How to Track AI Referral Traffic in GA4 for Your Hotel

GA4's default channel grouping was built before AI assistants sent meaningful referral traffic, and as of 2026 it still doesn't have a distinct "AI assistant" or "LLM referral" channel. A click from a ChatGPT answer typically lands in "Referral" (or sometimes "Unassigned"), mixed in with every other kind of referring site, with no separation from AI traffic specifically unless you build one yourself.

The main ones worth tracking: chatgpt.com and chat.openai.com (ChatGPT), perplexity.ai (Perplexity), gemini.google.com (Gemini), and copilot.microsoft.com (Microsoft Copilot). Claude's web interface (claude.ai) can also send referral clicks, worth including if you're tracking all five engines consistently with your Staylight scans.

In GA4, go to Explore, then start a blank Exploration. Add "Session source / medium" as a dimension and "Sessions" plus "Conversions" as metrics. Add a filter on "Session source" using a regex match: chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai. Set the date range to at least the last 90 days so you have enough data to see a trend rather than noise from a handful of sessions.

Save the Exploration with a clear name ("AI Assistant Referral Traffic") so it's easy to find again, and pin it or bookmark it, since you'll want to check it on a recurring basis, not just once.

Add "Session source" (not the combined source/medium) as a second dimension in the same Exploration, and GA4 will split the regex-matched total into chatgpt.com, perplexity.ai, gemini.google.com, and so on individually. This is worth doing before drawing conclusions from the aggregate number: a hotel getting all its AI referral traffic from one engine and none from the other four is a different situation than one with a small, even spread across all five, and it points to a different next step.

A GA4 Exploration works fine for a manual monthly check, but if you want this to show up automatically alongside other reporting, connect GA4 as a data source in Looker Studio (or whatever reporting tool your hotel already uses), add a time-series chart using the same regex filter on session source, and a scorecard showing total AI-referred sessions for the current period versus the prior one. That takes the same underlying filter and turns it into something a GM or owner can check without opening GA4 directly.

This is the most important limitation to understand before acting on the number: most AI-generated answers name a hotel in text without including a clickable link at all. A guest who reads "Golden Well Hotel is a great choice" in a ChatGPT answer and then searches the name directly on Google shows up in GA4 as organic or direct traffic, not as an AI referral, even though the AI assistant is exactly what drove that visit. Referral-traffic tracking only catches the subset of AI-influenced visits that came through an actual clicked link, a real floor on AI's impact, not the ceiling.

AI referral traffic and detection rate answer two different questions and are most useful read together, not separately. Detection rate (from your weekly Staylight scan) shows whether AI engines are naming your hotel at all. Referral traffic shows whether any of those mentions are turning into an actual site visit through a clicked link. A rising detection rate with flat referral traffic suggests AI is naming you but not linking to you, worth checking whether your site is the source being cited, or whether an OTA listing is getting the click instead. A rising referral trend alongside a rising detection rate is the clearest real signal that visibility work is translating into actual traffic.

This isn't a check-daily metric, sessions from AI referrers are still a small slice of most hotels' total traffic, and daily numbers will be noisy. A monthly review, alongside your Staylight detection-rate trend, is enough to see whether the two are moving together over time.