Measuring traffic from AI chatbots in GA4 & key metrics to analyze

5-minute setup

Quick one today: just sharing a GA4 setup that helps you track traffic coming from AI chatbots.

If you don’t have Google Analytics set up for your site, you definitely should. Here’s how to do it.

Before we dive in - let’s connect on LinkedIn.

How to set up custom view/exploration in GA4

1. Create new custom exploration:

2. Add dimensions & metrics:

3. Add them to exploration:

   1. Rows: Page location

   2. Columns: Session source

4. Add a metric to values/exploration:

   1. Start with Total users, and swap it out or experiment with other metrics as needed

5. Filter out AI traffic

   1. Filter: Session source matches regex

   2. Copy this:

(?i)(.*gpt.*|.*chatgpt.*|.*openai.*|.*neeva.*|.*writesonic.*|.*nimble.*|.*outrider.*|.*perplexity.*|.*google.*bard.*|.*bard.*|.*edgeservices.*|.*gemini.*google.*)

You’ll end up with something like this:

Different metrics to include:

If you're an SEO or already analytics-savvy, this section might not offer much new. But for my non-savvy friends, here are some useful metrics you should check out:

• Total users – how many users came from AI tools

• Event count – are those users interacting with your site?

• Engagement rate – how engaged are their visits? Compare this to your site's average to assess quality.

• Average session duration – same idea as above: longer sessions may mean more meaningful visits

• Ecommerce purchases – how many purchases these users made

• Total revenue – total revenue generated from these visits

Extra tips for measuring AI traffic:

Export the list of URLs and run it through Screaming Frog (or a similar SEO tool) to check if all the links are valid and not returning 404 errors.

AI can often be wrong or hallucinate links, so this step lets you identify any issues and take action—like setting up redirects—to ensure traffic lands where it should.

Let’s connect on LinkedIn and please do share any tips you might have!