How to set up and analyze AI-referred traffic in Google Analytics 4

On October 31, 2024, OpenAI launched web search, marking a major shift in how users find information online. While Perplexity had already integrated web search into its chat interface, OpenAI’s release made the concept mainstream. With AI-driven search on the rise, marketers and business owners must start treating AI-referred traffic as its own channel. In this article, we explore three ways to analyze and track AI-driven traffic effectively.

AI chatbots like ChatGPT and Perplexity aren’t just growing—they’re changing how people discover content. Search engines have long been the gateway to websites, but AI-driven recommendations are now reshaping traffic patterns faster than expected. Just a few weeks ago, I predicted this shift, but little did I know the future would arrive this quickly. AI-driven referrals are already happening, and websites are seeing more traffic coming from answer engines than ever before.

The question is: Are you tracking and analyzing this new source of traffic effectively?

In this article, we lay out 3 steps you can take in your GA4 account to start analyzing AI-referred traffic. Here is what we are covering:

Who and when should you analyze AI referred traffic?

Tracking AI-referred traffic makes the most sense if your audience is already engaging with AI-powered tools. If your users are tech-savvy, professionals, or early adopters, AI-driven referrals could already be shaping their online journeys.

On the other hand, if your audience is less AI-inclined or you don’t see AI traffic in GA4 yet, it’s not an immediate priority—but you should still keep an eye on the trend.

Check if you’re getting AI Traffic in GA4

In your GA4 account, go to Reports > Acquisition > Traffic Acquisition, and change the Primary Dimension to Session Source.

Do you see traffic from sources like chatgpt.com or perplexity.ai?
✅ If yes, it’s time to start tracking AI traffic separately.
❌ If not, you may not need to take action yet—but it’s worth checking periodically.

An incomplete list of several popular answer engines are given below – it includes the current most popular ones.

How should you analyze AI referred traffic?

Like with any other channel, there are several ways you could isolate AI traffic for analysis. Here are some of the ways:

  1. Add filters to your exploration reports
  2. Create a Custom channel grouping (mostly for attribution reports, works on historical data too)
  3. Create Segments
  4. Bonus: Create audiences

Before we dive into these specific implementations, let’s talk a little bit about which AI tools are major and most likely sending traffic to websites (popular opinion). As mentioned earlier, this isn’t an exhaustive list but at the time of writing these are some of the most popular tools. Moreover, the top 4 are more popular than the rest. Here is the list:

List of Popular AI tools/Answer Engines (including Multimodal Engines). (Feb 2025)

A single Regex (Regular Expression) to capture all AI-referred traffic in GA4

To group all AI-referred traffic in GA4, we’ll use Regular Expressions (Regex) to match known AI referrers.

Here’s the Regex pattern we’ll use:

chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|meta\.ai|you\.com|blackbox\.ai|chat\.mistral\.ai|felo\.ai|arc\.net|andisearch\.com|komo\.ai|waldo\.fyi

This pattern captures multiple AI platforms in a single rule, allowing GA4 to filter them under one category.

Why Regex Works?

Regex (Regular Expressions) is a pattern-matching technique used in analytics, programming, and search functions. GA4 supports regex-based filtering, which is why we’re using it here.

Two key Regex characters we’re using:
Pipe (|) → Acts as an OR operator, allowing us to match multiple AI sources.
Backslash (\) → Escapes special characters like . to ensure accurate matching (e.g., chatgpt\.com).

Pro Tip:
Always test your Regex before applying it in GA4. Try Regex101 for quick validation.

Question for you!
If a new AI tool starts sending traffic (e.g., deepseek.com), how would you change the Regex? Write in comments ;).

Now that our regex is ready, lets go ahead and start implementing

Adding filters to GA4 Exploration

This is the quickest way to analyze AI traffic in your existing reports. Simply head over to the exploration you are interested in. Add the Session Source dimension (if you don’t already have that in the report), drop it into the filter as shown below:

Session source
matches regex
And now just paste the regex (we created above) in the blank field.

chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|meta\.ai|you\.com|blackbox\.ai|chat\.mistral\.ai|felo\.ai|arc\.net|andisearch\.com|komo\.ai|waldo\.fyi

Click Apply and your report is now filtered for only AI traffic!

Create Custom Channel Grouping

Companies already seeing high volumes of AI-referred traffic are paying attention. And they should—because unlike Paid or Organic, where we have control over visibility (budgeting, targeting, SEO tweaks), AI-driven traffic is unpredictable.

We don’t know what prompts triggered the visit. We don’t even know if it’s possible to optimize for AI visibility, the way we do with SEO. Answer Engine Optimization (AEO) is emerging, but no clear playbook exists yet.

So, should we track this traffic separately? Yes—and here’s why.

