How to use this calculator
Plan with baseline and target rates before launch, or analyze final visitor and conversion counts after the planned horizon.
Formula and method
Analysis: pooled two-proportion z test; difference interval: Newcombe Wilson. Planning: equal-group normal approximation using α, power and target effect. SRM: χ² with 1 df.
Worked example
A 100/1000 and B 130/1000 give +0.03 absolute and +30% relative lift; two-sided p≈0.0355.
Assumptions and supported scope
Two independent binary-outcome groups. Fixed horizon and 1:1 allocation for planning; analysis accepts a stated planned allocation. No Bayesian inference, sequential correction, clustered users or multiple-metric correction. SRM is diagnostic, not a repair.
Method references
Method and worked example checked against the calculator implementation. About our methods · Report a correction