Studio Slate

How to Rank in Google AI Overviews (2026)

10 July 2026By Chris Raad

Google AI Overviews answer queries above the blue links. Getting cited is a distinct discipline from ranking. Here is how Google builds them and what works.

Key Takeaway

  • To get cited in an AI Overview, rank in the organic top 20 and put a self-contained answer in the first 50 words of each section, under a question-shaped heading. Google grounds Overviews in its own index (97% cite at least one source from the organic top 20) and lifts a roughly 200-word passage, so ranking is the entry ticket and the extractable passage wins the citation.
  • AI Overviews appear on roughly 15 to 18% of Google searches and far more on informational queries. They launched in Australia on 29 October 2024 and sit above the blue links.
  • When an AI Overview appears, the top-ranking page loses about 58% of its clicks (Ahrefs, 300,000 keywords). Pew found users click a traditional link 8% of the time with a summary versus 15% without.
  • There is no AI Overviews submission, no special markup, no llms.txt file that helps. Google's own guidance calls it "still SEO."

Your rankings held and your traffic still fell. That is the pattern most Australian business owners are hitting in 2026. You still sit at position two for a query, but a Google AI Overview now answers it above your listing, and the click that used to be yours never happens.

Getting into that Overview is a different job from ranking the blue link. Ranking rewards the whole page and the authority of your domain. A citation rewards one passage that the model can lift, ground in a named source, and attribute. This guide explains how Google assembles an Overview, then gives a checklist of what actually moves citation odds, each item sourced.

What AI Overviews are and how often they show up

An AI Overview is the generative answer block Google renders above the organic results, built on its Gemini models. It launched in the US in May 2024 under the earlier name SGE, then reached Australia on 29 October 2024. Each block carries a handful of cited source links beside or beneath the text.

Prevalence figures vary because every tracker uses a different keyword set. Semrush recorded AI Overviews on 6.49% of queries in January 2025, peaking near 25% in July, then settling to 15.69% by November after Google pulled back. Pew's browsing-behaviour study put it at about one in five US searches (18%) in March 2025. Commercial trackers focused on high-value verticals report higher rates still. The share climbs sharply on informational and how-to queries and stays low on navigational and transactional ones.

The traffic effect is measured and consistent. Ahrefs analysed 300,000 keywords and found the presence of an AI Overview correlates with a 58% lower click-through rate for the top-ranking page. Pew's panel of 900 adults clicked a traditional search result 8% of the time when an AI summary appeared, against 15% when none did, and only 1% clicked a link inside the summary itself. The cited sources inside the block pick up a small offsetting click, so being one of them is now a measurable traffic lever rather than a vanity metric.

How Google chooses the sources

Google assembles an Overview through a pipeline, not a single ranking pass. Understanding the four stages tells you where citation is won.

Query fan-out

Google decomposes your query into several narrower sub-queries and searches each one separately. Its own documentation describes this "query fan-out" technique, issuing multiple related searches concurrently across subtopics and data sources, then combining the results. For "best link building tactics for SaaS," the fan-out generates entity-specific and comparative variations, and a cited page can come from the top results of any of them.

This is the mechanism that broke the old top-10 assumption. A page ranking position 40 on your head term can be cited if it ranks near the top for one of the sub-queries. Analysis of 173,902 URLs and 33,000 fan-out queries found pages ranking for both the main query and at least one fan-out query accounted for 51% of AI Overview citations. Comprehensive content that covers the related angles of a topic wins here, not the hyper-narrow single-keyword page.

Grounding in the organic index

Google grounds Overviews in results it already trusts. Its documentation says AI Overviews are built to surface information backed up by top web results, with its core web ranking systems integrated into the experience. The data confirms it. A seoClarity analysis of 500 million keywords found 97% of AI Overviews cite at least one source from the organic top 20, with the number-one position appearing more than half the time. Originality.ai found 52% of citations come from the top 10 and nearly 90% from the top 30.

