Google Search Console is inadvertently recording fragments of user conversations with AI tools as search queries, providing site owners with a new, albeit limited, window into AI-driven traffic.

Key Points

• Search Console records AI follow-up questions and responses—such as "yes" or "what about"—as standard search queries within the general performance report.
• These conversational fragments fall into seven distinct categories, including reply artifacts, pivot follow-ups, conversational questions, and automated tracker probes.
• Google’s official Generative AI performance reports exclude query and click data, making these leaked fragments the only available insight into specific user prompts.
• A new, free classifier tool for the Search Console MCP allows users to identify and categorize these conversational queries within their own site data.
• Large websites can use BigQuery bulk exports to capture a more comprehensive view of these rare, non-repeating conversational strings.

Why it Matters

This discovery reveals that standard SEO performance data is increasingly contaminated by AI-generated conversation logs, which can skew traditional keyword analysis and traffic reporting. By isolating these fragments, site owners can better understand how their content is being consumed within AI answer blocks and optimize specific passages to improve visibility in generative search results.