What is Minimum Detectable Effect (MDE)?

MDE is the smallest improvement (lift) you want to be able to detect. Smaller lifts require much larger sample sizes to prove they aren't random noise.

Why do I need to calculate sample size upfront?

In traditional Frequentist testing, checking results before reaching the sample size increases the 'False Positive' rate. This calculator tells you the fixed horizon you must reach for valid results.

A/B Test Sample Size Calculator (Pre-Test Planning)

Baseline Conversion Rate (%)

Your current conversion rate before testing.

Minimum Detectable Effect (MDE) %

The smallest lift you care about detecting (e.g. 10%). Smaller MDE = Much more traffic needed.

Statistical Power (%)

Likelihood of finding a winner if one exists. Standard is 80%.

Daily Traffic (Optional)

Used to estimate the duration in days.

Why Sample Size Matters

[Image of sample size significance curve]

If you stop a test too early, you risk a "False Positive" (thinking a variation is a winner when it's actually random noise). If you run it too long, you waste traffic on a losing variation.

This calculator uses the standard Frequentist formula to determine the "Fixed Horizon" sample size. This means you must commit to this number of visitors before you look at the results.

The "Fixed Horizon" Problem

Traditional A/B testing is rigid. If you see a winner on Day 3, you are technically not allowed to stop until Day 20 (based on the calculator). This wastes money.

Zyro uses Sequential Bayesian Statistics. This modern approach allows you to check results daily and stop as soon as a winner is statistically proven, often cutting test duration by half.