How to Track How Many Times ChatGPT (and Other AI Bots) Hit Your Website

Understanding how often AI systems like ChatGPT interact with your website is becoming increasingly important for marketers, SEOs, and developers. While you can’t directly see "ChatGPT users visiting your site" in the traditional sense, you can track when AI crawlers access your content—and get reliable, "legit" numbers using tools like Cloudflare. This guide walks you through how to do exactly that.

First, Know What You’re Actually Tracking

There’s an important distinction:

• AI crawlers (bots) → These are automated systems (like GPTBot) that scan your site to train or improve AI models.
• AI-driven traffic (users clicking links) → These are humans using tools like ChatGPT or Perplexity who click through to your site.

This article focuses on tracking crawler activity, because that’s what you can measure reliably today.

Why Cloudflare Is the Best Option

Cloudflare sits in front of your website and logs every request. That makes it one of the most accurate ways to measure bot traffic. With Cloudflare, you can:

• Inspect raw request data
• Filter by user agent
• Separate real users from bots
• Build queries for specific AI crawlers

The Key: Filtering by User Agent

AI crawlers identify themselves using specific user agent strings. If you want clean, declared AI traffic only, you can filter for these:

(http.user_agent contains "GPTBot") or
(http.user_agent contains "ClaudeBot") or
(http.user_agent contains "Google-Extended") or
(http.user_agent contains "PerplexityBot")

These represent:

• GPTBot → OpenAI’s crawler
• ClaudeBot → Anthropic’s crawler
• Google-Extended → Google’s AI training crawler
• PerplexityBot → Perplexity AI crawler

Step-by-Step: How to Track AI Bot Hits in Cloudflare

1. Go to Cloudflare Analytics

• Log into your Cloudflare dashboard
• Select your domain
• Navigate to: Analytics & Logs → Logs (or Security Analytics depending on plan)

2. Use the Query Builder

In the query or filter bar, paste:

(http.user_agent contains "GPTBot") or
(http.user_agent contains "ClaudeBot") or
(http.user_agent contains "Google-Extended") or
(http.user_agent contains "PerplexityBot")

3. Set Your Time Range

Choose a timeframe depending on your goal:

• Last 24 hours → quick snapshot
• Last 7 days → trend view
• Last 30 days → reporting baseline

4. Count Requests (Your "Hits")

Each request = one hit from an AI crawler. You can now see:

• Total number of requests
• Requests per bot
• Requests over time

This is your clean, declared AI crawler traffic.

Getting More Granular (Highly Recommended)

To make your data more useful, break it down further:

By Bot Type

Group results by http.user_agent to see which AI system hits you most.

By URL

Add a filter or grouping for:

http.request.uri.path

This shows which pages AI crawlers care about.

By Frequency

Look at requests over time to identify:

• Crawl spikes
• Content updates triggering recrawls

Important Limitations (Be Honest About the Data)

Even with Cloudflare, this is not the full picture.

1. Not All AI Traffic Is Declared

Some bots may:

• Not identify themselves clearly
• Use generic user agents
• Route through third parties

2. This Is NOT ChatGPT User Traffic

You are tracking: ✅ Crawlers indexing your site ❌ People clicking links inside ChatGPT Those require separate tracking (UTMs, referrers, etc.).

3. One Page Load ≠ One "Use"

A single crawl session may generate multiple requests.

Pro Tip: Combine With Referral Tracking

To understand actual traffic from AI tools, combine this with:

• Referral sources (e.g., perplexity.ai, chat.openai.com if available)
• UTM-tagged links
• Server-side analytics

What "Good" Numbers Look Like

There’s no universal benchmark, but generally:

• Low traffic sites → a few dozen AI hits/day
• Growing sites → hundreds/day
• High-authority sites → thousands/day

Spikes often correlate with:

• Publishing new content
• Updating popular pages
• Gaining backlinks

Final Thoughts

Tracking AI bot traffic isn’t about vanity metrics—it’s about understanding how your content is being ingested by the systems that increasingly shape discovery. Using Cloudflare with user agent filtering gives you the cleanest, most defensible numbers available today. If you want accuracy, stick to:

• Declared crawlers
• Transparent filters
• Consistent time windows

That’s how you turn a fuzzy concept like “AI is using my content” into real, measurable data.

Copy-Paste Filter (Quick Use)

(http.user_agent contains "GPTBot") or
(http.user_agent contains "ClaudeBot") or
(http.user_agent contains "Google-Extended") or
(http.user_agent contains "PerplexityBot")

  If you want, I can also help you track actual traffic from ChatGPT users (not just crawlers)—that’s a different setup and a bit more nuanced.