How to track visitors from ChatGPT and other AI assistants in Google Analytics

You see ChatGPT traffic in Google Analytics by looking at session source, which is the website a visitor came from at the start of their visit. When someone clicks a link in a ChatGPT answer, their browser usually tells your site that the visit came from chatgpt.com. ChatGPT also adds a tag to many of the links it shows, which marks the source as chatgpt.com even when the browser doesn't pass it along. Either way, Google Analytics 4 records the visit with chatgpt.com as the session source.

The quickest place to find it is the Traffic acquisition report. Open Reports, then Acquisition, then Traffic acquisition, and change the first column from Session default channel group to Session source / medium. Type "chatgpt" into the search box above the table, and you'll see how many sessions, engaged sessions, and key events ChatGPT sent you in the date range you picked. Older visits may show chat.openai.com, which was ChatGPT's address before it moved.

The same method works for other assistants, because each one has its own domain:

• ChatGPT: chatgpt.com, and chat.openai.com for older visits
• Perplexity: perplexity.ai
• Gemini: gemini.google.com
• Claude: claude.ai
• Microsoft Copilot: copilot.microsoft.com
• DeepSeek, Grok, and Meta AI: chat.deepseek.com, grok.com, and meta.ai

To see all of them at once and find out which pages they land on, use an Exploration. Open Explore, start a Free form exploration, and add Session source and Landing page + query string as dimensions, with Sessions, Engaged sessions, and Key events as metrics. Then add a filter on Session source that matches this pattern:

.*(chatgpt|openai|perplexity|gemini\.google|claude\.ai|copilot|deepseek|grok|meta\.ai).*

The pattern catches any source that contains one of those names, so you don't need a separate filter for each assistant.

Since May 2026, the default channel group includes an AI Assistant channel for visits Google recognizes as coming from assistants such as ChatGPT, Gemini, DeepSeek, Copilot, and Grok. If you want to decide for yourself which sources count as AI visits, create a custom channel group. In Admin, under Data display, open Channel groups, copy the default group, and add a new channel called AI assistants whose rule is that the session source matches the same pattern. Move that channel to the top of the list, because Google Analytics assigns each visit to the first channel whose rule it meets. Custom channel groups apply to your past data as well as new visits, so you can compare this month with last quarter straight away.

Those steps get you a count, and the count is useful, but it isn't the full answer. A good result tells you which assistants send you visitors, which pages they arrive on, whether those visitors read and convert as well as your other traffic, and how all of that is changing month to month. It also says plainly what the count misses. Getting there means running several reports, lining them up against each other, and doing it again every month.

Kite does that work for you. It reads your Google Analytics property, builds the breakdown, and tells you what it means for the pages you publish.

Start by asking Kite a plain question:

"How much traffic are ChatGPT and other AI assistants sending to our website, and which pages are they landing on?"

Kite is an AI marketer. It researches your business, builds a growth strategy, and ships the work that brings you new customers. You can work with Kite in your browser or in Slack. Kite already knows which pages on your site matter most, which ones lead to signups or sales, and which questions your buyers ask AI assistants, so it can tell a visit to your pricing page apart from a visit to an old blog post without being told.

Connect your Google Analytics account through OAuth so Kite can read your reports. If your account can see more than one property, Kite lists them and confirms which one belongs to your website. Reading reports is routine work, so Kite starts as soon as the account is connected and posts what it finds in the same conversation.

Kite starts by confirming it's reading the right data. It lists the Google Analytics accounts and properties you have access to and picks the property for your website, checking the property's time zone so that "last month" means the same thing to Kite as it does to you. It reads the property's data streams to confirm the one for your website is sending data, and it runs a realtime report to make sure visits are still arriving today.

Kite also reads how your property is set up. It lists your channel groups to see whether someone on your team has already created an AI channel, and if so, it uses your definition so its numbers match the reports your team already looks at. It lists your key events, such as a signup, a demo request, or a purchase, so it knows which actions count as a conversion on your site. It also reads your reporting annotations, because a note about a site relaunch or a tracking change can explain a sudden jump or drop.