The practical reading: an AI Overview is not a discovery surface. It re-ranks pages Google already trusts on the topic. If you are not in the top 20 organically, the chance of citation is close to zero.

Passage extraction and corroboration

Inside that candidate pool, the model retrieves passages of roughly 200 to 600 tokens, scores them on how cleanly they answer a sub-question, and keeps the two or three that survive. An average Overview cites around four sources, with a range of two to nine. A page can rank well and still fail this stage if its answer is buried mid-paragraph or if the passage never names the entity it describes, leaving the model unable to attribute the claim.

The checklist: what actually moves the needle

Google says there are no special requirements to appear in AI Overviews, and no markup you can add to force it. That is true and also incomplete. Independent studies keep finding the same set of factors that raise citation odds. Here they are, ordered by impact.

1. Rank in the organic top 10 for the query

This is the precondition, not one option among several. With 52% of citations from the top 10 and near-zero from outside the top 20, pages in positions one to three are cited roughly four times more often than positions four to ten. If your target page sits at position 23, spend the budget on classic on-page SEO, internal linking, and topical authority until you clear the top 10. Schema and answer formatting will not rescue a page the classifier never puts in the candidate pool.

2. Lead each section with a self-contained answer

The model lifts a passage, so the passage has to stand alone. Open every major section with the direct answer in the first two or three sentences, then expand with detail and examples. Burying the answer 400 words down is the single most common reason a strong page fails to be cited. Analysis of ChatGPT citations found 44.2% came from the first 30% of a page, and the front-loading principle holds across AI answer engines. A clean answer unit of roughly 40 to 60 words is what gets paraphrased into the block.

3. Use question-shaped headings

Google matches your H2 text against the sub-questions its fan-out produced. A heading like "How do AI Overviews pick sources" beats "Source selection mechanics" because it mirrors the language a user types. Phrase headings, and where sensible your URL slugs, as the questions your buyers actually ask.

4. Name entities in full and keep content current

Each section should introduce its subject by full name, not "the company" or "the product," so the retrieval layer can ground the passage and attribute it. Link once or twice to a canonical reference, such as an official documentation page or a Wikipedia entry, to disambiguate the entity. Freshness matters more than in classic search: a page last updated in 2022 rarely earns a citation on any query with a temporal dimension, and across AI engines 76.4% of top-cited pages were updated within the previous 30 days. Real content updates count; changing a timestamp alone does not.

5. Earn brand mentions across the web

AI systems recommend brands that show up consistently across independent sources. Reviews on industry platforms, mentions in reputable publications, and a consistent name, address, and phone number across directories build the corroborated signal the model looks for. On the technical side, Organisation schema with sameAs links to your Google Business Profile and LinkedIn helps Google connect your pages to a verified entity, which supports both ranking and attribution.

Want your content appearing in AI Overviews?

Our AI SEO service pairs traditional search optimisation with AI-citation structure. Sourced strategy, measured results, no guesswork.

See AI SEO services

How to measure it

Search Console is the starting point, with a caveat. Google folds AI Overview and AI Mode clicks and impressions into the standard Performance report without a dedicated filter, so you can see aggregate movement but cannot isolate Overview traffic directly. Watch for queries where impressions hold or rise while clicks fall: that gap is often an Overview sitting above you.

Analytics platforms miss the citation click almost entirely. When a user clicks a source inside an Overview, Google strips the referrer, so GA4 buckets the session as Direct or (none) with no campaign tag. Recovering it takes server-side first-party tracking, not an out-of-the-box report.

For visibility itself, use a rank tracker that flags AI Overview presence and citations. Semrush, Ahrefs, and seoClarity all report whether an Overview appears for a keyword and whether your domain is cited. Track two things: the share of your target queries that now trigger an Overview, and the share of those where you are one of the cited sources. Test the queries yourself in Google and in AI Mode to see which passages get lifted.

