, calling its query set unrepresentative.

Not every study is as bleak. Semrush found that for the same keywords, the zero-click rate actually fell slightly, from 33.75% to 31.53%, after an AI Overview appeared, and argued that AI Overviews don’t automatically increase zero-click behavior (Semrush).

So judge informational pages by more than clicks, make sure cited pages lead somewhere (our clinic’s office syndrome guide should make booking easy for the few who click), and remember that commercial and local searches trigger AI Overviews far less often, so pages built for them still win clicks the old way.

We cover the traffic side in full, including how to find which pages lost clicks, in how AI Overviews affect traffic and how to recover. The bigger trend is in our explainer on zero-click search.

Should you ever block your site from AI Overviews?

For most businesses, no. Blocking removes your pages as sources, but the AI answer still appears, built from your competitors’ pages instead.

Since 31 August 2026, every site can switch on the Search generative AI control in Search Console (Search Engine Roundtable). It removes your content from AI Overviews, AI Mode and AI Overviews in Discover, both as a link and as grounding. Google says your pages still appear in regular results and that the control “won’t be used as a ranking signal” (9to5Google). It also doesn’t control AI training. For that, Google points to the separate Google-Extended controls.

Opting out can make sense for a publisher whose content is the product and whose readers come to it directly. It rarely makes sense for a business that wants customers. For our physio clinic, a citation for “how long does office syndrome take to heal” is free exposure to exactly the right people, and opting out would hand that slot to a competitor.

If you only want to protect certain pages or passages, such as paid content or a proprietary price calculator, use page-level controls instead. A nosnippet robots meta tag keeps a page out of snippets and AI features. A data-nosnippet attribute hides a specific section. Both also limit your regular snippets, so test on a few pages first. Google lists them as ways to limit what Search shows from your pages (Google Search Central).

Technical fixes can take effect as soon as Google recrawls the page, which can be days to weeks. Rewritten content usually takes weeks to be picked up. Building the trust and off-site presence to win competitive questions takes months.

There’s no official timeline, and anyone who promises a citation by a certain date is guessing. What we can give you is the order of work.

Week 1: access. Run the five checks from Step 1 on your most important pages. It’s the fastest win.

Weeks 1 to 2: targets. Build your list of 10 to 30 questions that show AI Overviews, note who gets cited, and pick the ones where your expertise is strongest.

Weeks 2 to 6: pages. Rewrite or create the pages for those questions, then request indexing.

Weeks 4 onward: tracking. Check your question list on a regular schedule and watch the Search generative AI report for new impressions on the pages you changed.

Months 2 to 6: reputation. Work on reviews, expert answers, YouTube, directories and press on the domains AI Overviews already cite for your topic.

For our physio clinic, that might mean fixing a noindex problem on day one, five strong guides in month one, stretch videos in month two, then steady work on reviews and expert answers.

If you’d like help, we can check your key questions in AI Overviews, AI Mode, Gemini and ChatGPT and show you exactly who gets cited instead of you. Book a free AEO audit and we’ll walk you through what we find.

1. Google Search Central: AI features and your website (updated 10 Dec 2025)
2. Search Engine Land: Google shares 8 ways to be successful with AI search experiences (May 2025)
3. Google Search Central Blog: Introducing Search generative AI performance reports in Search Console (3 Jun 2026)
4. Search Engine Roundtable: Search Console generative AI report and AI control live for all sites (1 Sep 2026)
5. 9to5Google: Google adds a Search Console opt-out for AI Mode and AI Overviews (2 Jun 2026)
6. Search Engine Roundtable: Search Console now counting AI Mode clicks, impressions and positions (Jun 2025)
7. Ahrefs: 76% of AI Overview citations pull from the top 10 (21 Jul 2025, 1.9M citations)
8. Ahrefs: Only 38% of AI Overview citations come from the top 10 (2 Mar 2026, 863K SERPs)
9. Ahrefs: What triggers AI Overviews? 86 factors and 146 million SERPs analyzed (10 Nov 2025)
10. Ahrefs: Update: AI Overviews reduce clicks by 58% (4 Feb 2026, 300K keywords)
11. Semrush: AI Overviews study, 10M+ keywords (refreshed 15 Dec 2025)
12. seoClarity: AI Overview citations and organic rankings overlap (Oct 2025, 362K keywords)
13. BrightEdge: AI Overviews at the one-year mark: presence, size and what they cite (12 Feb 2026)
14. Conductor: AI Overviews vs organic rankings, 168M citations (updated 25 Sep 2026)
15. Pew Research Center via Search Engine Roundtable: Google users click less when AI summaries appear (Jul 2025)
16. 9to5Google: Gemini 3 becomes the default model for AI Overviews (27 Jan 2026)
17. 9to5Google: Google Search rolls out AI Overviews in the US (14 May 2024)
18. TechCrunch: Google's AI Overviews have 2B monthly users, AI Mode 100M (23 Jul 2025)
19. Neowin: 13 Google products with more than 1 billion users, including AI Mode and AI Overviews (I/O, May 2026)
20. Business Today: AI Overviews expand to over 200 countries and 40+ languages (Google I/O, 21 May 2025)
21. Search Engine Roundtable: Google AI Mode expands to 35 new languages, including Thai (9 Oct 2025)

Keep reading

• AI platformsHow to Optimize for AI Search: A Step-by-Step Guide
• AI platformsHow AI Overviews Affect Traffic (and How to Recover)
• AI platformsHow to Optimize for Gemini
• AI platformsHow to Get Cited by AI: Mentions and Citations Explained