Skip to main content
Measurement

Tracking ChatGPT and Perplexity Referral Traffic in GA4

By · Updated · 7 min read

Why AI Search Referral Traffic Goes Uncounted in Standard GA4 Setups

ChatGPT and Perplexity send referer headers when a user clicks a cited link inside an AI-generated answer. GA4 receives these hits and logs them as referral traffic. The problem is recognition: GA4 has no built-in channel grouping for AI search engines, so this traffic gets scattered across the Referral channel or, in some configurations, lumped into Direct when the referer is stripped by the browser or platform.

For ecommerce operators, this creates a measurement gap. Traffic from AI answers behaves differently from traditional search clicks. Users arrive with higher intent context because the AI has already pre-qualified the answer. Treating these sessions as generic referrals or direct visits hides their true conversion value and makes it impossible to attribute revenue to content that earns AI citations.

Search Console does not solve this. It reports clicks from Google Search results, not from Google AI Overviews as a separate line item, and it has no visibility into ChatGPT or Perplexity. The only durable signal lives in the referer header captured by GA4, which means the analytics setup carries the full burden of identification.

The Exact Referrer Domains AI Search Engines Send

ChatGPT sends referer headers from two domains: chatgpt.com and chat.openai.com. The chatgpt.com domain is the current primary, while chat.openai.com still appears for older sessions and legacy bookmarks. Both should be treated as the same source for analysis purposes.

Perplexity sends referer headers from perplexity.ai. Citations inside Perplexity answers are clickable links that pass the referer cleanly, which makes it the most measurable AI search engine in GA4 by default. Gemini and Google AI Overviews are more complex because they live on google.com subdomains and blend into existing Google referral data.

In GA4, these appear under Reports > Acquisition > Traffic acquisition, filtered by Session source. The raw source names will read chatgpt.com, chat.openai.com, and perplexity.ai. No transformation happens automatically, which is why custom configuration is required to group them into a single AI Search channel.

Building a Custom Channel Group for AI Search in GA4

GA4 supports custom channel groups under Admin > Data display > Channel groups. Create a new group called AI Search and define a channel with a rule matching Session source containing chatgpt.com, chat.openai.com, or perplexity.ai. Place this channel above the Referral channel in the priority order so AI traffic is classified before it falls into the generic bucket.

Extend the rule set to include other AI surfaces as they emerge: copilot.microsoft.com for Microsoft Copilot, gemini.google.com for direct Gemini traffic, and you.com for You.com. Each new domain added to the rule keeps the AI Search channel comprehensive without disturbing historical data.

Custom channel groups apply to standard reports going forward and to exploration reports retroactively. This means once configured, every Traffic acquisition report, every funnel, and every attribution view will display AI Search as a first-class channel alongside Organic Search, Paid Search, and Direct.

Using UTM Parameters to Track Explicit AI Citations

Referer headers are reliable but passive. For pages specifically designed to earn AI citations, add UTM parameters to the URLs referenced inside the content. When the AI quotes the URL with parameters intact, every click carries explicit source data into GA4 regardless of what the referer header contains.

A working pattern: utm_source=chatgpt&utm_medium=ai_search&utm_campaign=content_hub_name. The utm_medium value of ai_search becomes the anchor for all AI-driven traffic regardless of which engine cited the page. This survives referer stripping, browser privacy modes, and any future changes to how AI engines pass referers.

The limitation is that AI engines decide whether to preserve URL parameters when they render citations. Some strip query strings, some keep them. Use UTM tagging as a complement to referer-based detection, not a replacement. The combination of channel grouping plus UTM tagging captures both implicit and explicit AI traffic.

Segments and Explorations for Isolating AI Traffic Behavior

In GA4 Explorations, build a segment defined as Session source matches regex ^(chatgpt\.com|chat\.openai\.com|perplexity\.ai)$. Apply this segment to a free-form exploration with dimensions for Landing page, Device category, and Country, and metrics for Sessions, Engaged sessions, Average engagement time, and Conversions.

This view answers the questions that matter for content strategy: which pages earn AI citations, which queries route users through AI answers, and how AI-sourced sessions convert compared to organic search. Save the segment for reuse across funnel explorations and path analysis reports.

Layer in a comparison segment for Organic Search traffic to the same landing pages. The side-by-side comparison reveals whether AI-sourced visitors convert at higher or lower rates, how their session depth differs, and whether the content earning citations is the same content driving organic clicks or a different subset entirely.

