AI search platforms sent an estimated 400 million referral clicks to external websites in 2025. Most of those clicks are invisible in Google Analytics. This guide covers the practical steps to start capturing them.
Step 1: Create a Referral Exclusion List (What Not to Exclude)
GA4 has a referral exclusion feature designed to prevent your own domains from inflating traffic. By default, some analytics setups accidentally exclude AI platforms. Check Settings → Data Streams → Configure Tag Settings → List Unwanted Referrals and make sure you are not blocking chat.openai.com, perplexity.ai, claude.ai, or gemini.google.com.
Step 2: Set Up a Channel Group for AI Search
In GA4, go to Admin → Data Display → Channel Groups → Create New Channel Group. Add a channel called "AI Search" with the following rules:
- —Source contains: perplexity.ai
- —Source contains: chat.openai.com
- —Source contains: claude.ai
- —Source contains: gemini.google.com
- —Source contains: copilot.microsoft.com
- —Source contains: you.com
This captures sessions where the AI platform does pass a referrer. It will not capture dark traffic (sessions with no referrer), but it gives you a baseline to work from.
Step 3: Add UTM Parameters to AI-Readable Content
In your llms.txt file, append UTM parameters to any links you include. For example:
https://yoursite.com/services?utm_source=ai&utm_medium=llms-txt&utm_campaign=ai-visibility
When an AI reads your llms.txt and a user clicks a link from the AI's response, the UTM parameters persist in the session. You will see these sessions in GA4 under the Source/Medium report as ai / llms-txt.
Step 4: Use JSON-LD to Embed Trackable URLs
AI crawlers read JSON-LD structured data directly. Include your primary URL with UTM parameters in your Organization and Service schema:
"url": "https://yoursite.com?utm_source=ai&utm_medium=schema"
This is a small but effective way to tag AI-sourced traffic that flows through structured data citations.
Step 5: Create a Server-Side AI Traffic Log
The most reliable method is server-side. Log the raw HTTP_REFERER header for every session and flag any request where the referrer contains a known AI domain. Most web frameworks can write this to a custom log file or database table.
Run a weekly query against this log and compare the AI-sourced session count to your GA4 "AI Search" channel. The gap between the two numbers is your dark traffic estimate.
What to Do With the Data
Once you start seeing AI referral numbers, look at three things: which pages receive AI-sourced traffic, what the bounce rate and time-on-site is compared to organic search, and whether conversion rates differ. In most sites, AI-sourced traffic converts at 1.5x to 3x the rate of organic search because the user already received a recommendation before they arrived.
Before you can track AI referrals, AI systems need to be able to find your site. Check your AI visibility score free at zygur.com.