Google's Gemini model now powers AI Overviews - the AI-generated summaries that appear at the top of search results for a growing share of queries. By mid-2025, AI Overviews appeared on more than 30% of all Google searches in the US, with particularly high penetration for informational, how-to, and comparison queries. For any business that relies on Google organic traffic, optimising for Gemini-powered results is no longer optional.

The key insight that most SEO commentary misses: Google AI Overviews do not simply quote the #1 ranking page. They synthesise across multiple sources and select citations based on a combination of traditional ranking signals and additional quality signals specific to the AI synthesis layer. Ravve Jay Prevendido at Through The Glass Creatives has been mapping these citation patterns for clients across service industries, and the signal set is materially different from what drove rankings in 2022.

For the broader GEO context, read generative engine optimization. For the strategy specific to Perplexity, see perplexity seo strategy.

What signals drive AI Overview citation

Google's own Search documentation describes AI Overviews as designed to surface "the most useful information from across the web" for a given query. In practice, the citation selection draws heavily on: (1) the existing quality signals used in core ranking (E-E-A-T, PageRank, Core Web Vitals); (2) content that directly and unambiguously answers the specific question posed; (3) pages with clean structured data (FAQ, HowTo, Article schema) that help the model understand content type; (4) pages with clear authorship and publication/update dates.

What AI Overviews exclude

The practical Gemini optimisation checklist

To maximise AI Overview citation probability: add a direct answer to the target query within the first 150 words of the page; use H2 and H3 headers that match the natural language sub-questions users ask; implement Article or FAQ schema; attribute content to named authors with rel="author" linking to their profile page; add a visible "last updated" date; cite your sources in the body of the article with links to the original research. These are the same best practices that good journalism has always demanded - Gemini's selection model has simply created a direct traffic incentive to apply them.

Gemini's citation model rewards pages that read like credible reference material, not pages optimised to rank. The good news: those two things are now converging.

Monitoring your AI Overview presence

Google Search Console introduced AI Overview impression and click data in 2025 - this is the most direct measurement of your Gemini SEO performance. Third-party tools including Ahrefs and SEMrush also added AI Overview tracking in their rank monitoring modules. Run weekly queries on your target keywords directly in Google to spot-check when AI Overviews appear and whether your content is cited. For a full view of SEO performance measurement, read how to measure SEO success.