Minimum Detectable Effect Calculator
Before running an A/B test, you need to know how many visitors to collect in each variant. Running a test with too few visitors means you might miss a real improvement (low statistical power), while collecting far more than necessary wastes time and delays decisions. The minimum detectable effect (MDE) is the smallest lift you need your test to reliably detect. This calculator takes your baseline conversion rate, target MDE, confidence level, and statistical power to compute the required sample size per variant and the number of days to reach that sample at your current traffic volume.
Sample size formula (two-proportion z-test)
p2 = p1 * (1 + MDE/100)
n = (z_alpha + z_beta)^2 * (p1*(1-p1) + p2*(1-p2)) / (p2-p1)^2
Days = n / daily_visitors_per_variant
z_alpha = 1.96 for 95% confidence; z_beta = 0.84 for 80% power. MDE is expressed as a relative lift percentage.
Sample size guidance
- A 10% relative MDE on a 3% baseline means detecting lifts from 3.0% to 3.3% CVR.
- Smaller MDE requires much larger samples - halving MDE roughly quadruples sample needed.
- At 80% power and 95% confidence, use z_alpha = 1.96 and z_beta = 0.84.
- Always run tests for whole-week multiples to avoid weekday seasonality bias.
MDE: frequently asked questions
What is minimum detectable effect (MDE)?
The minimum detectable effect is the smallest change in conversion rate that a test can reliably detect at a given statistical power and confidence level. It determines how many visitors you need to collect before your test is conclusive.
What is statistical power in A/B testing?
Power is the probability of correctly detecting a true effect. At 80% power, a correctly designed test will find a real difference 80% of the time. A test with too little traffic runs at low power, meaning many real improvements go undetected.
How many visitors do I need for an A/B test?
Required sample size depends on baseline conversion rate, minimum effect you care about, confidence level, and statistical power. The calculator computes this using the standard formula for two-proportion tests.
What MDE should I target?
Target an MDE that is practically meaningful for your business. If a 0.5% lift in conversion rate would justify the change, plan your sample size around a 0.5% MDE. Smaller MDEs require larger samples.
What happens if I run a test with insufficient sample size?
Tests with too little traffic are underpowered. They may correctly detect large effects but miss smaller genuine improvements. They also inflate false negative rates, causing real winners to go unrecognized.
Sources
- National Institute of Standards and Technology (NIST): Two-Sample z-Test for Proportions.
- U.S. Census Bureau - Statistical Methods: Statistical Research.
Reviewed by the CalculatorHub team, edited by James Graham, 14 June 2026. See our methodology.