When an AI summary appeared, users clicked a traditional result on 8% of visits, against 15% when it did not. Sources cited inside the summary were clicked on about 1%. Here is the full set of numbers and what they imply.
Almost every claim about AI Overviews traces back to a small number of studies, and most of them are quoted without their qualifications. The most rigorous public dataset is Pew Research Center's, based on 68,879 Google searches from 900 US adults.
The numbers
- About 18% of the searches studied produced an AI summary.
- 8% of visits to a results page with a summary included a click on a traditional search result — against 15% where no summary appeared.
- Around 1% of visits included a click on a source cited inside the summary.
- 26% of visits with a summary ended the browsing session entirely, against 16% without.
- 88% of summaries cited three or more sources.
What this does and does not prove
It shows that on affected queries, clicks roughly halve. It does not show that search traffic halves overall, because most searches in the dataset produced no summary, and because summaries appear disproportionately on informational queries rather than commercial ones.
It also does not show which queries are worth having. The queries most likely to be answered in a summary are the ones least likely to be followed by a purchase.
Three implications worth acting on
- Segment your reporting by intent. Informational clicks falling while commercial clicks hold is a normal, healthy pattern in 2026. Reported as one line, it looks like decline and gets the wrong response.
- Watch clicks per impression, not clicks. A falling click-through rate at a stable position is the fingerprint of an answer appearing above you.
- Invest in what cannot be summarised — original data, tools, first-hand experience and anything that changes weekly — and in channels you own outright, because a business with one route to demand is fragile regardless of what that route does next.
Looking for more? Browse all resources.