// Concept
What are Google AI Overviews?
AI Overviews are Google’s AI-generated answer summaries that appear above the traditional search results, citing sources inline.
// Mechanics
How AI Overviews work.
When a user submits a query Google judges suitable, the search stack runs the query through a generative model that synthesizes an answer from a small set of retrieved web sources. The result is a paragraph or bulleted summary, displayed at the very top of the results page, with a row of source cards beneath it. The model picks the sources; the user sees the answer first and the citations second — if they look at them at all. Coverage expanded sharply through 2024 and 2025; today AI Overviews appear on a majority of informational queries in supported markets.
// Impact
Why they cratered organic CTR.
The blue-link economy assumed the user had to click to get the answer. AI Overviews remove that requirement on a huge fraction of informational queries. Industry CTR studies from 2025 consistently showed organic clicks on AI-Overview-bearing queries down 30–60% versus the same query without one. The pages didn’t lose rank; they lost the click. The asymmetry is what makes this a strategic problem rather than a tactical one: even ranking #1 no longer guarantees traffic on the query you used to own.
// Selection
How citations are chosen.
Google hasn’t published the full ranking signal. From observed behavior, citation selection appears to weight:
- Entity authority for the topic. Sites the engine already trusts as authorities in the relevant entity cluster are cited disproportionately.
- Direct topical match. Pages that answer the specific sub-question, not just the broad keyword.
- Clear structure. Schema markup, semantic HTML, and discrete citable claims are easier for the model to extract.
- Recency where it matters. Time-sensitive queries reward freshly updated pages; evergreen queries don’t.
- E-E-A-T signals. Author bylines, organizational identity, citations to primary sources.
// Playbook
How to optimize for AI Overviews.
- Answer the question in the first 60 words. Models prefer extractable, self-contained answers near the top of the page.
- Ship the schema. Article, FAQ, HowTo, Product, DefinedTerm. Structure is a citation accelerant.
- Define your entities cleanly. One canonical name, consistent across your site, Wikipedia, and the open knowledge graph.
- Publish citable claims. Specific, attributable, verifiable. Vague marketing copy doesn’t get cited; precise data does.
- Monitor by query, not by keyword. Track which prompts trigger an AI Overview in your category and which sources Google chose.
// Related
Keep reading.
FAQ
Common questions.
What are AI Overviews on Google?
AI Overviews are Google’s AI-generated answer summaries that appear above the traditional ten blue links. They synthesize information from multiple web sources and cite those sources inline with linked source cards.
Did AI Overviews kill organic traffic?
They didn’t kill it, but they meaningfully compressed it. Click-through rates on informational queries dropped sharply through 2024–2025 because users no longer needed to click to get the answer. Buying-intent and high-consideration queries are less affected, but no category is untouched.
How does Google choose which sources to cite in an AI Overview?
Google has not published the full ranking, but observed behavior suggests strong weight on entity authority, structured data, direct topical match, recency on time-sensitive queries, and the same E-E-A-T signals that drive organic ranking. Sites with clean schema, clear entity definitions, and citable claims are cited disproportionately often.
Can you opt out of being used by AI Overviews?
Partially. The nosnippet meta tag and Google-Extended user agent rules limit some use, but if your page is indexed for normal search, fragments can still appear. The realistic strategy is to optimize for inclusion with attribution rather than opt out.
Do AI Overviews appear for every query?
No. They appear most often on informational, definitional, comparison, and how-to queries. They are rarer on navigational queries, on transactional queries where Google prefers shopping units, and on highly local queries dominated by the map pack.