Technical SEO
How to Track LLM Traffic in Google Analytics
ChatGPT, Perplexity, Gemini, Claude. They’re starting to recommend products and link to sites. The problem? Google Analytics doesn’t show “AI search” traffic out of the box.
25 August 2025
7 min read
Most LLM clicks either appear as Direct / None (no referrer) or get lumped in with Bing/Google if the model used a search integration. You have to tag and trap this traffic yourself.
Here’s how
1. Create an “LLM Trap” Page
Build a page that only appears in places you know LLMs crawl and quote from. Give it a unique URL and UTM parameters.
If you see traffic hitting it, you know it came from an AI-generated recommendation.
This is especially useful for proof-of-concept tracking in the early stages.
2. Don’t Block the Referrers You Want
In GA4, go to:
Admin > Data Streams > Configure Tag Settings > List unwanted referrals
Make sure you don’t accidentally exclude:
chat.openai.com
perplexity.ai
gemini.google.com
If they do pass a referrer string, you want to see it.
3. Build an “AI Search” Channel in GA4
Once you’ve tagged your links:
Go to Admin > Data Settings > Channel Groups.
Create a new rule:
Medium contains “llm” OR Source contains “chatgpt” OR Source contains “perplexity”.
Name it “AI / LLM Search”.
Now all your LLM-tagged traffic is grouped in one place, so you can track volume and performance over time.
4. Use Server Logs to Catch the Blind Spots
Some LLM-driven traffic will always look like “Direct” because no referrer is passed. To get closer to the truth:
Check server logs for IP ranges linked to LLM browser modes.
Look for sessions that land on deep product or content URLs without a prior session, a good sign they came from an AI recommendation.
Why This Matters
If you’re serious about AI search optimisation, tracking is non-negotiable.
It proves whether your LLM visibility work is driving clicks.
It gives you the before-and-after data to show ROI.
It helps you decide which platforms and source types are worth the effort.
At LMO7, we don’t just make brands visible in LLM answers, we make sure you can measure it.
Here’s how
1. Create an “LLM Trap” Page
Build a page that only appears in places you know LLMs crawl and quote from. Give it a unique URL and UTM parameters.
If you see traffic hitting it, you know it came from an AI-generated recommendation.
This is especially useful for proof-of-concept tracking in the early stages.
2. Don’t Block the Referrers You Want
In GA4, go to:
Admin > Data Streams > Configure Tag Settings > List unwanted referrals
Make sure you don’t accidentally exclude:
chat.openai.com
perplexity.ai
gemini.google.com
If they do pass a referrer string, you want to see it.
3. Build an “AI Search” Channel in GA4
Once you’ve tagged your links:
Go to Admin > Data Settings > Channel Groups.
Create a new rule:
Medium contains “llm” OR Source contains “chatgpt” OR Source contains “perplexity”.
Name it “AI / LLM Search”.
Now all your LLM-tagged traffic is grouped in one place, so you can track volume and performance over time.
4. Use Server Logs to Catch the Blind Spots
Some LLM-driven traffic will always look like “Direct” because no referrer is passed. To get closer to the truth:
Check server logs for IP ranges linked to LLM browser modes.
Look for sessions that land on deep product or content URLs without a prior session, a good sign they came from an AI recommendation.
Why This Matters
If you’re serious about AI search optimisation, tracking is non-negotiable.
It proves whether your LLM visibility work is driving clicks.
It gives you the before-and-after data to show ROI.
It helps you decide which platforms and source types are worth the effort.
At LMO7, we don’t just make brands visible in LLM answers, we make sure you can measure it.