Free Calculator · Instant Results

A/B Test Sample Size Calculator

How many users do you need before your experiment is statistically valid? Enter your parameters and get a result instantly — no signup, no email.

Your current conversion rate for the metric you're testing (e.g. 3% signup rate)
10% relative lift = detecting a change from 3.0% to 3.3%
Probability of detecting a real effect. 80% is the industry standard.
Acceptable false positive rate. 5% (95% confidence) is the industry standard.
Used to estimate test duration. Leave at 1,000 if you're not sure.
Use two-tailed for most tests. One-tailed gives smaller sample sizes but is only appropriate when you're confident the effect can only go in one direction.

Per variant

—
users needed

Total users

—
across both variants

Test duration

—
estimated days

How to read this

Common scenarios

The 3 most common A/B testing mistakes

  1. Peeking: Stopping the test early when results look good. This inflates your false positive rate dramatically. Always run to your pre-calculated sample size.
  2. Setting MDE based on what's detectable, not what's meaningful: A 1% relative lift might be detectable with a huge sample — but is it worth shipping? Set your MDE based on commercial significance first.
  3. Ignoring seasonality: Always run tests for a full week (minimum) to account for day-of-week effects, even if you reach sample size sooner.

Running experiments at scale?

Adasight helps scaling D2C and SaaS teams build systematic experimentation programs — from hypothesis to velocity to statistical governance.

Talk to Adasight →

A/B Testing FAQ

How do I calculate A/B test sample size?

Sample size depends on: baseline conversion rate, minimum detectable effect (MDE), statistical power (typically 80%), and significance level (typically 5%). The formula uses the normal approximation to the binomial distribution. This calculator handles the math automatically.

What is minimum detectable effect (MDE) in A/B testing?

The MDE is the smallest change in conversion rate you want to be able to detect reliably. Choose your MDE based on what change would be commercially meaningful — not what's convenient. A 10% relative lift on a 3% baseline means you're looking for a change from 3.0% to 3.3%.

What statistical power should I use?

80% is the industry standard. This means an 80% chance of detecting a true effect of your MDE size. For high-stakes decisions, use 90% — it requires a larger sample but reduces missed effects.

How long should an A/B test run?

Always run for at least 7 days regardless of when you hit sample size, to control for day-of-week effects. For most consumer products, 2–4 weeks is a healthy experiment window. Longer is fine; stopping early is not.

Can I run more than two variants at once?

Yes — multivariate and A/B/n tests are valid, but they multiply your sample size requirement. For an A/B/C test, you need the per-variant sample size multiplied by 3. Keep variant count low unless you have high traffic.