Measuring LLM Traffic: Source Analysis in GA4

Why LLM Traffic Matters

Users increasingly ask their questions to generative AI tools like ChatGPT, Perplexity and Gemini. When those tools cite sources, a user can click through to your site. This creates a new traffic channel that is distinct from classic organic search.

You cannot measure this traffic without isolating it; by default GA4 lumps these visits into the generic 'Referral' channel. To make strategic decisions you need to track AI-driven traffic as its own channel.

Defining Referral Sources in GA4

In the GA4 admin panel open Data Streams and select your web stream. Under 'Configure tag settings', configure things so referral data is preserved for AI domains rather than excluded. The goal is for these domains to appear correctly in the session_source and session_medium dimensions.

• chatgpt.com - Clicks from the ChatGPT web interface

• perplexity.ai - Source links inside Perplexity answers

• gemini.google.com - Referrals from Gemini chats

• claude.ai - Links shared in Claude answers

Custom Segments and Channel Groups

Under Admin > Channel groups, create a custom channel group and add an 'AI Referral' rule that matches session_source against these four domains. Now AI traffic shows up as its own row in standard reports.

Looker Studio Dashboard

Connect the GA4 data source to Looker Studio and limit the session_source filter to the AI domains. Build scorecards for session count, conversion rate, average engagement time and top landing pages. A time-series chart visualizes the monthly growth of AI traffic.

Interpreting the Data

AI traffic tends to be low volume but high intent; the user clicks through to go deeper after already getting an answer. Compare conversion rate and engagement time against organic traffic to gauge the channel's real value.