Anatomy of a Modern Page Load
summary
Audit of a Next.js WNBA site uses WebPageTest and RUM to compare lab versus field: lab TTFB ~886 ms while RUM shows different distributions, LCP lab ~3.5 s vs RUM ~1.7 s, and TBT ≈7.6 s with many long tasks. Findings push concrete fixes: serve HTML from a CDN, emit HTTP 103 early hints, preload the LCP image, use fetchpriority or <link rel=preload>, break up JS long tasks to reduce blocking. LCP request #32, 70% of LCPs are images
AI-written summary · about
chapters (13)
- 0:00Welcome and intro
- 4:27Modern page load challenges
- 7:35Time to First Byte explained
- 10:47HTTP 103 for faster loads
- 12:56Understanding Largest Contentful Paint
- 16:49Network limits on performance
- 18:57Cumulative Layout Shift basics
- 20:02Total Blocking Time and JavaScript
- 24:17Speculative Rules API demo
- 26:04Lab vs real user metrics
- 27:05Mobile device performance realities
- 29:15Key performance tips summary
- 31:02Closing and performance literacy
related talks
39mUnderstanding Page Load
28mR is for React Performance
31mEvery Millisecond Counts: Performance & UX
26mDon't Make Users Wait: Secrets to Supercharge Server-Rendered Page
49mDemystifying web performance tooling and metrics
48mThink Fast, First: Empowering a Performance Culture
46mOptimising Largest Contentful Paint
32mEverything you need to know about Web Performance