?How to Track Clicks Coming from ChatGPT

The Invisible Referrer: How to Track Clicks Coming from ChatGPT

ChatGPT has rapidly become a significant source of information for millions of users. It answers questions, recommends products, and provides links to resources.

For website owners, this presents a new challenge: attribution.

When a user clicks a link provided by ChatGPT, that visit often doesn't show up in Google Analytics 4 (GA4) or other tracking tools as coming from "OpenAI" or "ChatGPT." Instead, it gets dumped into the dreaded "Direct/None" bucket, a phenomenon known as "Dark Traffic."

Why? Because unlike a standard website or search engine, ChatGPT’s interface does not usually pass a "referrer header" when a link is clicked.

If you want to know how much traffic AI is actually driving to your site, you need to employ specific strategies to de-anonymize these clicks. Here is a comprehensive guide on how to do it.

Scenario A: Tracking Links You Control (The Precise Method)

This is the most reliable method, but it only works for links that you feed into ChatGPT.

Use cases for this scenario:

• You are creating prompts for yourself or your team that include links to your site.

• You are building "Custom GPTs" that point users to your specific resources.

• You are distributing PDFs or documents uploaded to ChatGPT’s knowledge base that contain links.

The solution here is the standard digital marketing workhorse: UTM Parameters.

What are UTM Parameters?

UTM (Urchin Tracking Module) parameters are short snippets of code added to the end of a URL. They don't change where the link goes, but they tell your analytics tool exactly where the click came from.

Step-by-Step Guide:

1. Build Your Tagged URL You need to create a unique URL for the links you intend to use within ChatGPT environments. You can use a free tool like Google’s Campaign URL Builder, or just type them out if you know the format.

You should define at least three parameters:

• utm_source: Identify the platform. (e.g., chatgpt or openai)

• utm_medium: Identify the type of traffic. (e.g., ai_chat, chatbot, or referral)

• utm_campaign: Identify the specific context. (e.g., custom_gpt_finance_bot or prompt_library_v1)

Example of a regular link: https://yourwebsite.com/pricing-page/

Example of a tracked link for ChatGPT: https://yourwebsite.com/pricing-page/?utm_source=chatgpt&utm_medium=ai_chat&utm_campaign=custom_gpt_pricing

2. Deploy the Link Whenever you paste a link into a ChatGPT prompt, upload a document to its knowledge base, or configure a Custom GPT action, use this long, tagged URL instead of the naked link.

3. Analyze the Data in GA4 When someone clicks that tagged link inside ChatGPT, GA4 will capture those parameters.

• Go to Reports > Acquisition > Traffic acquisition.

• Change the primary dimension in the table to Session source/medium.

• Search for the parameters you set (e.g., chatgpt / ai_chat). You will now see exactly how many sessions, users, and conversions originated from those specific links.

Scenario B: Tracking Organic Mentions (The Detective Method)

This is the harder challenge. What happens when a random user asks ChatGPT a question, and the AI organically decides to recommend your website as a source?

You cannot add UTM parameters to these links because you aren't the one providing them. ChatGPT is generating them on the fly based on its training data or browsing capabilities.

As mentioned, these clicks usually appear as "Direct" traffic. While you cannot track these with 100% accuracy, you can use deductive reasoning to estimate the impact.

Step-by-Step Guide:

1. Identify "Hidden Gem" Pages Look for pages on your website that receive traffic but rank poorly on Google Search and aren't heavily promoted on social media or email newsletters.

If a deep, specific blog post suddenly sees a spike in "Direct" traffic, and you know you haven't promoted it recently, there is a strong chance that an AI tool has started citing it as an answer to a common prompt.

2. Isolate Direct Traffic in GA4

• Go to Reports > Engagement > Pages and screens.

• Add a filter to show only traffic where the "Session source / medium" matches "direct / none".

• Look for anomalies. Are specific URLs getting a disproportionate amount of direct traffic compared to your homepage?

3. Correlate with "Bing / Organic" and Copilot Microsoft’s Copilot (formerly Bing Chat) is integrated heavily with ChatGPT’s technology. Unlike the standard ChatGPT interface, Bing's AI chat sometimes passes referrer data.

Keep an eye on your Bing / Organic traffic. If you see a rise there, it is highly likely that some of that traffic is coming from their chat interface rather than traditional blue links. While this doesn't track pure ChatGPT, these two sources often correlate strongly.

4. Use Third-Party SEO Tools Some advanced SEO tools (like Ahrefs or Semrush) are beginning to try and estimate "generative AI visibility," though these features are still in their infancy.

Furthermore, you can use social listening tools to search for screenshots people post of ChatGPT conversations that mention your brand. If you see users sharing screenshots of ChatGPT recommending your product, check your analytics for "Direct" spikes at that same time on that specific product page.

Summary

Until OpenAI and other AI providers decide to adopt a standard "referrer header" for outbound clicks, marketers must rely on a mix of precise tagging and clever detective work.

By implementing UTM tags wherever you have control, you can turn at least a portion of that "dark traffic" into actionable data.