RESEARCH · LAST CHECKED SEPTEMBER 5, 2026

AI SEO statistics with sources you can check.

Six findings on AI Overviews, search clicks and impressions. Each number includes its original report, measurement period and scope so you can decide whether it applies to your site.

BY THE SEO AGENT TEAMUPDATED 2026-09-055 MIN READ
AI OVERVIEWS AND SEARCH BEHAVIOR

What the studies measured

8%

1. Traditional-result clicks on pages with an AI summary

In Pew Research Center’s observed browsing panel, a traditional search result was the next visited link on 8% of visits to Google pages with an AI summary.

Scope: 900 US adults; March 2025 browsing, search results collected in April.

Pew Research Center · July 2025

15%

2. Traditional-result clicks on pages without an AI summary

The same analysis found a 15% traditional-result click rate when no AI summary appeared. This comparison is observational; it is not a controlled estimate of the loss for an individual website.

Scope: Same browsing panel and definition as the 8% finding.

Pew Research Center · July 2025

1%

3. Visits that led to a link inside the AI summary

Only 1% of visits to a Google results page with an AI summary were followed by a visit to a cited link in that summary.

Scope: Visit-level behavior; not the share of all websites receiving AI traffic.

Pew Research Center · July 2025

~58%

4. Lower estimated position-one CTR for AI Overview queries

Ahrefs’ February 2026 update estimated that the top organic result’s click-through rate was about 58% lower than its modeled baseline on queries showing an AI Overview. The analysis compares December 2023 and December 2025 data.

Scope: A modeled comparison of keyword groups, not a forecast for every site or search result.

Ahrefs · February 2026

>49%

5. Growth in search impressions in BrightEdge’s tracked data

BrightEdge’s May 2025 review reported more than 49% growth in impressions since AI Overviews launched in May 2024. More impressions did not translate into the same growth in visits.

Scope: BrightEdge’s observed dataset; not a census of all Google searches.

BrightEdge · May 2025

~30%

6. Decline in click-throughs in the same BrightEdge review

The report described a nearly 30% reduction in click-throughs over the same period. Treat this as historical context alongside your own Search Console clicks, rather than a universal traffic-loss assumption.

Scope: A different dataset and metric from Pew’s visit rates and Ahrefs’ position-one estimate.

BrightEdge · May 2025
METHODOLOGY

Use the number with its denominator

We checked these statements against their linked reports on September 5, 2026. The studies were published in 2025 and 2026; the page date does not turn historical observations into measurements of today’s search results. These are third-party findings, not measurements from our customers.

A search impression, a visit to a results page and a click on a position-one result measure different things. Do not average these percentages, add them together or apply one as a revenue forecast. The studies do not establish that an AI Overview caused every observed change.

What to measure on your own site

Compare relevant query groups in Search Console, then follow organic landing pages through signup, a verified payment and renewal. Separate branded searches, informational pages and buying decisions. Read our answer-engine optimization guide for the content workflow and our SEO automation comparison to evaluate the tools.

Useful content needs a clear audience, sources that support its claims and a reason to visit the original page. See how our SEO automation workflow handles researched articles, or send a source correction.

PUT THE RESEARCH TO WORK

Publish articles for the buyers you want to reach.

Research, drafting, fact-checking and publishing in one workflow. Compare the included output, supported CMS and price before you start.