What not to waste time on

Skip the tactics that testing has already ruled out.

llms.txt files. SE Ranking tested 300,000 domains and found zero measurable impact on citation rates. Google has confirmed it does not use them.

FAQ schema as a shortcut. SE Ranking's analysis found pages with FAQ schema received slightly fewer citations than those without. Schema supports entity clarity and ranking, but it will not lift a thin page into an Overview on its own.

Keyword stuffing. The Princeton and IIT Delhi GEO study found keyword stuffing performed worse than the unoptimised baseline. AI engines retrieve by meaning, not exact-match repetition, so stuffing degrades the signals they actually use.

Optimising a page that does not rank. Adding an answer capsule and schema to a position-23 page will not produce citations, because the classifier sources from the top of the organic pool first. Fix the ranking, then layer on citation hygiene.

How this fits with classic SEO and GEO

For Google AI Overviews, classic SEO is the foundation, not a competing channel. The Overview draws from Google's own organic index, so the work that ranks a page is the same work that makes it eligible to be cited. Google's May 2026 guidance says as much, framing AI optimisation as still SEO. The added layer is passage-level: rank the page, then make sure a self-contained answer sits near the top of each section so the model can lift it.

Other AI answer engines behave differently. ChatGPT and Perplexity draw from a wider, different set of sources and weight freshness and third-party consensus more heavily than Google's organic rankings, which is why a page cited by ChatGPT may never appear in an Overview and vice versa. Optimising for citation across all of them is a broader discipline called generative engine optimization, and the passage structure that wins Overviews is a strong start for the rest.

Generative Engine Optimization: How to Get Your Business Cited by AI

The Princeton GEO paper tested nine strategies across ChatGPT, Perplexity, and Google. Here is what works across every AI engine, not just Overviews.

Read more

If your rankings are holding but the clicks are not, the fix is not more of the same SEO. It is making your best answers liftable, on pages that already rank, so that when Google builds the Overview your name is on it. That is the difference between owning a position and owning the answer.

Sources

Frequently Asked Questions

How do I rank in Google AI Overviews?

Google AI Overviews cite pages that already rank in the organic top 20 for the query, then favour the ones that answer the specific question in a clean, self-contained passage near the top of the page. Get into the top 10 organically first, lead each section with a direct answer in the first 50 words, use question-shaped headings, and keep the page current. There is no separate AI Overviews submission or markup.

Is ranking in AI Overviews different from ranking a blue link?

Yes. Ranking a blue link is about the whole page and site-level authority. Getting cited in a Google AI Overview is about whether a single 200-word passage can stand on its own as a defensible answer that the model can lift and attribute. A page can rank number one and never be cited, and a page ranking position 14 can be cited every time if its passages are cleaner.

How often do AI Overviews appear in Australia?

AI Overviews launched in Australia on 29 October 2024. Across 2025 their presence on Google searches rose from about 6.5% of queries to a peak near 25%, then settled around 15 to 16% after Google recalibrated. They appear most on informational and how-to queries and are expanding into commercial ones.

Do AI Overviews reduce my website traffic?

For most informational queries, yes. Pew Research found users clicked a traditional link 8% of the time when an AI summary appeared, versus 15% when it did not. Ahrefs measured a 58% lower click-through rate for the top-ranking page when an AI Overview is present. Being the cited source recovers part of that loss; being uncited absorbs the full hit.

Does structured data help me get into AI Overviews?

Google states no special schema is required to appear in AI Overviews. Third-party studies find schema correlates with higher citation rates, most likely because it improves organic ranking and helps Google identify the entities on your page. Treat schema as good hygiene that supports ranking, not as a direct lever for AI Overviews.

Chris Raad

Written by

Chris Raad

Founder of Studio Slate. Law degree from Macquarie University. Fell in love with programming at law school when he discovered he could automate his study workflows. Now builds digital infrastructure for professional services firms on the same technology as TikTok and Uber.

More about Chris
Latest Post