AI Visibility

Track ChatGPT, Perplexity & Gemini Traffic in GA4 (2026)

Q: What regex pattern should I use to track all AI search traffic in GA4?

GA4 misses 35-70% of AI referral traffic by default. Complete setup guide: native AI channel, regex filters for 9 platforms, and a working dashboard.

Jaxon ParrottJan 28, 2026

Yes, you can attribute traffic from ChatGPT, Perplexity, and Gemini in GA4. The setup requires two layers: GA4's native "AI Assistant" channel (rolled out May 2026, broadly available since June 7) and a custom channel group with regex filters. The native channel recognizes ChatGPT, Gemini, DeepSeek, Copilot, and Grok automatically. It does not recognize Perplexity. And across all platforms, 35 to 70% of AI referral sessions arrive without referrer headers and land in "Direct" regardless. The custom channel group closes both gaps.

AI traffic attribution is the measurement layer of Machine Relations. Without it, you cannot connect earned media placements to the AI sessions they generate or calculate ROI on citation architecture. This guide covers the complete setup: native channel configuration, custom regex filters for every major AI platform, per-platform referrer patterns, and a dashboard that shows which content AI engines actually cite.

Key Takeaways

GA4's native "AI Assistant" channel does not cover Perplexity. Google's confirmed list includes ChatGPT, Gemini, DeepSeek, Copilot, and Grok. Perplexity traffic still lands in generic "Referral," making a custom regex channel group mandatory for complete tracking.
ChatGPT dominates AI referral traffic at 76.85% global share (April 2026 Statcounter data). Gemini holds 9.0%, Perplexity 7.73%, Copilot 3.76%, Claude 2.66%.
ChatGPT crossed 1 billion monthly active users in June 2026, the fastest app in history to reach that milestone. 900 million weekly active users. Over 2 billion daily queries processed.
AI search traffic converts at 4.4x the rate of traditional organic search, making accurate attribution essential for ROI measurement on earned media and content investment.
35 to 70% of AI referral sessions arrive without referrer headers and default to "Direct" in GA4. A custom regex channel group catches what native detection misses across ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, and Grok.

Why GA4 Misclassifies AI Search Traffic

GA4's default configuration categorizes most AI referral visits as direct or unassigned traffic. When a visitor clicks a link inside ChatGPT, Perplexity, or Gemini, the referrer header is often stripped by the AI platform's embedded browser, mobile app, or link-handling behavior. GA4 sees no referrer and classifies the session as "direct." The same bucket as someone typing your URL manually.

This creates a measurement blind spot at scale. According to Adobe Analytics research, AI referral traffic converts at 4.4 times the rate of traditional organic search traffic. When that high-intent traffic is invisible in your analytics, you cannot calculate ROI on the content and earned media placements that drive AI citations.

Three specific failure modes cause GA4 to misclassify AI traffic:

Stripped referrer headers. ChatGPT's mobile app and Perplexity's in-app browser frequently omit referrer data when opening external links. The visit registers as direct.
Unrecognized source domains. GA4's default channel definitions did not include AI platform domains until May 2026. Even now, the native AI Assistant channel does not cover every platform. Perplexity traffic with valid referrers still lands in generic "Referral," not "AI Assistant."
UTM parameter inconsistency. ChatGPT appends utm_source=chatgpt.com to some outbound links, but Perplexity, Gemini, and Claude do not consistently use UTM parameters. Relying on UTMs alone misses the majority of AI traffic.

The net result: your GA4 reports undercount AI search traffic, often by more than half. Gartner predicted that traditional search engine volume would drop 25% by 2026 as AI chatbots absorb query volume. That shift is happening. ChatGPT crossed 1 billion monthly active users in June 2026 and processes over 2 billion queries daily. AI-referred sessions now account for 0.9% of all web traffic, up 5x year-over-year. Most marketing teams still cannot measure the traffic this shift generates.

ChatGPT's share of AI referral traffic dropped to 76.85% in April 2026, its lowest level on record. Gemini holds second place at 9.0%, Perplexity third at 7.73%. These shifts matter for attribution setup because each platform handles referrer headers differently, and GA4's native channel does not cover all of them.

