How Alert Fatigue Destroys MTTR (and Burns Your Error Budget)

As false-positive ratio increases above 60%, engineers begin pattern-matching alerts to past false positives before investigating. This adds 2-5 minutes to MTTA per page as they mentally categorise the alert before acting. At 80%+ false positive rate, engineers begin acknowledging without investigating until they receive a second page or customer report.

After 6+ months of high page volume, engineers begin acknowledging night pages with a 10-30 minute delay or miss them entirely. The incident.io 2024 survey found 62% report weekly sleep disruption; the natural response is to silence the phone between 11pm and 6am. Real P1s in this window are delayed by hours.

Fatigued engineers conducting simultaneous investigations on multiple concurrent alerts miss the correlating signals that indicate a common root cause. Investigation proceeds on each alert independently. MTTR degrades because root-cause identification takes longer when the engineer cannot see the pattern across 15 simultaneous pages.

Fatigued teams skip post-mortems or conduct cursory ones. Root causes recur. Without a post-mortem culture, alert rules are never updated based on incident learnings, the false-positive ratio continues to climb, and MTTR degrades with each cycle.