⏱ Sequential A/B Testing
URLs A and B are measured on your real device and network, and the lead order alternates every round so neither side gets a first-mover advantage. Each is measured at the CDN's warm steady state — exactly what a returning visitor experiences — so the difference reflects delivery, not a one-time cold-start penalty.
10-Run Methodology
Each image is measured over multiple trials. The first (connection-setup) trial is discarded, and the headline is the median (p50) — not a mean — so a single outlier cannot swing it. p75 is shown alongside so a wide spread stays visible.
Video Testing
Videos measure Time to First Frame — the metric that matters most for perceived speed. Both URLs are warmed identically until each edge reaches its steady state; warm-up is decided by measured latency, never by which CDN is being tested.
Use Cellular, Not WiFi
Testing on 4G/5G shows the true CDN advantage. CDN optimisations matter most on constrained networks. Use cellular for the most meaningful results.
What the Numbers Mean
Avg — average of 9 runs (outlier removed). p50 — median. p75 — 75th percentile worst-case. TTFB — server response time. File size — actual bytes received.