How Google AI Overviews Pick Sources — and How to Get Cited

The most important shift in Google search since featured snippets

Google AI Overviews now appear on more than half of certain query types, and their share is still growing. When an AI Overview shows up, it answers the question directly, often without the user clicking through to any cited source. That makes "getting cited" the new "ranking number one," and the rules for how Google picks those citations are different from how it ranks blue links.

The single biggest factor: Google's top 10

A 2025 Writesonic study of more than a million AI Overviews found an 81.1% chance that at least one URL from Google's top 10 organic results gets cited in the Overview. Position #1 has a 33% chance of being included. Below position 10, your odds drop sharply.

The implication is uncomfortable but clear: AI Overview optimization starts with traditional ranking. If you're not in the top 10 for a query, you're mostly invisible to the AI Overview for that query, no matter how well-structured your content is.

Domain authority is the second filter

The same Writesonic study analyzed the top 200 most-cited domains and found that domains with a Domain Rating between 88-100 averaged 6,000+ AI citations. Domains below 63 DR were rarely cited at all. Google AI Overviews lean heavily on established trust signals, the same authority signals that drive traditional ranking, but with even less tolerance for low-DR sites.

This is why new sites struggle to break into AI Overviews. The threshold for participation is much higher than for traditional search. (This matters more than most people realize when they're planning a GEO strategy from scratch.)

Content depth and the 100-300 word zone

About 60% of AI Overviews draw from content sections that are 100-300 words long. Not 100-300 word total pages, 100-300 word extractable sections. Each H2 should have a focused answer in that range. One wall of text makes it hard for the model to extract anything useful.

The keyword myth

Roughly 85% of AI Overviews don't contain the exact original search query. Only about 15.8% include the keyword the user typed. Semantic relevance matters far more than keyword matching. A page that answers the underlying question well, in different language than the query, beats a page that repeats the keyword 12 times.

What to actually do

1. Get into the top 10 first. AI Overview optimization is layered on top of organic ranking, not separate from it.
2. Build domain authority. If your DR is below 65, focus on link building and topical authority before chasing AI citations.
3. Write extractable sections. Each H2 should answer one question in 100-300 words.
4. Add schema markup. FAQ, HowTo, and Article schemas help Google parse your content cleanly.
5. Cover the topic semantically. Stop optimizing for the exact keyword. Optimize for the underlying intent and the related entities.

AI Overview citations don't replace traditional rankings, they're a layer on top of them. Win in traditional Google search first. The AI layer follows.

For the step-by-step strategy on earning and keeping AI Overview citations, see How to Get Cited in Google AI Overviews.