Kite finds AI-assistant visits by running a report on session source and session medium, filtered to the domains of the assistants listed above. It reads the sources one by one rather than trusting a single pattern, because assistants change domains and new ones appear. When it sees a source that looks like an AI assistant but isn't on its list, it includes it and tells you why.

It keeps the assistants separate in the results, so you can see that ChatGPT sends most of your AI visits while Perplexity sends fewer visitors who read more. It also separates real visits from noise. A source with dozens of sessions and almost no engagement, all arriving within minutes, looks more like a test or a bot than like people reading an answer, and Kite flags it instead of counting it.

Kite breaks AI-assistant sessions down by landing page, because the page a visitor arrives on tells you which of your content the assistants are recommending. For each page, it reports sessions by assistant, engaged sessions, engagement rate, average engagement time, and key events.

Kite then sorts the pages into groups that call for different action. Some pages draw AI visitors and convert them, which makes them the pages to protect and keep up to date. Others draw AI visitors who leave quickly, which usually means the page doesn't deliver what the answer promised. A third group gets almost no AI visits at all, even though they cover topics your buyers ask about.

Kite compares AI-assistant visitors with your organic search, referral, and direct visitors over the same dates. It looks at engagement rate, engagement time, pages per session, and key event rate for each group, so you can see whether a visitor from ChatGPT is worth more or less to you than a visitor from Google.

It also compares this period with the one before it. Kite reports how much AI-assistant traffic grew or shrank, which assistants account for the change, and which landing pages gained or lost visits, so a new page that started getting cited stands out by name.

Kite tells you how much AI-assistant traffic your reports can't see, so you don't read the count as the whole picture. Visits from the ChatGPT and Perplexity mobile apps often reach your site without any source attached, and Google Analytics files them under Direct. So does anyone who copies a link out of an answer and pastes it into a browser. Visits from Google's AI Overviews and AI Mode arrive as google / organic, the same as an ordinary search click.

Kite looks for signs of this hidden traffic. When direct visits to deep pages, such as a single blog post or a comparison page, rise in step with AI visits to the same pages, it points that out as likely AI traffic that arrived without a label. It keeps that inference separate from the counted visits, and it says how confident it is.

Kite connects your Google Analytics numbers to its own AI-visibility checks, which run your buyers' questions through ChatGPT, Gemini, Perplexity, and Claude and record which pages each answer cites. Putting the two side by side shows you whether the pages assistants cite are the pages that bring you visitors.

When an assistant cites a page often but sends few visits to it, the answer is probably giving readers everything they need without a click, and Kite may suggest adding something to the page that's worth clicking through for, such as a tool, a template, or pricing details. When a page gets AI visits but isn't cited for the questions you care most about, Kite looks at the pages the answers cite instead and suggests what to change or publish.

Kite posts its findings in the conversation where you asked. It opens with the answer: how many sessions AI assistants sent you in the period, how that compares with the period before, and what share of your total traffic it represents. It follows with the breakdown by assistant, the top landing pages, how AI visitors compare with your other channels, and what the count likely misses. Every figure carries the dates it covers. When the findings are too long for one message, Kite publishes them as a findings page with charts and tables, or builds a dashboard that reads your Google Analytics data fresh each time you open it.

From there, Kite can act on what it found. It can rewrite a page that draws AI visitors who leave quickly, and on a site Kite hosts or one it updates through a supported CMS such as Webflow, it can publish the change once you approve it. If you haven't created an AI assistants channel yet, Kite gives you the exact pattern and steps so the channel shows up in every report your team uses. Ask Kite to run the same report every month, and it delivers each update where you choose: a Slack channel, an email, or a new conversation in the Kite web app.

When Kite finishes, you know how much traffic AI assistants send you, where it lands, and what it's worth, along with a clear next step.