Something new started appearing in GA4 referral reports in 2025: traffic from chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, and other AI assistant platforms. Users ask ChatGPT a question, get a response that mentions your site, click the link, and arrive on your page — and GA4 records it as a referral.
For most sites, this AI-sourced traffic is still a small percentage of total visits. But it’s growing, it behaves differently from other referral channels, and the default GA4 setup largely misses it or groups it in ways that obscure its value.
This guide covers what AI traffic looks like in GA4, how to properly track and segment it, and how to set up a custom channel grouping to monitor it as a distinct channel.
Why AI Traffic Is Different
Traditional referral traffic comes from a user clicking a link on a webpage. AI traffic comes from a user clicking a link in a chatbot response.
Key behavioural differences:
Higher intent: Users who find your site through an AI recommendation are typically deeper in their research phase. They asked a specific question; the AI judged your content relevant; they clicked through to verify or act on that recommendation. Anecdotally and in the early data we’ve seen across client properties, AI referral sessions show higher engagement time and lower bounce rates than average referral traffic.
Zero-click is the norm: For every user who clicks through from an AI response, many more read the AI’s summary of your content and never visit. GA4 sees only the click-through users — but your content is influencing many more people who aren’t tracked.
Referrer data is inconsistent: Different AI platforms handle referrer data differently. Some pass a full referrer URL; others pass a stripped version or no referrer at all.
How AI Traffic Currently Appears in GA4
What Shows Up in Referral Reports
In GA4 → Acquisition → Traffic Acquisition, filter for Session Medium = “referral”. You’ll find AI platforms appearing as Session Source values:
| AI Platform | What GA4 typically records as source |
|---|---|
| ChatGPT (chatgpt.com) | chatgpt.com |
| Perplexity | perplexity.ai |
| Google Gemini (web) | gemini.google.com |
| Claude.ai | claude.ai |
| Microsoft Copilot | bing.com or copilot.microsoft.com |
| You.com | you.com |
| Phind | phind.com |
The problem with default channel grouping: GA4 classifies most of these as “Referral” or “Unassigned” depending on the medium and source combination. There’s no “AI Traffic” or “AI Referral” channel in GA4’s default groupings.
The Dark Traffic Problem
Some AI-generated traffic arrives with no referrer information at all. When a user copies a URL from a ChatGPT response and pastes it into a new tab, GA4 sees this as Direct traffic. There’s no way to distinguish this from a user who typed your URL from memory.
This means GA4’s AI traffic figures are undercount estimates. The actual volume of AI-influenced visits is higher than what’s trackable.
Creating a Custom “AI Referral” Channel Grouping
Custom channel groupings in GA4 let you reclassify traffic into your own named channels. Creating an “AI Traffic” channel captures all the trackable AI referrals in one place.
Step 1: Go to Channel Groups
GA4 Admin → Channel Groups → Create new channel group
Name it: Custom Channel Grouping — with AI Traffic
Step 2: Add the AI Traffic Channel
Add a new channel with these conditions:
Channel name: AI Referral
Conditions (Any of the following):
Session source exactly matches chatgpt.com
OR Session source exactly matches perplexity.ai
OR Session source exactly matches claude.ai
OR Session source exactly matches gemini.google.com
OR Session source exactly matches you.com
OR Session source exactly matches phind.com
OR Session source contains copilot.microsoft.com
OR Session source contains openai.com
OR Session source exactly matches poe.com
OR Session source exactly matches character.ai
Add any other AI platforms you want to include.
Medium condition (AND): Session medium exactly matches referral
Step 3: Order It Appropriately
Drag the AI Referral channel above “Referral” in the channel order. Channel grouping rules apply in order — the first matching rule wins. Placing AI Referral before the general Referral channel ensures AI traffic gets classified correctly.
Step 4: Apply to Reports
Custom channel groupings can be applied in:
- Reports → Acquisition → switch from “Session default channel group” to your custom grouping using the dimension picker
- Explorations → available as a custom dimension
Note: Custom channel groupings apply going forward from creation. Historical data is reclassified when you view reports (GA4 applies groupings at query time), so you can see historical AI traffic once the grouping is saved.
Building an AI Traffic Exploration
For deeper analysis, build an Exploration specifically for AI traffic:
GA4 → Explore → Blank
Setup:
- Rows: Session source
- Columns: Date (weekly)
- Metrics: Sessions, Users, Engaged Sessions, Conversions, Average Engagement Time
Filter:
- Include: Session source CONTAINS regex
chatgpt|perplexity|claude|gemini|copilot|you\.com|phind|poe\.com
This gives you a view of which AI platform sends the most traffic and how it trends week-over-week.
What AI Traffic Data Tells You
Once you have AI traffic isolated, the useful questions to answer:
Which pages do AI platforms recommend most?
GA4 → Exploration → rows: Landing Page, filter by AI sources → which pages are AI chatbots driving users to?
High AI referral traffic to a specific page suggests:
- That page is well-cited in AI training data
- The content answers questions that users are asking AI assistants
- The page has strong topical authority that AI models recognise
What’s the quality of AI traffic?
Compare AI referral sessions vs. all referral sessions:
- Engagement rate
- Conversion rate
- Average session duration
- Pages per session
If AI referral traffic has materially better engagement metrics, that’s evidence that AI-recommended content reaches higher-intent users — which should inform your content strategy.
Which queries are bringing AI traffic?
GA4 can’t tell you what the user typed into ChatGPT. But you can infer it from landing page content and from search console data. If a blog post about “Shopify GA4 setup” sees a spike in AI referrals, it’s likely that users asking AI assistants about GA4 for Shopify are being directed there.
Optimising Content for AI Discoverability
Tracking AI traffic is the first step. The second is understanding what makes content AI-recommendable.
AI platforms tend to recommend content that:
- Directly and completely answers a specific question
- Is structured clearly with headers, bullet points, and code examples
- Has factual accuracy that can be verified (citations, official documentation links)
- Comes from domains with strong topical authority
- Is comprehensive enough to be the single definitive source on a topic
This isn’t dramatically different from good SEO — well-structured, authoritative, comprehensive content performs well both in search engines and AI recommendations. The emphasis on factual accuracy and verifiability is stronger for AI platforms, which are increasingly penalising content that makes unverifiable claims.
Practically: Your llms.txt and llms-full.txt files (if you’ve set these up, as recommended in our SEO setup guide) explicitly communicate your site’s content and expertise to AI crawlers. These help AI systems accurately understand and reference your content.
Monitoring AI Traffic Month-Over-Month
Add a simple monthly AI traffic tracking habit:
- Save the Exploration you built above as a named Exploration
- At the start of each month, check: total AI referral sessions, top referring AI sources, top landing pages, and conversion rate from AI traffic
- Track the trend — is AI referral traffic growing, stable, or declining?
For most content-driven sites, AI referral traffic has been growing steadily through 2025–2026 as AI chatbot adoption increases. Monitoring this gives you early signal of whether your content is being picked up and recommended by AI systems.
If you want help setting up AI traffic tracking and custom channel groupings for your GA4 property, book a free consultation — we set this up as part of our standard GA4 implementation service.