Triage delivery alerts
The Alerts view explains delivery risks as they fire and recommends the next operational action. Use it when you need a reason and response, not only a risk score.
Open Alerts
Alert types
| Type | What triggered it | Default action |
|---|---|---|
stuck | The shipment's no-scan risk crossed its carrier-specific threshold. | Contact the carrier. |
late_risk | Predicted late-delivery risk is elevated while there is still time to act. | Notify the customer. |
misroute | The observed facility or route is unusual for the destination. | Contact the carrier. |
eta_slip | The predicted ETA moved materially later or crossed the promised date. | Notify the customer. |
Alerts use carrier and model-specific evidence. Read the Top reason shown in the row before acting; the recommended action is a starting point, not a replacement for the shipment context.
Filter the queue
The header has two filter groups:
- Severity:
all,high, ormedium. - State:
fired,resolved, orexpired.
Start with high + fired for the smallest urgent queue. Use resolved to review risks that cleared, and expired for alerts that aged out without a new outcome. Refresh reloads the selected view.
The summary shows the number of rows currently displayed and how many are high severity.
Read an alert
Each row includes:
| Column | Meaning |
|---|---|
| Sev | High or medium operational priority. |
| Type | Stuck, late risk, misroute, or ETA slip. |
| Tracking | The shipment to investigate. |
| Carrier | The carrier currently handling it. |
| Fired | When the alert crossed its threshold. |
| Top reason | The strongest evidence behind the alert. |
| Recommended action | The next operational response. |
| Outcome | tp when the alert proved useful, fp when it did not, or — while ungraded. |
Outcomes are graded nightly after later tracking and delivery evidence becomes available.
Work an alert
- Filter to high + fired.
- Open the tracking number in your shipment workflow and verify the latest carrier scan.
- Read the top reason and compare it with the promised date and customer context.
- Take the recommended action or apply your team's escalation policy.
- Return later to review the resolved or graded outcome.
Use the Intervention Queue when you want risks ranked by cost of inaction. Use Alerts when you want explicit reasons, states, and recommended actions.
Empty and error states
No alerts means nothing matches the selected state and severity. Switch to
all, or confirm shipments are reaching Ingest & Scoring.
If loading fails, refresh once, confirm the correct organization is selected, and then follow the Intelligence troubleshooting checks.