Answer
The first-alert set should cover the conditions most likely to create business disruption ... failed or stalled jobs, storage thresholds, subsystem stress, message queue issues, replication health if HA exists, and security events that need immediate review. Buyers do not need every metric on day one. They need dependable coverage of the few signals that actually predict downtime.
That is why alert ownership matters as much as metric coverage. A monitoring platform only helps when each serious event reaches a real person with enough context to act quickly.