The analytics engine returned HTTP 503 Service Unavailable from multiple regions, causing the monitor to trigger. The service recovered across regions within about 9 minutes, the incident was acknowledged and fixed by Jamie Chennells, and it resolved automatically after the brief outage.