How to Track Traffic from ChatGPT and Perplexity to Your Website in 2026

To track traffic from ChatGPT and Perplexity, open Google Analytics 4, go to Reports → Acquisition → Traffic acquisition, and look at the Session source / medium dimension: ChatGPT web appears as `chatgpt.com` or `chat.openai.com` (medium `referral`) and Perplexity appears as `perplexity.ai` (medium `referral`). That's the quick answer. The hard part — the reason most people think AI sends them no traffic — is that a big slice of real AI visits never shows up as a referral at all: the ChatGPT and Perplexity mobile apps strip the referrer, so those visits are logged as `Direct`, and Google Search Console still doesn't separate AI-engine clicks. So the honest method is two-part: (1) capture the referrals you can see with a dedicated AI channel group, and (2) account for the ones you can't by watching Direct traffic and tagging every link you place yourself with UTM parameters.

If you've added your site to ChatGPT or Perplexity answers and want proof it's working, this is how you get it. Below is the exact GA4 setup, the free Explore report that shows which of your pages the AI engines are actually quoting, and the blind spots to adjust for — written for founders and marketers, not analysts. We build and measure AI-visible content every day at FRPROTECH as a Top Rated Plus studio on Upwork (8+ years, 3,000+ projects across 30+ countries, 100% Job Success), and the method here is the one we use on client sites. AI tools and analytics interfaces change often, so treat the menu paths as of late 2026 and verify the current layout in your own GA4 property before you rely on a number.

Where does ChatGPT and Perplexity traffic show up in Google Analytics?

By default, AI referral traffic lands in your normal GA4 reports — it just isn't labelled as "AI", so you have to go looking for it by source. In Reports → Acquisition → Traffic acquisition (or Engagement → Pages and screens for page-level detail), switch the primary dimension to Session source / medium and scan for these values:

Two things jump out. First, Perplexity is usually clean: it passes a referrer, so `perplexity.ai` visits are easy to find. Second, ChatGPT is split — web sessions are visible, but the app (which a huge share of people now use) sends no referrer, so those visits disappear into Direct. And AI Overviews inside Google search aren't a separate source at all; a click from an AI Overview is still counted as `google / organic`, which is why you can't cleanly isolate it — the broader picture of ranking inside those answers is covered in our guide to ranking in Google AI Overviews.

How do I create an AI traffic channel in GA4?

The cleanest way to measure AI as one line is a custom channel group — a rule that buckets all the AI sources together so you can see total AI traffic trend over time, instead of hunting for each source by hand. Set it up once:

1. In GA4, go to Admin → Data display → Channel groups, and create a new custom channel group (you can clone the default and add to it).
2. Add a new channel at the top of the list called "AI Assistants" so it's evaluated before Referral and Organic.
3. Set the condition to: Session source matches regex, then paste a pattern that lists the AI hosts, for example: `chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|edgeservices\.bing\.com`.
4. Save, name the channel, and apply the channel group. New data will now roll these sources into one "AI Assistants" channel.
5. Revisit the regex every month or two and add new engines as they appear — the list of AI sources is still growing fast.

A note on GA4's own handling: through 2026 some properties began showing an AI-related channel automatically, but reports on its exact behaviour and coverage conflict, the source list is incomplete (Perplexity, notably, is often missing), and it isn't retroactive. So check whether your property has one, but don't rely on it alone — a custom channel group you control is more complete and more predictable. Custom channel groups also don't apply to historical data, so the sooner you create one, the longer your trend line.

Which of my pages are ChatGPT and Perplexity citing?

This is the report that actually changes what you do next, because it tells you which content the AI engines trust enough to quote — and that's where you double down. Build it as a free-form Explore:

• Open Explore → Free form (blank).
• Add Session source / medium and Landing page + query string as rows (or use your new AI Assistants channel as a filter).
• Add Sessions and Engaged sessions (or Key events) as values.
• Filter Session source to the AI hosts (or filter the channel to AI Assistants).
• Read it top-down: the landing pages with the most AI sessions are the pages ChatGPT and Perplexity are citing most — your AI-visibility winners.

