Did the variant actually convert better?
Observed lift is the start. Uncertainty decides what you can claim.
Observed experiment counts
Representative experiment counts are ready.
Lift without uncertainty is still hype.
Calculate rates, intervals, and a two-proportion test before treating a variant as a conversion win.
Canonical experiment evidence
Variant lift audit
Decision evidence is visible first.
Control
0.00%95% CI 0.00-0.00%Variant
0.00%95% CI 0.00-0.00%0.00 ppabsolute lift
0.00%relative lift
0.0000z statistic
0.0000two-sided p
Interpretation
The two-proportion z test evaluates a fixed sample under independent assignment. Wilson intervals show plausible rate ranges without relying on a fragile normal interval for each arm.
Evidence boundary
This local calculator does not verify randomization, sample-ratio balance, novelty, attribution, repeated peeking, bot traffic, revenue quality, or causal validity. Plan stopping rules before the experiment.