Enter the visitors and conversions for your control and variation to see whether the difference is real — with confidence, p‑value, relative uplift and the probability your variation wins.
| Visitors | Conversions | |
|---|---|---|
| Control | ||
| Variation |
Each curve is the plausible range for that version's true rate. The less they overlap, the more real the difference.
Method: confidence and p‑value use a two‑proportion z‑test (one‑sided for the “Better” objective, two‑sided otherwise). Chance‑to‑win uses a Bayesian Beta‑Binomial simulation with uniform priors. Sample‑size estimate targets 80% power at your chosen confidence.