Context
A randomized field experiment using a custom Chrome extension compared Google Search with AI Overviews against an AIO-hidden condition, measuring organic outbound clicks, zero-click searches and perceived search quality.
What changed
In the experiment, conditional on an AI Overview appearing, outbound organic clicks were 39.8% lower and zero-click searches 34.5% higher with AIO exposure. Sponsored clicks and overall search frequency did not detectably change; the study found no measurable survey improvement in perceived search quality.
Why it matters
AI search citation visibility does not necessarily translate into publisher referral traffic. Controlled exposure evidence helps distinguish observed changes in search clicks from correlation in platform-level analytics.
Visibility OS interpretation
Visibility OS interpretation: track downstream clicks and zero-click behavior alongside AI Overview exposure and citation frequency; the experiment does not establish a universal traffic reduction or direct publisher revenue losses.
What remains uncertain
How do estimates generalize to other countries, devices, search intents and native rollout conditions? What longer-run behavioral adaptations occur? How robust is the exploratory AI Mode arm?