Google's Gemini chatbot has started adding UTM parameters โ€” the small tracking tags marketers normally attach to their own links โ€” to the URLs it sends users to. For small business owners who run a website, this means traffic arriving from Gemini should now show up clearly in analytics tools instead of vanishing into the vague 'direct traffic' bucket.

What happened

When Gemini answers a question and includes a link to a website, that link now carries UTM codes identifying Gemini as the source. Site owners using Google Analytics or similar tools can filter for this traffic and see, for the first time with real clarity, how much of their visits are coming from people asking Gemini questions rather than searching Google directly.

This addresses a problem that has frustrated marketers for the past two years: AI chat tools send real traffic to websites, but that traffic often arrives with no referrer information at all. Analytics platforms then lump it into 'direct' visits โ€” the same bucket used for people typing a web address straight into their browser. The result is that businesses have had no reliable way to measure whether AI assistants are actually sending them customers.

Gemini's move brings it in line with standard referral-tracking practices that have existed in digital marketing for over a decade. UTM tagging itself isn't new technology โ€” it's the same system email newsletters and paid ads have used for years. What's notable is that Google is applying it on the outbound side of an AI chatbot, rather than leaving attribution entirely up to guesswork or third-party tools.

Why it matters

The broader AI search landscape has been grappling with a trust problem around traffic data. Publishers and SEO professionals have pushed back against AI platforms, including Google's own AI Overviews, arguing that these tools summarize content and discourage clicks while leaving no trace of the interaction. Perplexity and OpenAI have faced similar criticism, and some platforms have responded with publisher programs or partial attribution fixes rather than full transparency.

Gemini's UTM rollout fits into this pattern of AI companies making incremental concessions on measurement without addressing the larger question of whether AI answers reduce the total number of clicks websites receive in the first place. Better tagging doesn't mean more traffic โ€” it means better visibility into whatever traffic already exists.

What this means for small businesses

If your business depends on website visits for leads, bookings, or sales, this is a genuine upgrade in measurement. You'll be able to check Google Analytics (or whatever platform you use) and filter by source to see Gemini-driven visits separately from organic search, social, or email traffic.

The practical catch is that most small business owners don't currently have their analytics set up to catch this. Default GA4 channel groupings may still bucket Gemini traffic oddly unless you create a custom channel or segment specifically looking for the new UTM parameters. This is a configuration task, not a plug-and-play feature.

It's also worth remembering that visibility into traffic volume doesn't tell you about traffic quality. A spike in Gemini-referred visits is only useful if those visitors convert into customers, so pairing this new data with your existing conversion tracking matters more than the raw numbers.

What to watch

Check your analytics dashboard this week for a new source labeled something like 'gemini' under referral or UTM campaign fields โ€” if you see nothing, your current setup may need a custom report or segment to surface it. Also watch whether other AI platforms, including ChatGPT and Microsoft Copilot, follow with similar tagging, since inconsistent standards across platforms would still leave businesses stitching together data from multiple tools.

The bottom line

Gemini's UTM tagging gives small businesses a clearer, though not complete, picture of how much traffic AI chat tools are actually sending them โ€” the next step is making sure your own analytics setup is configured to catch it.