When your website gets recommended by conversational engines like ChatGPT or Perplexity, those high-value visitors often disappear into your direct traffic reports. You can accurately identify AI traffic in Google Analytics (GA4) by drilling down into session source and medium data, building custom channel groups, and filtering out non-human crawler bots. Isolating these specific referral sessions lets you attribute real human signups to the exact AI models recommending your product.

Navigate to Reports > Acquisition > Traffic acquisition in GA4, change the primary dimension to Session source/medium, and filter for domains like chatgpt.com, perplexity.ai, or claude.ai. You can permanently group these sources together by creating a custom channel group with a matching regex filter.

What Counts as AI Traffic in Google Analytics

Before modifying any Google Analytics 4 (GA4) settings, you must distinguish between the different types of AI-driven activity hitting your website. Not all AI interactions represent human visitors, and GA4 handles each type differently.

AI referral traffic is defined as sessions initiated by real human users who click on a link recommended within an AI assistant's chat response—such as ChatGPT, Claude, Gemini, or Perplexity—and land on your website.

This is fundamentally different from AI crawler or LLM trainer traffic, which consists of automated bots scraping your site content to train large language models. As discussed by analytics professionals on Reddit, you can track active AI referral traffic from human users, but crawler traffic represents non-human background activity that requires different handling. Clicks from Google's AI Overviews generally cluster under standard organic search traffic rather than appearing as a distinct AI source.

When marketers talk about identifying AI traffic to measure ROI, they are referring to the human visits generated by AI recommendations.

How to Identify AI Traffic in Google Analytics 4

Google Analytics 4 does not classify AI referral traffic into a dedicated channel by default. These valuable visits are mixed in with your standard referral data, or masked as direct traffic. To find them, you must examine the source and medium level of your acquisition data directly.

Navigate to Reports > Acquisition > Traffic acquisition inside your GA4 property. Change the primary dimension of your table to Session source/medium. From there, use the search bar to filter for known AI assistant domains:

Group AI referrers into a custom channel

Searching for individual domains every time you run a report is inefficient. By building a custom channel group, you can bundle all AI traffic sources under a single label. For a comprehensive walkthrough of this configuration, refer to a step-by-step setup for tracking AI referral traffic in GA4.

  1. Go to the GA4 Admin panel.
  2. Select Data display and click on Channel groups.
  3. Create a new custom channel group based on your default group.
  4. Define a new channel named "AI Referral" within the group.
  5. Set the channel rules to match the source using a regular expression (regex) that includes your targeted AI domains, such as chatgpt|perplexity|claude|gemini|copilot.
  6. Save your changes to make the new group available across your standard reports.

Build an exploration report for AI sources

To analyze visitor behavior past the initial landing, standard reports are often too rigid. A custom exploration report allows you to break sessions down by both source and landing page.

Create a blank exploration report. Add Session source/medium and Landing page as your dimensions, and select Sessions and Active users as your metrics. Drag the dimensions into the rows section and the metrics into the values section. This grid shows you exactly which of your pages AI models are recommending.

Why AI Traffic Sometimes Shows Up as Direct Traffic

AI assistants often operate within secure chat interfaces or native mobile applications. When a user clicks a recommended link, some AI platforms fail to pass the full referrer headers. Because this referral data is stripped during the handoff, GA4 defaults to logging the visit as (direct)/(none).

This data loss means you cannot rely purely on standard referral reports. A sudden, unexplained rise in direct sessions to specific deep-linked pages—with no matching email campaign, social push, or offline marketing activity—is a strong signal that AI recommendations are driving the traffic. You can start troubleshooting this by confirming ChatGPT visits aren't hiding in your direct traffic report.

To verify these spikes, cross-check the timing and landing pages against your known AI-referral activity. If you run campaigns to boost your visibility, visits are tracked in the PerkFuel dashboard. Comparing the timestamped trends in your dashboard against the timing of unexplained direct-traffic spikes in GA4 allows you to confirm the real source of those visitors.

Filtering Bot and Crawler Traffic Out of Your Reports

Google Analytics 4 automatically filters out traffic from standard, known search engine bots. However, GA4's default exclusion lists are not built to dynamically isolate every automated crawler or LLM-training bot deployed by AI companies. This can skew your acquisition reports if left unaddressed.

Many LLM-trainer crawlers do not execute JavaScript. Consequently, they never register as a session inside GA4 at all and instead only appear within your web server logs. For the automated agents that do trigger JavaScript, you must rely on behavioral signals to separate them from genuine, human AI referral visits.

Use the following criteria to determine if a suspected traffic spike is automated bot activity rather than a human user referred by an AI assistant:

Turning Identified AI Traffic Into a Growth Channel

Once you can reliably identify AI referral traffic in GA4, the priority shifts from basic diagnostics to intentional scaling. Knowing how to isolate these sessions is only valuable if you have enough traffic to move your core business metrics. You can increase this volume by sending trackable AI referral traffic to your site using PerkFuel.

PerkFuel sends real, targeted website visitors from popular AI assistants, including ChatGPT, Gemini, Claude, and Perplexity. When setting up a project, you choose the specific AI sources you want traffic from, target countries, landing pages, and the overall delivery pace. This control ensures your new visitors land exactly on the high-value pages you configured in your GA4 custom channel groups.

All delivered visits are tracked directly within the PerkFuel dashboard. With analytics tracking enabled, they also populate in Google Analytics so you can measure their behavior alongside your organic and paid acquisition channels. New accounts receive 10 free visits with no credit card required.

Isolating your AI referral sessions in GA4 requires moving past default channel settings and actively monitoring unclassified traffic. Remember that secure mobile apps and chat interfaces often strip referrer data, meaning you must cross-reference direct-traffic spikes with your active campaign timelines. Once your tracking framework is configured, you can test and scale this channel by utilizing PerkFuel to send targeted, trackable AI visitors directly to your optimized landing pages.