Topic
Statistics and A/B testing
Reading an experiment with SQL: lift, pooled rates, sample-ratio checks and significance from a critical-values table. Candidates who can write SQL often freeze here, which is exactly why it's worth practising.
New to this? Read the Reading an experiment readout chapter.
easyConversion rate per variantShopee · experimentationFree→easyUsers assigned to both variantsNinja Van · experimentationBasic→mediumAbsolute difference and relative liftMaybank · experimentationFree→mediumCumulative conversion rate by dayTraveloka · experimentationBasic→mediumGuardrail metric regressionsAirAsia · experimentationBasic→mediumNovelty effect over four weeksShopee · experimentationFree→mediumSample ratio mismatchGrab · experimentationBasic→hardPooled standard error and z-scoreGrab · experimentationBasic→hardRequired sample sizeTokopedia · experimentationBasic→hardRevenue per user, before and after trimmingZalora · experimentationBasic→hardSignificance at 5%GoTo · experimentationBasic→hardSimpson's paradox in an A/B testLazada · experimentationBasic→