Custom Channel Grouping for AI Traffic in GA4

Why GA4 Misclassifies AI Traffic

Google Analytics 4 was built with a fixed set of default channel groupings: Organic Search, Paid Search, Direct, Referral, Social, Email, and a handful of others. These categories were designed for a world where traffic came from search engines, social platforms, and links on other websites. AI-driven traffic did not exist when these defaults were established.

The result is a measurement blind spot. When someone clicks a link in a ChatGPT response and arrives at your website, GA4 classifies that visit as "Referral" from chatgpt.com. When someone arrives from a Perplexity citation, it appears as another referral from perplexity.ai. And when Google Gemini or AI Mode sends traffic, it often blends into "Organic Search" because the traffic originates within Google's own ecosystem.

This misclassification creates three problems. First, you cannot see total AI traffic at a glance because it is scattered across multiple default channels. Second, you cannot compare AI traffic growth against other channels over time. Third, your reporting tells stakeholders nothing about the fastest-growing acquisition channel on the internet -- one where referral traffic from ChatGPT grew 326% year-over-year and converts at 4.4x the rate of organic search.

A custom channel grouping solves all three problems. It pulls every AI-sourced visit into a single "AI Search" channel that sits alongside your existing channels in every GA4 report. If you have already set up basic AI referral tracking using our GA4 AI referral traffic setup guide, this custom channel takes your measurement to the next level.

Prerequisites Before You Start

Before creating the custom channel grouping, verify that you have the following in place:

GA4 property access. You need Editor or Administrator access to the GA4 property. Viewer access is not sufficient to create or modify channel groupings. Check your role under Admin > Property Access Management.

Data already flowing. Your GA4 property should already be collecting data with the standard tracking snippet or Google Tag Manager container installed on your site. If you are starting from scratch, complete the basic GA4 setup for AI referral traffic first.

Understanding of your current referral sources. Open your GA4 Traffic Acquisition report and look at the Source/Medium dimension. Note which AI platforms already appear. Common sources include chatgpt.com, chat.openai.com, perplexity.ai, claude.ai, and copilot.microsoft.com. This tells you which AI platforms are already sending traffic and confirms your tracking is working.

A test plan. Before modifying channel groupings in a production property, consider testing in a staging or development GA4 property first if one is available. While custom channel groupings do not affect data collection, they do change how data is categorized in reports.

Step-by-Step: Creating the AI Search Channel Group

Follow these steps to create your custom channel grouping in GA4. The entire process takes approximately 10-15 minutes.

Step 1: Navigate to Channel Groups

Open your GA4 property. Click Admin (gear icon) in the bottom-left corner. Under the Property column, scroll down to Data display and click Channel groups. You will see a list of any existing custom channel groups. Click Create new channel group.

Step 2: Name Your Channel Group

Name it something clear and descriptive. We recommend "AI-Aware Channels" or "Channels with AI Search". Avoid names like "Custom" -- you may create additional custom groupings in the future.

Step 3: Copy Default Channels

GA4 gives you the option to start from scratch or copy the default channel definitions. Always start by copying the default channels. Click "Copy from Default Channel Group" (or the equivalent button). This preserves all your existing channel definitions and lets you add the AI Search channel on top.

Step 4: Create the AI Search Channel

Click Add new channel. Name it "AI Search". Now you need to define the conditions that identify AI traffic. Set the following condition:

• Condition: Source matches regex (see the complete pattern in the next section)
• Match type: matches regex

Step 5: Set Channel Priority

This is critical. Move the AI Search channel above the Referral and Organic Search channels in the priority list. GA4 evaluates channels from top to bottom and assigns traffic to the first matching channel. If Referral is above AI Search, ChatGPT traffic will match Referral first and never reach your AI Search rule.

The recommended order for the top of your channel group:

1. Direct
2. Paid Search
3. AI Search (your new channel)
4. Organic Search
5. Referral
6. (remaining channels)

Step 6: Save and Verify

Click Save. The channel grouping is now active and will retroactively apply to all historical data in your property.