What Good Tracking Looks Like vs Poor Tracking

Poor tracking: AI referrals sit inside the Referral channel mixed with link clicks from forums, newsletters, and partner sites. The chatgpt.com source name appears in raw reports but is never aggregated. Conversion attribution shows Referral as a vague bucket worth some revenue, with no way to isolate the AI portion. UTM tagging is absent, so any referer stripping erases the data entirely.

Good tracking: a custom AI Search channel group classifies chatgpt.com, chat.openai.com, perplexity.ai, and emerging AI domains into a single channel visible in every standard report. UTM parameters on key cited URLs provide a backup signal. A saved exploration segment makes AI traffic queryable on demand, and conversion paths show AI Search as a discrete touchpoint in multi-channel funnels.

The difference is operational. Poor tracking forces guesswork about whether content investment in AI visibility is producing revenue. Good tracking produces a defensible number for AI-sourced sessions, conversions, and revenue every reporting cycle.

Implementation Checklist for the Next 30 Minutes

Open GA4 Admin > Data display > Channel groups and create a new custom channel group named AI Search Tracking. Add a channel called AI Search with the condition: Session source contains chatgpt.com OR chat.openai.com OR perplexity.ai. Set this channel above Referral in the priority order. Save and apply.

Build a saved exploration segment using the regex ^(chatgpt\.com|chat\.openai\.com|perplexity\.ai)$ on Session source. Name it AI Search Sessions. Add it to a free-form exploration with Landing page, Sessions, Engaged sessions, and Conversions. This becomes the recurring report for AI traffic performance.

Audit the top 20 pages by organic traffic and add UTM parameters to any internal canonical URLs referenced in structured data, FAQ schema, or content explicitly designed for citation. Use utm_medium=ai_search as the consistent identifier. The full setup takes under an hour and produces durable measurement from the moment it ships.

Frequently asked questions

Does GA4 automatically track ChatGPT and Perplexity referrals?

GA4 captures the referer headers from chatgpt.com, chat.openai.com, and perplexity.ai and logs them as referral traffic. It does not group them into a dedicated AI Search channel by default. The raw source names appear in Traffic acquisition reports, but they sit inside the generic Referral channel alongside unrelated link traffic. A custom channel group is required to separate AI search referrals from other referral sources.

Why does some AI traffic show up as Direct instead of Referral in GA4?

Referer headers are stripped by certain browser privacy settings, mobile app contexts, and HTTPS-to-HTTP transitions. When the referer is missing, GA4 has no source data and classifies the session as Direct. This affects a portion of AI search clicks, particularly from mobile ChatGPT app users. UTM parameters on cited URLs solve this by embedding source data directly in the URL, surviving any referer stripping.

Can Search Console show clicks from Google AI Overviews?

Search Console does not break out AI Overviews clicks as a separate metric. Clicks from AI Overviews are aggregated into the standard Search performance report alongside regular organic results. There is no filter, dimension, or export that isolates AI Overviews traffic in Search Console. GA4 referer data and UTM tagging remain the only reliable methods for measuring AI-sourced traffic, and even GA4 cannot fully separate AI Overviews from standard Google referrals.

What UTM medium value should be used for AI search tracking?

Use utm_medium=ai_search as the standard value. This creates a consistent medium across all AI engines, making it queryable as a single segment in GA4. Pair it with utm_source set to the specific engine (chatgpt, perplexity, gemini, copilot) and utm_campaign set to the content hub or page cluster. The combination supports both aggregate AI traffic analysis and per-engine performance breakdowns.

Is AI search referral traffic worth tracking if volume is low?

Volume from AI search is smaller than organic search for most ecommerce stores, but conversion intent runs higher because the AI has pre-qualified the user with a specific answer. Tracking it produces a per-session value comparison against other channels and identifies which content earns citations. Without measurement, content investments aimed at AI visibility have no attribution, making it impossible to justify continued investment or identify which pages drive results.

MG
Written by

Matt is the founder of RunOctopus. He built All Angles Creatures from zero to page-1 rankings in reptile feeder insects in under 60 days using exactly this method — turning a hard, entrenched niche into RunOctopus's proof store for programmatic SEO and AI search citation.

Connect on LinkedIn →

See what Otto would build for your store

Free architecture preview. No card required. Five minutes.

Generate Preview →