According to Statcounter's global AI chatbot referral data for April 2026:

AI Platform	Referral Share (April 2026)	Change from April 2025
ChatGPT	76.85%	Down from 84.21%
Google Gemini	9.00%	Up from 2.31%
Perplexity	7.73%	Down from 12.07%
Microsoft Copilot	3.76%	Stable
Claude	2.66%	Up from ~1%
DeepSeek	<1%	New entrant

Two dynamics matter for attribution setup. First, Gemini's surge from 2.31% to 9.0% year-over-year means Gemini referrer patterns now warrant dedicated tracking. Second, Claude's growth to 2.66% (with weekly peaks above 3% in mid-March) makes Anthropic's platform a material traffic source for B2B content.

For B2B-specific referrals, the distribution skews differently. ChatGPT's share drops to roughly 63% while Claude reaches approximately 18% of measurable B2B AI referrals, according to Goodie's 2026 AI Search Traffic Report. B2B buyers use Claude and Perplexity for longer research queries at higher rates than the general population. Similarweb's 2026 AI visibility analysis confirms that AI-assisted sessions are growing at over 80% year-over-year, even as they still represent a single-digit share of total web traffic.

GA4's Native AI Assistant Channel

As of May 2026, Google Analytics automatically assigns an "ai-assistant" medium value and groups recognized AI chatbot traffic under a new "AI Assistant" default channel. The channel reached broad availability across GA4 properties on June 7, 2026. No configuration is required.

Here is the critical detail most guides miss: GA4's native AI Assistant channel does not include Perplexity. According to GA4's live Default Channel Group documentation, the confirmed platform list is ChatGPT, Gemini, DeepSeek, Copilot, and Grok. Perplexity, one of the highest-intent AI traffic sources (especially for B2B), still routes to generic "Referral" traffic. Claude's inclusion is unconfirmed. This means a custom regex channel group is not optional. It is the only way to track Perplexity and Claude traffic under a unified AI channel.

What the native channel does:

Automatic classification. Sessions from recognized AI chatbot referrers are tagged with medium ai-assistant and grouped into the "AI Assistant" channel. No configuration required.
Forward-only application. The channel applies to new data from the rollout date forward. Historical data before May 2026 remains classified under old channel assignments.
Zero setup. Every GA4 property gets the channel automatically. It parallels Google's 2022 addition of "cross-network" as a default channel for Performance Max traffic.

What the native channel does not do:

No coverage for stripped referrers. Traffic arriving without a referrer header, which accounts for 35 to 70% of AI referral sessions depending on the platform and month, still lands in Direct.
No Perplexity coverage. Perplexity sessions with valid referrers are classified as generic "Referral," not "AI Assistant."
No historical reclassification. Sessions prior to May 2026 stay wherever GA4 originally bucketed them.
No Google AI Overviews. Clicks from Google's own AI Overviews and AI Mode count as Organic Search, not AI Assistant.

The native channel is a useful baseline, but it is not sufficient for complete AI traffic attribution. You still need a custom channel group to catch Perplexity, Claude, and the 35 to 70% of sessions with stripped referrers.

How to Create a Custom AI Search Channel Group in GA4

A custom channel group with a regex filter captures AI traffic that GA4's native channel misses, including Perplexity sessions and traffic from platforms Google has not yet added to its recognized list. This setup takes under 10 minutes and works alongside the native AI Assistant channel.

Step-by-step setup:

Open channel settings. Go to Admin > Data display > Channel groups in your GA4 property.
Create a new channel group. Click "Create new channel group" and name it something distinct from GA4's native channel. "AI Search (Custom)" works well to differentiate from the built-in "AI Assistant" channel.
Add a new channel. Click "Add new channel" and name it "AI Search."
Set the rule. Choose "Source" matches regex, then enter the regex pattern from the next section.
Position the channel above Referral. GA4 evaluates channel rules top-down. If your AI channel sits below Referral, AI visits with valid referrers get classified as generic referrals before your AI rule is checked.
Save and verify. Apply the channel group to your reports. New sessions will begin classifying under the custom channel immediately.

You can run both the native AI Assistant channel and your custom channel group simultaneously. The custom group gives you explicit control over which platforms are tracked and lets you include AI sources before Google adds them to its recognized list.

AI Platform Referrer Patterns and Regex Filters for 2026

Each AI platform uses different referrer domains, and the pattern set has expanded significantly since 2025. A production-ready regex filter must cover the current major platforms plus recent entrants like DeepSeek and Grok.

