Moneyball for Performance Metrics
summary
Claim: measuring first usable time beats page weight or onload for real user impact. Uses concrete tools: Speed Index from WebPageTest, automated phantomjs HAR captures, the performance timing and user timing API to mark render-ready events. Shows numbers: TCP slow start limits first response to 14.65 KB, Etsy’s +160 KB images cost 12% more bounce, one-second drops yield ~2% higher conversions, Amazon loses ~1% revenue per 100 ms. Recommends inlining critical CSS, server rendering and custom real-user metrics and budgets.
AI-written summary · about
chapters (26)
- 0:00Start
- 5:16slow websites lose
- 7:13people expect mobile to be fast
- 8:19what do you mean by "win" and "lose"?
- 11:00measure things
- 12:18easy, traditional measurements?
- 12:49page weight
- 12:59request/response
- 14:00first usable time
- 17:04render blocking
- 17:18so how?
- 17:48HAR file?
- 18:35round trips
- 19:57monitor whether or not your site is usable in 14.65K
- 20:19timing differences?
- 20:35performance tining. LoadEventEnd - performance. timing.navigation Start
- 21:34different websites need different measurements
- 22:32UserTiming API
- 24:55now what?
- 25:30critical path
- 26:18server-side rendering
- 27:11best practices
- 27:32concatenate
- 27:49minification
- 29:27performance budget
- 30:04final thought
related talks
20mReal metrics for front-end performance
18mWebsite performance from user perspective
31mEvery Millisecond Counts: Performance & UX
11m"Time is Money - The Measurable Value of Performance by Design"
35mWeb Application Performance Testing from Users Perspective
30mMeasuring Web Performance - Metrics & Tools
36mHunting the Unicorn
29mWeb Performance Kinetosis