Minimum Detectable Effect Calculator

Estimate the smallest conversion-rate lift an A/B test plan can reliably detect.

At a glance

Computes
Effect at chosen power, or the nearest attainable boundary for the selected design.
You supply
Baseline rate, total sample size, split, alpha, power, tails, and direction.
Use when
Traffic is fixed and you need the chosen-power effect or nearest attainable boundary.
Not for
Choosing a sample size for an effect you already decided matters. Power and Sample Size Calculator

1 or more reads as a percent, so 1 means 1%. Below 1 reads as a proportion, so 0.5 means 50%.

Same reading as the baseline: 1 means 1%, 0.5 means 50%.

Tails

Minimum detectable effect

Enter values

Uses total sample size and visible split.

Absolute effect against Total sample size

1,0002,0000.040.060.080.100.12Entered sample size 1000Minimum detectable effect
Planning details
MetricValue
Export

Solve d where power(pA, pB, nA, nB, alpha) equals requested power How?

How this is calculated

This calculator solves the two-independent-proportions z-test power equation. It uses pooled variance under the null, non-pooled variance under the alternative, and the visible traffic split to derive A and B sample sizes.

Formula: Solve d where power(pA, pB, nA, nB, alpha) equals requested power

Sources

  1. statsmodels power_proportions_2indep. statsmodels. Retrieved .
  2. statsmodels samplesize_proportions_2indep_onetail. statsmodels. Retrieved .

Method last reviewed