There's an extremely well defined industry surrounding conversion lift. Which is exactly what you're talking about. Advertisers use holdout groups to test what

There's an extremely well defined industry surrounding conversion lift. Which is exactly what you're talking about. Advertisers use holdout groups to test what % lift was correlated to an ad. Using large enough datasets, it's possible to have strong statistical confidence in the effects of ads on user behavior.