GA4 now recognizes some AI referral traffic automatically through its native AI Assistant channel, but it only covers a handful of sources and never reclassifies data collected before it existed. This guide shows you how to build your own custom channel group, reorder your evaluation rules, and isolate human visits from every AI platform that matters to your traffic -- including the ones Google's own channel leaves out. By placing a custom rule above your standard referral category, you separate real AI referral sessions from standard website backlinks, for sources old and new alike.

Quick answer: To track all AI referral traffic in GA4, create a custom channel group with rules defining your specific AI sources—including platforms like Perplexity that Google's default settings miss—and position this rule above your standard referral category. This configuration retroactively and prospectively separates real human visits initiated by AI assistants from standard website backlinks.

Why GA4 doesn't separate AI referral traffic by default

Google Analytics 4 (GA4) added a native AI Assistant default channel in 2026, which automatically classifies visits from a specific list of recognized sources -- currently ChatGPT, Gemini, Claude, DeepSeek, Copilot and Grok. It's a real improvement, but it has two gaps that matter for most sites: Perplexity isn't on the list and can't be added to it, and the classification only applies going forward -- sessions GA4 recorded before the channel existed stay in their original bucket. A custom channel group closes both gaps: it covers whichever sources you choose, in your own naming, and -- unlike the native channel -- applies retroactively to data you've already collected.

AI sources that GA4's native AI Assistant channel doesn't recognize -- Perplexity chief among them -- still fall into the generic Referral channel. This default behavior, as noted in an analytics guide by Analytics Mania, means that traffic gets mixed in with directory links, blogs, and spam. A custom channel group lets you classify these additional AI sources alongside the ones GA4 already recognizes, so nothing falls through the cracks.

AI crawlers vs AI referral visits: what you actually need to track

AI referral visits are sessions initiated by real human users who click on a link recommended by an AI assistant and land on your website with active referrer data.

Distinguish these human visits from AI crawlers. Crawlers and bots fetch your web pages to index content or train large language models, but they do not execute tracking scripts the way a human browser does. Because they do not behave like real GA4 sessions, they will not show up in your standard traffic reports.

The configuration steps in this guide focus on measuring real human interactions. For instance, PerkFuel sends real, targeted website visitors from AI assistants such as ChatGPT, Gemini, Claude, and Perplexity. Knowing what counts as LLM referral traffic allows you to isolate these genuine prospective customers from background bot activity inside your analytics dashboard.

How to track AI referral traffic in GA4: step-by-step setup

To build a broader, fully customizable view of AI referral traffic -- covering more sources, your own naming, and full reporting control -- you can create a custom channel group instead of relying only on GA4's default classification. GA4 does not allow you to modify default settings directly, so copying the existing structure maintains a clean baseline while adding custom logic.

Create the custom channel group

Begin by copying your existing channel rules. This retains all the standard traffic classifications, such as Direct and Organic Search, while adding your new rules.

  1. Navigate to your GA4 Admin panel.
  2. Click on Data display and select Channel groups.
  3. Click Create new channel group to copy the default setup as your template.

Define the AI source rule

Next, define the specific criteria that GA4 will use to identify and group incoming sessions originating from AI platforms.

  1. Click Add new channel and name the channel "AI Referral".
  2. Select Session source as your matching condition.
  3. List the specific AI assistant referrers you want to group, or use a regular expression to match multiple sources.

Set the rule order

GA4 assigns traffic to the first matching rule it encounters from top to bottom. According to a GA4 guide by Swydo, you must place your custom AI Traffic channel above the standard Referral channel. If you do not prioritize it, AI visits will match the generic Referral rule first and fall into that default bucket.

  1. Drag your custom AI Referral rule so that it sits above the standard Referral rule in the evaluation order.
  2. Review the entire list sequence to make sure your custom rule is evaluated first.

Save and apply it to your reports

Once the sequence is correct, save your changes and apply the new grouping to your active analytics dashboard.

  1. Click Save group to store your custom channel configuration.
  2. Open your standard Traffic acquisition report.
  3. Change the primary dimension from the default group to your new custom channel group to begin analyzing the data.

Where to see AI referral traffic once it's set up

Once you save the custom channel group, GA4 applies it retroactively -- your existing historical sessions get re-grouped under the new rules, not just traffic from this point forward. Navigate to the Traffic acquisition report and switch the primary dimension to your custom channel group. You will see AI Referral listed as its own dedicated row alongside default channels like Organic Search and Direct, covering both past and future sessions.

To break this channel down by individual AI assistants, build a free-form exploration. By adding Session source/medium as a dimension, you can isolate exactly how much traffic originates from specific platforms such as ChatGPT, Gemini, Claude, or Perplexity.

Look beyond raw session counts. Focus on user quality metrics, such as the engagement rate and conversion rate per assistant. AI-referred users often arrive with specific intent, and these engagement metrics show which assistants drive valuable actions on your site.

Why your AI referral numbers will always undercount reality

Even after configuring your custom channel group, the traffic numbers you see in GA4 will be lower than the actual volume of visitors coming from AI platforms. This discrepancy is a structural limitation of modern web tracking rather than a mistake in your analytics configuration.

A major cause of this gap is how users interact with AI assistants. When a user copies a link from an AI chat window and pastes it directly into their browser address bar, the session begins without any referrer data. GA4 categorizes these visits as Direct traffic because there is no source website to pass along to your tracking script.

Many native AI applications and privacy-focused mobile browsers strip out referrer headers entirely before the user lands on your site. Without this header, your analytics platform has no way of identifying the session origin. Treat your tracked AI referral data as a baseline trend rather than an absolute, literal count of every single visitor.

How to confirm your AI channel group is actually working

Do not wait for weekly or monthly reports to compile before verifying your new GA4 configuration. Check the real-time report immediately after setup by triggering an AI referral click and watching the session register. This immediate feedback ensures your matching rules are properly aligned before you rely on the data for long-term marketing decisions.

Open your Realtime report and monitor the incoming traffic right after generating the click. Confirm that the session source and medium match your new rule exactly and fall into the custom AI Referral channel. If your brand does not have an active, organic AI mention to test with yet, you can send real AI referral visits to test the setup on a specific landing page.

These test visits are tracked in the PerkFuel dashboard and, with analytics tracking enabled, in Google Analytics. While PerkFuel guarantees delivery of the visits included in your plan, using them for initial verification lets you see exactly how the traffic registers under your newly created channel group.

Growing AI referral traffic once you can measure it

With a validated channel group, you can measure and optimize your growth decisions based on real performance data, isolating which assistants, countries, and landing pages yield the highest conversions.

View your AI traffic as distinct categories. For example, you can focus on targeting ChatGPT specifically to evaluate user behavior. You can drive high-intent users by targeting Gemini referrals. You can also expand your reach by targeting Claude referrals. Finally, targeting Perplexity referrals lets you capture users looking for direct, citation-backed answers.

When setting up a project in PerkFuel, you choose your AI sources, target countries, landing pages, and delivery pace. The platform supports 12 AI sources: ChatGPT, Gemini, Claude, Perplexity, Copilot, Grok, Meta AI, DeepSeek, Mistral, Manus, Poe, and Kimi. You can target 65+ countries per campaign, matching your GA4 data to deliberate, localized growth campaigns.

Implementing this custom channel group gives you the data needed to measure how visitors from AI platforms interact with your site. Keep in mind that privacy settings, native apps, and direct copy-pasting mean your analytics will slightly undercount the true volume of these sessions. Once your tracking is validated, you can scale your traffic by running targeted AI referral campaigns aimed at your highest-converting landing pages.