Google AI Overviews: How They Pick Sources and How to Get Included
AI Overviews pull from pages that already rank, but inclusion is not automatic. The structure, markup and freshness signals that increase your odds.
AI Overviews appear above traditional results and answer the query in a paragraph with a small set of supporting links. The sources are drawn from pages Google already considers relevant, which means the first requirement is boring and non-negotiable: rank.
How sources are selected
Google has been consistent in public statements: AI Overviews use core ranking systems plus query fan-out, where one question is broken into several sub-queries answered by different passages. Pages that cover the sub-questions inside clearly separated sections are far easier to include than pages that bury one answer in a long narrative.
Signals that help
- Direct answer near the heading. Forty to sixty words that resolve the sub-question.
- One topic per section. Clean chunk boundaries for fan-out retrieval.
- Tables for comparisons. Compressed, structured, easy to lift.
- Visible dates and honest updates. Recency matters for fast-moving topics.
- Author and organisation clarity. Schema plus a real about page.
- Fresh internal links. Show that the page is part of a maintained site.
What does not help
| Tactic | Reality |
|---|---|
| Keyword stuffing | Ignored at best, harmful at worst |
| AI-generated filler at scale | Low value; usually fails fan-out relevance |
| Hidden text for models | Cloaking; a policy violation |
| Copying competitor outlines | Produces the same answer with no reason to cite you |
A page template that works
- H1 that names the exact topic.
- A two-sentence definition or answer, no preamble.
- A table summarising the facts the query implies.
- H2 sections, each answering one sub-question, each starting with the answer.
- A worked example with real numbers.
- Four to six genuine FAQs.
- Author box, sources, and an updated date.
Check the “People also ask” box for your query. Those are the sub-questions fan-out is likely to expand into, and they are free research.
Measuring inclusion
Search Console does not separate AI Overview clicks, but you can infer movement: a jump in impressions with flat clicks often means inclusion in an AI surface. Pair that with monthly manual prompt checks for the queries you care about most.
The structural parts of this checklist are exactly what the AI Visibility Checker scores — run it before you start rewriting so you can see what changed.
Frequently asked questions
Do I need a special tag to appear in AI Overviews?
No. There is no AI Overviews opt-in tag; eligibility follows from conventional indexing and ranking.
Can I opt out?
Google offers controls for some AI features and for training, but AI Overviews are part of Search itself.
Does schema guarantee inclusion?
No. It removes ambiguity, which improves the odds, but ranking and content quality dominate.