A new idea deserves a test, not a promise. Change one limited action, observe the result, and use that evidence to decide whether to continue.