What Is ChatGPT Referral Traffic?

ChatGPT referral traffic is the set of visits to your website that begin when a person clicks a link inside ChatGPT. For a health practice, it is the measurable slice of AI influence, because it shows up in your website analytics. It counts clicks only, so it understates AI's reach. A patient who reads your name in an answer and later searches for you directly appears under a different source.

How ChatGPT referral traffic works

A referral is a visit that arrives from another website or app, and analytics tools record where it came from. Google Analytics 4 calls this the session source. For ChatGPT, the visit starts when a patient reads an answer, sees a link to your site and clicks it.

Three steps have to happen before that click exists, and a gap at any one of them explains a low count.

• ChatGPT has to be able to show your site. OpenAI says OAI-SearchBot is the crawler used to surface websites in ChatGPT's search features, and that sites opted out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. So a blocked bot closes that path.
• ChatGPT has to choose your page. Whether it names or links you depends on your content and on what the rest of the web says about you. A source link in an AI answer is called an AI citation.
• The patient has to click. Some people take the name from the answer and skip the link.

Where does the data live? Google describes the Traffic acquisition report in Google Analytics 4 as a report that shows where your visitors come from. You reach it from Reports, then Acquisition, then Traffic acquisition. Session source and medium is a dimension in that report, defined as the source and medium associated with a new session. If you do not see the report, Google says it may have been removed or hidden from your default view, and an editor or administrator can add it back.

A caution on labels. When visits do arrive, your analytics may list the source with a name like chatgpt.com. I have not confirmed from an OpenAI page whether ChatGPT adds a tracking tag to its links, so I will not claim one. The reliable approach is to look at the source names in your own reports, rather than build a filter on an assumption.

The data also has limits. If you want a routine to pair it with, see prompt tracking. Together they show both what the tools say and what patients clicked.

Judge these visits by what they do, not only by how many there are. In my judgment, visitors who click through from an answer may have already read a summary of your practice, so a small group can behave differently from ordinary search traffic. In Google Analytics, compare the chatgpt source against organic search on engaged sessions and on booking requests, using the same 90-day window. Small counts support no firm conclusion, but a pattern across a few pages shows which content is doing the work.

A click is the end of a chain, and some AI influence does not reach it. For the bot that opens the first link of that chain, see OAI-SearchBot.

For the wider picture of how often you are named, see AI visibility.

The source-link idea is explained under AI citation.

Example: ChatGPT referral traffic at a health practice

Think of a med spa that checks its Google Analytics 4 account once a month. This example is illustrative, not a transcript of a real account, and the numbers are made up to show the method.

In the Traffic acquisition report, with the date range set to 90 days, the owner searches the source column for "chatgpt" and finds one row. It shows six visits, all landing on the Botox page, and the owner's conversion event shows two of those visitors requested a consultation. The owner writes the figures in a spreadsheet next to organic search, paid ads and referrals, with the month and the landing page.

Six visits is small, and the owner treats it that way. A count that small can swing from month to month. What it does tell the owner is useful. One page is being linked, and it is the page with the clearest answer about pricing and recovery. That suggests building the same kind of answer into the filler and laser pages.

Now some real context on what pages get used. The MGMA published an analysis in June 2026, written by the founder of a commercial service, of 33,340 ChatGPT citation events across 200 practices in 20 metros and 10 specialties, run with GPT-5.5 through the API in May 2026. In that data, practice-owned websites were about 23% of cited sources, behind hospital staff rosters at 27.5% and specialty association credentialing pages at 22.1%. Google Business Profile had zero direct citations, which the author attributes to Google not exposing reviews as crawlable URLs. This was an API study from one author with a commercial interest, so treat the percentages as a hint, not a rule.

The takeaway for the med spa is practical. Your own website is one of several places ChatGPT cites, so the chatgpt source in your analytics is worth tracking. But other surfaces get cited too, and clicks from those do not show up here.

The med spa also adds one question to its intake form, "How did you hear about us?", with an option for an AI tool, to catch the patients who saw its name in an answer and searched later.

How to check yours in 60 seconds

1. Open Google Analytics 4 for your website.
2. Go to Reports, then Acquisition, then Traffic acquisition.
3. Search the source column for "chatgpt".
4. Set the date range to the last 90 days and note the visit count and the landing pages.

If nothing appears, that may be real, so check your robots.txt for OAI-SearchBot next. Also check that the report has data for the full range, since a tracking tag installed last month will not show older visits.

What to do about it

1. Confirm OAI-SearchBot is allowed in your robots.txt.
2. Save the chatgpt source count monthly, in a spreadsheet next to your other channels.
3. Record which landing pages receive visits, since those are the pages being linked.
4. Set up a conversion event for booking requests and calls, so you can see what those visits do.
5. Add a "How did you hear about us?" field to your intake form with an AI tool option, to catch patients who saw your name and searched later.
6. Compare the numbers with your prompt tracking results each month.
7. Look at the source report by name, and do not let these visits sit unlabeled in direct or other traffic.
8. For the rules behind step 1, read AI bot blocking.
9. Review the monthly numbers with whoever runs your site and your front desk, so both teams know which pages bring in ChatGPT visits.