We're setting up our category ontology now and are completely rethinking what we did with our old system. Whereas we once listed every app, server, network device, etc, we're now moving towards a more generic list. We're trying to move to an ontology that reflects only what we need for management reporting rather than the more comprehensive list that feels good because everything has a place but adds ambiguity (do I choose Server | Reboot, Messaging | Exchange or Mobile | Blackberry if I need to bounce the BES server?) and time to the analysts' day. We have 25 IT staff (~10 front-line) servicing 400 users.
We have created three categories but only require the first two. We have sixteen items in Category1, including Backup, HR (for new hires, e.g.), Security, etc. Category 2 is linked to Cat1 and contains about 100 items currently, though I suspect we'll prune another 10-15 out before we go live. Category 3 is also linked to Cat1 (it was linked to Cat2 but we've found it works better this way).
Software is turning out to be a special challenge, but we're probably going to solve it by linking Cat2 to a new parameter in Cat1 rather than to the Cat1 name, as is done now. We can then have multiple Cat1s for Software: SW Install/Uninstall, SW Patch, SW Config, SW Troubleshoot, etc, all of which would link to all the software apps we choose to represent.
Anyway, the purpose of this note isn't so much to describe what we're doing but to understand how everyone else handles categorization. Do you get deep into the weeds with different subcategories for every model of desktop? Have you gone in the opposite direction and opted for a single category? If so how do you handle trending and report requests? What are your criteria for determining when to add a new category or subcategory?