Logistics report pitfalls

Start with Logistics reports — eight screens and what they measure to choose a report and the right unit. The rules below help distinguish a data limitation from an error.

Empty report or error

1. Check the period and filters. In the four parcel reports, an empty result has an informational message. Change the range or filters and check whether the orders were imported.
2. Check the date range. The maximum is 186 days, including both ends. The Lead Time, Carrier Performance, and SLA report interfaces no longer offer non-working year or all-history options. Shorten a custom range that is too wide.
3. Check the shipment type. The default SEND does not include returns. Outbound shipments and returns are not identical processes and should not be compared as such.
4. Check the calculation time and source freshness. Refresh bypasses the report cache, but does not fetch new tracking automatically. The latest status in a parcel group does not tell you when synchronisation last ran.
5. If the problem remains, report the report name, company, range, filters, and error text.

The report cache lasts 10 minutes; for anomalies and problem orders it lasts 5 minutes.

No measurement does not mean zero

A parcel can be delivered without a recorded pickup or dispatch. In that case, the outcome may be known while transport time is not. The report does not substitute the delivery time for pickup or create a 0-hour transport.

Check the number of valid measurements beside the result and the tracking coverage. Parcels without tracking still belong to the parcel population. A valid dispatch date can contribute to the measurement of time waiting to ship. Orders without parcels are listed separately.

Some events may be from before the order entered Noxti, for example when historical data is imported. The delivery result may be known, but the time from import to that event cannot be calculated as a positive fulfilment interval. Inconsistent chronology needs its source checked; recalculation does not repair historical records.

Parcel, order, and case are different units

Lead time, transport time, and the E2E target describe parcels. One order with three parcels can provide three measurements. Problem orders count orders, and anomalies count detected cases, including aggregated groups of errors. The forecast describes incoming orders.

That is why comparing counters across these screens can be misleading even with similar filters.

Time since import is not time since purchase

The starting point is when the order enters Noxti. A delay between purchase and import shortens the E2E time visible in the report compared with the time experienced by the customer.

A long time before dispatch does not measure only warehouse work. It may include waiting for payment, stock, or pickup. Likewise, a recorded dispatch and the carrier’s physical receipt of a parcel are not always the same event. A time breakdown does not determine blame.

The 24/48/72-hour thresholds are measured in calendar hours, including weekends and holidays.

On-time delivered does not describe every commitment

The result among delivered parcels excludes open parcels. Read the count of open parcels past the target and the result for resolved or overdue cases beside it. Missing data and parcels still within the target are shown separately.

The E2E target is a threshold chosen for analysis. The shipping deadline saved on an order means something else: it is used for the problem-order list, not a promised delivery date to the customer.

The selected period does not always mean delivery date

Parcel reports select orders by the date they entered Noxti. They can therefore include deliveries that happened later. The trend also groups by order-entry date. Results change as new events arrive for the same group.

In period comparisons, the newer group has had less time for transport to finish. A higher share of open parcels does not prove that carrier performance has worsened.

WMS backlogs on the anomalies screen reflect the current state, regardless of date range. Other anomalies concern the group of orders in the period, while scan anomalies use the scan date.

Older backlogs and partial shipments

In Problem orders, the default Created in the last 90 days excludes older orders. Choose All dates to include older backlogs. This is a separate current-list mode, not a way around the date limit on period-based parcel reports.

Cancelled orders are excluded. A recorded shipping deadline takes precedence over the age threshold. The minimum waiting-time filter may additionally hide a new order that has already passed a short deadline; check it when looking for urgent cases.

Partial shipment is assessed using recorded shipments and parcels. The metric does not check whether all required product quantities have been fulfilled.

Counters cover the entire filtered result; the list page is only a slice. Amounts in different currencies are shown separately. An order value is not automatically the value of a loss or compensation.

Longer analysis requires combining results correctly

For history longer than 186 days, choose non-overlapping periods. Do not add averages, percentages, medians, or percentiles.

Counts can be added only for non-overlapping populations with the same filters and definition. Calculate a combined percentage from the summed numerators and denominators. An average must be weighted by the number of valid measurements, not all parcels. Medians and percentiles must be recalculated from the combined measurements; they cannot be reconstructed from period medians alone. An approximation from rounded averages shown on screen will not be exact.

An anomaly is not a confirmed incident

A valid transition from status 50 to 60 is not a regression. An actual step backward remains a signal to check mapping and chronology. Several successful scans for the same parcel may represent different valid actions. Missing tracking after a recorded dispatch requires checking both the integration and the parcel handover.

The list shows up to 50 examples per category; full counts and the number omitted are shown separately. The report suggests an action but does not record an owner or whether a case was closed. No alert is not a certificate that the process is correct.

The forecast has no confirmed accuracy; FDA remains unknown

The forecast is the average order inflow over the four weeks before and including the selected period’s end date. The horizon is the following seven days. If you select a historical period, the forecast also covers a historical horizon, not the current week.

No error assessment has been run on previously unused periods. Four similar weeks do not prove high accuracy; without history there is no forecast. The report does not forecast deliveries. Carrier and non-standard shipment-type filters can narrow the inflow to work that already has an assigned shipment.

FDA, or first-attempt delivery success, requires mapped delivery-attempt events and their outcomes. The report currently marks it unavailable and gives the number of unknown cases. No failed-attempt wording in the history is not proof of success.

Permissions

Each of the seven reports has its own permission. Access to the overview does not open the other screens. See Report permissions — two layers and four mismatches.