Field notes, mathematical derivations, telemetry forensic analyses, and statistical methodologies from the consultants at Flow Harbor Point.
Why subtle allocation imbalances between test variants corrupt experiment validity, how to compute chi-square goodness of fit, and four common mobile client mechanisms that trigger silent SRM.
Read Technical Note →
How Controlled-experiment Using Pre-Experiment Data (CUPED) shrinks confidence intervals by 30-45% without requiring larger sample sizes or extended test runtimes.
Read Technical Note →
Why checking dashboard p-values daily inflates false positive error rates from 5% to over 30%, and how sequential testing boundaries protect decision validity.
Read Technical Note →
How to select, categorize, and statistically evaluate non-inferiority guardrail metrics to prevent technical performance and customer churn regressions.
Read Technical Note →