AI Overviews auto-expand: the citation data that changes your SEO
AI Overviews now auto-expand on nearly half of Google searches. Ahrefs data shows 62% of AI citations come from brands not in the top 10. Here is what the citation collapse means and what to fix.
B
Stephan Ochse
11 min read
September 20, 2026
Share
By Stephan Charles | Last fact-checked: 2026-09-20
BrandCited is an AI brand visibility platform that monitors how often your brand gets named across nine AI engines: ChatGPT, Perplexity, Claude, Gemini, Copilot, Grok, DeepSeek, You.com, and Brave. It assigns an AI Visibility Score from 0 to 100 and runs 30+ audit checks to show exactly what to fix. This article analyzes the data behind Google AI Overviews' auto-expansion and what it means for brands competing to get cited.
Google confirmed in September 2026 that AI Overviews now auto-expand into longer AI Mode-style answers, removing the "Show More" button for some queries. AI Overviews now answer approximately 47% of qualifying Google searches, up from 32% in Q1 2025. The organic link list starts below the first screen or does not appear at all. For most brands, the right response is not to optimize for position one. The data shows position one no longer predicts whether Google's AI cites you.
What Google confirmed about AI Overviews auto-expansion#
Google confirmed in September 2026 that AI Overviews can dynamically expand into longer, AI Mode-style answers for queries where its systems determine a fuller response is more useful. The change removes the manual "Show More" button. Google's generative UI feature—which builds custom visual layouts on demand—simultaneously began rolling out from AI Mode into AI Overviews.
AI Overviews that auto-expand push the conventional organic list below the first screen entirely, with an expanded AI response, supporting links, and an "Ask anything" box all appearing before any blue link. Google has not defined which query categories receive the expanded treatment, but the pattern concentrates on informational and research queries—exactly the queries where brands build awareness and capture consideration.
The BrandCited team covers GEO, AI search optimization, and brand visibility strategy. We publish research, practical guides, and product updates every week.
Spot an error in this article? Report it.
Why your Google ranking no longer predicts an AI citation#
A 2025 Pew Research Center analysis of 68,879 Google searches found that users clicked a traditional result in 8% of visits when an AI summary appeared, compared with 15% when no summary appeared. Traffic concentrates in cited brands, not distributed evenly.
The problem for brands relying on SEO rankings: citation is a separate selection process from ranking.
The economics favor acting on this. Brands cited inside an AI Overview earn 120% more organic clicks per impression than uncited brands on the same query. Cited brands see 35% more organic clicks and 91% more paid clicks than uncited competitors. The economic case for AI citation is clear. Most brands are trying to win it by improving their Google rankings, which now predict citation at a fraction of the rate they once did.
The multi-surface gap: AI Overviews and AI Mode cite different brands#
The citation divergence does not stop at the gap between rankings and citations. Within Google's own AI products, citation sources have almost no overlap.
Only 10.7% of URLs cited in Google AI Overviews also appear in Google AI Mode citation sets, based on an analysis of roughly 540,000 matched query pairs. Domain-level overlap reaches 16%. A brand cited in AI Overviews on a given query is statistically unlikely to appear in AI Mode results for the same query. Two Google products, one search intent, nearly zero source overlap.
This is the core challenge for GEO (Generative Engine Optimization) in 2026. Optimizing one page for one AI product does not carry over. Each surface—AI Overviews, AI Mode, Perplexity, ChatGPT, Claude—draws from a different pool of sources. The brands with the broadest AI visibility are the ones with consistent entity authority across all surfaces, not the ones who optimized a single landing page.
The JavaScript rendering gap most brands have not fixed#
One specific technical failure accounts for a large share of the "my content does not get cited" problem: JavaScript rendering.
An analysis of over 500 million GPTBot fetches found zero evidence of JavaScript execution. OpenAI's GPTBot, Anthropic's ClaudeBot, and Perplexity's bots all send HTTP requests and download the HTML the server returns—without executing JavaScript or waiting for components to mount. None of the major AI-native crawlers render JavaScript per their vendor documentation.
The fix is server-side rendering (SSR) or static generation for any content you want AI engines to see. React-rendered product pages, dynamically loaded FAQs, and client-side comparison tables will not appear in the training or retrieval pool for AI engines, regardless of how well they rank.
The FAQ schema reversal brands need to understand#
Google retired FAQ rich results on May 7, 2026, removed the documentation in June, and dropped Search Console API support in August. FAQ schema markup no longer produces any SERP feature. Some brands removed FAQPage markup entirely as a result—a mistake that reduces AI citation rates.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What is the exact question your customer types into an AI chatbot?",
"acceptedAnswer": {
"@type": "Answer",
"text": "A complete answer in 50-150 words. Names your brand, states a specific claim, requires no surrounding context to make sense."
}
},
{
"@type": "Question",
"name": "A second question targeting a different query type.",
"acceptedAnswer": {
"@type": "Answer",
"text": "A second complete answer. Each Question and Answer pair stands alone as a citable unit."
}
}
]
}
Each acceptedAnswer text field is a direct-extraction target for AI retrieval. The answer must be self-contained—no pronouns that reference other answers, no phrases that depend on the page title for context.
