Rule number 1: Always favor your system receiver. A system's coverage area is primarily determined by how well it receives stations in the field, not by how much transmitter power it has.
Rule number 2: This goes hand in hand with Rule number 1: Maximize your received signal. Maximize the isolation from receiver to transmitter. Concentrate on minimizing the insertion loss on the receive side of the system. Test for Effective Sensitivity frequently (at least twice a year) - this includes tests with the transmitter on (but keep in mind the effective sensitivity can be a dynamic value unless the environment is a broadcast site with a constant RF environment). Keep good records - a site logbook is important. If something changes between visits you need to find out what changed and why.
Repeater performance is defined by RF performance as well as audio
performance. You can have the best sounding repeater on the
planet, but if the users can't get into it or hear it then it's
worthless. Audio performance is covered elsewhere on this
web site. Good RF perfomance is a balance between receiver and
transmitter... you want neither an "elephant system" (all ears and no
mouth) or an "alligator system" (all mouth and no ears). It gets
complicated when some users use flea-powered handhelds, some use 10w
mobiles, some use 50w mobiles, and others use hilltop remote
bases. What set of users do you use as a reference for
"balanced"? You have to consider the demographics of your users.