AI & Development
Google AI Overviews cite sources that get 35% more clicks than organic position 1. Here's what Google looks for when selecting sources to cite
Google AI Overviews appear for a growing share of informational queries. When they do, organic click-through rates for results below the Overview drop significantly - Seer Interactive measured a drop from roughly 15% to 8% in September 2025. But sites cited inside the AI Overview receive more clicks than organic position 1. The new success metric in AI Overview-eligible queries is citation, not ranking position. Here is what determines which sources get cited.
Google AI Overviews are AI-generated summaries that appear above traditional organic results for certain queries. They synthesize information from multiple sources, cite those sources with attribution links, and provide a direct answer to the query without requiring the user to click through to a source page.
Google generates AI Overviews using a combination of its large language model (Gemini) and real-time web retrieval. The sources cited in an AI Overview are selected from the current top-ranking pages for the query, not from a separate pool. A page that does not rank in the traditional results will not be cited in the AI Overview for that query - citation is a secondary outcome of ranking, not an independent path.
Being cited in a Google AI Overview requires satisfying two conditions simultaneously:
Condition 1: The page ranks in the traditional results. Google selects AI Overview sources from the pages already ranking for the query. Pages that rank higher are more likely to be cited. Pages that rank but are not selected are typically excluded because their content structure does not facilitate extraction - the information is present but not formatted in a way the AI can extract cleanly.
Condition 2: The content is extractable. The AI Overview's generation process extracts specific factual passages from source pages and synthesizes them into an answer. Pages whose key information is buried in long paragraphs, written with dense context dependencies, or structured in ways that require reading the full article to understand are less likely to be cited even when they rank. Pages whose key information is in short, self-contained, clearly-headed paragraphs are cited at higher rates.
The content structure that maximizes AI Overview citation probability is the same structure that optimizes for featured snippets - direct answers at the top of relevant sections, clear question-like headings, and factual information in self-contained paragraphs. This is not coincidental: Google AI Overviews and featured snippets are produced by similar extraction mechanisms.
Direct answers under question headings. Structure sections so that the heading is close to the actual user query ("How long does the French citizenship exam take?" not "Exam Duration") and the first 40 to 60 words under that heading directly answer the question. The AI Overview extraction mechanism prioritizes the first content under each heading.
Short, autonomous paragraphs. Each paragraph should make sense without the surrounding context. AI Overview extraction takes paragraphs in isolation. A paragraph that requires reading the previous three paragraphs to understand will not be accurately extracted. A paragraph that is self-contained and complete in one to three sentences will be.
Factual, affirmative language. Write "The French citizenship test has 50 questions drawn from a pool of 180" not "The French citizenship test may include around 50 questions." AI systems are more confident in extracting affirmative statements with specific details than hedged statements with approximations.
Google AI Overviews appear most consistently for: informational queries with a clear factual answer, how-to queries with step-by-step procedures, comparison queries where multiple options are being evaluated, and definition queries. They appear less commonly for: navigational queries (users looking for a specific website), transactional queries (users ready to purchase), and queries with significant recent news relevance where real-time information is more important than synthesis.
Identifying which of your target keywords trigger AI Overviews is the first step in optimizing for them. Search each target keyword and note whether an AI Overview appears. For keywords where an AI Overview consistently appears, the content structure requirements above are mandatory. For keywords where no AI Overview appears, traditional SEO optimization is sufficient.
Article JSON-LD schema provides Google with machine-readable metadata about your content: the author, publication date, the organization behind the article, and the article's subject matter. While structured data is not a confirmed factor in AI Overview citation selection, it provides clear entity signals that help Google correctly attribute and contextualize content during extraction.
FAQPage schema is particularly relevant for AI Overviews. Pages with valid FAQPage schema - structured question-answer pairs in the HTML - provide pre-formatted Q&A content that the AI can extract without needing to parse prose. FAQPage schema on a page that answers common questions about your topic is one of the most direct ways to make your content's key information accessible to AI Overview extraction.
Google Search Console does not yet provide a dedicated AI Overview impression and click report (as of mid-2026). The current measurement approach is: search your target keywords manually in Chrome (logged-in, private browsing to reduce personalization), note which queries trigger AI Overviews, and check whether your domain appears in the citation list under the Overview. Do this monthly for your top 20 target keywords and track changes over time.
Third-party tools including SEMrush and Ahrefs have added AI Overview tracking features that automate this monitoring at scale. For sites with large keyword portfolios, these tools are more efficient than manual checking.