Overview
Cluster detection runs automatically when new complaints are logged and on a nightly batch. It identifies when complaints of the same failure category are arriving faster than your historical baseline.
How It Works
The detection engine compares recent complaint rates to your configured thresholds:
- Minimum complaint count — how many complaints before triggering (default: 3)
- Time window — the period to look at (default: 14 days)
- Baseline multiplier — how much above normal triggers an alert (default: 2×)
- Baseline period — how far back to calculate "normal" (default: 90 days)
Configure these in Settings → Configuration.
Responding to Clusters
- Review the linked complaints to understand the pattern
- Click Create NCR to initiate a formal investigation
- Transition the cluster status: Open → Investigating → Resolved