Lily Ray's 2026 B2B software study found that AI Overviews cited self-promotional "best" listicles yet excluded the publisher from recommendations 69% of the time. FAQPage markup needs to answer what users actually ask AI assistants, not what brands want to say about themselves.
Google AI Overviews auto-expand confirmed for September: Google confirmed AI Overviews can dynamically expand for some queries, removing the manual "Show More" step and pushing organic links below the first screen. AI Overviews now appear in nearly half of qualifying Google searches. (Harri Digital)
Gmail AI Overviews reach full global rollout: AI Overviews in Gmail Search hit full deployment for Scheduled Release domains today, following the Rapid Release rollout that began September 3. (Google Workspace Blog)
GPT-6 Astra as ChatGPT stable engine: The September 8 ChatGPT stable release runs on the GPT-6 Astra engine. Voice mode invokes GPT-6 Astra for search and reasoning queries. GPT-5.5 retires from ChatGPT on October 14, 2026. (OpenAI release notes)
Perplexity signs multi-year infrastructure deal: Perplexity announced a partnership with Crusoe to run its full model lifecycle on Crusoe's GB300 NVL72 clusters, announced September 16. (HPCwire)
AI citation top-10 overlap at historic low: 5W Research found overlap between top Google rankings and AI-cited sources dropped from 70% to under 20%. Ahrefs data independently puts citations from top-10 pages at 38%, down from 76%. (PR Newswire)
How BrandCited audits AI Overview citation readiness#
BrandCited's audit engine checks the technical signals that AI Overview and AI Mode citation systems evaluate. The Engine Compatibility score covers schema completeness, JavaScript rendering exposure, FAQPage markup, content freshness signals, and entity resolution through sameAs links.
The AI Overview citation rate check runs your brand across 15 question types—informational, comparison, and branded queries—and shows your citation rate by query category on each AI surface. Most brands find their category-query citation rate is zero while their branded citation rate is healthy. The category gap is where citation growth comes from.
Run the full audit at brandcited.ai. Your score across nine AI platforms in 30 seconds, with every issue ranked by impact.
1Audit which key pages depend on JavaScript rendering. Any content behind a React, Vue, or Angular render step is invisible to GPTBot, ClaudeBot, and Perplexity bots. Migrate product descriptions, FAQ sections, and comparison content to server-side rendering or static generation. This is the highest-impact structural change for multi-engine AI visibility.
1Add FAQPage schema back to every page with Q&A content—even if you removed it after May 2026. Google deprecated the SERP feature, not the AI citation signal. Keep the JSON-LD markup. Add 4-6 questions per page, each with a self-contained 50-150 word answer that names your brand and states a specific claim.
1Stop using Google rankings as a proxy for AI citation. The citation selection process is separate from the ranking algorithm. Run a BrandCited scan to see your AI citation rate directly—it does not correlate with your Google position.
1Check your entity presence across surfaces. Your Organization schema needs sameAs links to Crunchbase, LinkedIn, and Google Business Profile at minimum. This is the entity resolution gate for both AI Overviews and AI Mode. Missing sameAs links make the brand absent from entity lookup before content quality is evaluated.
1Submit updated pages to both Google Search Console and Bing Webmaster Tools. Retrieval-augmented engines—Perplexity, Copilot—pick up new content from Bing's index within 7-21 days. AI Overviews draw from Google's index. Both submission paths matter for multi-surface coverage.
The auto-expansion of AI Overviews means more searches return AI-generated answers instead of a ranked list. The citation gap means most brands getting cited are not in the top 10. These two facts together define the GEO opportunity in Q4 2026: citation optimization is no longer a secondary SEO concern.
Run a free AI visibility audit at brandcited.ai. Your score across nine AI platforms in 30 seconds, with every issue ranked by impact.
How do AI Overviews and Google AI Mode differ in what they cite?
Only 10.7% of URLs cited in Google AI Overviews also appear in Google AI Mode citation sets, based on analysis of approximately 540,000 matched query pairs. Domain-level overlap reaches 16%. The two Google AI products draw from almost entirely different source pools for the same queries.
Do AI crawlers render JavaScript?
No. GPTBot, ClaudeBot, and Perplexity's bots do not execute JavaScript. They download the HTML the server returns and stop. Content that requires JavaScript execution—React-rendered text, dynamically loaded FAQs, client-side comparison tables—is invisible to AI crawlers. Onely's February 2026 analysis found 42% of JavaScript-rendered content never gets indexed by AI systems.
Should I remove FAQPage schema since Google deprecated the rich result in May 2026?
No. Google removed the FAQ rich result feature, not the AI citation signal. AI Overviews, ChatGPT browsing, Perplexity, and Gemini all parse FAQPage schema for citation extraction. Sites with properly implemented structured data get cited 3.2 times more often in AI responses than those without.
How do I check whether AI Overviews are citing my brand?
Run a free BrandCited scan at brandcited.ai. It queries your brand across nine AI platforms including Google Gemini and shows your citation rate by engine and query type. Results in 30 seconds.
What is GEO and how does it differ from SEO?
GEO—Generative Engine Optimization—is the practice of structuring content so AI models cite it in generated answers, rather than simply ranking it for clicks. SEO focuses on ranking pages in blue-link results; GEO focuses on getting named in AI-generated answers. Citation is a separate selection mechanism from ranking, which is why brands that rank well often have low AI citation rates.
