WRITE TO US
Get IN TOUCH

If users switch to a website from ChatGPT, Perplexity, Gemini, or other similar systems, GA4 does not always highlight them in a separate category. Part comes via regular links, while part looks like direct visits. You can check these data via two options. The first is a quick view of sources in the report if you need to evaluate everything at once. The second is creating your own channel group via a regular expression for detailed and regular monitoring.

Why Tracking AI Search Traffic Matters in 2026

More and more people look for answers not on Google, but on ChatGPT, Perplexity, Gemini, or other similar services, after which they switch to websites. For business, this is a great stream of new users, yet in GA4 it often merges with the general mass. Because of this, if clicks from AI fall into regular referrals, it is almost impossible to properly assess their efficiency.

Separate analysis helps answer practical questions:

  • which AI services already bring visitors;
  • which pages they recommend;
  • how many of such sessions convert into requests or purchases;
  • whether the behavior of these users differs from organic or paid traffic;
  • which topics and materials are worth developing further.

This is especially relevant for B2B, local services, consulting, and other areas where a client can first ask AI to pick several options, and only then switch to a specific website. For such companies, AI Promotion becomes another vector of work with brand visibility in generative search.

At the same time, GA4 will not show all AI search traffic. If a service or browser does not transmit data about a source, Analytics will not be able to determine it retroactively. The setup goal is to separate that part of AI traffic that GA4 can actually recognize.

Step-by-step diagram showing how AI search traffic flows from generative tools like ChatGPT into Google Analytics 4

How to Check AI Traffic in GA4 Using Default Referral Reports

If you are wondering how to check AI traffic in GA4, it is enough to look into data by sources and see if there are already transitions from AI platforms on the website. This is a quick way to assess the current state of affairs before taking on more complex settings.

Step 1: Navigating to Traffic Acquisition in GA4

In the left menu of GA4, consistently open “Reports”, “Acquisition”, and “Traffic Acquisition”. It is this GA4 traffic acquisition report that is responsible for showing sources of traffic entry.

Initially, the table is grouped by standard channels like Organic Search or Referral. To check specific transitions, such a generalization will not be enough.

You need to enter the upper part of the table and select the parameter “Session source / medium”. The GA4 interface sometimes updates or changes depending on the language, but the essence remains clear: you need detailed sources, not summary categories.

Step 2: Filtering Session Source/Medium for AI Referral Domains

Now you can look in the table at the domains that interest you. To begin with, check:

  • chatgpt.com;
  • perplexity.ai;
  • claude.ai;
  • gemini.google.com;
  • copilot.microsoft.com.

You can see in the list something like chatgpt.com / referral – obviously, in this case, GA4 recognized the visit from ChatGPT as a referral visit. Similarly, you can review metrics for other sources as well. This is a simple way to analyze GA4 referral traffic and see if there are transitions from AI resources.

 

This method is convenient for a first analysis, but constantly searching for domains manually is inconvenient. Besides, the list of AI services changes, so for regular work, it is better to create a separate category.

How to Create a Custom AI Search Channel Group in GA4

If the first visits from AI services are already recorded, it is worth separating them into an independent channel via Google Analytics 4 custom channel group. This method is useful for specialists looking for how to track AI traffic in GA4 and wanting to keep these data in order, separately from regular receipts.

Such a group allows uniting several sources under one rule. Instead of separate lines for ChatGPT, Perplexity, Claude, and Gemini, one category will appear in the report, for example “AI”. At the same time, inside the analysis, you can continue to look at specific sources.

This is convenient if AI traffic needs to be regularly compared with Organic Search, Referral, or Paid Search.

Step-by-Step Setup Using Regex Rules

To gather transitions from AI services into one channel, first create a separate rule for them in GA4.

  • Open “Admin” – “Channel groups” and create a new custom channel group. Give it a clear name, for example AI Traffic.
  • Add a new channel to the group, for example AI Assistants.
  • As a parameter for the condition, choose Session source, and for the match type – matches regex.
  • Insert a regular expression with domains of the required AI services:
  • .chatgpt.com.|.perplexity.ai.|.claude.ai.|.gemini.google.com.|.copilot.microsoft.com.
  • The vertical bar | means “or”, while .* allows accounting for additional characters before or after the specified domain.
  • Check the channel order in the group. AI Assistants must stand higher than general Referral so that corresponding transitions fall into the new channel instead of being classified as regular referrals.
  • Save the group. After this, it can be used for channel analysis and comparing AI traffic with other categories, for example Organic Search, Referral, or Paid Search.

