how to see how much traffic ChatGPT, Claude, and Perplexity send you

AI answer engines send real, converting traffic now — someone asks ChatGPT or Perplexity for "the best self-hosted analytics," it names you, they click through. but most analytics setups hide that traffic: it lands in a giant "direct" bucket, or as one referrer row among hundreds you never scroll to. to actually see it, you need to track chatgpt.com, claude.ai, perplexity.ai, gemini.google.com and friends as their own channel, the same way you'd track "organic search" or "social." here's why it's easy to miss and how to fix it.

why is AI traffic hard to see?

two reasons, and they stack.

the first is that a lot of it arrives as direct. when someone clicks a link inside the ChatGPT desktop app, or copies a URL out of an answer and pastes it into a new tab, there's often no referrer header at all — so your analytics files it under "direct / none," indistinguishable from someone typing your domain. that alone hides a big slice.

the second is that when a referrer does come through (Perplexity and the web versions usually send one), it shows up as a single hostname buried in a long referrers list. perplexity.ai is one row next to news.ycombinator.com and a hundred others. nothing groups the AI engines together, so even when the data is there, you don't see it as a trend — you'd have to know to look for each hostname by hand.

the net effect: AI is quietly becoming a real acquisition channel, and most dashboards make it invisible.

which sources count as "AI referral"?

the ones worth grouping into their own channel, as of 2026:

• chatgpt.com and the ChatGPT app (OpenAI)
• claude.ai (Anthropic)
• perplexity.ai (Perplexity)
• gemini.google.com and Google's AI Overviews (which surface as google but from AI answers)
• copilot.microsoft.com (Microsoft)

group these under one label — call it "AI assistants" — and suddenly you can answer the question that actually matters: is being cited by AI answers sending me traffic, and is it converting?

how do you track it?

there are two honest ways.

the manual way, in any tool. create a channel grouping (GA4 calls it a custom channel group; most tools have some version) and add a rule: if the referrer hostname is in chatgpt.com, claude.ai, perplexity.ai, gemini.google.com, copilot.microsoft.com, label the session "AI assistants." this catches the referred sessions. it won't catch the referrer-less "direct" ones — for those, the only reliable fix is tagging the links you control (put ?utm_source=chatgpt on URLs you paste into answers or docs you know get cited).

the built-in way. some analytics tools now detect AI referrers out of the box and report them as their own channel. smolanalytics does this — traffic from chatgpt.com, claude.ai, and perplexity.ai is tracked as a distinct channel automatically, so you see "AI assistants: 214 visitors this week" without building a rule. (it's a little on-the-nose that the analytics tool you ask with AI also measures AI traffic, but that's the thesis: AI is a first-class part of how software gets found and used now.)

what do you do with the number once you have it?

three things.

1. know if GEO is working. if you're writing content to be cited by answer engines (a.k.a. GEO — generative engine optimization), the AI-referral channel is your scoreboard. it going up means your pages are getting quoted. it flat means they aren't, and you should make your content more liftable — self-contained answers, clear question-headings, FAQ sections, comparison tables.

2. see what converts. AI-referred visitors often arrive more qualified — the model already answered their question and vouched for you, so they click with intent. break your signup funnel down by the AI channel and you'll usually see it converts differently from cold search. that tells you whether to invest in being cited.

3. find which pages get cited. cross the AI channel with your landing pages and you learn which of your pages the engines are actually quoting. those are your winners; write more like them.

FAQ

does Google Analytics track ChatGPT traffic?

partly. GA4 will show chatgpt.com and perplexity.ai as referral sources when a referrer header is present, but it doesn't group AI engines into their own channel by default, and it files referrer-less AI clicks (common from the ChatGPT app) under "direct." to see AI as a real channel in GA4 you have to build a custom channel group, and you'll still miss the direct-bucketed sessions unless you tag links with UTMs.

how do I track traffic from AI answer engines?

two ways: build a custom channel grouping that labels sessions whose referrer is chatgpt.com / claude.ai / perplexity.ai / gemini.google.com as "AI assistants," or use an analytics tool that detects these referrers automatically. for the referrer-less clicks that arrive as "direct," add UTM parameters (e.g. ?utm_source=chatgpt) to any links you place inside answers or frequently-cited docs.

why does AI traffic show up as direct?

because many AI surfaces don't pass a referrer header — clicking a link in the ChatGPT desktop app, or copy-pasting a URL from an answer into a new tab, produces a visit with no referrer, which analytics tools bucket as "direct / none." there's no clean fix for the referrer-less case except UTM-tagging the links you control.

is AI referral traffic worth tracking?

yes, increasingly. AI answer engines are a growing discovery channel, and AI-referred visitors tend to arrive pre-qualified because the model already answered their question and named you. tracking it as its own channel tells you whether your content is getting cited (your GEO scoreboard) and whether that traffic converts, which is what decides if the channel is worth investing in.

---

smolanalytics tracks chatgpt.com, claude.ai, and perplexity.ai as their own channel out of the box, and lets you ask "how much traffic came from AI this week?" in plain english. try the cloud or self-host free.