Current referrer patterns by platform:

AI Platform	Referrer Domains	UTM Behavior	GA4 Native Channel
ChatGPT	`chatgpt.com`, `chat.openai.com`	Appends `utm_source=chatgpt.com` on some links	Included
Perplexity	`perplexity.ai`	No consistent UTM parameters	Not included
Google Gemini	`gemini.google.com`, `bard.google.com`	No UTM parameters	Included
Claude	`claude.ai`	No UTM parameters	Unconfirmed
Microsoft Copilot	`copilot.microsoft.com`, `bing.com/chat`	Inconsistent UTM usage	Included
DeepSeek	`deepseek.com`, `chat.deepseek.com`	No UTM parameters	Included
Grok	`grok.com`, `x.com/grok`	No UTM parameters	Included
Meta AI	`meta.ai`	No UTM parameters	Unconfirmed
You.com	`you.com`	No UTM parameters	Unconfirmed

Full regex pattern for GA4 source matching:

This pattern covers the platforms responsible for over 99% of measurable AI referral traffic as of mid-2026. Review and update quarterly as new platforms emerge.

How to Track ChatGPT Traffic in GA4

ChatGPT is the easiest AI platform to track because it appends UTM parameters to some outbound links and uses consistent referrer domains. ChatGPT holds 76.85% of all AI referral traffic according to Statcounter and crossed 1 billion monthly active users in June 2026, the fastest app in history to reach that milestone. Accurate ChatGPT attribution is the single highest-leverage fix for AI traffic measurement.

ChatGPT sends traffic from two primary domains:

chatgpt.com (the consumer product)
chat.openai.com (the original domain, still active)

Additionally, ChatGPT appends utm_source=chatgpt.com to many outbound links. This means you can track ChatGPT traffic through three independent signals:

Source domain matching. The regex pattern chatgpt\.com|chat\.openai\.com captures both referrer domains.
UTM parameter. Filter for utm_source=chatgpt.com in GA4 explorations to isolate ChatGPT sessions even when the referrer is stripped.
GA4 native channel. ChatGPT is a confirmed recognized platform in GA4's AI Assistant channel.

For the most accurate ChatGPT measurement, create a GA4 exploration that combines all three signals: source domain regex OR utm_source match. This captures sessions that any single method would miss.

ChatGPT also uses OAI-SearchBot as its web crawler for search features. Check your server logs for OAI-SearchBot user-agent strings to see which pages ChatGPT's search index is crawling. This is a leading indicator of which content may generate future AI referral traffic.

How to Check Perplexity Referral Traffic in GA4

Perplexity is the most important AI platform to track manually because GA4's native AI Assistant channel does not include it. Perplexity traffic with valid referrers lands in generic "Referral," not "AI Assistant." Without a custom channel group, Perplexity sessions are invisible as AI traffic in your default reports.

Perplexity sends traffic from perplexity.ai. The web interface passes referrer headers reliably, but the mobile app strips them frequently. Perplexity's global referral share dropped from 12.07% to 7.73% year-over-year, but it remains the primary AI research tool for long-form queries. Perplexity users spend more time on-site than ChatGPT referrals on average, making them a higher-value audience for B2B content.

To check Perplexity referral traffic in GA4:

Custom channel group (recommended). Add perplexity\.ai to your custom AI channel regex. This is the only way to see Perplexity alongside other AI platforms in a unified channel view.
Exploration filter. In any GA4 exploration, filter Session source contains "perplexity" to isolate Perplexity sessions.
Traffic acquisition report. Go to Reports > Acquisition > Traffic acquisition, then search for "perplexity" in the source/medium column. Perplexity sessions appear as perplexity.ai / referral unless you have a custom channel group reclassifying them.

Perplexity also crawls your site with PerplexityBot. Checking your server logs for PerplexityBot hits shows which pages Perplexity indexes and may cite in answers.

How to Track Gemini and Claude Traffic in GA4

Google Gemini sends traffic from gemini.google.com and the legacy bard.google.com. Gemini's referral share surged from 2.31% to 9.0% in the year ending April 2026, making it the second-largest AI referral source globally. Gemini is confirmed in GA4's native AI Assistant channel, so tracking is automatic for sessions with intact referrers. Validate by checking your GA4 Realtime report while clicking through from a Gemini response.

One caveat: Google AI Overviews and AI Mode clicks do not register as AI Assistant traffic. Those sessions count as Organic Search because Google treats its own AI features as part of the search product. There is no native way to separate "clicked from an AI Overview" from "clicked from a blue link" in GA4.

