December 12, 2025

10 min read

Comparisons & Analysis

Measuring AI traffic: How to track visitors from ChatGPT & Co.

More and more traffic comes from AI systems – but how do you measure it? Google Analytics shows '(direct)' or '(not set)' while ChatGPT users land on your site. Learn how to properly track and analyze AI traffic.

The problem: AI traffic is invisible

You look at Google Analytics and see: Organic Search, Direct, Referral. But where is the traffic from ChatGPT, Perplexity or Claude?

The truth: It's hiding. Depending on how the user clicks your link, AI traffic appears as:

→Direct: When the referrer isn't transmitted
→Referral: When Perplexity or others send the referrer
→(not set): When GA4 can't assign the source

This means: Without targeted setup, you're massively underestimating your AI traffic.

'What you don't measure, you can't optimize. AI traffic is the blind spot in most analytics setups.'

The most important AI traffic sources

These platforms (potentially) send traffic to your website:

ChatGPT (chat.openai.com)

Referrer rare

Traffic often appears as 'Direct' – hard to track without special measures

Perplexity (perplexity.ai)

Referrer usually present

Good referrer support – appears as referral traffic

Google AI Overviews

As Organic Search

Counted as normal Google traffic – not separately identifiable

Microsoft Copilot (copilot.microsoft.com)

Varies

Different referrer behavior depending on integration

Claude (claude.ai)

Referrer rare

Similar to ChatGPT – often classified as Direct

You.com

Referrer present

Appears as referral from you.com

Method 1: Set up GA4 referral filters

The simplest way: Filter in Google Analytics 4 for known AI referrers.

Step 1: Open traffic acquisition

GA4 → Reports → Acquisition → Traffic Acquisition

Step 2: Add secondary dimension

Click '+' → Session source or Session source/medium

Step 3: Filter by AI sources

Search for these domains:

perplexity.ai chat.openai.com openai.com claude.ai anthropic.com copilot.microsoft.com bing.com/chat you.com phind.com

Method 2: Custom channel grouping for AI

Create a custom channel group to automatically categorize AI traffic.

In GA4 Admin

Admin → Data Settings → Channel Groups → Create new channel group

Create new 'AI Referral' channel

Define conditions:

Source contains: perplexity OR Source contains: openai OR Source contains: claude OR Source contains: anthropic OR Source contains: copilot OR Source contains: you.com OR Source contains: phind

Advantage: From now on you'll see 'AI Referral' as its own channel in all reports.

Method 3: Regex filters for Explorations

For deeper analysis, use GA4 Explorations with regex filters:

Regex for all AI sources

Use this regex as a filter on the 'Session source' dimension

Method 4: Looker Studio Dashboard

For regular monitoring, create a dedicated AI traffic dashboard:

1.
Create new Looker Studio dashboard
Connect your GA4 property as data source.
2.
Calculated field for AI traffic
Create a field that identifies AI sources (CASE statement with regex).
3.
Add widgets
AI traffic over time, top AI sources, landing pages, conversions.

The problem with 'Direct' traffic

A large portion of AI traffic hides in the Direct channel. Why?

→ChatGPT desktop app doesn't send a referrer
→Copy-paste of links from AI answers = Direct
→Some browsers block referrer transmission
→HTTPS → HTTP switch loses referrer

Rule of thumb for estimation

If your Direct traffic suddenly rises without offline marketing, part of it is likely AI traffic. Estimate: 10-30% of 'unexplainable' Direct traffic could come from AI.

Server log analysis as alternative

When GA4 isn't enough, some go deeper – into server logs:

User agent analysis

Some AI systems have recognizable user agents:

ChatGPT-User PerplexityBot ClaudeBot

Note: These are often crawlers, not user traffic

Referer header in logs

Server logs show the referer header even when GA4 doesn't capture it. Grep for AI domains in your access logs.

Evaluating AI traffic quality

It's not just about quantity. These metrics show AI traffic quality:

Engagement Rate

Goal: >60%

How many AI visitors interact with your site?

Avg. Engagement Time

Compare with Organic

AI traffic often has higher dwell time – users are pre-qualified

Conversion Rate

Compare with other channels

Perplexity traffic often converts 15-25% better than search

Pages per Session

Goal: >2

Do AI visitors explore beyond the landing page?

Bounce Rate

Goal: <50%

High bounce rate may indicate intent mismatch

Analyze AI traffic by landing page

Which pages receive AI traffic? This shows you which content AI systems cite:

Create GA4 Exploration

Dimensions:

→ Landing page + query string
→ Session source

Metrics:

→ Sessions
→ Engagement rate
→ Conversions

Filter:

Session source contains AI regex

Insight: If certain blog posts get a lot of AI traffic, you know: This content is being cited by AI. Optimize it further and create similar content.

Tracking setup checklist

Basic setup

☐ GA4 correctly implemented
☐ Referrer list for AI sources defined
☐ Custom channel grouping created

Advanced

☐ Looker Studio dashboard
☐ Automatic alerts for traffic spikes
☐ Conversion tracking for AI traffic

Expert

☐ Server log analysis
☐ BigQuery export for deep analysis
☐ Attribution modeling incl. AI touchpoints

AI traffic benchmarks

What's 'normal'? Here are reference values (as of 2025):

Share of AI traffic in total traffic

1-5%

Still low for most websites, but growing fast

Perplexity as largest AI source

60-70%

Most trackable AI traffic comes from Perplexity

AI traffic growth YoY

+200-400%

AI referrals are growing exponentially

Conversion rate vs. organic

+15-25%

AI traffic often converts better than classic search traffic

Common AI traffic tracking mistakes

✗Only relying on recognizable referrers: Most hides in Direct/Unassigned.
✗Confusing crawlers with users: GPTBot is a crawler, not user traffic.
✗Viewing AI traffic in isolation: It's part of the customer journey, not the end.
✗No conversion attribution: Measure not just traffic, but also value.

The future: Better AI traffic tracking

Tracking will improve:

→Standardized AI referrers: Platforms will send uniform referrers
→GA4 updates: Google will add AI as its own channel
→Specialized tools: AI Visibility tools like art8 bridge the gap

Point of Truth

AI traffic is no longer a mystery. With the right filters, channel groups and dashboards, you make it visible.

Yes, some will always disappear into Direct traffic. But you can identify the trend, assess quality and make data-driven decisions about where to invest in AI Visibility.

The most important thing: Start measuring today. The earlier you have baseline data, the better you can prove growth and ROI.

Track more than just traffic

art8.io shows you how often and where your brand appears in AI answers – before traffic happens.

Try now