You're evaluating a new recommendation idea on a product surface where users may see either the current experience or a personalized one. You need a simple, credible way to measure whether the new treatment improves user behavior.
How would you design a simple experiment to test a new recommendation or personalization idea?
Set up a basic randomized A/B testChoose a primary metric and guardrailsEstimate sample size from baseline rate and target liftInterpret lift with confidence intervals and significance