Claude sends traffic from claude.ai. Claude's global referral share is 2.66%, but in B2B contexts it reaches approximately 18% of measurable AI referrals according to Goodie's 2026 AI Search Traffic Report. Claude does not append UTM parameters. Claude's inclusion in GA4's native AI Assistant channel is unconfirmed. Add claude\.ai to your custom regex channel group to ensure Claude traffic is captured regardless.

For each platform, validate your tracking by opening the AI tool, searching for a query that cites your content, clicking through to your site, and checking GA4's Realtime report to confirm the session is classified correctly.

How to Build an AI Traffic Dashboard in GA4

A dedicated AI traffic exploration in GA4 lets you monitor AI referral trends, compare platform performance, and identify which content AI engines cite most. This dashboard takes five minutes to build and becomes the operational layer on top of your channel group setup.

Create a free-form exploration in GA4 with these parameters:

Dimensions:

Session source (to identify which AI platform sent the traffic)
Session medium (to confirm ai-assistant vs. referral classification)
Landing page (to see which content AI engines are citing)
Date (for trend analysis)

Metrics:

Sessions
Engaged sessions
Engagement rate
Key events (conversions)
Average engagement time per session

Filters:

Session source matches regex: chatgpt|openai|perplexity|gemini|bard|claude|copilot|deepseek|grok|meta\.ai
OR Session medium equals ai-assistant

This combined filter catches traffic from both your custom channel group and GA4's native AI Assistant channel, giving you a unified view of all AI traffic regardless of how GA4 classified it.

Add a comparison period (previous 28 days vs. current 28 days) to track whether your AI traffic is growing or declining after content changes. Pages that appear frequently in this report are the ones AI engines trust and cite. These are your highest-value assets for AI visibility optimization.

What AI Traffic Data Reveals About Content Strategy

AI traffic attribution exposes which content AI engines trust enough to cite, and which content is invisible to AI discovery despite ranking in traditional search. This data directly informs both content optimization and earned media investment.

Once your AI traffic attribution is live, three patterns typically emerge:

Pattern 1: High AI traffic, moderate organic traffic. These pages are being cited in AI-generated responses. They tend to have clear answer-first structure, specific claims with named sources, and extractable data (tables, lists, definitions). Double down on these pages. They are your citation architecture foundation.

Pattern 2: High organic traffic, zero AI traffic. These pages rank in traditional search but AI engines are not citing them. Common causes: vague headings without query-specific language, narrative-heavy prose without extractable claims, or missing structured data (FAQ schema, comparison tables). According to the Princeton GEO study (Aggarwal et al., ACM KDD 2024), adding statistics and authoritative citations to existing content improves AI citation visibility by up to 41%. These are your highest-leverage optimization targets for GEO, a distribution tactic within Machine Relations.

Pattern 3: Zero AI traffic, zero organic traffic. These pages are not visible anywhere. Before investing in optimization, verify the page has a genuine query target and source-backed claims worth citing. If not, the page may need a fundamental premise upgrade rather than technical fixes.

The gap between Pattern 1 and Pattern 2 pages is often structural, not topical. A page ranking position 5 for a competitive query but receiving zero AI traffic is almost always missing extractable claim blocks, comparison data, or FAQ schema. Those are the elements AI engines parse for citation.

How Earned Media Drives AI Search Referrals

AI engines cite third-party publications, not brand websites, when generating answers. Earned media placements are the primary driver of AI search referrals to brand sites. According to Muck Rack research, 82 to 89% of AI citations in generated answers come from earned media sources: Forbes, TechCrunch, The Wall Street Journal, Reuters, and industry publications. BrightEdge research corroborates this pattern, finding that AI engines disproportionately cite authoritative third-party content over brand-owned pages.

The AI search referral path works in sequence:

Your brand secures an earned media placement in a publication that AI engines index and trust.
The placement mentions your brand with context, claims, and links.
An AI engine (ChatGPT, Perplexity, Gemini) retrieves that placement when answering a relevant query.
The AI engine synthesizes the placement's content and may cite it with a link.
The user clicks through to your site from the AI-generated response.
GA4 records the visit as AI referral traffic (if properly attributed).

Without AI traffic attribution, step 6 is invisible. You cannot measure which earned media placements generate AI search traffic, which publications AI engines trust most for your category, or whether your earned media investment is producing measurable AI visibility returns.

AI traffic attribution closes this measurement loop. When you can trace a visit from ChatGPT back to a specific landing page, and cross-reference that page with your earned media calendar, you know exactly which placements drive AI search results. This is the earned media ROI measurement that traditional PR metrics (impressions, reach, AVE) never provided.

