One root cause triggering dozens of alerts gets ignored on-call. Aggregating by site plus root cause into a single alert with impact makes alerting stay credible. De-duplication and aggregation decide whether on-call is sustainable.
Tag alerts with site and root cause, merge same-cause events within a time window into one with affected page count, first/last times and status distribution. Cross-check security events against the cheat sheet to confirm they're one event, and don't aggregate scan noise into a fake major alert. Record merges and closure conclusions.
Alerting is a recurring action: if numbers don't move, record them as-is — no direction change, no extra spend. This article issues no verdict on any single alert.