How to Track Gemini Referral Traffic in GA4 After Google Added UTM Tags

Google's Gemini has started adding UTM tags to some outbound links, and Google has not documented it. To track the traffic in GA4, check where those visits land by default, because GA4 sorts them by exact rules. A tag with the medium organic would count as Organic Search, not the AI Assistant channel.

This guide is for B2B SaaS marketers who report on AI search traffic and want numbers they can defend. Cekpr is skeptical of any tactic or metric that arrives without data, and that applies here. Below is what is confirmed, what is only reported, and a short test plan you can run in your own GA4 property before you put a Gemini number in a deck.

What Is Actually Confirmed About Gemini UTM Tags?

Very little is confirmed, and that is the honest starting point. Search Engine Journal reported on October 1, 2026 that a Reddit user noticed Gemini adding UTM parameters and that Google's John Mueller replied that he saw it too.

The report does not name the parameters or their values, and it calls Google's implementation undocumented. Mueller also asked whether Gemini does not already pass a referrer, and said he was happy to forward reproducible details to the team.

Some third-party blogs say the tags are utm_source=gemini and utm_medium=organic. Treat that as a lead, not a fact. Check your own landing-page URLs and logs to see what actually arrives, in a clean browser, before you write the values down in a spec.

How Does GA4 Sort Traffic by Default?

GA4 sorts traffic into channels using exact rules, and Google publishes them. The Default channel group help page says Direct requires the source to exactly match (direct) and the medium to be (not set) or (none). Organic Search requires the medium to exactly match organic, or the source to match a listed search site.

The same page defines an AI Assistant channel. It covers visits from AI chat sources such as ChatGPT, Gemini, DeepSeek, Copilot or Grok, excludes Google's AI Overviews and AI Mode, and requires the medium to exactly match ai-assistant. Google's May 13, 2026 release note says this value is assigned automatically when the referrer matches a recognized AI assistant.

Put those rules next to the reported tags and a problem appears. If a Gemini link carries a medium of organic, GA4 would file the visit under Organic Search. It would not go under AI Assistant. That is a hypothesis you can test, not a result.

Why Could Your Organic Search Numbers Change?

Organic Search could rise from Gemini visits that used to be Direct or AI Assistant. That would look like an SEO win when part of it is a relabeling. The risk is a false story: leadership sees organic growth and credits a content push that had nothing to do with it.

The opposite can happen too. If your AI Assistant channel shrinks because some visits now carry a different medium, a report may show AI traffic falling when it only moved. Annotate the date, and compare like with like.

Mobile makes everything murkier. Visits that start in an app often lose their referrer, and GA4 puts them in Direct. A UTM tag can rescue some of those visits. Untagged visits remain invisible, which is why tagged traffic should be treated as a minimum.

How Do You Test Where Gemini Visits Land in Your Property?

Run a simple before-and-after test in an exploration. The aim is to see whether any traffic matches the reported tag and which channel GA4 assigns it. You need no new tools, only an hour and your own data.

Use this plan.

Agency write-ups suggest similar steps. Derivatex's GA4 guide to Gemini UTM parameters recommends comparing the weeks before and after the change, and it tells readers not to add Gemini UTMs to their own links or sitemaps. That second point is worth following, since it keeps your own URLs clean.

How Do You Report Gemini Traffic Honestly?

Report three numbers and label them. The first is sessions that carry a Gemini tag. The second is sessions that GA4 files under AI Assistant with a Gemini referrer. The third is the combined total, with a note that it is a floor.

The write-up from Vizup makes the same point in its title. Gemini UTM tagging is a better attribution signal, but not a complete one, because Google has not said which Gemini surfaces add tags or when. So a visit with no tag is not proof that it came from somewhere else.

Say so in the report. A footnote that reads "tagged and referrer-detected visits only, coverage undocumented" costs nothing and protects you from a hard question later.

Should You Build a Custom Channel Group?

A custom channel group is a reasonable fix if you want one AI line in your reports. GA4's default group cannot be edited, but a custom group can place a rule above Organic Search that matches Gemini sources. That keeps tagged Gemini visits from disappearing into organic.

Build it carefully. Order matters, because GA4 applies rules from top to bottom. Match on source, and document the rule so the next analyst knows why Gemini traffic is separate from Organic Search.

Also keep a copy of the old numbers. Changing channel logic mid-quarter breaks comparisons unless you note the date and show both views for a period.

Can Server Logs Settle the Question?

Server or CDN logs can confirm what actually arrived, which makes them a useful tiebreaker. They record the full request, including any query string and the referrer header, before GA4 applies its channel rules. If GA4 and the logs disagree, the logs show the raw input.

Look for two kinds of requests. First, requests that carry a Gemini UTM tag. Second, requests with a Gemini referrer but no tag. Comparing the two tells you how often tags appear and how much traffic the tags miss.

This takes some engineering time, so use it when the numbers matter, such as before a budget request. For a quick read, the GA4 exploration is enough.

What Should You Tell Leadership?

Tell leadership what you know, what you do not and what you will do about it. A short note beats a long dashboard. Say that Gemini began tagging some links around October 1, that the tags are undocumented and that your reports will separate tagged traffic from the rest.

Avoid promises. Do not say Gemini traffic grew by a specific percentage unless you can show the before-and-after comparison. A small base makes percentages jump, and a relabeling can look like growth.

Finally, set a review date. Check again in a month, once more data has come in and Google may have said more. If the tags turn out to be documented by then, update the report and note what changed.

What About Conversions and Sales?

Traffic is the easy part, and conversions are what the business cares about. If you can capture the source and medium at sign-up, you can compare Gemini visitors with other channels on demo requests or trials. Many B2B teams store UTM values in a first-party cookie and pass them into a hidden form field.

Keep samples in perspective. AI referrals are still a small slice of most B2B sites, so a handful of conversions can swing a percentage wildly. We explored that tension in whether AI visibility tracking is worth it, using three new traffic studies.

The aim is a useful signal, not a perfect one. A rough, honest estimate that you can defend beats a precise number that falls apart under one question.

How Does This Fit With Your Other Measurement?

It is one input among several. Search Console shows how Google Search itself performs, and our guide to measuring AI search visibility in Search Console covers what that report can and cannot show. GA4 adds the visit side, and server logs can confirm what arrived.

Data you collect from third parties needs care too. After recent reports about altered rank-tracker data, we wrote about how to verify rank tracker data. The same habit applies here: compare at least two sources before you act.

Conclusion

Gemini's UTM tags may make AI referrals easier to measure, but nothing about them is documented. GA4's own rules suggest tagged visits could land in Organic Search, which could inflate SEO reports. Test it in your property and report what you find.

Cekpr will update this guide when Google documents the tags. Until then, run the before-and-after test, report tagged traffic as a floor and write the caveat in the report where your readers will see it.