How to See Visits From ChatGPT and Other AI Assistants in Your Analytics

A new client mentions that ChatGPT suggested you. You open your analytics to see how many others came the same way, and there is no line for it. Just "direct", "organic search", "referral" and a long list of sources you have never read.

Here is how to see ChatGPT traffic in analytics, and visits from the other assistants, using reports you already have. The first look takes about five minutes. Setting it up to stay visible takes about fifteen more.

Below is the short answer, where each kind of AI visit lands in your reports, the quick check, how to give those visits their own line in Google Analytics, what Google's own AI answers hide, and how to read a number that will always be too low.

How do you see ChatGPT traffic in analytics?

Open the report that lists where your visitors came from and search it for "chatgpt". Visits from links inside ChatGPT answers arrive labeled chatgpt.com. Then search for the other assistants by name. That is the whole method. Everything else is making the result easier to find next month.

It works because OpenAI labels the links. Its help page for publishers says ChatGPT "automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs". A UTM parameter is a short tag on the end of a link that tells your analytics where the visitor came from. You did not have to add it. It is already there.

The same page ties this to letting OpenAI's search crawler read your site. If your site blocks it, you may have little to count.

Where an AI visit lands in your reports

A visit that began with an AI answer lands in one of three ways: labeled as coming from an assistant, counted under another name, or counted with no AI label at all. Only the first is easy to see. Knowing which is which stops you reading the visible part as the whole.

ILLUSTRATIVE: HOW EACH VISIT IS LABELED, NOT HOW MANYTHE LINE OF SIGHT

1. Labeled as AI

   Clicks a link in a ChatGPT answer

   Shows up aschatgpt.com

   OpenAI says it tags these links.

2. Labeled as AI

   Clicks a link in another assistant, in a web browser

   Shows up asReferral from that assistant, when one is passed

   Look for perplexity, claude, gemini, copilot.

3. What your analytics can label ends here

4. Counted under another name

   Clicks a link in a Google AI Overview or AI Mode

   Shows up asgoogle / organic

   Counted, but not separated from ordinary search.

5. No AI label

   Arrives with no referrer and no tag

   Shows up asDirect

   Nothing says an assistant was involved.

6. No AI label

   Reads your name in an answer, searches it later

   Shows up asOrganic search or Direct

   The answer did the work. The report credits the search.

AnswerHalo answerhalo.com

CLICK TO ENLARGE

The figure carries no counts on purpose. We have not measured what share of AI-started visits falls in each group, and it will differ by business. What it shows is the labeling, which is the same for everyone.

The bottom two rows are why the visible number is a floor. Why that part of the trail breaks, and what to do about it, is the subject of whether you can see how many clients come from AI answers. This post stays with the part you can see.

The five-minute check

Open your source report, widen the date range to the last 90 days, and search for each assistant's name in turn. Write down the visits and the pages they landed on. That gives you a first count before you change any settings.

In Google Analytics, that report is under Reports, then Acquisition, then Traffic acquisition. Change the first column from the default channel group to "Session source / medium" and use the search box above the table. Other analytics tools have the same report under a name such as Sources or Referrers.

Finding nothing is a real result. It can mean no assistant links to you yet, or that people read the answer without clicking. The report cannot tell those apart.

Give AI visits their own line in Google Analytics

Create a custom channel group with a channel called "AI assistants", so those visits stop being scattered across referral and other lines. Google publishes the steps and an example pattern, and the new channel applies to past data as well as new visits.

The steps, from Google's help page on custom channel groups:

1. In Admin, under Data display, click Channel groups, and create a new group.
2. Add a channel and name it "AI assistants".
3. Set its condition to source "matches regex", and paste a pattern that lists the assistants.
4. Move the new channel above Referral in the list.

A regex is a search pattern. A simple one for this job is chatgpt|openai|perplexity|claude|gemini|copilot, which means "any source containing one of these words". Google's page gives a longer pattern of its own. Whichever you use, check that it lists every assistant you care about, and add any that are missing.

Step four matters. Google's page says traffic "is included in the first channel whose definition it matches", in the order the channels are listed. Leave your new channel at the bottom and Referral claims those visits first.

Three limits from the same page. You need the Editor role on the property. A standard property allows two custom channel groups. And the page says custom channel groups "can be applied to your reports retroactively", so you do not lose history by setting this up late.

What Google's own AI answers hide

Visits from Google's AI Overviews and AI Mode cannot be separated from ordinary Google search visits. They are counted, in both your analytics and Search Console, but under Google's name with nothing to mark them as coming from an AI answer.

Google says so directly. Its guide to AI features states that sites appearing in AI Overviews and AI Mode "are included in the overall search traffic in Search Console", in the Performance report "within the 'Web' search type". There is no filter for them.

So no setting fixes this. What you can do is watch the pages. If a page that answers one specific question starts getting more search visits while its position in ordinary results has not moved, an AI answer may be part of the reason. That is a hint, not proof, and we would not report it to anyone as a measured number.

What the number means, and what it misses

The number you now have counts people who clicked a link inside an AI answer and landed on your site. It is real and it is a floor. It does not count buyers who read your name and came back another way, and it says nothing about answers you were left out of.

Expect it to be small. For a solo practice, a handful of visits a month is normal, and a handful is too few to read a trend from. One week with six visits against one with two is not growth. Look at a quarter at a time, and read how to tell if your AI visibility is getting better or worse before you draw a conclusion from a small count.

The landing pages are worth more than the total. They tell you which of your pages an assistant chose to link, which is the closest thing you will get to seeing what it found useful. Add them to the same monthly log described in how to track your AI visibility without buying software.

And some businesses can skip this. If you take on four clients a year and each one tells you how they found you, your intake conversation already beats any report.

What analytics can never show is the answer you were not in. The buyer asked, three other names came back, and nobody visited you. Seeing that takes asking the questions yourself. The free check does it for you: it asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout with your score, who got named in your place, and the one fix to start with. It runs as soon as you ask for it, and the report lands within the hour between 7am and 5pm Central, or by 10am the next morning outside those hours.