SEO & AI Search
How to Rank in AI Overviews: What Google Actually Says
The honest answer to how to rank in AI Overviews is that you cannot optimize for them directly, and Google says so in its own documentation. A page becomes eligible by being indexed and allowed to show a snippet. There are no extra technical requirements, no special markup, and no AI file to publish.
What you can influence is whether Google picks your page over someone else’s once you are eligible. That comes down to two things most teams are not working on: covering the subquestions Google generates behind the scenes, and writing passages that still make sense when they are lifted out of the page.
Check your eligibility before you change anything
Google’s guidance is specific on this point. According to its documentation on AI features, a page that wants to appear as a supporting link in AI Overviews or AI Mode must be indexed and eligible to appear in Search with a snippet. That is the whole requirement.
That makes a snippet control you forgot about the fastest way to disappear from AI Overviews. The nosnippet and max-snippet directives, along with data-nosnippet on individual sections, remove or limit what Google can show. Many publishers added these during the early backlash against AI answers and never revisited the decision.
Check your robots meta tags before you spend anything on content. If a page carries a snippet restriction, no amount of optimization will get it cited. It is also worth confirming nobody on your team has switched on the AI features opt-out in Search Console, which we covered in Google Now Lets You Block AI Overviews.
Rankings no longer decide who gets cited
The old assumption was that a top 10 position handed you the citation. That link has weakened considerably. An Ahrefs analysis of 863,000 keywords and roughly 4 million AI Overview URLs, reported by Search Engine Journal, found that about 38% of cited pages also ranked in Google’s organic top 10 for the same query. The previous version of the same study put that figure at 76%. The rest of the citations split almost evenly between pages ranking 11 to 100 and pages ranking beyond 100.
Treat the direction as more reliable than the exact number, since measurement methods for AI citations are still maturing. But the direction is clear. Roughly three in five citations now go to pages a searcher would never see on page one for that keyword. Ranking helps. It no longer decides. We made the broader case in Ranking #1 Doesn’t Mean What It Used To.
Optimize for the fan-out, not the keyword
The reason for that gap is a technique Google documents openly. AI Overviews and AI Mode use what Google calls query fan-out, issuing multiple related searches across subtopics before assembling an answer. Google’s own example: a search for how to fix a lawn full of weeds might fan out into searches for the best lawn herbicides, removing weeds without chemicals, and preventing weeds in the first place.
So the unit of competition is no longer the keyword. It is the cluster of questions sitting underneath it. In practice:
- List the follow-up questions a real buyer asks after the head term, then confirm each one is answered somewhere on your site.
- Give each distinct subquestion its own H2 or its own page rather than burying three answers in one paragraph.
- Cover the awkward angles competitors skip, including cost ranges, failure modes, and when your solution is the wrong choice.
- Answer comparison and alternative queries directly, since those are prime fan-out territory.
- Keep every answer self-contained, because a fanned-out subquery retrieves a passage, not your whole page.
- Publish the boring operational detail, such as timelines and prerequisites, which is exactly what thin competitor content leaves out.
A site that covers twelve related subquestions well will get pulled into overviews it never ranked for. That is the whole mechanism.
Tactics that do not move AI Overviews
Plenty of agency proposals right now sell work Google has explicitly said is unnecessary. Judge any proposal against the documentation.
| Tactic being sold | Does it affect AI Overviews? | What to do instead |
|---|---|---|
| Publishing an llms.txt file | No, Google does not use it | Check robots.txt and snippet controls |
| Adding special AI schema | No such schema exists | Use standard Article, FAQPage, and Organization markup |
| Stuffing in question phrasing | No, unless the answer is genuinely useful | Answer the question in the first 100 words |
| Chasing a single head keyword | Partially, since rankings still help | Build subtopic coverage for fan-out |
| Buying AI citation tracking before fixing content | No, measurement alone changes nothing | Fix eligibility and coverage first, then measure |
The work that actually pays off is less exciting than anything in the left column: make sure your important content exists as plain text Google can read, with internal links pointing at it. If you use structured data, standard schema markup is all you need.
Measuring it
Search Console now has a dedicated generative AI performance report that shows your impressions inside AI Overviews and AI Mode by page, country, device, and date. It launched in June 2026 and reached every site worldwide on August 31. Data starts from mid-May 2026, with no earlier history.
The catch is that the report shows impressions, not clicks. You can see that you appeared, but not what it earned you. To fill that gap, compare impressions against clicks on your informational pages in the standard performance report. When impressions hold steady while clicks fall, you are probably being shown inside an overview and not clicked through. That is a visibility change, not a ranking loss, and treating it as a demotion usually makes things worse. Our guide to finding AI traffic in GA4 covers the referral side.
See the fan-out behind your best keyword
Pick the one keyword your business should own. Search it, read the AI Overview it returns, and list every source cited. If you are not among them, look at what those pages answer that yours does not. That comparison usually explains the gap in ten minutes.
If you want the full map instead, OgreLogic will take one keyword and send back the subquestions Google is fanning out to, who currently gets cited for each, and the gaps your site could fill first. Send us the keyword or call +1 (512) 861-8471. It is the first step in our AI Search Optimization work, and it is useful on its own whether or not you continue.
Before you touch content, make sure crawlers can read your pages at all. LLM SEO covers that, and GEO vs SEO explains how the two disciplines divide up.
FAQs
Can I pay to appear in AI Overviews?
No. Citations are not sold, and no tool or agency can place you in one. Anyone promising guaranteed AI Overview placement is describing something that does not exist. What you can pay for is the work that makes citation more likely: eligibility fixes, subtopic coverage, and clearer page structure.
Does blocking AI crawlers remove me from AI Overviews?
Not the way most people assume. AI Overviews run on Googlebot’s crawl of your site, so Googlebot access, your snippet settings, and the AI features opt-out in Search Console are the controls that matter. Google-Extended governs how Google’s other AI systems, such as Gemini, use your content. It does not control whether you appear in AI features in Search.
How long does it take to get cited after publishing?
Usually faster than a ranking change, because eligibility depends on indexing rather than accumulated authority. A page that is indexed and answers a fanned-out subquery can be pulled in within days. Holding that position is the unpredictable part, since the set of sources shifts from query to query.
