PLATFORMS · Chapter 2

Getting cited in Google AI Overviews and AI Mode.

AI Overviews and AI Mode are two surfaces sharing one engine. Both ground on Google's regular index. Both use Gemini to compose an answer. Both pick sources by fanning the user's query into sub-queries and matching each to a passage. The work that earns citations is the same — passage-shaped content, an Article and author schema stack, and ruthless primary-source discipline.

KEY TAKEAWAYS

AIO and AI Mode share Google's regular index — there is no separate AI index to optimise for.
Query fan-out means a page that anticipates the next three questions a buyer asks beats a page that only answers the headline query.
Opting out of Google-Extended does not affect citation eligibility. It only prevents future Gemini training on the site.
The eligibility list below is the minimum bar — meeting it is necessary, not sufficient.

AIO and AI Mode eligibility checklist

Requirement	Why it matters
Indexed in regular Google (not noindex, not blocked).	AI Overviews and AI Mode ground on Google's main index. No index, no citation.
Self-contained answer paragraph in the first 100 words.	Gemini extracts a citable passage, not a whole page. The lead paragraph is the most-tested location for that passage.
Sub-headings phrased as the questions a user would ask.	Query fan-out — Google decomposes a question into sub-queries and matches each to a passage. Section headings that read as questions get matched first.
Article + Author + Publisher schema stack.	AIO and AI Mode show source attribution. Schema makes the identity unambiguous and increases the chance the citation displays with author and brand name attached.
Cited primary sources for any factual claim.	Gemini visibly prefers pages that themselves cite primary sources, especially in YMYL topics. This is the same Tier-1 source discipline classical SEO has always rewarded — Google's AI surfaces just penalise its absence more visibly.
Page passes a manual fact-density check.	If a paragraph has fewer than one verifiable claim per 100 words, AI surfaces will prefer a denser source.

Source for AI Features in Search: Google Search Central. Field observations on what moves citation share are from José's own implementations.

Running a query fan-out audit

The single highest-leverage AIO fix for most pages is a fan-out audit. The process takes 30 minutes per page and looks like this.

Write down the headline query. The one search the page is built for. Example: "what is a fractional CMO".
List 8 to 12 follow-up questions a buyer would naturally ask next. Cost, comparison to alternatives, hiring process, contract length, who it is for, who it is not for, how to evaluate one, red flags.
Check each against the current page. Mark each follow-up Covered, Mentioned, or Missing. Anything Missing is a fan-out gap.
Add an H2 per gap. Each section opens with a 40 to 80 word self-contained answer to the sub-query. The model should be able to lift any of them as a citation.
Re-run the priority query in AI Mode after 4 to 6 weeks. Track whether the site moves from uncited to cited, or from cited-once to cited-multiple-times.

Why most SMB pages lose the AI Overview

Three failure patterns account for most "we rank but never get cited" complaints I see in audits.

Marketing-speak lead paragraph. The first 100 words sell the company instead of answering the query. Models skip the section entirely and pick a competitor's introduction instead.
H2s phrased as features, not questions. "Comprehensive solutions" is not a sub-query anyone runs. Headings that match a real fan-out question (e.g., "how much does a fractional CMO cost") get matched and lifted.
Zero outbound citations. A page that asserts statistics without linking to primary sources is treated as a weaker citation candidate than a page that cites the same source the model would have cited anyway.

Best for

Pages that already rank top 10 in Google for the priority query — AIO eligibility starts from a strong classical SEO base.
Definitional pages, comparison pages, and pricing pages — the page types AIO is most likely to feature.
Sites with a credible author and organisation schema stack already in place.

Fails when

The page is noindexed, blocked by robots.txt, or returns JavaScript-only content on first response.
The lead paragraph reads as a brochure rather than a direct answer.
Every claim is unsourced — AIO visibly prefers pages that themselves cite primary sources.

Verify before

The page is indexed in Google (check via site: search or Search Console URL inspection).
The first paragraph would stand alone as a citation if quoted.
Every factual claim links to a primary source.

Frequently asked

How do I get my site cited in Google AI Overviews?

Google AI Overviews cite sources from the regular Google index. Eligibility requires being indexed without a noindex tag, opening every page with a self-contained answer paragraph in the first 100 words, using question-shaped sub-headings that match how users ask, marking up Article and Author schema, and citing primary sources for any factual claim. The same fundamentals apply to Google AI Mode, which grounds on the same index.

Does AI Mode use a different index than regular Google search?

No. AI Mode and AI Overviews both ground on Google's main search index. The difference is the surface: AI Overviews appears inside the classical SERP above the blue links; AI Mode is a standalone conversational interface. Both use Gemini to compose answers and cite sources back to the same set of indexed pages.

What is query fan-out and why does it matter?

Query fan-out is how Gemini decomposes a single user question into multiple sub-queries, then assembles an answer from passages matching each sub-query. A page that only answers the headline question loses to a page that anticipates the fan-out. Running a manual fan-out audit on a priority page — listing every sub-question a buyer would naturally ask next — and ensuring each has its own H2 is the highest-leverage AIO fix for most SMBs.

Does opting out of Google-Extended affect AI Overviews citations?

No. Google-Extended is the opt-out token for Gemini's training on a site's content. AI Overviews and AI Mode ground on the regular Google index, so disallowing Google-Extended does not affect citation eligibility. It only affects whether the content can be used as training data for future Gemini models.

Why does Google AI Overviews cite Reddit and Quora so often?

AIO weights pages that directly answer the user's question and pages that show genuine first-hand experience. Reddit and Quora threads do both — the question is the page title, the answer is the top comment, and the experience is unmistakably human. SMBs that want to compete have to write pages that answer one specific question per page with comparable directness, and back claims with real implementation details.