I've never seen a document, but I can tell you what I do.
All my devices are grouped by location - State, City, Facility, Floor, Room (Each group being a folder). Each folder that relies on another folder has that folder added as a dependency. This helps in suppressing alerts. Each device has the PING monitor set to fail and alert before any other monitor. Again, this helps in suppressing alerts. All my routers are set to alert before switches, and switches alert before servers.
For naming I use the standard DEVICENAME :: MONITORNAME.
For alerts I have a few alert groups. One main group that emails for any alert that goes down. Then i use Smart Groups to split up which monitors alert which tech. I use smart groups so that I don't have to remember to add the monitor to the alert group. It's done for me. For example, all monitors for PING are in a Smart Group which alerts the Network Admin. Anything with SQL in the name alerts the DBA.
I hope this is what you are looking for.