暂无此语言版本： ZH. 正在显示英文版本。
计算器 已上线
Sample size calculator
How many users per variant you need to detect an effect with confidence — without opening a spreadsheet.
更新于 2026年5月12日
Enter your baseline conversion rate, the minimum detectable effect (MDE) worth caring about, and the significance level. The calculator returns the sample size per variant and the estimated test duration given your weekly traffic.
When to use it
Before shipping any experiment. Running an underpowered A/B test is worse than not running one: you spend the traffic and still draw the wrong conclusion.
What it assumes
- Two-tailed test, two variants (control + treatment).
- 80% statistical power by default (adjustable).
- Independent conversions — for per-session metrics with repeats, treat the result as a floor, not a ceiling.