Most people search for “Google AI Overview ranking factors” expecting a tidy checklist, the way we once talked about title tags and backlinks. There is no such list. Google does not publish a separate algorithm for AI Overviews, and it has said plainly that there are no special files, tags, or tricks that get you in. What actually exists is a set of signals that make your page easy to find, easy to trust, and easy to quote, plus a hard prerequisite most guides skip: you have to be eligible in normal Search first.
This is the version I wish I had when I started tracking which of our pages got pulled into AI Overviews and which got ignored. It is built on what Google documents, what large citation studies measure, and what we see in the data every week.
The short answer
To appear in a Google AI Overview, a page needs three things working together. First, technical eligibility: the page must be indexed and eligible to show with a snippet, because that is the literal gate Google applies. Second, topical strength: it should rank well for the query and the related sub-questions Google generates, since that is where most citations still come from. Third, extractability: the answer to a specific question should sit in a clean, self-contained passage a model can lift and attribute. Authority, freshness, and useful structure raise your odds, but none of them replace those three.
Notice what is missing from that list: there is no “AI Overview schema”, no llms.txt requirement, and no secret formatting. Google is explicit that none of those are needed.
”Ranking factors” is the wrong mental model
A classic ranking factor sorts a list of blue links. An AI Overview does something different. Google takes your query, breaks it into several related sub-queries (a process usually called query fan-out), gathers candidate passages across all of them, writes a synthesized answer, and attaches a handful of supporting links. So a page is not “ranked” into the box. Its passages are selected to support claims in a generated paragraph.
That reframing changes what you optimize. You are not trying to be the single best page for one head term. You are trying to be a strong, citable source across a cluster of related questions, with passages clean enough to quote. I cover the broader strategy in how to rank in AI Overviews, and the underlying discipline in what GEO actually is. This post is the factor-by-factor view.
The factors that actually move AI Overview citations
1. Be technically eligible (the prerequisite nobody mentions)
Google’s documentation is blunt: “to be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet.” It also states there are “no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary,” and that you do not need machine-readable files or new markup, per Google’s AI features documentation. If you block snippets with nosnippet or a tight max-snippet, you can quietly exclude yourself. So step one is not clever, it is hygiene: confirm the page is indexed and snippet-eligible.
2. Rank for the query and its fan-out
Ranking still matters, but less absolutely than it did. In a study of 863,000 keyword SERPs and 4 million AI Overview URLs, Ahrefs found that 38% of AI Overview citations come from pages ranking in the top 10, down sharply from about 76% in their July 2025 analysis. The rest now split almost evenly between positions 11 to 100 and beyond position 100.
Two practical reads. A top-10 position is still the single strongest lever, so traditional SEO is not dead. But because of fan-out, a page that ranks modestly for the head term can still be cited if it answers an adjacent sub-question better than anyone else. Breadth across a topic beats a single ranking.
3. Answer one specific question in a liftable passage
AI Overviews quote passages, not pages. The pages that win give a direct, two to three sentence answer near the top of the relevant section, in plain language, with the question echoed in the heading. Bury the answer under three paragraphs of preamble and a model has nothing clean to extract. Lead with the answer, then expand.
4. Show real first-hand expertise (E-E-A-T)
Google’s own optimization guidance tells you to create content with unique expertise and to not “just recycle what others on the internet have already said, or could easily be produced by a generative AI model,” from Google’s AI optimization guide. Original data, first-hand testing, named authors with real credentials, and concrete examples are the signals that separate a citable source from interchangeable filler. This is also the cheapest moat, because most competitors are publishing the recyclable version. In our own tracking, the pages that earned citations almost always carried something a model could not synthesize from the rest of the web: a number we measured ourselves, a screenshot from a real test, or a clearly credentialed author byline. The generic explainer, however well written, rarely got picked.
5. Keep it fresh, multimodal, and well-built
Freshness helps in fast-moving topics, and Google recommends supporting text with relevant images and video and delivering a good page experience: fast, mobile-friendly, clearly structured. Structured data does not “rank” you in an AI Overview, but it helps Google understand the page and qualify it for rich results, so use it where it genuinely fits (products, FAQs, how-tos) and skip it where it does not.
What the data says, at a glance
Here is how the main signals net out, with the numbers I would actually quote in a meeting:
| Factor | What the evidence shows | What to do |
|---|---|---|
| Top 10 ranking | 38% of citations come from top-10 pages, down from 76% in mid-2025 | Win the head term, but do not stop there |
| Query fan-out | About 62% of citations now sit outside the top 10 for the original query | Cover the whole question cluster, not one keyword |
| Snippet eligibility | A hard requirement; no snippet means no citation | Audit nosnippet and max-snippet settings |
| First-hand expertise | Google tells you to avoid easily-AI-generated content | Add original data, testing, named authors |
| Clean answer passages | Overviews quote passages, not whole pages | Put a direct answer under each question heading |
The headline is the trend in that first row: citations are spreading down and across the SERP, which is good news if you have depth and bad news if your whole strategy was one ranked page.
