AI Traffic

Overview

LLMs like ChatGPT, Claude, and Perplexity now refer real human visitors to websites — someone asks an AI a question, it mentions your site, and the person clicks through. Peepmetrics automatically detects these referrals and surfaces them on a dedicated AI Traffic page.

At the same time, the same LLM providers send automated crawlers to index and train on web content. These bot visits are detected and excluded from your regular analytics so your visitor counts stay accurate.

AI referral detection

When a visitor arrives via a known AI platform, their referrer URL is matched against a list of recognized AI domains. No configuration required — this works automatically for all tracked page views.

The following platforms are recognized:

Ambiguous domains like bing.com and google.com are intentionally excluded — these carry regular search traffic and cannot be reliably attributed to AI-generated referrals.

Bot crawler detection

LLM providers also send automated crawlers to index and gather training data from the web. These visits are detected at ingest time by matching the request's user-agent against known crawler signatures.

Recognized crawlers include:

• GPTBot, ChatGPT-User, OAI-SearchBot — OpenAI
• ClaudeBot, anthropic-ai, Claude-Web — Anthropic
• PerplexityBot — Perplexity
• YouBot — You.com
• PhindBot — Phind
• Bytespider — ByteDance

Bot page views are stored with is_bot: true and the crawler's display name. They are automatically excluded from all standard analytics — total views, breakdown tables, the traffic graph, and daily rollups — so they never inflate your real visitor counts.

The AI Traffic page

Each domain has a dedicated AI Traffic page at /domains/:id/ai-traffic. It includes:

• A time-series chart of AI referral visits over 7, 30, 90 days, or 1 year
• A breakdown of visits by AI platform, ranked by volume
• A bot crawler activity table showing which crawlers have visited and how often

All sections respond to the time range selector at the top of the page.

Limitations

AI referral detection is based entirely on the Referer HTTP header sent by the visitor's browser. This means:

• Visits where the browser strips the referrer (e.g. private browsing, strict referrer policies) will not be attributed to any source, including AI platforms.
• There is no way to detect that your site was mentioned in an AI response if the user typed your URL directly rather than clicking a link.
• Attribution is limited to the platforms in the table above. Newer or less common AI platforms will not appear until their domains are added.

These are fundamental constraints of passive referrer-based attribution — no tracking script, cookie, or server-side integration can fully work around them. The data shown is a reliable lower bound, not a complete count.