How do I know what needs attention without checking every system?

Create one exception view that pulls only overdue, failed, high-risk, or unusual items from the systems that own them. Give each item an owner and next action. Staff should spend time resolving exceptions, not opening systems just to see whether something is wrong.