Documentation / Understand results
Set the analysis plan
Fix the decision rule before launching. A planned experiment needs both time and enough observations.
Step by step
- In Goals & plan set the baseline conversion rate from comparable historical traffic. Enter the minimum effect you need to detect and the analysis end.
- Read the calculated visitors per variant. A smaller detectable effect typically requires more traffic. Do not choose the plan after looking at a promising result.
- Launch only when the goal and plan are correct. The plan is locked after first launch; a pause does not extend the horizon.
A WordPress lead form converts at a hypothetical 5%. A 20% relative MDE means detecting roughly 6% conversion, an increase of 1 percentage point; it does not mean 25% conversion.
How to verify
Use your own baseline, not the example as a benchmark. Reaching the sample bar early is not permission to stop early. If traffic is too low, the outcome may be inconclusive.