Adding ChatGPT, Perplexity, Claude, and Gemini Sources

The provided regex is only a basic option. A set of sources should be selected for real website traffic, as the audience can transition not only from the most popular AI platforms.

To start, it is enough to check the following sources:

  • ChatGPT– chatgpt.com
  • Perplexity – perplexity.ai
  • Claude – claude.ai
  • Gemini – gemini.google.com
  • Copilot – copilot.microsoft.com

If you see transitions from other AI services in reports, you can also append the domains that interest you to the very same regex. As a result, you will see all AI transitions in one channel, but at the same time, if needed, you will be able to look at the number of ChatGPT referrals, or at the features of Perplexity AI traffic or other AIs, separately from others.

This list is better to check from time to time. Services develop, change domains and principles of marking sources, so a custom channel group is worth updating in time so that statistics on AI remain correct.

Measuring Conversions and Lead Quality from Generative AI

The number of transitions by itself does not show how useful a channel is. Two hundred sessions can be more valuable than a thousand if the first group more often leaves requests or buys.

After creating an AI channel, it is worth comparing it with other sources across several metrics:

  • number of sessions;
  • active users;
  • engagement rate;
  • average session duration;
  • key events;
  • number of leads;
  • purchases or revenue, if they are tracked.

Separate transitions from real results

AI conversion tracking should show not only how many people came from a specific service, but also what happened after the transition.

For an online store, this can be a purchase or adding an item to the cart.

For a clinic – making an appointment.

For a B2B company – submitting a form, requesting a commercial offer, or booking a consultation.

For a deeper comparison in GA4, it is convenient to set up a separate exploration to see how the AI channel looks against the backdrop of Organic Search or Referral. In practice, it happens that the number of visits from AI is smaller, yet conversion into target actions is noticeably higher.

If a sale does not happen right away

For B2B or services with a long sales cycle, one GA4 session is not enough. A user can transition from ChatGPT today, leave a request in a week, and become a client even later.

In such a case, the source of the first visit should be passed to CRM along with lead data. This allows checking later where the client came from, even if the deal occurred in several weeks or months.

 

Common Pitfalls When Tracking AI Referral Traffic

Even a properly created channel group will not show the whole picture. Part of the limitations is related to GA4 settings, and part – to the way a user transitions from an AI service.

Do not put the AI rule lower than Referral

If Referral is checked earlier than the AI channel, a session can fall into referrals before GA4 reaches the necessary rule. Therefore, a specialized AI rule must stand higher than general Referral.

Do not limit Regex to a single domain

If you set a rule only for chatgpt.com, other AI services will not be counted. Domains and redirection methods can also change, so review and update the list from time to time.

Do not perceive Direct as completely “direct” transitions

If GA4 does not see a referral source, the system can count the session as Direct. This sometimes happens during transitions from embedded browsers or AI applications. If you do not see chatgpt.com / referral in the list, this does not mean there are no transitions from there at all.

Do not mix GA4 data with data that can be controlled

For your own links, where possible, use UTM parameters – they help mark the transition source in advance. But external mentions and transitions via an environment without a referrer cannot be controlled by a business.

Optimize Your Business for AI Search Visibility

Tracking transitions shows what is already happening. The next question is how to make AI services find and recommend a website more often.

Here generative engine optimization (GEO) comes into play – preparing content and site structure so that information is easy to understand, verify, and use in answers of generative search engines.

To begin with, it is worth checking:

  • whether there are clear answers to typical client questions on the site;
  • whether services, products, and operating terms are clearly described;
  • whether there are specific facts, numbers, and examples, rather than just general ad phrasing;
  • whether pages and headings are logically structured;
  • whether company information is relevant;
  • whether brand data across different platforms are consistent.

Next, combine these data with GA4 analytics: check which pages are already receiving AI traffic, which topics attract attention, and where potential for growth lies.

Want to determine what exactly is worth changing on the site? Contact our team and order Marketing Consulting – together we will define priorities and prepare a plan for further action.