Unfortunately, this is almost certainly a space where "absent any other variables" is too simplifying an assumption. Some confounding factors to consider: * Is

Unfortunately, this is almost certainly a space where "absent any other variables" is too simplifying an assumption. Some confounding factors to consider: * Is mobile growth rate so high that it outstrips the reduced expected CTR for desktop users? * Is mobile use not displacing desktop use, meaning desktop CTR is not falling as fast as anticipated? * Is mobile use growing in places where there was previously no desktop use? Those users previously had a functional CTR of 0; onboarding them should increase the total clicks even if it reduces the mean CTR.