Split Testing
Create
Compare two versions to see which one is better.
A research facility developed an interactive online tool to help calculate sample sizes for clinical studies. Unsure about the best layout, they compared two interface designs. Version A used a step-by-step form, guiding users one input at a time. Version B showed all fields on a single page. Each version was shown to half of the visitors. The team measured completion rates—how many users reached the final calculation and downloaded their report.
Steps
Select one element—like a headline, button, layout, or message—that you want to compare.
Develop version A (the current or baseline) and version B (the variation you’re testing).
Show version A to one group and version B to another, ideally with equal, random distribution.
Track performance using a clear metric—clicks, sign-ups, purchases, or another key behavior.
Use the data to determine which version works best. Apply the winning version, or test again with a new variation.
Based on the ideas of Ronald Fisher (1920s).
Related Building Cards
