Modern Metrics
summary
Shows concrete RUM results and measurement caveats: billions of experiences, Chrome‑biased data, and most metrics measured only up to onload. Key numbers: 75th percentile CLS 0.28 → 0.21 (improved), LCP ~3.3s (75th), FID ~92ms (75th). Explains tool differences (web-vitals, uses Boomerang and Impulse RUM), how sampling and timing bias results, and announces a public BigQuery export at rumarchive.com public dataset so researchers can reproduce and slice the data.
AI-written summary · about
chapters (25)
- 0:00Modern performance metrics intro
- 1:05What defines modern metrics
- 3:33Web performance evolution
- 6:02Creating new metrics
- 7:48Metric categories
- 8:52Best metric for your business
- 10:15Data biases in RUM
- 13:13When to stop measuring
- 16:21CLS trends and insights
- 19:09Core Web Vitals deep dive
- 21:34CLS across devices and industries
- 25:02CLS versus business metrics
- 27:30CLS case studies
- 29:15LCP performance analysis
- 31:01LCP data bias
- 32:46LCS versus user frustration
- 33:06Individual website analysis for LCP impact
- 34:09FID performance insights
- 35:55FID versus bounce rate
- 37:18Introduction to Interaction to Next Paint
- 38:44Evaluating metrics for your own website
- 39:47Public performance data sources
- 40:07Rum Archive project
- 41:37Querying Rum Archive
- 42:20Q&A
related talks
30mWeb Vitals
34mWhy Google’s CrUX Results Are Not Reproducible With Your Real-User Monitoring
30mMeasuring Web Performance - Metrics & Tools
20mReal metrics for front-end performance
38mThe Metric System: Making Correct Performance Measurements
27mChrome User Experience Report
15mWhat's new in Web Vitals
57mAn Opinionated Guide to Web Performance Budgets