Count how many exceptions your on-call person can clear well in a shift. Design flags to land under that number on a normal day. Burst days need a overflow lane, not a larger panic.
Separate volume from severity
A spike in failed logins may be noisy but low severity; a single payroll role change may be quiet and high severity. Your board should sort by both axes.
Name an owner
Unowned flags rot. Assign a role, not a hero. Rotate coverage so knowledge does not live in one inbox.
Exception Queue Design is our short course dedicated to this problem; the flagship programme covers it in module three.