Skip to main content

Anomaly Classification

An anomaly flags a tradeflow whose computed transport plan contains an internal inconsistency worth a human look. It is a "check this one" signal, not a verdict that the shipment is wrong.

An anomaly is binary

A tradeflow either has an anomaly or it does not. The dashboard flags any tradeflow with an anomaly present (anomaly > 0). There is no "small" or "large" anomaly to tune past - if the flag is on, something in the plan is worth reviewing. The underlying score is only a magnitude for sorting the worst first.


What triggers an anomaly

Anomalies come from the transport-plan builder detecting that the story it assembled does not hold together. The common triggers, in plain terms:

TriggerWhat it meansTypical cause
Multiple shipments mergedMore than one distinct journey appears to have been folded into one tradeflowTwo bookings tracked under one reference; reused container number dragging an old journey in
Multiple bills of ladingMore than one distinct BL detected on the tradeflowSeveral BLs entered together; a BL reference that should have been split out
Container with no transport planA container on the tradeflow could not be placed into any planMissing or unmatched events for that container
Milestones out of orderConfirmed events sit in an impossible chronological sequenceConflicting source timestamps; a late correction landing out of order
Origin equals destinationThe load port and final discharge port resolve to the same locationMissing POL/POD information producing a circular plan; two journeys merged (see note)
Transshipment after final portA transshipment step appears after what was marked the final dischargeA location wrongly classified as final POD
Orphan eventsEvents that could not be attached to any point in the planLocation or sequencing data that does not line up
API locations with empty datesLocations supplied via API arrive without usable dates while live data exists elsewhereIncomplete API input alongside live tracking

Most anomalies trace back to the input the plan was built from - references that should have been split, a reused container, a missing port - rather than to a container genuinely doing something impossible.

A container never runs a round trip

No single container ships out of a port and back to that same port within one tradeflow. So when the load port and final discharge port of the same container resolve to one location, it is always a defect - a phantom leg, two journeys merged, or missing port information - never a legitimate there-and-back. (Two different containers on one tradeflow travelling opposite directions, sharing a location, is normal and is not flagged.)


What you are expected to do

An anomaly is an invitation to check, in this order:

  1. Open the tradeflow and look at the plan the anomaly is pointing at.
  2. Check the references - is more than one booking or BL riding on one tradeflow that should be separate? Is a container number reused from an earlier shipment?
  3. Check the ports - is a POL or final POD missing or mis-resolved, producing a circular or truncated plan?
  4. If the input is correct and the plan is still inconsistent, that is a case worth reporting to support - it points at something to fix on the platform side, not on yours.

You do not need to act on every anomaly immediately. Sort worst-first and work down; a tradeflow that is already delivered and correct despite a flag can be left alone.


What an anomaly is not

  • Not an SLA breach. Anomalies are about plan consistency, not timeliness. The timeliness metrics are Freshness, TTFF and Confirmation Delay.
  • Not automatically your fault or ours. It is a neutral signal that the assembled plan is inconsistent; the cause could be input data or platform handling, which is what the review step establishes.
  • Not gradable. There is no threshold to set. Present or absent.