DEV Community

Edward Calanni
Edward Calanni

Posted on

How to Track AI Referral Traffic from ChatGPT, Claude, Gemini and Perplexity in GA4 ?

AI assistants are becoming a measurable source of website traffic.

People now use ChatGPT, Claude, Gemini, Perplexity and other AI assistants to research products, compare tools, solve problems and discover websites.

When someone clicks a link suggested by one of these assistants, the visit may appear inside Google Analytics 4 as referral traffic.

That means you can already start identifying AI traffic and analyzing how those visitors behave on your website.

Here’s how.

Where AI referral traffic appears in GA4

Open Google Analytics 4 and go to:

Reports → Acquisition → Traffic acquisition

Then change the primary dimension to:

Session source / medium

This gives you a more detailed view of where individual sessions originated.

Depending on the assistant and how the visit was referred, you may see sources associated with platforms such as:

  • ChatGPT
  • Perplexity
  • Claude
  • Gemini
  • Copilot
  • other AI assistants

Not every visit influenced by an AI assistant will necessarily appear as an AI referral.

For example, someone might discover your company through ChatGPT, remember the brand name, and search for it on Google later.

GA4 would likely attribute that session to search rather than to ChatGPT.

So AI referral traffic should be seen as the observable portion of AI-driven discovery, not a complete measurement of AI influence.

See how AI referral traffic appears inside Google Analytics

Here’s a quick walkthrough showing where AI referral traffic can appear inside GA4 and how to find it:

Once you know where to look, it becomes much easier to separate AI referrals from your other acquisition sources.

Why separate AI traffic from other referrals?

If every referral is grouped together, you lose information about how individual sources perform.

Separating AI traffic lets you analyze things like:

  • which AI assistants are sending visitors
  • which pages those visitors land on
  • how long they stay
  • whether they continue browsing
  • which sources generate engaged sessions
  • which AI assistants send the most valuable visitors

This becomes particularly useful when several AI assistants are referring traffic to the same website.

For example, visitors arriving from ChatGPT may behave differently from visitors coming from Perplexity.

The number of sessions might be similar, while engagement or landing-page behavior can be completely different.

Look at landing pages, not just traffic volume

Traffic volume alone doesn’t tell you very much.

The landing page often gives you a much better idea of the visitor’s intent.

Imagine an AI assistant sends someone directly to:

/pricing

That visitor probably has a very different intent from someone landing on:

/blog/what-is-seo

That’s why it makes sense to analyze the combination of:

AI source + landing page + engagement

rather than simply counting visits.

A relatively small number of highly targeted visits can sometimes be much more interesting than a larger amount of generic referral traffic.

Compare AI traffic with other acquisition channels

Once you’ve identified your AI referral sources, you can compare them against channels such as:

  • Organic Search
  • Direct
  • Referral
  • Organic Social
  • Paid Search

Useful metrics to monitor include:

  • total users
  • new users
  • sessions
  • engaged sessions
  • average engagement time
  • events
  • conversions
  • landing pages

You can also create a dedicated GA4 exploration or comparison containing only the AI sources you want to monitor.

Over time, this makes it easier to understand whether AI assistants are becoming a meaningful acquisition channel for your website.

Keep a list of AI sources to monitor

The number of AI assistants people use continues to grow.

I find it useful to maintain a simple list of platforms worth monitoring.

For example:

ChatGPT
Gemini
Claude
Perplexity
Copilot
Grok
Meta AI
DeepSeek
Mistral

PerkFuel has published a more detailed guide on tracking AI referral traffic in GA4, based on its work analyzing how traffic from different AI assistants appears inside analytics platforms.

The observations and examples in that research are also useful when building your own AI referral tracking setup.

AI attribution is still imperfect

This is probably the most important limitation to understand.

A visitor’s journey could look like this:

  1. Someone asks an AI assistant for a recommendation.
  2. The assistant mentions a website.
  3. The person doesn’t click immediately.
  4. They remember the brand.
  5. They search for it later.
  6. They visit through Google.

The AI assistant influenced the discovery, but GA4 may never classify that visit as AI referral traffic.

The same problem already exists with podcasts, social media, word of mouth and many other discovery channels.

So AI referrals shouldn’t be treated as a perfect measurement of AI visibility.

They’re better understood as one measurable signal.

Why AI referral traffic is worth watching

The interesting question isn’t necessarily whether AI assistants will replace Google.

It’s that they’re becoming another way people discover websites.

Search traffic already has its own analytics.

Social traffic has its own analytics.

Paid traffic has its own analytics.

Referral traffic does too.

AI traffic is starting to deserve the same treatment.

As ChatGPT, Gemini, Claude, Perplexity and other assistants become a larger part of how people research products, companies and information online, understanding how that traffic behaves will become increasingly useful.

If you’re already seeing traffic from AI assistants in GA4, take a look at the landing pages and engagement metrics rather than focusing only on the number of sessions.

That’s where the interesting patterns usually start to appear.

Top comments (1)

Collapse
 
lunarose profile image
Luna Rose

Great point. AI discovery is becoming another channel worth analyzing.