AI Traffic

Available onPlusScale

AI assistants read websites for the people using them, and AI companies crawl the web to build search indexes and train models. None of that is a person reading your page.

The AI Traffic section of your dashboard shows how much of your traffic is AI, and where it comes from.

AI traffic is not counted in your pageviews, visitors, bounce rate or anything else, and it's not billed.

What it shows

• The share of AI versus human traffic for the selected dates.
• A list of AI traffic in three tabs:
  • User: an AI fetching a page because someone just asked it something. ChatGPT, Claude, Perplexity and agentic browsers.
  • Search: crawlers building an AI's search index.
  • Training: crawlers collecting pages to train models.

Search and Training need Cloudflare connected.

What Cabin sees by default

Cabin's tracking script runs in the browser, so it sees AI agents that load a page and run its JavaScript. That covers most User traffic.

It can't see:

• Crawlers such as GPTBot and ClaudeBot. They download the HTML and never run JavaScript.
• Blocked requests. If your server, host or CDN blocks a bot, the page never loads, so the script never runs.

If you want to see all AI traffic, including crawlers, connect Cloudflare below.

Connect Cloudflare

If your site is behind Cloudflare, open your site's settings, find AI traffic and click Connect Cloudflare. You sign in to Cloudflare and approve read-only access. There's no API token to create. Cabin then reads AI crawler requests from your Cloudflare account, including the ones Cloudflare blocked, and adds them to your dashboard. Blocked bots are labelled, for example GPTBot (blocked).

• It's read-only. Cabin never changes your Cloudflare settings.
• The last 30 days are filled in when you connect. After that it updates every hour.
• Only crawlers verified by Cloudflare or by the operator's published IP ranges are counted. Requests pretending to be an AI crawler are left out.
• Cabin checks a crawler's IP against those ranges and never stores it.
• The site's owner or a manager can connect it. It runs on the owner's plan.

If you see This domain isn't in that Cloudflare account, you signed in to a Cloudflare account that doesn't hold the site. Connect again and pick the account that does.

Disconnecting

Click Disconnect on the same card. Cabin gives up its access at Cloudflare and stops syncing. The crawler data leaves your dashboard but isn't deleted, so if you connect again later, your history comes back with it.

How agents are identified

Agents are matched on the User-Agent string they send, against the community-maintained ai.robots.txt list. Ordinary search and social crawlers such as Googlebot aren't counted as AI; they're filtered out as bots, the same as always.

If you see an agent you think is missing, tell us.