Skip to main content

Intervene before a shipment is late

Turn Intelligence risk into a repeatable operations workflow: find urgent shipments, understand the reason, take action, and learn from the outcome.

1. Start with urgent alerts​

Open Intelligence Alerts

Filter Alerts to high + fired. For each row, note:

  • alert type,
  • tracking number and carrier,
  • fired time,
  • top reason,
  • recommended action.

Use the recommendation as the default response:

AlertFirst response
stuckConfirm the latest scan and contact the carrier.
late_riskCheck the promise and prepare a customer notification.
misrouteConfirm the unexpected facility and contact the carrier.
eta_slipCompare the new ETA with the promise and notify the customer when needed.

2. Prioritize by cost of inaction​

Open the Intervention Queue

Use the queue when several shipments need attention. Start at the top and read:

  • P(breach) for SLA-breach probability,
  • elapsed transit against the SLA limit,
  • Act by for the last responsible intervention time,
  • shipment value and customer context where available.

A high-risk low-value shipment may require a different response from a high-value customer order with the same probability.

3. Verify the shipment context​

Before contacting anyone:

  1. Open the shipment and confirm the newest carrier scan.
  2. Compare the predicted ETA with the promised delivery date.
  3. Check whether weather, a hub, or a lane issue is visible on the Network Map.
  4. Confirm the customer has not already been contacted.
  5. Choose the escalation from your team's playbook.

4. Take and record action​

Typical actions include:

  • request a carrier trace,
  • ask a fulfillment partner to investigate,
  • notify the customer before they ask,
  • prepare a replacement or refund decision,
  • monitor without contacting when the risk is medium and the promise has room.

Record the action in the system your team uses for customer or operational history. Include the alert reason and the next review time.

5. Review the outcome​

Return to Alerts and review resolved alerts. Outcomes are graded nightly:

  • tp means later evidence showed the alert was useful.
  • fp means the predicted issue did not materialize.
  • — means the outcome is not graded yet.

Use repeated false positives or missed risks as evidence for model and policy review, not as a reason to ignore one shipment without checking its context.

Daily operating rhythm​

  1. Review high + fired alerts at the start of the shift.
  2. Work the Intervention Queue in Act by order.
  3. Recheck urgent shipments after carrier cutoff times.
  4. Review newly resolved outcomes.
  5. Escalate carrier, facility, or lane patterns visible across several shipments.

Success checklist​

  • High fired alerts have an owner.
  • Every intervention uses the latest tracking evidence.
  • Customer communication happens before the promise is missed when possible.
  • Actions and next-review times are recorded.
  • Outcomes are reviewed for recurring carrier or model patterns.