SearchGPT and ChatGPT Referral Tracking : Why your Analytics Can Be Missing It

With the news yesterday that ChatGPT is getting 400million visits a week, Webmasters are scratchy and itchy for a fresh source of traffic. Lets remember, that it was only a few weeks ago that OpenAI released SearchGPT to non-logged in users. Clearly there is an adoption phase required before it rains referrals.

Yesterday we marked a day of our first 200 referral day from ChatGPT. Meanwhile  Bing sent half that, and Google sent apparently ten fold that.  That is what we could find trackable. I don’t believe that was all the traffic they sent us of course, because of Chromes referral policies often will completely ignore sending referrals (especially from JS driven links like those from OpenAI).

It is also clear from some blog and forum posts, that there are Webmasters that don’t know what to look for in order to track traffic from ChatGPT. It’s time to figure out how to track, optimize for, and leverage this new ecosystem.

How to Spot ChatGPT and SearchGPT Traffic in Logs

Unlike traditional search engines that use clear user agents and referral headers, traffic from ChatGPT and SearchGPT is harder to identify. Here’s what to look for:

• • User Agents: OpenAI-powered browsing and link-following activity may have distinct user agents such as:

    Mozilla/5.0 (compatible; OpenAI; +https://openai.com/bot)


    That is a referral from ChatGPT itself  – at the request of a user – as it looks on the web for information.

    User-generated traffic from ChatGPT users clicking on links will likely appear as a plain old browser user agents.

    See more indepth article on OpenAI bots.

  • Referral Data: Reminder, ChatGPT is mostly JavaScripted site. The JS engine in Chrome (Vivaldai, Opera, Edge, Brave, and other chromium variants) have different referral and Cross Domain settings. Therefore, there is not always a  clear referrer sent.  Depending on how links are handled, referrals may appear:

    • As direct traffic (no referrer – type in)
    • From chat.openai.com
    • As an embedded browser referrer like:

Referrer-Policy: no-referrer-when-downgrade

• That header is not tracked by traditional logging systems. Only by parsing headers, and writing the pertinent headers to disk, can you track all OpenAI generated traffic.

• Custom UTM Parameters: OpenAI can employ UTM parameters when linking out from AI-generated answers. Watch for terms like:

  utm_source=chatgpt
utm_medium=ai_search
utm_campaign=openai_suggestions


  These would indicate OpenAI-generated referrals.

  It seems like UTM parameters are reserved for larger sites that may have partnered with OpenAI and data access.

Why It’s Hard to Spot ChatGPT Traffic

As mentioned, several factors make tracking ChatGPT and SearchGPT referrals challenging:

1. Lack of Standard Referral Data – Unlike Google, Bing, or other search engines, ChatGPT often does not pass referrer headers, making it look like direct traffic.
2. AI Summarization & Answering – Users may never visit your site directly if their query is fully answered in the AI response.
3. User Behavior – When users do click, they may open the link in a new tab, further obscuring referral data. Chrome will block a referral in a new tab, from an https site.

How SEOs Can Track ChatGPT Referrals

• Check for OpenAI-related user agents in logs
• Analyze spikes in direct traffic alongside relevant keyword trends (meh)
• Look for UTM parameters associated with AI-generated traffic (nice work if you can get it)
• Use JavaScript tracking (e.g., document.referrer) to capture potential referrer info before it gets stripped (winner-winner chicken dinner)

Robots.txt Considerations for OpenAI Crawlers

Webmasters can allow or block OpenAI crawlers just like other bots. OpenAI has a dedicated user agent:

To allow OpenAI’s bot to crawl your site:

To block it completely:

This gives site owners control over how much of their content AI models access for training and search purposes.

Why SearchGPT is a Big Win for Webmasters

1. A New Referral Traffic Source

SearchGPT represents a completely new channel for discovery. Webmasters should see this as an opportunity to gain traffic in a way that wasn’t possible before.

2. Less Dependency on Traditional Search

For years, SEOs have been at the mercy of Google’s algorithm changes. AI-driven search like SearchGPT provides an alternative traffic stream, reducing risk from Google updates.

3. Potential for Higher Intent Clicks

AI-assisted users may arrive further down the conversion funnel, as AI search models filter out irrelevant queries and send more qualified traffic.

4. Direct Answers & Authority Building

Being cited by ChatGPT or SearchGPT increases brand visibility and positions your site as an authoritative source in AI-driven searches.

What’s Next?

• Monitor your logs for ChatGPT-related traffic.
• Adjust robots.txt settings to match your AI content strategy.
• Experiment with content formats that AI prefers (structured data, Q&A).
• Track AI-generated referral patterns and optimize accordingly.

SearchGPT isn’t a threat—it’s an opportunity. SEOs who understand how AI search works will be ahead of the curve in leveraging this new wave of traffic.

<added>

Why document.referrer Can Show a Referral When the HTTP Header Doesn’t

1. Referrer-Policy Settings

   • When a user clicks a link from an external site, browsers determine what referrer information to send based on the Referrer-Policy of the referring page.
   • If the policy is set to no-referrer, then no referrer data is sent in the HTTP request headers at all.
   • However, if it’s set to strict-origin-when-cross-origin, then the origin (domain) might still be available to JavaScript inside the destination page (document.referrer), even if it’s not passed in the HTTP header.

2. Differences Between HTTP Headers and JavaScript Environment

   • HTTP request headers are controlled by the server and browser policies.
   • document.referrer is accessible only after the page loads in the client-side JavaScript environment, meaning the browser may still expose the referrer information within the page’s DOM.

3. Cross-Origin Requests and Referrer Trimming

   • If a site sets Referrer-Policy: same-origin, the full URL is only passed for same-origin requests. However, document.referrer may still capture a trimmed version (like just the domain).
   • Some browsers modify referrer headers for privacy reasons but still allow document.referrer access under certain policies.

4. How This Relates to ChatGPT and AI Referrals

   • If ChatGPT-generated links don’t explicitly set a strict Referrer-Policy, document.referrer might still contain some identifying information, even if the HTTP headers omit it.
   • SEOs can use JavaScript tracking (like Google Tag Manager) to capture document.referrer and store it in analytics for deeper insights.

How to Capture document.referrer for Tracking AI Traffic

Since document.referrer can sometimes capture AI-generated traffic even when the server logs don’t, you can log this data using JavaScript:

  if (document.referrer) {
    console.log("Referral Source:", document.referrer);
    // Send data to analytics
    fetch('/log-referrer', {
      method: 'POST',
      body: JSON.stringify({ referrer: document.referrer }),
      headers: { 'Content-Type': 'application/json' }
    });
  }

This script will log the referral source and send it to your analytics system.

Conclusion

Even if ChatGPT and SearchGPT don’t pass referrer headers, document.referrer might still capture some referral data within the browser. This makes it a valuable workaround for tracking AI-driven traffic that otherwise appears as “direct” in server logs.