How to Rank in AI Overviews: Google's 7 Claims Against the Citation Data

Google's official answer to how to rank in AI Overviews is that there is nothing extra to do. Its documentation says there are "no additional requirements" and no "special optimizations necessary". The best independent data disagrees on one point that matters: in March 2026 only 37.9% of the pages AI Overviews cited ranked in Google's top 10 for the same query. Ranking still gets you eligible. It no longer gets you picked.

What Google's AI features documentation actually says

Google publishes one page on this subject, titled "AI features and your website," in Search Central. The version we read was last updated on 10 December 2025. It covers AI Overviews and AI Mode together and treats them as part of Search rather than as a separate product.

The central sentence is blunt. "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary". The page then rules out the tactics most vendors sell: "You don't need to create new machine readable files, AI text files, or markup to appear in these features".

It also states something most optimisation guides skip. AI Overviews "are only shown when our systems determine that it is additive to classic Search, and as such, often don't trigger". That is Google telling you that a large share of your queries will never show the feature, whatever you do.

The eligibility rule is the only hard requirement

To be a supporting link, Google says, "a page must be indexed and eligible to be shown in Google Search with a snippet". Then it repeats itself: "There are no additional technical requirements". It also notes that meeting every requirement does not guarantee crawling, indexing or serving.

The page lists seven SEO practices it calls relevant. Allow crawling in robots.txt and at your CDN. Make content findable through internal links. Provide a good page experience. Keep important content in text. Support text with images and video. Make structured data match visible text. Keep Merchant Center and Business Profile data current.

None of those is new. That is the point Google is making, and it is also why the page frustrates anyone looking for a lever. In our view the page is honest about eligibility and silent about selection, and the gap between those two words is where every study below lives.

Seven claims in Google's AI Overviews documentation, seven verdicts

We pulled every testable claim from the documentation and from Google's August 2025 traffic statement. Then we matched each one to the best published measurement we could open. Where no study tests a claim, the verdict says so rather than guessing.

Sources: Google Search Central "AI features and your website," last updated 10 December 2025. Liz Reid, Google, 6 August 2025; Ahrefs, July 2025 and March 2026; Pew Research Center, July 2025; Seer Interactive, November 2025. The 62.2% figure is our sum of Ahrefs' 31.2% and 31.0%.

Read the table as a scorecard of two different documents. Google's page is a policy statement about what it will accept. The studies are measurements of what it actually chose. Those can both be true and still point you in different directions.

Rank position no longer predicts AI Overviews citations the way it did

In July 2025, Ahrefs analysed 1.9 million citations from 1 million AI Overviews. It found 76.1% of cited pages ranked in the top 10. The median organic position of the first cited URL was 2. On that evidence, Google's advice to do normal SEO looked almost complete.

In March 2026 the same team repeated the work at more than double the scale: 863,000 keyword SERPs and 4 million AI Overview URLs. The top-10 share fell to 37.9%. Counting only standard blue links, it was 37.1%, with 36.7% of cited pages ranking nowhere in the top 100.

The broader history of that fall is covered in our breakdown of the top-10 citation overlap collapse. The question here is narrower: does the fall contradict Google's documentation? It does not. Google never claimed that top rankings drive selection. It claimed only that no extra requirements exist.

What 37.9% means for how to rank in AI Overviews

It means ranking for your head term is a weak proxy for being cited on it. If 62.2% of citations come from pages outside the top 10, then most cited pages earned their place on something other than the query you are tracking.

Ahrefs' own interpretation is that AI Overviews increasingly draw from the results of fan-out sub-queries. A page that ranks first for "how to price an AI agent" may be cited on "AI pricing models" without ranking for it. That reading is plausible. Ahrefs did not quantify it, so treat it as a hypothesis.

Why fan-out is the claim that reconciles both sides

Google's documentation describes "query fan-out" as "issuing multiple related searches across subtopics and data sources". It says this lets AI features "display a wider and more diverse set of helpful links" than a classic search. The Ahrefs distribution is what that sentence would look like in data.

So the documentation and the tests agree once you read them carefully. Normal SEO remains the route. The query you rank for is just not the only query that counts. We think this is the most useful sentence in Google's page, and it is buried in a paragraph about how the feature works rather than how to appear in it.

Freshness in AI Overviews: Google's silence matches the data

Google's AI features page says nothing about content age. Most optimisation guides fill that gap with advice to update pages often. On AI Overviews specifically, the best available measurement says the guides are wrong.

Ahrefs measured the age of 16.975 million cited URLs across six surfaces. Pages cited by ChatGPT, Perplexity, Gemini and Copilot were younger than pages in organic results. AI Overviews were the exception. Their top 3 citations averaged 1,432 days, against 1,416 days for organic results in the study's table.

The difference between AI Overviews and organic results is 16 days on an average of almost 4 years. That is noise. The study's own conclusion was that "Google is the least influenced by content freshness," and we agree with it.

Why changing the date will not help

The same Ahrefs piece repeats Google's John Mueller on changing publish dates without changing content, and it calls frequent updating a weaker investment than new pages. We would go further for AI Overviews. A refresh programme justified on AI citation grounds alone has no supporting evidence for this surface.

Refresh pages because a figure went stale, a price changed or a product shipped. Those are reasons a reader cares about. "AI Overviews like fresh content" is not one, at least not on the only large dataset that separates AI Overviews from other assistants.