AuthorityTech guarantees Tier 1 placements in publications AI engines cite (Forbes, TechCrunch, The Wall Street Journal) or you pay nothing. Start with a free AI visibility audit to measure your current citation footprint across ChatGPT, Perplexity, and Gemini.

How GEO, AEO, and SEO Fit Within Machine Relations

These disciplines are not competing alternatives. They represent different layers of the same system. Machine Relations is the full architecture that contains each of them:

Discipline	Optimizes for	Success condition	Scope
SEO	Ranking algorithms	Top 10 position on SERP	Technical + content
GEO	Generative AI engines	Cited in AI-generated answers	Content formatting + distribution
AEO	Answer boxes / featured snippets	Selected as the direct answer	Structured content
Digital PR	Human journalists/editors	Media placement	Outreach + storytelling
Machine Relations	AI-mediated discovery systems	Resolved and cited across AI engines	Full system: authority, entity, citation, distribution, measurement

AI traffic attribution is a Layer 5 (Measurement) function in the Machine Relations stack. Without it, the entire system operates without feedback. You cannot tell which authority signals, entity associations, citation architectures, or distribution tactics actually produce AI search visibility. AuthorityTech's AI visibility research shows that brands with complete AI traffic attribution identify 3 to 5x more optimization opportunities than those relying on traditional analytics alone.

Frequently Asked Questions

Does GA4 now track AI traffic automatically?

Partially. As of May 2026, GA4 automatically classifies traffic from recognized AI assistants under a native "AI Assistant" channel. The confirmed list includes ChatGPT, Gemini, DeepSeek, Copilot, and Grok. Perplexity is not included, and Claude's status is unconfirmed. The native channel only works for sessions with intact referrer headers. Between 35% and 70% of AI referral sessions arrive without referrers and still land in "Direct" traffic. For complete coverage, you need both the native channel and a custom channel group with regex filters that explicitly include Perplexity and Claude.

Can I attribute traffic from ChatGPT or Perplexity in GA4?

Yes. ChatGPT traffic is the easiest to attribute because it uses consistent referrer domains (chatgpt.com, chat.openai.com) and appends utm_source=chatgpt.com to some outbound links. GA4's native channel also recognizes it automatically. Perplexity traffic requires a custom regex channel group because GA4's native AI Assistant channel does not include Perplexity. Add perplexity\.ai to your custom channel group's source regex to capture it.

What regex pattern should I use to track all AI search traffic in GA4?

Why does ChatGPT traffic show as direct in GA4?

ChatGPT's mobile app and some desktop interactions strip the referrer header when opening external links. Without a referrer, GA4 defaults to classifying the session as "direct." ChatGPT does append utm_source=chatgpt.com to some outbound links, which helps, but this parameter is not present on all click-throughs. A custom channel group using both source regex and UTM parameter matching captures the broadest set of ChatGPT sessions.

How do I check Gemini traffic in GA4?

Google Gemini traffic is automatically classified under GA4's native AI Assistant channel. Go to Reports > Acquisition > Traffic acquisition, then filter for Session source containing "gemini" or look at the AI Assistant channel. Gemini sends traffic from gemini.google.com and the legacy bard.google.com. Note: clicks from Google AI Overviews and AI Mode count as Organic Search, not AI Assistant. There is no native way to separate AI Overview clicks from standard organic clicks in GA4.

How much does AI search traffic convert compared to organic search?

AI search traffic converts at approximately 4.4 times the rate of traditional organic search traffic, according to Adobe Analytics. This higher conversion rate occurs because AI search users have already asked a specific question, received a synthesized answer citing your brand or content, and clicked through with higher intent and trust than a typical SERP visitor.

Who coined Machine Relations?

Jaxon Parrott, founder of AuthorityTech, coined the term Machine Relations in 2024 to describe the discipline of earning AI citations and recommendations by making a brand legible, retrievable, and credible inside AI-driven discovery systems. Machine Relations encompasses SEO, GEO, AEO, and digital PR as component layers within a unified five-layer stack.

Where do GEO and AEO fit inside Machine Relations?

GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) are distribution tactics within Layer 4 of the Machine Relations stack. They optimize how content is formatted and delivered for AI extraction and citation. Machine Relations contains GEO and AEO but extends beyond them to include authority building (Layer 1), entity clarity (Layer 2), citation architecture (Layer 3), and measurement (Layer 5), including the AI traffic attribution covered in this guide.