Analyzing the business value of site speed improvements
summary
Project A measures session-level load time via the navigation timing API (stored in Google Analytics 360 and queried from BigQuery), uses DOMContentLoaded − navigationStart as the page-speed metric, and finds longer loads correlate with lower conversion rates. Sampling defaults to 1% so increase it. A simulation suggests making pages ~1s faster yields roughly a 3% revenue uplift; to prove causation the team runs controlled A/B variants that add +0.5s, +1s and +2s to loading time and measures conversion shifts.
AI-written summary · about
chapters (13)
- 0:00Einleitung
- 1:09What is the value of our website performance?
- 1:52What you'll need (data set)
- 2:58Understanding the critical rendering path
- 4:55Getting timestamps via Navigation Timing API
- 5:23Getting your hands on the data set
- 6:51Visualize distribution of site speed
- 9:59Causation: Sleeping with your shoes on
- 11:57The A/B/C/D testing approach
- 13:20Determining causation of site speed on CR
- 15:16Finalizing test & exploring UPWT
- 17:33Thank you. Let's connect!
- 20:07Simulate potential revenue uplifts
related talks
42mMobile Site Speed and the Impact on E-Commerce
21mWeb Performance @ NBrown It's a Business Affair
55mLet's make the web faster - tips from trenches @ Google
39mGiving you ways to increase, but also explain site performance, to your boss.
39mFrom Milliseconds to Millions
52mWhy Fast Matters
36mMoneyball for Performance Metrics
12mJacob Groß at performance.sync()