Randomized Controlled Trial (A/B Test)

Splitting people at random into a group that gets the change and a group that doesn't, then comparing outcomes. It's the cleanest way to know whether an intervention actually caused a result rather than something else.