The prompt
An experiment is too small to detect a 1% lift overall but you suspect power users react differently. Should you slice it? When does that help, and when is it just noise?
Concepts this question tests
How to practice it in PrepOS
PrepOS's adaptive queue surfaces this question when you calibrate to a matching target level and your weakest concepts overlap with what it tests. Open the practice simulator and select the round type to start a rep with this prompt.
Practice this rep →