AI-generated referrals represent a fundamentally different user journey than search, social, or ads. Unlike a Google searcher who actively selects a result, AI-referred users are handed a single or a few recommendations—which may influence their expectations, engagement, and conversion behavior.

Without separating this traffic, we risk misinterpreting performance metrics.

  • Are AI users more likely to bounce because they trust the AI’s summary instead of clicking deeper?
  • Do they engage with different types of content than search-driven visitors?
  • Are they more likely to return after their initial visit?

These patterns won’t be visible if AI traffic is lumped into, Direct, or Referral traffic.

Therefore, creating a Custom Channel grouping with AI as a separate group for your reports already makes sense ensures that we are not just reacting later—but preparing now.

Here is how to do that:

    1. Go to GA4 Admin > Channel Groups (under Data Display) > Click on the blue “Create a new channel group” button
    2. A new window pops up. Give the new channel group a name (something like Including AI Channel) or whatever you like.
    3. Add a description if you want to. “Traffic from AI tools/chatbots extracted from Referrals and grouped under AI Channel. Rest is the same as Default Channel grouping”.
    4. Click on the “Add New channel” button. Another window slides out. Give this a name – lets call it “AI Traffic”.
    5. Click on “Add conditions”.
      • Under “Match AT LEAST ONE rule in this group”, select Source dimension, choose matches regex and copy-paste the regex.
        chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|meta\.ai|you\.com|blackbox\.ai|chat\.mistral\.ai|felo\.ai|arc\.net|andisearch\.com|komo\.ai|waldo\.fyi
      • Click Apply. It’ll look a bit weird as shown in the image, but thats alright.
      • Click on “Save Channel” button
    1. As this window slides shut, the one previously slided-out reveals again with the list of all channels in your new grouping. Your newly created channel “AI Traffic” is at the end of the list.
    2. Click on the “Reorder” button next to the search box.
    3. Drag your “AI Traffic” channel aboveReferrals”.
    4. Click the “Apply” button
    5. Click “Save Group” button on the top right corner.

Why do we drag it above Referrals?

The order of various groups is important. Think of GA4’s Custom Channel Grouping as a set of traffic buckets where each visit to your website is placed into the first bucket (channel) that matches its rules. GA4 checks channels in order from top to bottom. When it finds the first match, it stops looking and assigns the traffic to that channel. Even if the visit could match other channels below, it won’t be checked further. If our new channel was below referrals, the traffic will get assigned to Referrals because that would be the first match. We know that AI traffic comes through as referrals.

Now that you have the new custom channel grouping created, you can analyze AI traffic by selecting the new Channel grouping you just created. For example, in your Traffic Acquisition report, you can now select your new grouping.

When you do that, your AI traffic channel shows up even for your historical data!

Create Segments

To create Segments, we again use the Regex.

Difference Between Analyzing via GA4 Segments vs. GA4 Channel Grouping

GA4 Channel Grouping is session-scoped, meaning it categorizes traffic sources per session and helps analyze traffic behavior at a session level. However, it does not track users across multiple sessions.

In contrast, GA4 Segments provide greater flexibility by allowing you to define rules based on user behavior across multiple sessions.

For example:

  • You could create a User Segment to find users who had more than one session, where at least one came from AI traffic.
  • Or, you could define a First User Segment to analyze users whose first-ever session originated from AI traffic.

Once you create a Segment, they are available across all tabs in GA4 Explorations, making them more versatile than filters. You can also create Segments from the admin section and make them available to all users of your GA4 account.

Before you create a Segment – pause and decide

Pause before proceeding and think of:

  • What do you want to analyze?
  • Should the segment focus on users or sessions?
  • Who else on your team (e.g., performance, content) can provide input?

It’s best to decide before setting up the segment to ensure you’re extracting meaningful insights.

In the example below, we’ll create a User Segment that isolates users who had at least one session referred by AI traffic across all their sessions. We’ll then apply this segment to our Traffic Acquisition report (Default Channel Grouping) and see what we find.

Let’s create a segment:
Go to an exploration tab where you would like to apply the AI traffic segment

  1. Click the + sign next to Segment in the left hand navigation – a new window will slide open
  2. Click Create New Segment – another window slides open.
  3. Click on the User Segment – another window slides open
  4. Give the segment a name – I am calling it “AI referred Users”. Add a description if you like. “Users that came through AI channels like chatGPT
  5. Right across “Include users when” on the other side of this box is a small person icon. Click on that and choose the first option “Across all sessions
  6. Click on “Add new condition” and search for “Session source” – you will see several results. Choose the one next to Traffic Source > Session source.
  7. Click on Add Filter > Choose the condition “matches regex” > copy paste the regex we wrote earlier
    chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|meta\.ai|you\.com|blackbox\.ai|chat\.mistral\.ai|felo\.ai|arc\.net|andisearch\.com|komo\.ai|waldo\.fyi
  8. Keep the checkbox “At any point in time” checked.
  9. As you hit Apply, the right sidebar will show you a preview of your applied segment.
  10. If satisfied, click on the blue “Apply” button on top right corner.

