How do you recognise visitors from ChatGPT in your analytics?
Recognising ChatGPT visitors by referral traffic, why that's a lower bound, and how to turn it into a fixed monthly figure.
Lees dit in het Nederlands →$ filter --source=chatgpt.com --period=month
How do you recognise visitors from ChatGPT in your analytics?
You recognise visitors from ChatGPT in your analytics as referral traffic from chatgpt.com and chat.openai.com; for Perplexity that's perplexity.ai and for Gemini gemini.google.com. Some arrive as direct traffic, so what you measure is a lower bound. Turn it into one fixed monthly figure alongside your search traffic. This piece is about measuring, not about getting more visitors.
You measure AI visibility in two places: in the answers themselves, and in your own visitor figures. This piece is about the second place, because that's where the proof lies that a mention actually brings someone to your site.
01 / the originWhere you see your AI visitors
In almost every analytics package you'll find AI visitors under referral traffic, also called referrals. A visitor who clicks your link in an AI answer arrives with the sender's name attached. These are the names to look for:
| Platform | Referrer in your analytics | Note |
|---|---|---|
| ChatGPT | chatgpt.com | Older traffic may still arrive as chat.openai.com; count both |
| Perplexity | perplexity.ai | Shows sources with links for every answer |
| Gemini | gemini.google.com | Separate from your regular Google search traffic; that remains visible separately |
This works the same in every package, from Google Analytics to cookieless analytics tools: the referrer is data the browser sends along, not a feature of the tool. So for this measurement it doesn't matter which package you run, as long as it can show referring domains.
The first step is to look up those names in your referral report. The second step is to turn it into a fixed segment or filter, so you don't have to search again every month. Every common package can group referrers by domain name; this doesn't require more technical skill than that.
Do keep the list up to date, though. AI services occasionally change domain name, as ChatGPT did with the switch from chat.openai.com to chatgpt.com, and new providers keep appearing. So once a quarter, go through your referral report for new names you recognise as AI services and add them to the segment. Note when you do this, so a jump in the figure stays explainable.
02 / the lower boundWhy part of it stays invisible
What you see in that report isn't everything. Some clicks from AI conversations don't send sender data along, for example from apps, and those visitors then land in the direct traffic category, as if someone had typed your address in themselves.
The measured AI traffic is therefore a lower bound: the actual number is higher, and no one can say exactly how much. That's no reason to skip the measurement. A lower bound that you measure the same way every month reliably shows whether the share is growing, and that movement is what you want to track.
What you see in your analytics is the lower bound of what AI does with your brand.
03 / the monthly figureOne number, measured the same way every month
Make it simple for yourself and build it into your monthly routine as a fixed set of steps:
- Create a segment or filter once with the AI referrers from the table above, so you don't have to search again every month.
- Each month, add up the visitors from that segment and note that one figure, with the date, alongside your organic search traffic.
- Use exactly the same list of referrers every month, otherwise you're comparing apples and oranges; only expand the list deliberately through the quarterly check, and note when you do.
- Each month, also check which pages those visitors land on. AI referrals usually don't land on your homepage, but on the page that carried the answer; that tells you which content is doing the work, and therefore where expanding it pays off most.
The absolute figure says little at first; for most businesses the share is still small right now. The trend over months is the number that counts. If AI traffic grows while your search traffic stays level, your visibility is shifting from results page to answer, and that's something you want to spot early.
04 / the interpretationWhat the figure does and doesn't tell you
The monthly figure proves that AI systems are referring people to you. What it doesn't show: how often you're mentioned without anyone clicking. An AI answer that names you but doesn't produce a click leaves nothing in your analytics, even though it does put your brand in front of someone. Why that happens more often is covered in you lose visitors to AI answers.
Your picture only becomes complete with two measurements side by side: your visitor figures for the clicks, and a fixed set of search questions for the mentions. How that second measurement works is covered in how do you measure AI visibility. If you get barely any visitors from AI and you're also not mentioned, you know where the work lies; if you are mentioned but without clicks, your visibility is in order and the visibility of your contact details is the point of attention.
Under what name do I see ChatGPT traffic in my analytics?+
Why does part of the AI traffic show up as direct?+
How much AI traffic is normal for a site?+
Can I see this in every analytics package?+
Does little AI traffic mean I'm invisible in AI?+
Do you know how many of your visitors already arrive via an AI assistant?
The scan
This is the baseline measurement: fixed search questions on ChatGPT, Gemini, Claude, Perplexity and Google. Current prices are on the pricing page. Even without a follow-up, you keep the report.
This text was produced with AI assistance and checked and approved for publication by a human.