How to Track if ChatGPT is Sending You Traffic

ChatGPT is already shaping how people discover websites, but most analytics setups still treat that traffic as invisible. The click lands, the session appears, and the source often dissolves into direct traffic or an unhelpful default bucket.

The Problem with GA4 and AI Traffic

By default, ChatGPT referrals can appear as direct or (not set) in GA4 because many clicks do not pass a reliable referrer header. That means a page can be recommended by ChatGPT, receive qualified visitors, and still look like it gained mystery traffic from nowhere.

Step 1 — Check for chatgpt.com in Your Source Report

Open GA4, go to Reports, then Acquisition, then Traffic Acquisition. Add the secondary dimension Session source, and filter for known ChatGPT sources. You are looking for the sessions GA4 does identify before you estimate the hidden share.

Session source contains: chatgpt.com
Session source contains: chat.openai.com

Step 2 — Check GSC for AI Referral Patterns

Google Search Console will not label ChatGPT traffic, but it does show which landing pages are being discovered. If a page suddenly receives more engaged traffic without a matching ranking or impressions change, it is often being referred from an AI answer.

Step 3 — UTM Tag Your Content for AI Platforms

Add UTM parameters to important internal links, templates, and reference pages. If AI platforms scrape and surface those links, more of the click path becomes measurable when visitors arrive.

What the Data Actually Looks Like

A site with little AI visibility usually has flat referral data and no recognisable AI sources. A healthy pattern looks different: a small but growing share from ChatGPT, Claude, Perplexity, and similar platforms, often concentrated around high-authority guides and comparison pages.