What does this report tell us after applying the Segment?

  1. Some users who arrived via AI referrals also visited through other channels like Organic or Paid at some point.
  2. User journeys are not linear—AI is just one touchpoint in a broader conversion path.
  3. AI referrals introduce a new touchpoint for multi-touch attribution, making cross-channel analysis even more delightful 😉

What can we investigate further?

Users who came from multiple channels within the selected date range can fall into two categories:

    • AI was their first touchpoint → Did they later search on Google before returning? Did they come back directly?
    • AI was a later touchpoint → What triggered their visit? Did they start from an ad or a search before AI led them back?

Understanding these patterns can reveal how AI traffic interacts with other marketing channels—and whether it accelerates conversions or simply supplements existing acquisition efforts.

There are a lot of options for creating a segment as you’ll notice in the same window – like adding sequences of actions, adding exceptions, two conditions at a time etc. Feel free to explore and set up whatever is most relevant for your use case.

 

For creating property level segments that other users can also use in their explorations,, head over to Admin > Segments (under Data display) > Click on “New Segment” button and follow the same process as above.

Bonus Tip: Create Audiences

Once you’ve analyzed AI-referred traffic and identified the best ways to engage these users, the next step is to create audiences for deeper insights and remarketing.

How to Create an Audience from AI Traffic in GA4

  1. Go to the report where you created your AI segment.
  2. Click the three dots next to the segment and select “Create Audience.

OR

  1. Go to Admin > Segments (under Data Display).
  2. Click the three dots next to your segment and select “Create Audience.

Taking AI Audience Targeting to the Next Level

At TrackFunnels, we help businesses go beyond standard GA4 tracking—like capturing events that don’t get tracked by default, such as email opens.

Why does this matter?

  • Knowing a user was referred by AI is useful.
  • But knowing they also opened your email? That’s a powerful retargeting opportunity.
  • Imagine an audience of users who were referred by AI AND engaged with your emails—your messaging to them would be very different than if you only knew their source.

With TrackFunnels, you can start tracking email opens in GA4 in just a few clicks—unlocking richer data for smarter audience building. Find out how to get set up today!

Good to know: GA4 backfills audiences with up to 30 days of data, so if you’ve already received AI-referred traffic in the past month, your audience will be populated within 24-48 hours.

Challenges

AI is no longer a futuristic concept—it’s already shaping how users discover content. While we don’t need to be philosophers to predict that AI-driven traffic will only grow, analyzing it today gives us a critical early advantage. By understanding how AI-referred users behave now, we’ll be far better prepared to engage them when AI becomes a dominant traffic source.
However, there are still some key challenges:

  1. We don’t know what Prompt led to the visit
    Unlike search engines, where we can analyze keyword data, AI engines do not pass the user’s original query when they refer traffic. Even if they could, prompts are often long and complex, making them harder to track in a structured way. Technically, an API solution could exist in the future, but this remains speculative for now.
  2. Many AI Platforms Do Not Send Referrer Headers
    Not all AI chat tools provide referrer data when sending users to websites. In our tests:

    • Only 6 AI tools we categorized under referrals by GA4
    • The rest were classified under Direct traffic or Unassigned, making AI attribution challenging.This lack of standardization creates a growing industry-wide issue, where more AI-driven visits will appear as Direct traffic—limiting visibility into AI as a source. A standardized protocol for AI engines to pass referral data (similar to UTM parameters in traditional search) would solve this, but no such standard exists yet.
  3. The Future of AI & Website Interactions is Still Unclear
    There are ongoing discussions about how websites should structure data for AI crawlers and agents. Just like we have sitemap.xml for SEO and robots.txt for crawler control, there’s speculation that AI engines might soon use a dedicated file or structured data format to interact with websites. Dharmesh Shah (HubSpot co-founder) recently suggested that websites should provide an interface for AI agents to interact—a concept that could reshape AI visibility.

Expect AI agents to dominate discussions in 2025.

Whether it’s structured AI feeds, API-based discovery, or an entirely new framework, businesses that monitor these developments closely will gain an early-mover advantage.

Okay, we now have three practical ways to start analyzing AI referred traffic today. You also know why it’s important and will be even more in the future.

Let me say this one more time – while AI-driven discovery is still evolving, businesses that begin tracking and understanding this traffic now will be better prepared for the future.

If you have any questions, think I missed something, or want me to cover a specific topic next—drop a comment! I’d love to hear your thoughts.

0 Comments

    Leave a Reply