Are ChatGPT, Perplexity, and other LLMs recommending your products and website? To answer this, we need to dive into GA4 and learn more about AI traffic analytics.
Wes say your site is referenced by ChatGPT, Claude AI, or Perplexity when a user wants to know how to fix their broken guitar neck. They click on the URL, and it counts as AI traffic.
Put simply, AI traffic is generated when an AI surfaces your site’s URL and a human clicks it.
Why does AI traffic appear as referrals in GA4?
From GA4’s point of view, AI traffic is simply traffic from a website that is not a search engine listing (organic) or paid search (PPC ads). Our goal now is to make it ‘understand’ AI traffic.
How can you use the data to get recommended by AI?
Tracking is the basis of optimising. We have worked extensively with AI SEO since ChatGPT launched and tracked AI traffic ever since - that was how we got our client’s product recommended by ChatGPT and Perplexity first.
What AI Traffic can’t be tracked by GA4?
A large share of AI‑influenced visits won’t show up as referrals at all. Most people copy and paste links from chatbots into a new browser tab; GA4 records those sessions as “Direct + New User” with no referrer.
Also, when AI tools summarise your content without linking to it, you get no measurable traffic. To approximate this invisible segment, we can segment your direct traffic by new users landing on FAQs, guides, and blog posts. Those spikes often indicate AI copy‑and‑paste behaviour.
That said, there are a few limitations to GA4 when tracking AI traffic:
- Traffic from AI apps does not count as a referral; it counts as direct. GA4 can’t distinguish, for example, traffic from the ChatGPT app from traffic from any other app.
- Traffic from Googlet be tracked as an AI source. A pity, considering there are ways to get your products recommended.
- GA4 can only track traffic from users who accept cookies. It can’t, for example, track traffic from AI agents or any form of agentic AI that doesn’t accept cookies.
This means the actual figures for AI traffic on your site are higher than GA4 suggests. While you can’t track it, what we can see on GA4 is enough to give us a picture of what to optimise - especially if you consider ChatGPT now supports checkout directly in the conversion.
How to track AI traffic from the acquisition tab
This method serves as a quick overview to set the tone. In GA4, go to Dashboard > Acquisition > Traffic. Change the data source to source/medium, and the various AI sources will appear.
In the Traffic Acquisition report, switch the primary dimension to Session Source / Medium, then search for your AI domains to isolate referral traffic from known chatbots.
To capture copy‑and‑paste visits, add an additional segment for Direct + New User sessions landing on content pages. If you’re seeding AI platforms with your content, add UTM parameters, e.g., utm_source=chatgpt) to your links so any click is unambiguously tagged.
This method is easy, but messy, and won’t take you far. You can compare AI traffic to other sources, but you still need to redo it manually, and you won’t be able to have all AI sources line up. There is a more effective way to achieve this.
How to make a custom AI traffic report on GA4
Here is where the gold is. We want a snapshot of all AI sources and add new ones as they come into existence.
Step 1 - Create a new Exploration from GA4 and set the parameters
We want to create an exploration dedicated to AI traffic and apply the following settings
- Set the Dimension to ‘referrers’ (we already know AI traffic counts are referrals)
- Set the Metrics to any you are interested in knowing. It could be sessions, key events, bounce rate, engagement, or anything else.
Beyond sessions, you can include Engaged sessions, Engagement rate, Average session duration, Active users, Conversions, and Revenue. These metrics tell you not just how many visitors arrived via AI, but whether they stick around and generate orders or leads.
Step 2 - Add breakdowns and values
You will simply drag and drop the dimensions and metrics from left to right to build your report, as shown in the screenshot below (step 3).
Step 3 - Filter for only AI traffic
This is the final step where we need to add our filter. The filter at this stage is a Regex - a string of instructions for GA4 we borrowed from our friends at Orbit Media, which you can copy and paste directly into the sections.
To apply our filter, go under Settings, add a filter on Session Source / Medium with the condition set to Matches Regex. Paste a pattern like chat\.openai\.com|chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai, and add any new domains you discover.
Remember that GA4 regex matching is case‑sensitive, so include both uppercase and lowercase variants if needed. For an always‑updated list of domains and a place to log these metrics, download our <AI traffic tracking template> and make a copy of it in Google Sheets.
The Result we have is the report below, which shows all AI sources from the regex. Now you can open Analytics and head to this report immediately to check any new AI activity.
Find out where AI sources land
Once you’ve isolated AI sessions, add the Landing page as a secondary dimension to see which products, categories, or articles are getting cited by chatbots. These insights help you refine page copy, adjust headlines, and optimise FAQs so that AI tools are more likely to reference your content.
Open another tab in the report and simply add a new Dimension>Landing page + query string. Make sure you repeat steps 2 and 3 of the process, and that the results you look for are as follows.
What can we do with the Data?
At this point, you can track what AI traffic comes into your website. The next step is making something out of the data.
You can compare the engagement rate, average engagement time, conversion rate, and revenue of AI referrals against other channels. If AI‑driven visitors convert well, invest in optimising content for LLMs.
You can also create a custom channel group (e.g., “AI Chatbots”) so AI referrals appear separately in your standard reports and dashboards. For ongoing visibility, build a Looker Studio dashboard that tracks AI referrals alongside Direct + New User proxies. This makes it easy to show leadership how AI search is impacting your KPIs.
The table below provides a broad starting point for the action to take based on your reports.
| Insight You Get | What To Do Next |
|---|---|
| AI sends traffic to specific pages | Improve those pages (rewrite copy, add FAQs, fix structure, update product descriptions). |
| AI traffic comes to certain products | Rewrite product pages: clearer benefits, specs, visuals, trust signals. |
| Some pages get good AI engagement | Strengthen CTAs and internal links to move users deeper into the site. |
| Certain questions/topics drive AI visits | Create more content answering those questions (blog posts, FAQs, comparison pages). |
| One AI assistant sends more traffic (e.g., Perplexity) | Optimise content formatting for that AI tool’s style (structured, concise answers). |
| AI visitors don’t convert well | Improve UX, trust elements, and clarity on those landing pages. |
The key is understanding what LLMs want and what humans ask them - a wider take on SEO than ever before.
Conclusion
As AI‑powered search and chat continue to influence how shoppers discover products, the brands that take a disciplined, data‑driven approach to measuring and optimising AI referrals will seize the advantage.
By defining what counts as AI traffic, acknowledging the blind spots in GA4 reporting, and building custom explorations with the right metrics and filters, you can surface high‑intent visitors that your competitors may not even know exist.
Pair this with a repeatable tracking template and a habit of updating your regex list, and you’ll turn AI referrals from a curiosity into a meaningful channel in your acquisition mix.
