What an AI Overview is
DOCUMENTED Google describes AI Overviews as a feature that shows “when our systems determine that generative AI can be especially helpful”, for example to “quickly understand information from a range of sources”, with “key information and links to dig deeper”. They are “a core Google Search feature, like knowledge panels” and cannot be turned off, though the Web filter shows only text links. Google also states plainly that “AI Overviews can and will make mistakes”.
For a site owner the important word is core. An AI Overview is not a separate product with its own index, submission path or ranking. It is Search, composing.
Eligibility and what Google says about optimisation
DOCUMENTED Google’s documentation for site owners is unusually direct. “The best practices for SEO remain relevant for AI features in Google Search. There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” To be a supporting link, “a page must be indexed and eligible to be shown in Google Search with a snippet”. “You don’t need to create new machine readable files, AI text files, or markup to appear in these features. There’s also no special schema.org structured data that you need to add.” And the standing caveat: meeting every requirement “doesn’t mean that Google will crawl, index, or serve its content. Indexing and serving isn’t guaranteed.”
The fundamentals Google lists are crawling allowed in robots.txt and by any CDN or hosting layer, content findable through internal links, good page experience, important content in text form, structured data matching the visible text, and current Merchant Center and Business Profile information.
How sources are retrieved
DOCUMENTED “Both AI Overviews and AI Mode may use a query fan-out technique, issuing multiple related searches across subtopics and data sources, to develop a response.” As the response is generated, Google’s models “identify more supporting web pages”, producing “a wider and more diverse set of helpful links” than a classic search. Google adds that “AI Mode and AI Overviews may use different models and techniques”.
INFERRED The inference this publication draws, and labels: a page can be retrieved by a fan-out for a sub-question the searcher never typed, which is why breadth of coverage around a subject can matter more for AI Overviews than a single page aimed at the head term. Google does not publish which sub-queries it issues or how it weighs the results.
What selection favours
Google has not published a ranked list of factors for which retrieved pages an overview draws on. Anything presented as one is inference. What can be said from the documentation is that ordinary Search eligibility is the gate, that preview controls narrow what is shown, and that Google’s advice for succeeding is its long-standing advice: “focus on your visitors and provide them with unique, satisfying content”. Everything more specific belongs to observation, which is what the planned Perth source study exists for.
What the click studies found
TESTED Pew Research Center analysed the browsing data of 900 US adults for March 2025. 18% of Google searches in the study produced an AI summary. Users who saw one clicked a traditional result link in 8% of visits, against 15% for users who did not; they clicked a link inside the summary in 1% of visits. 88% of summaries cited three or more sources, and only 1% cited a single source.
TESTED Ahrefs compared 300,000 informational keywords, half of which triggered an AI Overview in March 2025, using Search Console click-through rates for March 2024 (before the US rollout) and March 2025. Its estimate was that the presence of an AI Overview reduced the click-through rate of the top-ranking page by about 34.5%, relative to what the informational-keyword trend predicted. Ahrefs notes 99.2% of keywords that triggered an overview in its data were informational.
Both studies are US-based and pre-date AI Mode. No equivalent Australian panel study is known to this publication, and neither study says anything about Perth.
AI Overviews and AI Mode
DOCUMENTED AI Mode is the conversational surface: “an AI-powered search experience for those who want to go deeper”, where “people can ask nuanced questions that might have previously taken multiple searches”. The same eligibility rule applies, the same fan-out may be used, and Google says the two “may use different models and techniques”. Google Australia announced AI Mode’s rollout “in Australia in English” on 8 October 2025, “using a custom version of our advanced Gemini models”. This publication records the two surfaces separately in its methodology because their citation behaviour can differ.
Controls and measurement
DOCUMENTED Because AI is “integral to how Search functions”, the robots.txt rules for Googlebot are the control over crawling for these features. To limit what is shown, Google names nosnippet, data-nosnippet, max-snippet and noindex, and says “more restrictive permissions will limit how your content is featured in our AI experiences”. Google-Extended, the token for Gemini training and grounding, “does not impact a site’s inclusion in Google Search nor is it used as a ranking signal”.
DOCUMENTED Sites appearing in AI Overviews and AI Mode “are included in the overall search traffic in Search Console”, within the Performance report’s Web search type. There is no separate AI Overviews report, so a Perth site cannot read its overview clicks directly from Search Console.
How AI Overviews apply in Perth
DOCUMENTED AI Overviews are available in Australia and AI Mode has been available in Australia since 8 October 2025, so both appear for searches made in Perth on google.com.au.
HYPOTHESIS Which Perth queries trigger an overview, and which sources those overviews cite, is not documented anywhere and has not been measured by this publication yet. The planned study records, per Perth prompt, whether an overview appeared, every cited source, and the organic results beneath it, so the overlap between cited sources and organic rankings can be reported as a count rather than assumed.
What is documented and what is inferred
- Documented: eligibility; no special markup; query fan-out; preview controls; Search Console reporting; availability in Australia; AI Mode’s Australian launch date.
- Tested: Pew’s frequency and click figures; Ahrefs’ CTR estimate; both US.
- Inferred: that fan-out rewards coverage of sub-questions.
- Not established: any factor list for source selection; any Perth trigger rate or source mix.