We've been trying to do stuff like that for a long time, e.g., the impact of smartphones. My understanding though is that such changes are so widespread it's really hard to have a meaningful control group in the same cohort etc.