While I agree, the problem is in the data-driven decision-making. There is simply no good way to get good data for this long term type of effect. It's either g

While I agree, the problem is in the data-driven decision-making. There is simply no good way to get good data for this long term type of effect. It's either going to be user interviews, some arbitrary score like customer satisfaction, or maybe a really convoluted split test (long ago cohorts vs new ones... but let's try to remove criteria so it's not apples and oranges). If a company culture only values data in its product decisions, this is always going to happen. Good luck fighting an argument against someone who has a supporting data point, and a revenue increase!