Back to Tools

Statistical Power Calculator

Calculate the probability of detecting a true effect in your A/B test

When to use this calculator

Use this tool when:

  • You have fixed sample size constraints
  • You need to know what effects you can detect
  • You're planning experiment feasibility

Common scenarios:

  • Limited monthly traffic
  • Fixed experiment duration
  • Resource allocation decisions

Configure Your Test

Quick Presets

10.0%Current
0.1%30%
+10%Relative lift
1%50%

Target: 10.0% → 11.0%

Statistical Power

0%

Very low power

High risk of missing true effects

Detection probability

0.0% chance to detect a 10% lift

Type II error risk

100.0% chance of false negative

What Changes Can You Reliably Detect?

Power Analysis by Effect Size

Based on your sample size, here's your chance of detecting different improvement levels (effect sizes):

90%
50% lift
80%
50% lift
70%
50% lift
50%
50% lift

Your test: Looking for a 10% improvement → 0% chance of finding it

80% Power

Need ≥50% lift

Current Setup

10% lift = 0% power

Sample Size

n=5,000 per group

Recommendations

To achieve 80% power, you could:

  • Increase sample size: With a 10% effect, you'd need approximately users per group
  • Target larger effects: With 5,000 users, you can reliably detect effects ≥50%
  • Accept lower confidence: Though your current power is quite low for reliable decisions
Technical Details & Calculations

Current Parameters

Control (p₁)

10.00%

Treatment (p₂)

11.00%

Absolute Diff (Δ)

1.000%

Configuration

n=5000, α=0.05

Formula: Power = P(reject H₀ | H₁ is true)

Using normal approximation for proportions with pooled variance under H₀

When Power Matters Most

  • Product launches: Avoid missing performance regressions
  • Algorithm changes: Detect both improvements and degradations
  • Limited traffic: Know your detection limits upfront

Power Guidelines

Exploratory testing≥ 70%
Standard decisions≥ 80%
Critical changes≥ 90%