All AI Referral Traffic Sources

When someone asks ChatGPT, Perplexity, or Gemini a question and clicks through to your site, that visit lands in your analytics like any other referral — but under a hostname most teams have never filtered for. If you aren't looking for these referrers explicitly, AI-driven traffic is almost certainly being lumped into "Other" or mistaken for direct traffic.

This guide lists the referrer hostnames the major AI assistants send, and shows how to isolate them in your analytics so you can measure what your AI visibility is actually worth.

The referrer hostnames to track

Each assistant passes its own hostname in the HTTP referrer when a user clicks a cited link. Track these:

OpenAI / ChatGPT

• chatgpt.com — the current primary referrer for ChatGPT clicks
• chat.openai.com — the legacy hostname; still worth keeping in your filter for historical data

Perplexity

• perplexity.ai (and www.perplexity.ai) — Perplexity cites sources prominently, so it often produces a higher click-through rate per citation than assistants that summarise without linking

Google Gemini

• gemini.google.com — clicks from the Gemini assistant

Note that Google's AI Overviews inside normal search results are a separate case: those clicks generally arrive as ordinary google.com organic traffic, not as a distinct AI referrer.

Microsoft Copilot

• copilot.microsoft.com — the standalone Copilot experience
• bing.com — Copilot answers surfaced inside Bing arrive under the standard Bing hostname

Other assistants worth adding

• claude.ai — Anthropic's Claude
• meta.ai — Meta AI
• grok.com and x.com — Grok, in-product and via X
• you.com — You.com
• poe.com — Poe

How to isolate AI referrals in GA4

1. Open Reports → Acquisition → Traffic acquisition and switch the primary dimension to Session source. This alone will surface the hostnames above once you have any AI traffic.
2. For a persistent view, build an Exploration: add Session source as a dimension, then apply a filter using the matches regex condition with a pattern covering the hostnames you care about — for example chatgpt|openai|perplexity|gemini|copilot|claude|meta\.ai|grok|you\.com|poe\.com.
3. Save it as a comparison or audience so AI-sourced sessions can be trended over time and compared against organic search.
4. Watch engagement metrics, not just sessions. AI referrals tend to arrive with high intent because the user has already read a summary and chose to click through for more.

If you use a different analytics tool, the principle is identical: filter on the referrer hostname rather than relying on default channel groupings, which usually don't yet recognise these sources.

Why the referrer list is only half the picture

Referrer tracking tells you about the users who clicked. It says nothing about the far larger number of times an AI assistant crawled your content, used it to answer a question, and cited you without the user ever clicking. That gap — between how often you're read and how often you're visited — is where most of your AI visibility actually lives.

To see the whole funnel you need both halves: the agent crawls hitting your site, and the human referrals they eventually produce. AEO connects the two, so you can see which content earns AI citations and what that visibility is worth in real traffic.

Getting started

1. Add the hostnames above to an analytics filter or exploration today — it takes minutes and immediately reclassifies traffic you're already receiving.
2. Track the trend monthly rather than daily; AI referral volumes are still small for most sites but growing, and the trend matters more than any single day.
3. Make sure AI crawlers can actually reach your content in the first place — if your pages need JavaScript to render, many assistants will never see the text they'd otherwise cite.