How GA4 records traffic from Perplexity Comet and ChatGPT Atlas

AI browsers disrupt how GA4 records referral data, obscuring traffic sources and discovery patterns. Find out how to track them accurately.

As AI-powered browsers Perplexity Comet and ChatGPT Atlas reshape how users discover and interact with content, their influence is beginning to surface in web analytics. Understanding how these emerging traffic sources appear in Google Analytics 4 (GA4) is essential for accurate measurement and attribution. The challenge is that these browsers pass referral data in different ways, creating gaps and inconsistencies in reporting.

How traffic from Comet and Atlas appears in GA4

GA4 records visits based on referrer and session data, but the way this information is passed depends on how each AI browser handles outbound links.

When users click links in Perplexity Comet, sessions often appear in GA4 with a source like perplexity.ai and medium referral, making it relatively easy to identify in traffic reports.

ChatGPT Atlas, by contrast, operates more like an embedded browser within the ChatGPT ecosystem. Links opened through it often strip or block referrer headers, so sessions may appear as “Direct” or occasionally as (not set) in source/medium — limiting visibility into its true contribution to discovery and engagement.

Testing both browsers across multiple websites shows variable results. In some cases, sessions appear in GA4 or Microsoft Clarity in real time, while in others they fail to register entirely, both in live tracking and in retrospective data. These differences make it difficult to assess how reliably traffic from either AI browser is captured.

 

Why referrers might not appear in analytics platforms

Referrer data may not appear in analytics platforms for several technical and privacy-related reasons. AI browsers, privacy-focused modes and embedded environments often block or modify the transmission of referrer headers when a user clicks a link. Key reasons include:

• Embedded browsers and privacy controls: AI browsers such as Atlas or Comet may use sandboxed environments that suppress referrer headers to protect user privacy or maintain session integrity.
• HTTPS to HTTP transitions: When a link moves from a secure (HTTPS) environment to a non-secure (HTTP) one, most browsers remove referrer data for security reasons.
• Tracking prevention technologies: Browser-level features such as Safari’s Intelligent Tracking Prevention and similar privacy mechanisms in Chromium or Edge can strip referrers or truncate them to the root domain.
• App-based browsing: Mobile apps and AI-driven interfaces sometimes open links through webviews, which may not behave like standard browsers. These webviews frequently omit referrer information altogether.
• Crawling and pre-fetching: AI tools that prefetch or preview web content before showing it to users often bypass client-side analytics scripts entirely, meaning those impressions are never recorded.

Together, these behaviors explain why traffic from AI browsers may appear incomplete or missing in analytics systems, even when users are actively engaging with the content.

The risk of AI browsers inflating ad spend

Recent reporting has raised concerns about the potential for AI browsers to distort both advertising spend and analytics accuracy. In October 2025, Search Atlas highlighted that OpenAI’s ChatGPT Atlas browser can interact with web pages in ways that closely mimic human behavior, including clicking on paid advertisements. 

Because Atlas is built on Google Chrome, ad networks interpret these actions as genuine user activity.

This matters because:

• Every AI-triggered click on a sponsored link can consume budget as if it were from a real prospect.
• Analytics platforms may record these sessions as legitimate traffic, which can affect conversion rates and ROI analysis.
• Traditional bot-detection tools are ineffective at identifying Atlas-driven interactions, as they originate from standard browser environments.

What to look for:

• Unusual spikes in referral or direct traffic coinciding with campaign spend.
• Higher-than-expected click volumes but lower engagement or conversions.
• Patterns in session behavior that suggest automation, such as uniform visit lengths or repetitive navigation.

Search Atlas founder Manick Bhan warned that this issue could accelerate the creation of new industry standards for distinguishing human from AI traffic. As AI browsing agents become more active, accurate traffic verification will be critical for protecting ad budgets and maintaining data integrity.

 

Why the difference matters

These differences impact how AI-origin traffic is attributed and how performance data feeds into decision-making systems.

• Referrer data: Perplexity Comet behaves more like a traditional search or discovery engine, passing referrer details through to analytics. Atlas, by contrast, tends to obscure referral information due to its architecture and privacy handling.
• Attribution accuracy: Without referrer data or UTM tagging, AI-driven traffic from Atlas can inflate “Direct” traffic metrics, masking real patterns of discovery and engagement.
• Data integrity: GA4’s automated bot filtering can exclude some AI-driven link previews, meaning not every impression or interaction makes it into analytics reports.

Strengthening measurement and attribution

To maintain accuracy in reporting and executive dashboards, treat AI browser traffic as an emerging acquisition channel and adapt their analytics setup accordingly.

• Identify AI-origin traffic: In Reports > Acquisition > Traffic acquisition, set the primary dimension to Session source/medium and look for entries like perplexity.ai/referral or chat.openai.com/referral.
• Create a dedicated channel group: In Admin > Data settings > Channel groups, create a new category called “AI Tools.” Add a regex rule that captures likely AI sources and place it above the default “Referral” rule so these sessions are grouped correctly:
  • (chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com)
• Incorporate UTMs for experimentation: If your organization engages with these platforms through content partnerships or experiments, append unique UTMs to track performance.
• Build an Explore report: Use GA4’s Explore workspace to isolate and analyze engagement, conversion and session metrics for AI-driven traffic. This will provide clearer insight into early-stage discovery patterns.

Strategic considerations for CMOs and CTOs

AI browsers are introducing a new layer of discovery that sits between traditional search engines and social platforms. This shift demands a forward-looking approach to data architecture.

• Channel forecasting: Treat AI browsers as a distinct discovery layer and model their impact in attribution frameworks.
• Data pipelines: Ensure your analytics, CDP and marketing automation platforms can consistently capture and label AI-origin traffic.
• Measurement innovation: Push analytics teams to design frameworks that can accommodate opaque traffic sources where referrer data is limited.

 

Key takeaway

Perplexity Comet and ChatGPT Atlas mark the next frontier of search-driven traffic, but GA4 interprets them differently.

• Comet typically passes referrer information and appears as a clear referral source.
• Atlas often masks its origin, blending in with direct traffic.

Recognizing and adjusting for these differences is key to preserving data accuracy, understanding new discovery patterns and strengthening attribution in the AI browsing era.

The post How GA4 records traffic from Perplexity Comet and ChatGPT Atlas appeared first on MarTech.

MarTech

(1)