Amostra A/B conversão
- Created by
- Renato Passos, Eng. de Software
- Reviewed by
- Renato Passos, Eng. de Software
Last updated: Apr 18, 2026
About this calculator
The A/B test sample size calculator determines the necessary sample size for reliable results, using the formula 16·p(1−p)/Δ². Here, 'p' is the expected conversion rate and 'Δ' is the smallest detectable difference. This tool prevents underpowered tests (missing real effects) or overpowered ones (wasting resources), ensuring statistical validity for digital marketing, UX testing, or funnel optimization.
The formula accounts for conversion variability (p·(1-p)) and the minimum effect size (Δ) you want to detect. For example, if your current conversion rate is 5% and you aim to identify a 0.5% improvement, the calculator shows how many users are needed for 80% power at 95% confidence. This is crucial for experiments where small gains justify significant investments.
Key assumptions include random sampling and consistent data conditions. If the baseline conversion rate (p) is uncertain, use a conservative estimate. Adjustments are also required for multi-variant tests or when group allocations differ. Always validate results with real-world data, as theoretical calculations can't account for all variables.
Frequently asked questions
What is Δ in the formula?
Δ represents the smallest conversion rate difference you want to detect. For example, if your current rate is 5%, a Δ of 0.5% means the test will identify changes from 5.5%. Choose a value that balances practical significance and resource constraints.
Why use the formula 16·p(1−p)/Δ²?
This formula is a simplified approximation for A/B tests with 80% power and 95% confidence, assuming equal group allocation. More precise calculations require advanced statistical methods, but this approximation works for most practical cases.
Should I adjust the result for more than two variations?
Yes. For tests with N variations, multiply the result by (N-1) to maintain statistical power. This accounts for the increased number of possible comparisons between variations.
How do I interpret the calculator's output?
The number shown is the minimum sample size per group. For a two-variation test, multiply by 2 to get the total required. Ensure you collect data until reaching this threshold before analyzing results.