A blog owner implemented new Cloudflare Worker request-header and network rules to filter bot traffic, successfully reclassifying 74.5% of browser-User-Agent requests as non-human traffic.

Key Points

• The analysis compared 372 browser-User-Agent requests against 14 Cloudflare Web Analytics page loads over two days.
• New classification rules utilize network metadata, Fetch Metadata headers, and browser version thresholds to identify automated traffic.
• Repairs were made to address defects in HTML-acceptance checks and missing network-provenance markers in database logs.
• The study distinguishes between verified bot signatures, such as those from Ahrefs or DuckAssist, and unidentified automated requests.
• The author emphasizes that browser-based analytics often conflate automated scripts with human readers, requiring more granular evidence for accurate measurement.

Why it Matters

Distinguishing between human readers and automated bots is a significant challenge for website owners relying on standard analytics tools. By documenting these classification methods, the author provides a framework for improving data transparency and reducing the misinterpretation of traffic metrics.