I generally structure backups to handle all SIMPLE recovery model databases one way, all FULL and BULK_LOGGED recovery model databases another way, Master (which gets Full backups whenever SIMPLE databases get Differentials) by itself, and then special handling for special databases (rare). Tempdb is always excluded.

Unless you're very, very space or time constrained, back up everything except tempdb and databases you know you deliberately don't want to (they're restored from another server's backups on a frequent schedule and any changes have no value, etc.)

On most of my servers, there's several tiny databases (including master, msdb, and model), a few small ones, and perhaps one large one, which most often is the most critical. Thus, any exceptional rules for the large one that require more frequent backups have only a marginal cost to apply them to all other like databases (backing everything up more frequently takes little more time or space than backing only the big one up more frequently), especially if it's in the FULL recovery model, since T-Log backups don't back up the same data over and over like Full or Differential backups do.