A dataset of 846,000 Google searches now sits at the center of a Search Engine Journal analysis of AI Overviews and user behavior. The sample is large enough to move the conversation from anecdote to evidence.

What did the study look at?

The analysis examines how Google's AI Overviews change what searchers do after they see a result. AI Overviews are Google's AI-generated summaries that appear above traditional organic links for many queries. They use the company's Gemini model to synthesize answers from multiple web sources. The feature rolled out broadly in the United States in May 2024 after a limited test. It has since expanded to more countries and more query types. Publishers have watched referral traffic shift since launch. The analysis tracks behavioral signals across query categories, device splits, and result-page layouts. The takeaway: this is the rare study big enough to show real patterns.

Why does the sample size matter?

Most public studies of AI Overviews run on small datasets. Hundreds of queries, sometimes a few thousand, form the typical base. A pool of 846,000 searches gives researchers room to break down behavior by query type, by industry, and by SERP layout. Patterns that look noisy in small samples often show up cleanly at this scale. The volume also helps separate signal from seasonal noise. Holiday shopping spikes, news cycles, and product launches distort smaller studies. A dataset this size smooths much of that out. Bottom line: bigger samples turn noise into signal.

What is AI Overviews changing?

Search Engine Journal's title points to behavior shifts in the data. The full report holds the specific breakdown. The wider public record since the May 2024 US launch has tracked reduced clicks on source pages, more zero-click queries, and more attention paid to the AI block itself. A 846,000-query dataset can confirm or refine those patterns. For publishers, even AI-cited pages are seeing referral declines. For users, more queries end without a click. Search Console does not separate AI Overview impressions from regular search appearances, which makes outside data the only public view. Takeaway: clicks are moving, and the data will tell us by how much.

What is confirmed and what is speculation?

Confirmed by the source:

• A Search Engine Journal analysis exists.
• It covers 846,000 Google searches.
• The focus is AI Overviews and search behavior.

Speculation pending the full read:

• The exact percentages of zero-click shifts.
• The date range of the data.
• The geographic breakdown.
• The per-vertical impact.

Read the full Search Engine Journal report for the breakdown.

Which verticals feel this most?

Health queries, finance questions, and how-to searches see the highest AI Overview placement rates. These are also the categories where users most often accept an AI summary at face value. Publishers in YMYL verticals face the sharpest traffic shifts of any group in the search results. E-commerce and local search see fewer AI Overview placements so far. Transactional queries still tend to surface product cards, maps, and ads. Local-intent queries, like "plumber near me," still trigger Maps packs first, not AI Overviews. Bottom line: YMYL categories carry the heaviest shift.

What should publishers and SEOs monitor?

Watch three numbers over the next two quarters:

• Click-through rate on pages cited inside AI summaries.
• The share of informational queries that end without a click.
• The volume of branded follow-up searches after an AI Overview appears.

Google has not published official AI Overview impression data inside Search Console. Third-party tools and large studies like this one are the main public signal. Treat the 846,000-query dataset as a baseline, not a final word. Bottom line: track your own numbers against the dataset.

What is the bottom line?

A 846,000-query sample puts real weight behind claims about AI Overviews and search behavior. The next move belongs to publishers, who should compare their analytics to the patterns the Search Engine Journal report surfaces. Watch what happens when AI Overviews expand further into your vertical over the coming months.