How Fast is Fast Enough?

on YouTube ↗26m

" summary

Shows concrete tests and numbers: A multivariate throttling experiment produces fewer conversions and a throttled cohort returns less even after speed restores; RUM-based correlation charts from Speed Curve map conversion to LCP with a 1.1s LCP peak and a performance plateau at 2.8s for one site; ContentSquare aggregates show 2.0% vs 2.5% conversions by INP bands; independent site analyses find INP peaks between 50–150ms not always matching Google's 200ms guideline.

AI-written summary · about

" chapters (13)

1. 0:00Introduction
2. 1:23Revisiting fast enough
3. 2:46Magical vs pragmatic fast
4. 6:19Business goals vs user needs
5. 7:44A/B testing assumptions
6. 9:30Real user monitoring
7. 12:16Google’s Core Web Vitals
8. 14:45Core Web Vitals vs conversions
9. 17:11User expectations vs needs
10. 19:59Engagement vs productivity
11. 22:07Flow and instantaneous UX
12. 24:14Pragmatism vs magic
13. 24:56Key takeaways and closing thoughts

" related talks

• 52mWhy Fast MattersHarry Roberts · JS Fest 2018
• 45mWhy Fast MattersHarry Roberts · Beyond Tellerrand Munich 2018
• 12mJacob Groß at performance.sync()Jacob Groß · performance.sync() 2024
• 39mFrom Milliseconds to MillionsHarry Roberts · ffconf 2019
• 44mWhy Fast MattersHarry Roberts · CSSConf Budapest 2017
• 31mEvery Millisecond Counts: Performance & UXDavey Shafik · PyCon AU 2016
• 01h 01mUnveiling the Power of Web Performance Metrics for TestersSrinivasan Sekar · Spartans Summit 2025
• 16mFrom beginner to expert in web performanceTracy Lee · JSConf Mexico 2020 Virtual Edition