How much of your traffic is coming from AI: 2 methods to use today

AI chatbots are reshaping how users find and interact with your website. Instead of clicking through Google search results, users are having full conversations with platforms like ChatGPT and Perplexity, and by the time they land on your site, they already have a sense of who you are and what you offer.

That’s why it’s crucial to know how much of your traffic is coming from AI, and which pages those visitors land on. With that insight, you can tailor your content to meet users exactly where they are and convert more leads into paying customers.

This article gives you 2 methods that shows you how to track AI traffic, including how to tell if the AI traffic your site gets is from real people (or just bots).

Key insights

• Tracking AI referral traffic from tools like GA4 often requires the help of your engineering team to set up regex filters and build other custom reports that encompass every AI platform you need to track. However, Contentsquare offers an out-of-the-box solution where you can configure AI tracking yourself in just a few clicks. 

• 89% of B2B buyers use AI in their buying process, meaning teams that want to stay relevant to their ideal buyers need to track and monitor how much traffic is coming from LLMs. And with AI referral traffic projected to grow, knowing how to measure and analyze your site’s AI traffic gives you a competitive advantage. 

• AI traffic expects high content relevance. Because users receive hyper-personalized answers from AI tools, they expect your page to seamlessly continue that conversation. Understanding which pages attract the most AI visitors helps you provide a personalized journey that delights every visitor who lands on your site.

4 differences between AI and search traffic

There are 4 major differences between AI and search traffic:

1. Prompt-driven discovery: traditional search relies on keyword matching and SEO (think ‘gluten-free chocolate dessert recipes’), whereas AI platforms respond to conversational prompts (like ‘What is a good gluten-free dessert I can make for a large crowd that contains chocolate?’). This shift favors content written in a conversational tone, rather than content optimized for exact-match keywords. 

2. Complex (but shorter) user journeys: users arriving from AI tend to be deeper in the funnel because AI already gave them a lot of the information they needed. As a result, users can have fragmented, non-linear paths that are more difficult for marketers to track.

3. Higher-value visitors: because AI referral traffic is more qualified, visitors from AI are more likely to convert and are 4.4 times as valuable as the average visitor from traditional search (based on conversion rates). 

4. Smaller traffic share but higher conversion: currently, AI referrals count for less than 1% of referral traffic. However, AI traffic is forecast to grow and even potentially overtake organic search traffic by 2029.

Knowing these differences helps you review your product analytics in relation to your AI traffic to understand what’s going on. 

For example, you might notice a high percentage of AI traffic lands on a top-of-funnel (TOFU) page and leaves right away. Knowing that AI traffic tends to be deeper down the funnel, you might add some more information to that TOFU page to satisfy those who are near the middle (or bottom) of your funnel, like links to products.

How to measure how many visitors come from AI

You can track how many visitors come from AI using web analytics tools like Google Analytics and Contentsquare’s Product Analytics tool.

We’ll show you how to with both methods, but first let’s go over 2 important points:

1. Some AI traffic will be labelled as a direct or unassigned traffic source because AI platforms don’t always pass on referrer information. Analytics tools need referrer information to categorize traffic into different acquisition sources, and as a result, your AI traffic is likely higher than what an analytics tool tells you. 

2. Currently, you can’t track traffic from Google’s AI Overviews. Google doesn’t separate web traffic from AI Overviews, but it does separate it from AI Mode, so you can still see how much traffic you get from AI Mode.

Knowing that, let’s learn how to track AI traffic.

How to track AI traffic with Google Analytics

Track AI traffic with Google Analytics 4 (GA4) by configuring regex filters. Regex filters, also called regular expressions, are characters used to match string patterns. For example, ‘gr(a|e)y’ is a regular expression that, when executed, will search for (and find) instances of ‘grey’ and ‘gray.’

1. Start by clicking ‘Reports’ > ‘Acquisition’ > ‘Traffic Acquisition.’

2. Next click ‘Add filter +,’ and change the ‘Dimension’ to ‘Session source/medium’ and ‘Match Type’ to ‘matches regex.’ 

3. Finally, enter the filter below and click ‘Apply’:  

The above regex filter covers major LLMs like ChatGPT, Copilot, Claude, and Perplexity. You can customize the filter to contain other LLMs if needed.

4. After applying the filter, ensure the report is set to ‘Session source / medium.’ You’ll now be able to see how much traffic your site gets from the included LLMs.

There are multiple ways you can slice and dice your GA4 data to include LLM traffic. For example, you can adjust your regex filter to include social traffic from different social media platforms like Facebook and Pinterest, so you can compare how traffic behaves based on where it comes from.

Or you can create a custom AI channel group to compare traffic from AI sources with other channels like social, organic, or direct. However, you need access to ‘Data Display’ within the admin panel to do this, and if you don’t have admin permissions, you’ll need to speak to your engineers to have them set up custom AI channels.

How to track AI traffic with Contentsquare

An even easier way to track AI traffic is with Contentsquare’s Experience Analytics tool, which is an out-of-the-box method that lets you track AI traffic without any complex setup. 

1. Click "Acquisition Analysis” > “Analysis setup,” and click into the “Properties” tab. Click “Edit” and add different LLM referrers.

You’ll then be able to see how much of your traffic is coming from AI. With Contentsquare, you can track traffic from AI without relying on your engineers. You’ll get retroactive data that you can use right away to inform your strategy.

How to track human vs. agent traffic

Agent traffic is when LLM bots crawl your website for information—similar to how Google crawls websites to store in its index for search retrieval.

Analytics tools typically don’t track agent traffic, and, instead, you’ll need to access your site’s server logs located on your hosting environment (such as cPanel or your content delivery network).

Here’s an example of a server log showing a record of an agent bot crawling a website: 

But there’s a catch: sometimes LLMs will visit your site in such a way that triggers your analytics code, so you can sometimes view bot traffic via your analytics.

This happens when LLMs use headless browsers to visit your site. A headless browser is basically a browser that loads your page the same way a human would, but without a visible window. Since headless browsers fully load your page, they also execute JavaScript (i.e. your analytics tracking).

When this happens, you may see a spike in your AI referral traffic with behavior that differs from your human traffic, such as high bounce rates and low time spent on page. So, look for oddities in your traffic metrics to determine whether the traffic came from agents or humans. 

Audit and improve pages with AI referral traffic

Once you know which pages get the most traffic from AI platforms, audit those pages to see whether visitors are actually finding what they need.

Use Contentsquare’s free Heatmaps tool to determine how far users scroll down the page and what areas capture their attention.

Contentsquare gives you 5 different types of heatmaps to analyze traffic 

And launch surveys to ask visitors to rate their experience. Then, prioritize improving pages with low interaction and ratings, and apply any user feedback you receive from your surveys.

FAQs about tracking AI traffic