There is such thing as too much information, of course. Once you break down data to a certain level, its reliability becomes suspect. That's why polling companies choose a minimum number of people when conducting their surveys. The overall result is far, far more important than the responses of individuals.
A data analytics firm called Musicmetrics has released a system where artists can track responses to any campaign (radio, advertising, social media) hour by hour. The thinking is that the information will allow real-time tweaking of the campaign.
This reminds me of the extent to which data from PPMs (personal people meters)--the devices used to determine ratings for radio--can be misinterpreted. Available software allows for real-time tracking of what causes holders of these PPMs to stick around--and what caused them to tune out.
Panicky program directors will look at a dip in the graph and go "What happened at that moment? What song were we playing? What was the announcer saying that turned off so many people? FIX IT! FIX IT! FIX IT!"
What they fail to understand is that the dip may have been caused by the listening behavior of just two or three people. That's hardly enough data on which to make a rational decision.
In my opinion, this sort of granularity just adds to the overwhelming amount of data noise out there. Yes, this information is important. But if you break it down to minute-by-minute bits, it tells us...what, exactly?
Here's a good story on the Musicmetrics service. This can be valuable stuff--but only if it's interpreted properly and without panic.