Tammy Everts (SOASTA)
summary
Collaboration with Google feeds 1.1 million sessions of real RUM data (stored in perpetuity by the vendor) into machine learning that examines 93 attributes; models show page load time = 2nd most predictive metric for bounce rate while start render ranked 69, a surprising reversal of common intuition. Code and models are on GitHub and run on a laptop; recommendation is to run ML on your own site data to measure third‑party impact and build predictive revenue/traffic models that actually link performance to conversions.
AI-written summary · about
chapters (5)
- 0:00Start
- 2:30Page Load Time
- 5:38Time Is Money the Business Value of Web Performance
- 6:26Why I Wanted To Write this Book
- 8:20Predictive Analytics
related talks
08mLightning Demo: New RUM Resource From SOASTA
36mMoneyball for Performance Metrics
25mAnalyzing the business value of site speed improvements
11mData science: Next-gen performance analytics
38m"Performance Impact, Part Two: More Findings from the Front Lines..."
49mModern Metrics
36mHunting the Unicorn
21mWeb Performance @ NBrown It's a Business Affair