For most of the past two years, attributing traffic from AI search surfaces has meant a combination of guesswork, referrer parsing, and dark traffic assumptions. When a visit arrived from ChatGPT or Perplexity, you might catch it. When it came from Google Gemini, you often would not. That is now changing. Google Gemini is adding UTM parameters to links it sends to external sites, giving site owners cleaner attribution data for traffic originating from the platform.
This is not a minor reporting tweak. It is a structural shift in how AI search visibility can be measured and, critically, connected to business outcomes. If you are running an AEO or GEO programme and measuring performance through sessions and engagement rates alone, this opens the door to something more useful: revenue attribution.
Why Attribution From AI Search Has Been So Difficult
AI search platforms generate traffic in fundamentally different ways to traditional organic search. When Google returns ten blue links, clicks are tracked through a combination of referrer data and campaign parameters that GA4 uses to build the attribution chain - though the mechanism is more involved than referrer headers alone. AI platforms are less consistent. Some pass referrer data. Some strip it. Some generate traffic that lands in GA4 as direct, leaving no trace of where it came from.
Gemini, as a Google product, sits in an unusual position. Visits from Gemini do not appear in Search Console the way organic clicks do. They are AI-generated referrals, not traditional search clicks. Without explicit UTM tagging, those visits have been difficult to separate from other Google-origin traffic - let alone from direct sessions. The result is that Gemini's contribution to site traffic has been systematically underreported, even for teams actively trying to measure AI search performance.
What UTM Parameters Actually Tell You
UTM parameters are querystring values appended to a URL that tell your analytics platform where a visit originated and how it should be classified. The exact parameter structure Gemini uses - including which source, medium, and campaign values it passes - will need to be confirmed against what your own GA4 installation actually receives, as the specific values have not been publicly documented in full. Treat the implementation as active but still stabilising: verify the actual UTM values in your own data before building channel groupings or CRM integrations against assumed values. What matters is that the data is now structured and consistent rather than absent.
With clean UTM data flowing into GA4, you can build a dedicated Gemini channel grouping, track sessions and engagement by source, and - most usefully - connect those sessions to conversion events. If your GA4 goals are configured correctly and your CRM integration is in place, a Gemini referral that becomes a form submission, a trial sign-up, or a product enquiry is now traceable through the full journey. That is the measurement step most teams have been missing.
It also provides a feedback loop for your AEO programme. If Gemini is citing a specific page and referring traffic from it, you will be able to see which content earns citations that convert. That is qualitatively different from knowing a page was mentioned somewhere in an AI response. It tells you whether the citation drove behaviour.
Setting Your Measurement Stack Up to Use This Data
The UTM parameters only help if your analytics configuration is ready to receive them properly. The first step is to check what UTM values Gemini is actually passing by reviewing incoming referral traffic in GA4 directly - do not assume a particular source or medium label without confirming it from your own data. Once you know the actual values, check your channel grouping rules in GA4. By default, GA4 may not have a channel definition that correctly captures Gemini as a distinct source. If Gemini traffic is bucketed into a generic referral or unassigned group, you will get the data but lose the clarity. Create a custom channel definition that matches the UTM source and medium values Gemini is passing.
Second, check your conversion events. GA4 tracks what you tell it to track. If your key conversion actions - demo requests, contact form submissions, phone call clicks, account sign-ups - are not configured as conversion events, you will see Gemini traffic arriving but have no way to judge its quality. Get those events firing correctly before you start drawing conclusions from the referral data.
Third, and this is where most teams stop short, connect GA4 to your CRM. A Gemini referral that becomes a session is interesting. A Gemini referral that becomes a qualified lead, progresses through a pipeline, and closes as revenue is the number that justifies your AEO programme to a board. Tools like HubSpot, Salesforce, and others support GA4 integration via first-party data and UTM passthrough. The technical work is not trivial, but the measurement case for AI search depends on it.
What This Means for How You Report AI Search Performance
Until now, most AI search reporting has focused on visibility metrics - whether a brand appears in AI responses, how frequently it is cited, and which competitors appear alongside it. Those metrics have value as leading indicators. But they do not answer the question that commercial stakeholders actually ask: is this generating pipeline?
Gemini UTM data creates a path to answering that question, at least for the Gemini surface. Where earlier work on measuring AI referral traffic established the general principle of attributing AI search visits in GA4, Gemini's UTM implementation is distinct because it originates from within Google's own ecosystem - meaning the data flows differently to third-party AI platforms and requires specific channel configuration to separate it from standard Google organic and paid traffic. Combined with referrer data from Perplexity, which practitioners report tends to pass source information more consistently than some other AI platforms, and the growing body of ChatGPT referral data that analytics platforms are beginning to capture, you can start to build a more complete picture of AI search's contribution to the business. This is not about vanity metrics. It is about putting AI search alongside PPC and organic in a single revenue attribution model.
The framing matters for internal reporting too. When Gemini traffic has its own channel grouping, stakeholders can see it. When it is buried in direct or unassigned, it is invisible - and invisible channels do not get resource. Clean attribution is not just a measurement problem. It is a budget justification problem.
The Broader Signal: AI Platforms Are Becoming Accountable Traffic Sources
Gemini adding UTM parameters is part of a wider shift. AI search platforms are maturing from closed black boxes into traffic sources that can be measured with the same rigour as paid or organic channels. That shift has commercial consequences for both advertisers and publishers. If AI platforms can demonstrate that their referrals convert, they become defensible parts of the marketing mix. If they cannot, they risk being treated as noise.
For teams running GEO or AEO programmes, this raises the bar on what good programme management looks like. Earning a citation in Gemini is a start. Understanding whether that citation sent qualified traffic, and whether that traffic converted, is the actual job. The tools to do that are now becoming available. The teams that build the measurement infrastructure now will be better positioned when AI search referral volumes grow to a point where the data becomes statistically meaningful at a channel level.
Start with the basics: verify your GA4 channel groupings, confirm your conversion events, and check the UTM parameters Gemini is actually passing in your own analytics before configuring anything against assumed values. Then connect what you find back to CRM pipeline. That is the chain that matters.