Unlocking Web Speed: Billions of Real-User Data

Decoding Global Web Performance

Cloudflare's BEACON dataset is a monumental step towards democratizing web performance insights. The sheer scale and granularity of the data, covering billions of real-user measurements across 10,000 major websites and broken down into Core Web Vitals sub-parts, offer unprecedented clarity into what truly impacts user experience. The ability to analyze long-tail performance and identify bottlenecks beyond simple averages is invaluable for developers and researchers alike. The commitment to transparency by publishing this data in BigQuery, along with the methodology for anonymization and normalization, builds trust and fosters a collaborative environment for improving the web.

However, while BEACON provides a broad view, its scope is inherently tied to Cloudflare's network. While 10,000 of the largest websites offer significant diversity, the dataset might not fully capture the performance characteristics of smaller, niche, or emerging websites that don't rely on Cloudflare's services. Furthermore, the 'normalization' to the 10,000th website's traffic volume, while a clever approach to balance representation, might still subtly favor the architectural patterns prevalent in larger sites. Future iterations could explore alternative normalization strategies or provide more granular control over data selection to cater to diverse research needs. The challenge of fully isolating the impact of individual factors (website optimization vs. network infrastructure) remains, and while BEACON aids this, it doesn't entirely resolve it. Nevertheless, the dataset is a powerful tool that significantly advances the understanding of web performance from a real-user perspective.

Key Points

• Cloudflare has released BEACON, a massive, anonymized dataset of billions of real-user web performance measurements.
• The dataset covers 10,000 of the largest websites, updated daily in Google BigQuery, and follows RUM Archive standards.
• BEACON provides insights into Core Web Vitals (LCP, CLS, INP) and their sub-parts, enabling analysis beyond simple averages.
• Key findings include performance disparities across countries and device types (e.g., WebKit on iOS), and industry-specific performance differences.
• Analysis of LCP and INP sub-parts reveals that discovering LCP candidates and unblocking rendering are often bigger challenges than resource download, and JavaScript execution and CSS recalculations impact interaction responsiveness.
• Soft navigations in SPAs are significantly faster than hard navigations but come with a heavier initial landing page load.
• The dataset can be joined with external data (e.g., GDP per capita) for deeper correlations, revealing links between economic conditions and web performance.
• Cloudflare emphasizes privacy through anonymization and normalization, making the data useful without compromising user trust.
• BEACON is publicly available on Google BigQuery, encouraging researchers, developers, and standards groups to use it for web improvement.

---

📖 Source: How fast is the web? Explore billions of real-user measurements with BEACON