Start with up/down monitoring across all critical systems: network, infrastructure, and applications.

This gives you a baseline to see where problems start, not just where they surface.