Measure AI Traffic in GA4: Regex + Data Studio Guide

AI chatbots like ChatGPT, Claude or Perplexity now work like search engines: they recommend websites, cite sources, link to providers – and send visitors. In Google Analytics 4, this traffic hides inside the "Referral" channel, invisible among spam referrers and newsletter clicks. This guide shows how to isolate it, make it measurable, and keep it visible in a permanent dashboard.

1. Why measure AI traffic separately?

AI systems operate on a different logic to Google: they read content, evaluate sources and deliver concrete recommendations. Anyone mentioned by ChatGPT or Perplexity appears in the answer – often with a click. Anyone not mentioned simply does not exist for a growing share of users.

The problem: Google Analytics 4 does not recognise this traffic as a distinct channel. ChatGPT, Claude, Perplexity and the rest land under "Referral" in the default view – mixed in with newsletter services, partner sites and spam. The result: you cannot see whether AI systems are sending relevant visitors to your website, which pages benefit, or how this traffic compares to Paid Search or Organic.

2. Agency trick: one setup for all GA4 accounts

There is no technical "master setup" that automatically applies to all GA4 properties. Custom Channel Groups are always property-specific in GA4 – a deliberate decision by Google. For agencies and businesses with multiple GA4 properties, however, there are three practical approaches:

Option 1 (recommended): Data Studio as a central analytics layer

The most efficient approach: build a single Data Studio dashboard that connects to multiple GA4 properties. The visualisations, comparisons and all charts only need to be built once. The channel group setup in GA4 takes around 5 minutes per property using the prepared regex below. The dashboard itself is reused across all properties – you simply switch the data source.

Option 2: GA4 Admin API (for developers)

The Google Analytics Admin API allows the channel group setup to be rolled out to any number of properties via script. Once programmed, adding a new property takes seconds. Worthwhile for agencies with 10+ properties that regularly onboard new accounts.

Option 3: GA4 property template (manual quick setup)

Create an internal onboarding document with the regex, channel settings and a step-by-step checklist. Per new property: 5 minutes of setup. The most pragmatic solution for most agencies.

3. Creating a Custom Channel Group in GA4

The core of the setup: a new channel group that identifies and classifies AI traffic as a distinct channel – instead of hiding it in the Referral pool.

Step by step

1. Click the gear icon (Admin) in the left-hand menu.
2. Under "Data display", click Channel groups.
3. Select + Create new channel group in the top right.
4. Name the group, e.g. Custom Channel Group or MY AGENCY Analysis.
5. All existing default channels remain intact – you are simply adding one new channel.
6. Click Add new channel.

Configuring the AI channel

4. The AI regex: complete list of sources

This regex covers all relevant AI platforms: ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Microsoft Copilot, Grok, Mistral, Hugging Face and over 80 further sources.

5. Adapting the standard report: instant overview

Once the Custom Channel Group is saved, you can activate it in the Traffic Acquisition report. Important: the group does not appear automatically as a new row – it must be actively selected from the dimension dropdown. You are switching the rule set, not the dimension.

1. Click the dropdown above the table (currently showing "Session default channel group").
2. A list of all available dimensions and channel groups opens. Your group appears at the top – directly alongside "Session primary channel group" and "Session default channel group".
3. Select your group. The table reloads and shows "Artificial Intelligence" as its own row.
4. Review the metrics: sessions, users, engagement rate, session duration, conversions.
5. Set the date range to 3–6 months to identify the trend.

6. Exploration: which pages, how long, how deep?

The standard report shows the picture at channel level. To answer "Which pages does AI traffic visit?", "How long do these users stay?" and "How many pages do they view?", an Exploration is needed.

Step 1: Create the "AI Traffic" segment

1. In the left panel under "Segments", click + → select Session segment.
2. Condition: Session channel group → exactly matches → Artificial Intelligence.
3. Alternatively: Session source → matches regex → same regex as above.
4. Name the segment: AI Traffic → Save.
5. Drag the segment into the Exploration (top left into the segment field).

Step 2: Configure dimensions and metrics

Step 3: Build the Exploration

1. Add dimensions and metrics to the configuration by double-clicking.
2. Rows: Page name – Columns: leave empty (or Session source to break down by AI platform).
3. Values: drag all relevant metrics into the value fields.
4. Sort by Sessions descending → top AI traffic pages visible immediately.
5. Date range: at least 3 months for meaningful data.

7. AI vs. Paid: direct channel comparison

Now it gets strategic: how does AI traffic compare to Paid Search, Organic or Direct – not just in volume, but in quality?

Segment comparison in the Exploration

1. Create a second segment: Session segment → Session channel group → Paid Search. Name it: Paid Search.
2. Optional: create a third segment Organic Search.
3. Drag all segments into the Exploration – GA4 displays them side by side.
4. Compare: session duration, pages per session, engagement rate, conversions per channel.

Quick comparison in the standard report

Select two channels there – the time-series chart shows the trend for both. Faster than Explorations, but fewer metrics available.

8. Building a Data Studio dashboard

Since 11 April 2026, Google Looker Studio is officially called Data Studio again – Google reversed the rename after 3.5 years. All existing reports remain unchanged; no action is required.

Data Studio is the right place for a permanent AI traffic dashboard: it updates automatically, can be shared and sent as a PDF by email, and functions as a central analytics layer across multiple GA4 properties.

Connecting to GA4

1. Open Data Studio and create a new report.
2. Select Google Analytics as the data source (native connector, free of charge).
3. Choose your GA4 property and connect.
4. For multiple properties: add further properties via Add data source. A single report can use multiple data sources simultaneously.

Setting up the AI traffic filter

1. In the report toolbar: Resource → Manage filters → Add a filter.
2. Name: AI Traffic only.
3. Condition: Session source → matches regex (RE2) → paste regex.
4. Activate the filter in the desired charts under "Filter".

Recommended dashboard structure

1. Overview: KPI scorecards – total sessions, AI share %, engagement rate, session duration.
2. Channel comparison: AI vs. Paid vs. Organic – all key metrics side by side.
3. Page analysis: top pages by AI traffic volume, time series for the most important pages.

Setting up automated PDF reports

Data Studio sends the report automatically as a PDF – daily, weekly or monthly. Ideal for monthly reports to clients or management.

9. Which charts actually help

10. Exporting data

Directly from GA4

Contains all dimensions and metrics from the current Exploration view. Suitable for ad-hoc analysis and spreadsheet work.

From Data Studio

1. PDF download: Share → Download → As PDF. Ideal for client reports.
2. CSV from a chart: Right-click on chart → Export data → CSV.
3. Automated PDF delivery: Share → Schedule email delivery → set recipients and frequency.

BigQuery for raw data

GA4 can export all event data automatically to BigQuery, where session sources are also available at event level. For automated analyses and custom data pipelines; requires a Google Cloud account. The first 10 GB of storage and 1 TB of queries per month are free.

11. Frequently asked questions