Bots surpass humans on the web, Cloudflare says traffic hits 57.3%

Agencies just got a sharper reason to rethink the way they read traffic reports. Cloudflare said automated traffic now accounts for 57.3% of worldwide HTTP requests to HTML content, while human traffic makes up 42.7%, a shift that turns bot handling from a background technical issue into a client-facing economics problem.

That matters because machine visits do not behave like people. An AI agent can pull thousands of pages to answer one task, where a person might have checked only a few. That creates heavier load, noisier analytics, and a visibility problem for brands that still measure success only through clicks, sessions, and conversions. For SEO firms, the opportunity is no longer just ranking pages, but explaining which requests signal real demand and which ones are infrastructure churn.

Matthew Prince had warned in March at SXSW in Austin that AI bot traffic could overtake human traffic by 2027, and he said an agent might visit 5,000 sites where a person would visit five. The pace of change moved faster than that forecast. Cloudflare’s latest figures show that Google’s crawling bot was the single biggest source of automated internet traffic in 2025, while AI user-action crawling rose by more than 15 times. Other AI crawlers accounted for 4.2% of HTML requests, and Googlebot alone accounted for 4.5%.

The longer view reinforces the same trend. Cloudflare said global internet traffic grew 19% in 2025, post-quantum encryption secured 52% of all human traffic by early December, and the company observed more than 25 record-breaking DDoS attacks. Crawler traffic rose 18% from May 2024 to May 2025, with GPTBot up 305% and Googlebot up 96%. Even now, 14% of top domains are using robots.txt rules to manage crawlers.

Cloudflare has started to productize the problem. Radar now includes an Agent Readiness Scan, and Cloudflare’s documentation says managed robots.txt can tell known AI crawlers to stay away from content, though compliance is voluntary and some crawlers can ignore it. For agencies, that makes bot-heavy traffic a technical SEO upsell, not a measurement crisis: clean the analytics, separate machine behavior from human intent, and have the server-cost conversation before the next wave of automated browsing arrives.