How would you answer an interview scenario involving metrics and alerting?

In an interview, I would first define metrics and alerting and the problem it solves, then explain how I would choose meaningful SLIs, set thresholds with duration and context, reduce duplicate alerts, include runbook links, test alerts, and review noisy or unactionable rules regularly. I would also call out the main failure mode: alerting on every CPU spike or individual error creates fatigue, causing responders to miss the alerts that actually indicate user impact. Finally, I would describe how I would test, monitor, and safely roll back or recover the solution.