Confidence Interval Calculator
Calculate an interval around a conversion-rate or numeric-metric difference between two independent groups.
Why use this calculator
See the range around the observed difference
A point estimate says what happened in the observed sample. A confidence interval adds the missing context: how precise that estimate is and which positive, negative, or near-zero effects remain compatible with the data.
What does a confidence interval mean?
A 95% confidence interval is a range produced by a procedure that would contain the true value in 95% of repeated samples under its assumptions. For an A/B test, it is often the clearest way to show both the estimated lift and the uncertainty around it.
If the interval for treatment minus control includes zero, the data are still compatible with no difference at the selected confidence level. That does not prove the variants are equivalent; it may mean the interval is too wide to rule out effects that matter.
How to interpret confidence intervals in A/B testing
Start with the full range, not only whether it crosses zero. An interval from −0.2 to +1.8 percentage points has a very different decision implication from an interval tightly concentrated around +0.1 percentage points, even if both are statistically significant or non-significant in a simple threshold test.
Precision comes from sample size and variability
More observations usually make a confidence interval narrower. For numeric metrics, high variation widens the interval; for conversion rates, rare events and smaller groups make the range less precise.
Match planning and analysis
Use the same unit of analysis, metric definition, and treatment assignment population in your confidence interval that you used to plan the experiment. Ratio metrics, clustered randomization, and sequential designs can need a different interval method.