A/B Test Sample Size Calculator
A/B test planning needs a baseline conversion rate and a minimum effect worth detecting. The calculator estimates sample size per arm based on the chosen z-values for confidence and power.
How to use this calculator
- Use a defensible baseline from comparable traffic and choose confidence/power z-values appropriate to your experiment plan.
- Check that units, denominator definitions and time periods are consistent before comparing scenarios.
- Run a second scenario when an assumption is uncertain.
Formula used by ab test sample size calculator calculator
Equal-arm sample size uses a two-proportion normal approximation: [zα√(2p̄(1−p̄)) + zβ√(p1(1−p1)+p2(1−p2))]² / (p2−p1)².
- Equal-arm sample size uses a two-proportion normal approximation: [zα√(2p̄(1−p̄)) + zβ√(p1(1−p1)+p2(1−p2))]² / (p2−p1)².
- This approximation does not adjust for allocation ratios, multiple comparisons, clustering, sequential testing or low expected counts.
- The tool performs arithmetic on the values entered and does not validate assumptions against outside records.
Worked examples
Worked example
Baseline 3.5%, relative detectable lift 15%, z values 1.96 and 0.84.
The target rate is 4.025%; the formula estimates an equal sample size per arm from the two-proportion normal approximation.
Estimated required sample size per variant is shown.
Understanding the result
A/B test planning needs a baseline conversion rate and a minimum effect worth detecting. The calculator estimates sample size per arm based on the chosen z-values for confidence and power.
This approximation does not adjust for allocation ratios, multiple comparisons, clustering, sequential testing or low expected counts.
Important notes
This approximation does not adjust for allocation ratios, multiple comparisons, clustering, sequential testing or low expected counts.
Treat this output as an estimate tied to the stated inputs. For regulated, safety-critical, tax or professional decisions, verify applicable official guidance and specifications.
Frequently asked questions
Does this guarantee the experiment will be significant?
No. It estimates planning sample size under model assumptions; real outcomes remain uncertain.
What is minimum detectable lift?
It is the smallest relative improvement the experiment is planned to have power to detect under the chosen assumptions.
How should I interpret the result?
This approximation does not adjust for allocation ratios, multiple comparisons, clustering, sequential testing or low expected counts.
Related calculators
- Website Conversion Lift Calculator — Compare baseline and variant conversion rates, percentage-point difference and relative lift using two measured cohorts.
- Event Budget per Attendee Calculator — Estimate total event budget and cost per attendee from fixed costs and a variable per-person cost.
- Sample Size for a Mean Calculator — Estimate the sample size needed to estimate a population mean for a stated standard deviation, margin of error and confidence z-score.
- Office Space Cost per Employee Calculator — Estimate annual and monthly workplace cost per employee and cost per square metre.
- Sample Size Calculator — Estimate a simple random-sample size for a proportion using a chosen confidence z-score, assumed proportion and margin of error.
- Seed Germination Rate Calculator — Calculate the observed germination percentage from a seed test and estimate how many seeds germinated.
Related explanations
Learn the method behind the number or compare the terminology before using the calculator.
- How to Calculate Salary from an Hourly Rate — Convert hourly pay into weekly, monthly, and annual estimates and understand the assumptions behind the result.
- How to Calculate Standard Deviation — Learn what standard deviation measures, the population and sample formulas, and how to interpret the result.
- Basal Metabolic Rate (BMR) — Learn what basal metabolic rate means and how BMR estimates are used in calorie planning.
This page provides a calculation tool plus explanatory information so you can understand what the result means and when the underlying method may not fit your situation.