Why this matters even when nobody clicks
It is fair to ask whether an AI Overview citation is worth chasing. The honest answer: the click economics have shifted, so visibility itself is now part of the goal. Pew Research, tracking 68,879 real Google searches from 900 US adults, found that users click a result only 8% of the time when an AI summary is present, versus 15% when it is not, and that they very rarely click the sources inside the summary. Pew also found that longer, question-style queries are far more likely to trigger a summary in the first place.
So fewer people click, but more people see the answer, and your brand is either named in it or it is not. That is exactly why being the cited source, even without the click, has become its own form of distribution. If your organic traffic recently fell off a cliff, this is often the cause, and I broke down that pattern in why blog traffic disappeared after AI Overviews.
What Google says you can ignore
Plenty of “AI SEO” advice is noise. Google explicitly calls out tactics that do not help: creating llms.txt files, artificially “chunking” content, and rewriting pages specifically for AI systems. You do not need them. Spend that time on the eligibility check, the topic coverage, and the answer passages instead. Chasing rumored tricks is how teams burn a quarter and move nothing.
How to actually win this, and measure it
Factors only matter if you can see whether they are working. The workflow that holds up: map the real questions people ask an assistant about your topic, group them into a cluster, answer each one in a clean passage on pages that are already indexed and snippet-eligible, then watch which queries start surfacing you in AI answers and double down there.
That last step is the one most teams skip because the tooling is young. This is the gap we built SQSEO to close: you start from a seed keyword, fan it out into the longtail and question-level queries that actually trigger AI Overviews and assistant answers, and see where you already have a citable angle versus where a competitor owns the passage. The core research is free, which matters when you are testing a topic cluster before committing to a month of writing. The point is not the tool, it is the loop: pick the questions, ship clean answers, measure citations, repeat.
A practical checklist to earn AI Overview citations
Run this on any page you want cited:
- Confirm it is indexed and snippet-eligible. Check Search Console coverage and remove any nosnippet or restrictive max-snippet tag on the page. This is the gate, so clear it before anything else.
- Map the question cluster. List the head term plus the ten to twenty sub-questions a real person would ask an assistant about it. Those are your fan-out targets, and they are where most citations now live.
- Give each sub-question its own clean answer. One heading phrased as the question, then a direct two to three sentence answer, then the supporting detail. A model should be able to lift the answer without rewriting it.
- Add one thing only you can offer: original data, a tested workflow, a named expert opinion, a real example. This is the expertise signal Google rewards and most competitors skip.
- Support the text with a relevant image or short video where it aids comprehension, and make sure the page loads fast on mobile.
- Track citations weekly. Note which queries start naming you in an AI Overview or assistant answer, then expand the ones that work and rewrite the ones that do not.
The teams that win AI Overviews are not the ones with the most pages. They are the ones running this loop on a tight cluster until they own the answer to every question in it.
None of this is mystical. AI Overview “ranking factors” are mostly the SEO fundamentals you already know, re-pointed at passages instead of pages, with one extra prerequisite (snippet eligibility) and one strategic shift (cover the question cluster, not the single keyword). Get those right and you stop guessing why the box ignores you.
Frequently asked questions
What are the ranking factors for Google AI Overviews?
There is no separate ranked list. In practice a page needs to be technically eligible (indexed and eligible for a snippet), strong on the query and its related sub-questions, and written so a clear answer can be quoted and attributed, with first-hand expertise, freshness, and sensible structure raising the odds. If you want to find the exact questions that trigger AI Overviews in your niche and track whether you are cited, SQSEO maps those queries from one seed keyword for free, which is the fastest way to turn this list into an actual content plan.
Do I need special schema or an llms.txt file to appear in AI Overviews?
No. Google states there are no additional requirements and no special files or markup needed, and it specifically lists llms.txt and AI-only rewriting as things that do not help. Use normal structured data where it fits, and focus on eligibility and content quality.
Does ranking number one guarantee an AI Overview citation?
No. Ranking in the top 10 is the strongest single signal, but only 38% of citations come from top-10 pages now, down from about 76% in mid-2025. Because Google fans a query out into sub-questions, pages outside the top 10 are frequently cited when they answer an adjacent question better.
Is it still worth being cited if people do not click?
Yes. Click-through rates roughly halve when an AI summary appears, so being named inside the answer is increasingly the visibility itself. Brand presence in the Overview shapes the decision even when no click follows.
