Growth calculator

A/B Test Sample Size Calculator

Estimate sample size per variant and test duration for detecting a conversion-rate lift with deterministic two-proportion math.

Inputs

Defaults are visible and can be changed before calculation.

Inline validation messages appear here when a value needs to be corrected.

Results

Results are deterministic scenario outputs, not guarantees.

Enter values and calculate to see the summary, supporting metrics, warnings, and interpretation.

Interpretation

Deterministic interpretation rules will explain what the modeled result means once a calculation is available.

Detailed breakdown

Intermediate calculation rows will appear here after calculation.

Formula

Target conversion rate

p2 = p1 + absoluteEffect

Bonferroni adjustment

adjustedAlpha = alpha / (numberOfVariants - 1)

Sample size per group

n = [zα√(2p̄(1-p̄)) + zβ√(p1(1-p1)+p2(1-p2))]^2 / (p2-p1)^2

Daily traffic per variant

dailyTrafficPerVariant = dailyEligibleTraffic × trafficAllocationRate / numberOfVariants

Estimated duration

estimatedDays = nPerGroup / dailyTrafficPerVariant

Assumptions

  • Uses a standard normal approximation for two proportions.
  • Relative MDE is converted to an absolute lift from the baseline conversion rate before sample-size math is applied.
  • Bonferroni adjustment is applied when there are more than two variants.
  • Currency is not applicable because the calculator models traffic and conversion rates only.

Worked example

Example: detecting a 20% relative lift

With a 5% baseline conversion rate, a 20% relative lift means the target rate is 6%; the calculator then estimates the sample needed per variant and the duration based on available traffic.

FAQ

Is this a forecast of test success?

No. It estimates the sample size needed to detect the entered effect under the selected confidence and power assumptions.

What is the difference between absolute and relative MDE?

An absolute MDE adds percentage points to the baseline rate; a relative MDE multiplies the baseline rate by the entered lift percentage.

Why does adding variants increase sample size?

More variants split traffic and require a multiple-comparison adjustment, so each variant usually needs more time or traffic.

Related calculators

Want help interpreting the model?

Use this calculator as a deterministic planning tool, then talk with Propel Collective about which assumptions are worth validating first.