Patterns emerge quickly. Usually it's your clearest, most direct, well-structured answer pages that get cited — the ones that answer a specific question in the first sentence. That's not a coincidence; it's the same structure that earns citations in the first place, which we break down in our guides to getting your brand mentioned in ChatGPT and the wider shift from search to answers in AEO vs SEO. Once you know which pages win, you replicate their structure on the topics that still aren't getting cited.

Why is my AI traffic showing up as Direct — and how do I handle it?

Because the ChatGPT and Perplexity native apps don't pass a referrer, so GA4 has nothing to attribute the visit to and files it under Direct. This is the single biggest reason people underestimate their AI traffic. You can't convert those Direct sessions into labelled AI sessions after the fact, but you can account for them three ways:

• Watch the Direct trend as a proxy. A steady, unexplained rise in Direct traffic — with no email blast, no offline campaign, no app of your own — that tracks alongside your visible AI referrals is very likely hidden AI-app traffic. Trend it month over month.
• Tag the links you control with UTM parameters. Anywhere you place your own URL that an AI might read and surface — your Upwork profile, directory listings, your `llms.txt`, social bios, guest posts — add UTM tags (for example `?utm_source=ai&utm_medium=profile`). Those clicks then arrive correctly attributed instead of as Direct.
• Use server-log or bot analysis for the crawl side. Your server logs (or a log-analysis tool) show when `GPTBot`, `PerplexityBot`, `OAI-SearchBot` and similar crawlers fetch your pages. That's not the same as a human visit, but a page being crawled by AI bots is a leading indicator that it can be cited — pair it with the referral data for the full picture.

Does Google Search Console show AI traffic?

Not as a separate line. Clicks from Google's AI Overviews are folded into your normal Search Console performance data as ordinary Google clicks and impressions — there's no dedicated "AI Overview" filter that isolates them, and ChatGPT and Perplexity aren't in Search Console at all because they're not Google. So Search Console stays essential for your Google visibility, but it will not, on its own, tell you how much traffic the AI answer engines send. GA4 (with the AI channel group above) plus your server logs are where that story lives. The practical upshot: run both, and don't expect one tool to show the whole funnel — getting cited consistently across engines is a content-structure and brand-consistency discipline, which is the thread running through our work on AI brand consistency.

What should I actually do with AI traffic data once I can see it?

Measurement is only worth it if it changes the next decision. Once you can see AI traffic, use it to steer content and prove ROI:

• Find your cited pages and make more like them. Match the structure of your AI winners — direct answer first, question-shaped headings, tight lists and tables — across the topics you want to own.
• Set up a simple monthly view. Trend AI Assistants sessions, your cited landing pages, and your Direct-traffic baseline together, so you can show the AI channel growing even while Google organic is slow.
• Connect it to leads, not just visits. Mark your contact or enquiry as a Key event in GA4 so you can see whether AI-sourced sessions convert — this is how you justify investing in answer-engine visibility to a founder or client.
• Feed it into your wider plan. AI referrals pair naturally with social and organic; our social media marketing packages for small businesses guide covers how the channels reinforce each other.

That's the whole method: capture what GA4 can see with a custom AI channel, surface your cited pages in Explore, and adjust for the Direct/app blind spot with UTM tags and trend-watching. Do it once and you turn "I think AI is sending us traffic" into a number you can grow. If you'd rather have your content built and structured to be cited by AI answer engines — and measured properly from day one — that's exactly what our SEO and social media marketing service does, and you can see verified results on our Top Rated Plus Upwork profile.

Want this done for you?

FRPROTECH is a Top Rated Plus Upwork agency with 3,000+ projects delivered across 30+ countries. Tell us your goals and we'll handle the rest.

Written by the FRPROTECH design team. 8+ years building brands and websites for clients in 30+ countries, with a 100% Job Success Score on Upwork.