You are the PM for a digital product that runs frequent experiments to improve conversion. One recent test increased sign-up completion, but user satisfaction dropped in the same release window. The team now has to decide whether the change is a win, a temporary cost, or a sign the experiment should be rolled back.
What would you do if an experiment improved conversion but reduced user satisfaction?