If we run an experiment and it shows negative effect, should we count this sprint as a rework? On the one hand, it turns out to be a wasted effort, on the other — we get a validated learning.