Which content types AI Overviews cite more than standard results

Google's documentation does not say which kinds of pages it prefers. The most careful independent look at this comes from Pew Research Center, which matched AI summaries against the standard results on the same searches.

Pew found that 6% of sources linked in AI summaries were .gov sites, against 2% in standard results. Wikipedia, YouTube and Reddit together made up 15% of summary sources and 17% of standard result sources. News sites held 5% in both.

Two more figures shape what a cited page looks like. The median summary ran 67 words. And 88% of summaries cited 3 or more sources, while only 1% cited a single source. You are rarely competing for one slot. You are competing to be one of several.

The practical reading is that AI Overviews tilt toward primary, institutional sources. If your page restates a government figure, the summary may cite the government page instead. If your page publishes an original figure with its method, you become the primary source. That is a content strategy argument, not a markup one, and it is the case made in our analysis of original research as a content moat.

Vendor trackers publish sharper numbers on YouTube and Reddit shares inside AI Overviews. They conflict with each other by wide margins and none publishes a reproducible method, so we have not used them. Pew's denominator is narrower, but it is the only one we could verify end to end.

The click quality claim in Google's documentation, measured

This is where the documentation and the tests part ways. Google's page says that "when people click from search results pages with AI Overviews, these clicks are higher quality (meaning, users are more likely to spend more time on the site)". It publishes no figure, sample or period for that statement.

Liz Reid, head of Google Search, made the wider version in August 2025. She wrote that total organic click volume from Google "has been relatively stable year-over-year" and that Google sends "slightly more quality clicks". She defined quality clicks as ones where users "don't quickly click back". That post also gave no numbers.

What Pew and Seer measured on click volume

Pew tracked real browsing by 900 US adults across March 2025. On searches with an AI summary, users clicked a traditional result in 8% of visits. Without a summary, they clicked in 15%. Users clicked a link inside the summary itself in just 1% of visits, and ended their browsing session on 26% of summary pages against 16% of standard ones.

Seer Interactive looked at 3,119 informational search terms across 42 client organisations and 25.1 million organic impressions. In Q3 2025, queries with no AI Overview had an organic CTR of 1.45%. Where an AI Overview appeared and cited the brand, CTR was 0.70%. Where it appeared and did not cite the brand, CTR was 0.52%.

Our own calculation from Seer's figures: a cited brand keeps 48% of the organic CTR of a comparable query with no AI Overview, and an uncited brand keeps 36%. Pew's figures give a similar ratio: 8% over 15% is 53%. Two methods, two samples, and the answer lands in the same range.

Why Search Console cannot settle the argument

Google's documentation tells you AI feature traffic appears in the Search Console Performance report under the "Web" search type. That is accurate. It is also blended with every other web result, so you cannot isolate AI Overview clicks with Google's own tool.

The documentation suggests tracking conversions and time on site in Google Analytics instead. That helps test the quality claim on your own site, but only if you know which sessions came from a results page with an AI Overview, which Analytics does not tell you. Seer had to combine Search Console with two outside AI Overview trackers to do it. The wider measurement problem is the subject of our piece on the metrics that replace clicks in zero-click search.

Our position is simple. Google may be right that each click is worth more. It has not shown it, and the one claim it could prove with its own data is the one it leaves unquantified. Until it publishes a number, plan on fewer clicks per impression and treat "higher quality" as a hope to test.

Where this argument is weakest

A claim-by-claim comparison flatters the side with numbers. The studies here have real limits, and some of them cut in Google's favour.

The studies measure correlation, and most are vendor research

Ahrefs and Seer sell SEO tools and services, which makes both Tier 2 at best for this argument. Ahrefs itself says its 2025 and 2026 rank studies are not cleanly comparable, because the parser changed and the underlying model moved to Gemini 3. Some of the drop from 76.1% to 37.9% may be measurement rather than behaviour.

Seer is explicit that it "cannot definitively prove that citation causes higher CTRs." Brands with stronger authority may simply be cited more and clicked more. Its sample covers informational queries only, at 42 organisations that chose to hire an SEO agency.

Ahrefs' freshness study also contains small internal inconsistencies. Its text gives the organic average as 1,432 days and its table gives 1,416. We used the table, which makes the AI Overviews gap the 16 days the authors describe.

Google may simply be right about click quality

Fewer clicks and better clicks are not contradictory. If an AI Overview answers the simple question, the people who still click may be the ones with real intent. Pew counted clicks, not what happened after them, and Seer measured CTR, not conversions.

Pew's data is also from March 2025, before AI Mode expanded and before the Gemini 3 change. A click study from 2026 could look different. We would rather publish the gap with its date attached than pretend the question is closed.

Frequently asked questions

Where to start this week

Pull your top 50 informational queries from Search Console. For each one, check by hand whether an AI Overview appears and whether your page is cited. That gives you the three groups Seer used, on your own data, for the cost of an afternoon.

Then pick the 5 queries where you rank top 10 and are not cited. Read the pages that are. If they answer a sub-question yours skips, write that section. If they are primary sources you only summarise, publish the figure you can stand behind yourself. Leave the schema and the AI text files alone; Google says they do nothing here, and the schema tests we reviewed earlier found no reason to doubt it.

Related reading