The Complete AI Traffic Regex Pattern

The regex pattern is the engine of your custom channel grouping. It needs to match all known AI platforms that send referral traffic while avoiding false positives. Here is the comprehensive pattern we recommend:

chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|you\.com|phind\.com|deepseek\.com|grok\.x\.ai|poe\.com|huggingchat\.co|andi\.search|kagi\.com

Pattern Breakdown

| Source Domain | Platform | Notes | |---|---|---| | chatgpt\.com | ChatGPT | Primary domain since late 2025 | | chat\.openai\.com | ChatGPT (legacy) | Older domain, still generates some traffic | | perplexity\.ai | Perplexity | Second-largest AI referral source | | claude\.ai | Claude (Anthropic) | Growing with Apple Safari integration | | gemini\.google\.com | Google Gemini | May also appear as organic; include for safety | | copilot\.microsoft\.com | Microsoft Copilot | Integrated into Bing, Windows, and Office | | you\.com | You.com | AI-powered search engine | | phind\.com | Phind | Developer-focused AI search | | deepseek\.com | DeepSeek | Rising in Asia and technical communities | | grok\.x\.ai | Grok | X/Twitter's AI assistant | | poe\.com | Poe | Quora's multi-model AI platform | | huggingchat\.co | HuggingChat | Open-source AI chat | | andi\.search | Andi | Conversational search engine | | kagi\.com | Kagi | Premium search with AI features |

Handling Gemini and AI Mode Edge Cases

Google Gemini traffic presents a unique challenge. When users interact with Gemini through Google Search's AI Mode, the traffic often appears with a source of "google" and a medium of "organic." This makes it indistinguishable from regular Google organic traffic in many cases.

To capture Gemini traffic that arrives as a distinct referral (from gemini.google.com), the regex above is sufficient. For Gemini traffic that blends into organic, you currently need to rely on additional signals such as landing page parameters or session behavior patterns. This is an evolving area -- Google may introduce clearer attribution in future GA4 updates.

For a deeper discussion on measuring all forms of AI referral traffic, see our guide on measuring AI referral traffic.

Validating Your Channel Grouping

After saving your custom channel grouping, you need to verify that it is classifying traffic correctly. Do not assume it works -- test it.

Method 1: Check Historical Data

Open the Traffic Acquisition report in GA4. In the dimension dropdown, switch from the default channel grouping to your new custom channel grouping (e.g., "AI-Aware Channels"). You should see an "AI Search" row in the table. If the row appears with data, your regex is matching historical AI traffic correctly.

Method 2: Real-Time Verification

Open GA4's Realtime report. In a separate browser or incognito window, navigate to your website by clicking a link from one of the AI platforms (or simulate it by appending ?utm_source=chatgpt.com temporarily). Watch the Realtime report to see if the session appears under the correct channel.

Method 3: Explore Custom Reports

Create an Exploration in GA4. Set the dimension to your custom channel grouping and add metrics like Sessions, Engaged Sessions, and Conversions. Filter for "AI Search" only. This confirms that the channel is capturing data and that key metrics are being tracked.

Common Validation Issues

• No AI Search data appears: Check that the channel priority is correct. If Referral appears above AI Search, traffic will match Referral first.
• Some platforms missing: Verify the regex syntax. A missing backslash before a dot (. instead of \.) can cause unexpected matches or misses.
• Duplicate counting: Ensure your regex does not overlap with other custom channels. Each session should match exactly one channel.

Building an AI Traffic Dashboard in Looker Studio

Raw GA4 data is useful for analysts but not for stakeholders. A dedicated AI traffic dashboard in Looker Studio (formerly Google Data Studio) translates your custom channel data into clear, actionable visualizations.

Recommended Dashboard Components

1. AI Traffic Overview Scorecard. Show total AI Search sessions for the current period alongside percentage change from the previous period. Place this at the top of the dashboard for immediate visibility.

