What would you do if an A/B test shows a statistically significant increase in user clicks, but no change in overall feature adoption?
