How to Track Google Gemini Traffic to Your Website (Complete Guide)

According to Digiday's analysis of AI referral traffic, Gemini traffic grew 388% year-over-year in 2025. That's not a typo.

While ChatGPT still dominates AI referral traffic (77.97% of all AI traffic per the same report), Gemini is the fastest-growing platform. If you're not tracking it, you're flying blind.

This guide covers everything: GA4 setup, user-agent strings, the AI Overviews complication, and why some Gemini traffic is basically invisible.

Why Gemini Traffic Is Different

Gemini sends traffic through multiple pathways:

1. Direct Gemini conversations: Users ask Gemini a question, it provides citations, users click through
2. AI Overviews in Google Search: AI-generated summaries at the top of search results with source links
3. Gemini Deep Research: Autonomous browsing that accesses hundreds of sites to compile research reports

Each pathway has different referrer behavior. That's what makes this complicated.

The GA4 Problem: No Native Detection

GA4 does not natively detect Gemini traffic as a distinct channel.

When Gemini sends traffic through referral pathways, GA4 dumps it into the generic "Referral" bucket alongside traffic from every other website on the internet. You have to dig through your referrer list to find "gemini.google.com" manually.

Worse: traffic from AI Overviews often appears as "google.com" referral—indistinguishable from regular Google Search. GA4 can't tell whether someone clicked an AI Overview link or a traditional search result.

Google has indicated they're planning a dedicated AI traffic channel for GA4, but no timeline exists.

Step 1: Quick Manual Check

Before setting up custom channels, check if you're already getting Gemini traffic.

1. Open GA4 → Reports → Acquisition → Traffic Acquisition
2. Change the dimension from "Session default channel group" to "Session source / medium"
3. Search for "gemini"

If you see entries like "gemini.google.com / referral", you're getting Gemini traffic. The number shows how much.

This works for immediate visibility but requires manual repetition every time you want to check.

Step 2: Create a Custom AI Traffic Channel

For ongoing automated tracking, create a custom channel group.

Navigate to: Admin → Data Display → Channel Groups

Click: "Copy to create new" (duplicate the default channel grouping)

Name it: "Default + AI Traffic" or similar

Add a new channel called "AI Traffic" with these rules:

Set the condition to:

• Dimension: Session source
• Match type: matches regex
• Value: .*gemini.*google.*|.*perplexity.*|.*chatgpt.*|.*openai.*|.*claude.*|.*copilot.*

Important: Drag this new "AI Traffic" channel ABOVE the existing "Referral" channel. GA4 assigns traffic to the first matching channel, so position matters.

Save the channel group.

Now in any acquisition report, select your "Default + AI Traffic" channel group and AI traffic appears as its own segment.

Step 3: Understand the User-Agent Strings

Gemini uses different user-agent strings for different operations:

Gemini Deep Research Bot

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Gemini-Deep-Research; +https://gemini.google/overview/deep-research/) Chrome/135.0.0.0 Safari/537.36

This is the bot that autonomously browses sites to compile research reports. It's not user traffic—it's content ingestion. You'll see it in server logs but not in GA4 (which filters bot traffic by default).

Gemini iOS App

Mozilla/5.0 (iPhone; CPU iPhone OS 18_7_1 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E148 GoogleWv/1.0 (WKWebView) GeminiiOS/1.2025.4170000

This is real user traffic from the Gemini iOS app. The "GeminiiOS" identifier is the key marker.

Generic Google

Much Gemini traffic arrives with just "Google" as the user agent, making it impossible to distinguish from other Google services. This is a deliberate design choice by Google that limits analytics precision.

The AI Overviews Complication

AI Overviews now appear on 13.14% of all Google searches (up from 6.49% in January 2025).

When AI Overviews appear:

• Users click on traditional search results only 8% of the time (vs 15% without AI Overviews)
• Clicks on links within the AI Overview itself happen only 1% of the time
• Most users get their answer from the AI summary without clicking anything

Here's the measurement problem: when users DO click AI Overview links, that traffic often appears as "google.com / organic" in GA4—indistinguishable from regular search traffic.

There's currently no reliable way to separate AI Overview clicks from traditional organic search clicks in GA4.

Estimating Hidden Gemini Traffic

Like ChatGPT, some Gemini traffic arrives without referrer data:

• Users copying and pasting URLs from Gemini responses
• Gemini iOS app WebView sometimes stripping referrer headers
• AI Overview links occasionally losing attribution

To estimate hidden Gemini traffic, use proxy indicators:

1. Filter for: Direct traffic + New users + High engagement (session duration over 3 minutes)
2. Check landing pages: Are they informational content that would appear in AI responses?
3. Look at timing: Does Direct traffic spike after you're cited in AI Overviews?

This gives directional estimates, not precise measurements.

Server Log Analysis for Gemini Bots

For complete visibility, analyze your server logs for Gemini crawler activity.

Search for these patterns:

• Gemini-Deep-Research in user-agent strings
• google.com/gemini in referrer fields
• Request patterns consistent with AI content ingestion (systematic crawling of specific content)

Tools like Cloudflare Radar or log analysis platforms can help. This shows when Gemini is indexing your content—useful for understanding the delay between content publication and AI discovery.

What the Data Shows

Current Gemini traffic benchmarks:

Metric	Value
Share of AI traffic	6.40% globally, 8.41% in US
Year-over-year growth	563.6%
Median session duration	6-7 minutes
Referrer transmission rate	60-70%

Gemini users show engagement patterns similar to non-branded organic search, but with higher average session durations. They're genuinely engaged, not just clicking and bouncing.

Integration with Loamly

If manual GA4 configuration sounds tedious, Loamly handles this automatically.

Our tracker detects Gemini traffic using:

• Referrer analysis (when headers are present)
• User-agent pattern matching
• Behavioral fingerprinting for ambiguous cases

Plus we track what Gemini says about your brand—not just traffic, but perception. You can see whether Gemini recommends you, how you compare to competitors in Gemini's responses, and where optimization opportunities exist.

Action Items

Minimum viable setup:

1. Create the custom AI Traffic channel in GA4 (10 minutes)
2. Check your traffic weekly

Advanced setup:

1. Custom channel + server log monitoring
2. Behavioral proxy analysis for hidden traffic estimation
3. Loamly for brand monitoring and automated detection

Don't do:

• Block Gemini-Deep-Research bot (kills your visibility in Gemini)
• Assume GA4 organic search = only traditional search (AI Overviews are mixed in)
• Ignore Gemini because it's "only" 6% of AI traffic (fastest growing platform)

Should You Optimize for Gemini?

Gemini's 563% growth rate suggests it's worth attention.

Brands cited in AI Overviews earn 35% more organic clicks and 91% more paid clicks than non-cited competitors. Being visible in Gemini's responses—even when users don't click through—builds brand recognition that compounds over time.

The question isn't whether to track Gemini. It's whether you're willing to let 6% (and growing) of your AI visibility remain invisible.

---

Want to see how Gemini perceives your brand right now? Run a free AI visibility check at loamly.ai/check—includes Gemini, ChatGPT, Claude, and Perplexity analysis.