Quite an interesting discussion. In all honesty I don't know specifically what my institution collects and who is in charge of said collections.
What I do know is that information is stored in silos that aren't mutually accessible for any large scale analysis. Our campus doesn't have single-signon (which causes headaches) and some times the same (or similar) data is collected by many entities.
There are many places that this kind of data would be useful (in my immediate realm - library collection development) but the info just isn't available :-)
Academia still seems to be run as a conglomeration of many different silos as opposed to one monolithic organization