AI tracking is the process of detecting and recording visits made by artificial intelligence bots (GPTBot, ClaudeBot, PerplexityBot, etc.) to a website, as well as traffic coming from AI platforms (referrals).
What is AI Tracking?
AI tracking refers to the techniques used to detect, record and analyze interactions between a website and artificial intelligence systems. It covers two main aspects:
1. AI Bot Crawl Detection
AI engines send crawlers to index website content. Each bot has a specific User-Agent:
- GPTBot — OpenAI's crawler for ChatGPT
- ClaudeBot — Anthropic's crawler for Claude
- PerplexityBot — Perplexity AI's crawler
- Google-Extended — Google's crawler for Gemini
- Bytespider — ByteDance's crawler
AI tracking identifies these User-Agents in HTTP requests and records which pages are crawled, how often, and by which bot.
2. AI Referral Detection
When a user clicks a link in a ChatGPT, Perplexity or other AI response, the browser sends a Referer header containing the source domain (e.g., chat.openai.com). AI tracking detects these referrals to measure the actual traffic generated by answer engines.
How Does an AI Tracker Work?
An AI tracker is installed server-side (or via a JavaScript pixel) and works by intercepting each HTTP request. It analyzes:
- The User-Agent to identify AI bots
- The Referer header to detect referrals from AI platforms
- The visited URL to know which pages interest AI engines
Events are signed with an API key and an HMAC secret to ensure data integrity, then sent to the analytics platform API.
Why is AI Tracking Essential?
- Measure AI visibility — Knowing if AI bots crawl your site is the first step to optimizing your presence on answer engines
- Quantify AI traffic — AI referrals are a growing traffic source, often invisible in classic analytics
- Optimize content — By knowing which pages are crawled, you can adapt your content strategy
- GDPR compliant — A good AI tracker collects no personal data: no cookies, no IP, no fingerprinting