Understanding How to Track ChatGPT Traffic in GA4 has become important as the AI chatbot drives visits that often outperform traditional search sources.
Marketers note that, despite its modest share, ChatGPT referrals tend to stay longer and click more, making the traffic especially valuable.
To isolate these visits, open the Acquisition section and select Traffic acquisition. Below the graph, open the drop-down, choose “Session source / Medium,” type “gpt” in the search box and press Enter.
After the filter is applied, users can save the view as a custom report for quick access to the same data set later. Giving the report a clear name helps teams retrieve it without rebuilding the filter each time.
Because ChatGPT does not expose the exact user prompts, analysts infer intent by examining the landing pages that receive the most clicks from the filtered source.
Go to Engagement → Landing Pages, click “Add filter,” pick “Session source / Medium,” select “Contains,” and enter “gpt.” The resulting list ranks pages by the volume of AI-driven visits, allowing SEO specialists to see which content attracts the most conversational queries.
For site owners, isolating AI-originating traffic reveals which content directly answers the problems users pose to a conversational assistant. This insight lets teams prioritize updates that match real-world queries, potentially boosting relevance across all channels.
Technical quirks affect reporting: the AI chatbot often adds a “noreferrer” attribute to outbound links, which hides the referrer and leads GA4 to label many visits as Direct or Unassigned. Because of this, analysts may need to audit suspicious Direct sessions to confirm their true source.
Complementary tools can enrich the picture. In Search Console, look for “possible” AI-mode prompts that surface each page, and in Bing Webmaster Tools identify the grounding queries LLMs use. Using both services provides a broader view of how the bot discovers the site.
Across the reported accounts, the majority of AI traffic appears under the Unassigned category, confirming that the platform’s default attribution does not fully capture ChatGPT-driven sessions. Recognizing this gap helps marketers adjust their measurement strategy and avoid under-reporting the impact of conversational search.