2. AI Traffic Trend Line. A time-series chart showing daily or weekly AI Search sessions over the past 90 days. Overlay it with total site sessions to show AI traffic as a proportion of overall traffic. This reveals growth trends and seasonality.

3. AI Source Breakdown. A pie chart or horizontal bar chart showing sessions by source within the AI Search channel. This answers the question: "Which AI platforms are sending us the most traffic?" Expect ChatGPT to dominate with roughly 84% of the total.

4. Landing Page Performance. A table showing the top 20 landing pages for AI Search traffic, with columns for sessions, engagement rate, average engagement time, and conversion rate. This reveals which content AI platforms are citing most.

5. Conversion Comparison. A bar chart comparing conversion rates across channels: AI Search vs. Organic Search vs. Referral vs. Direct. This data point is critical for justifying AI SEO investment to stakeholders. AI traffic typically converts at 4.4x the rate of organic search.

Connecting GA4 to Looker Studio

In Looker Studio, create a new report and add GA4 as a data source. Select your property and ensure you choose the custom channel grouping dimension (not the default). Build each component using the AI Search channel as a filter or dimension. Set the dashboard to auto-refresh daily.

For a broader view of AI monitoring beyond just traffic, see our AI visibility monitoring guide.

Advanced Segmentation: Platform-Level Breakdowns

The single "AI Search" channel is sufficient for high-level reporting, but deeper analysis requires platform-level segmentation. There are two approaches:

Approach 1: Sub-Channels per Platform

Instead of one AI Search channel, create multiple channels:

• AI Search -- ChatGPT (source matches chatgpt\.com|chat\.openai\.com)
• AI Search -- Perplexity (source matches perplexity\.ai)
• AI Search -- Gemini (source matches gemini\.google\.com)
• AI Search -- Other AI (source matches the remaining AI platforms)

This gives you platform-level data at the channel level but adds complexity to your channel group. It works best when you have enough AI traffic volume to justify the granularity.

Approach 2: Drill-Down Within a Single Channel

Keep the single AI Search channel and use GA4's secondary dimensions or Explorations to break down by source within the channel. This is simpler to maintain and works well for most organizations.

Audience Segments for AI Traffic

Beyond channel groupings, consider creating GA4 audience segments for AI traffic users. This lets you:

• Compare behavior of AI-referred users vs. other users across your entire site
• Build remarketing audiences for users who arrived via AI recommendations
• Track how AI-referred users move through your conversion funnel

To learn more about what AI SEO is and why these measurement capabilities matter, start with our foundational guide.

Maintenance and Quarterly Updates

Your AI traffic channel grouping is not a set-and-forget configuration. The AI landscape evolves rapidly, and your measurement must keep pace.

Quarterly Review Checklist

1. Scan unclassified referral traffic. Open the Traffic Acquisition report, filter for the Referral channel in your custom grouping, and sort by sessions. Look for any AI-related domains that your regex is not capturing. New AI platforms launch frequently.

2. Update the regex pattern. Add new AI platform domains as they emerge. Remove domains that have been deprecated or renamed. Document every change with the date and reason.

3. Check for domain changes. AI platforms occasionally change their referral domains. OpenAI, for example, transitioned from chat.openai.com to chatgpt.com. Monitor for similar changes from other platforms.

4. Validate classification accuracy. Spot-check 20-30 sessions classified as AI Search to confirm they genuinely came from AI platforms. Similarly, spot-check Referral sessions to ensure no AI traffic is leaking into the wrong channel.

5. Review channel priority. If you have added other custom channels since your initial setup, verify that the priority order still places AI Search above Referral and Organic Search.

Staying Current

Subscribe to GA4 release notes and AI platform blogs. Google periodically updates default channel definitions, which could affect how AI traffic is classified. If Google introduces a native "AI Search" default channel in the future, you will want to evaluate whether to migrate from your custom grouping or keep both.