Don't Make Users Wait: Secrets to Supercharge Server-Rendered Page
summary
Production stack uses SvelteKit and Tailwind, Cloudflare fronting ECS containers running Node.js, and measured with Lighthouse, WebPageTest and a Web Vitals RUM plugin. Targets: reduce TTFB, FCP and LCP. Tactics: short-term API cache, internal VPC API calls, render only above-the-fold HTML, preload fonts/LCP images, limit preconnects, lazy-load and set fetchpriority. Results: payload 553KB→400KB, FCP 3.8s→1.45s, LCP 3.9s→1.75s, JS down 20–30%, and conversion lifts up to ~30% when p75 FCP drops ~270ms.
AI-written summary · about
chapters (9)
- 0:00The impact of performance
- 2:06UI architecture and flow
- 3:44Identifying bottlenecks
- 6:43Measuring web vitals
- 9:01Monitoring and tooling
- 11:40Server-side optimizations
- 14:24Asset and loading strategies
- 17:48Advanced front-end tweaks
- 23:22Performance results
related talks
42mBuilding quick apps in 2018
29mHigh Performance in the Critical Rendering Path
23mOptimizing page load speed on front end side
31mWe're All Doing the Best We Can
32mStay Away From My Path!
28mCSS and the Critical Path
38mCSS and the first meaningful paint
27mHigh Performance in the Critical Rendering Path