At Search Central Live Deep Dive 2025, Google Search analyst Gary Illyes said the company is still trying to balance user satisfaction with the long-term health of the open web as it rolls out AI Overviews.
Google aims to satisfy both users and publishers
During a Q&A moderated by search consultant Kenichi Suzuki, Illyes explained how Google weighs multiple data points before making public statements about AI Overviews.
- Ongoing user-satisfaction surveys rate the quality of AI-generated answers.
- Executive comments, including those from CEO Sundar Pichai, are vetted against survey findings.
- Market uptake of other AI assistants such as ChatGPT and Copilot informs how people prefer to consume information.
- Illyes stressed that AI Overviews were built to help searchers, not to undermine publishers.
- He acknowledged Google has not yet finalised a sustainable model that delivers traffic back to content creators.
As Suzuki wrote on LinkedIn, the discussion underscored that “we are still figuring things out” when it comes to aligning user experience with the health of the open web.
How Google measures impact
AI Overviews launched in United States search results on 14 May 2024 after a year in Search Generative Experience testing. The feature uses Gemini language models to summarise information above traditional web listings.
Historically, Google has evaluated search changes with metrics such as click-through rate, query reformulation and customer satisfaction. Events like Search Central Live give publishers and developers an additional forum to provide feedback.
Illyes said the company is still tracking how much traffic publishers receive when AI summaries appear above organic links and will continue to adjust the experience as data comes in.
Source citations
- Google Search blog, “AI Overviews: information about generative AI in Search” (14 May 2024)
- Kenichi Suzuki, LinkedIn post reporting Gary Illyes’ remarks at Search Central Live Deep Dive 2025
.svg)
.svg)
.svg)