Three signals drive exception severity in shipping data: ETA variance, carrier reliability history, and order commitment proximity. None of them is sufficient on its own. A large ETA slip may affect no immediate commitment. A small slip may be critical when a customer delivery is due tomorrow. A carrier's revised ETA may be dependable, or it may be an early indication of another delay.
Tenoro uses these signals together to produce a priority score for exceptions that have already occurred. The purpose is not to treat every late movement as equally urgent. It is to rank the exceptions most likely to require operational attention first, based on the data available when the score is computed.
Three signals describe severity from different angles
Each signal answers a different operational question:
- ETA variance: How much has the current arrival estimate moved compared with the original estimate, and how significant is that movement relative to the shipment's planned lead time?
- Carrier reliability history: When this carrier revises an ETA, how often does the shipment arrive within that revised estimate?
- Order commitment proximity: How close is a delivery commitment that depends on this shipment?
The first signal measures the change that has already happened. The second provides context for whether the current change is likely to be a dependable update or a warning of further uncertainty. The third connects the shipment to a business obligation.
ETA variance must be relative to lead time
Absolute delay is easy to calculate, but it is not always a useful severity measure. A two-day slip on a shipment with a 30-day lead time is different from a two-day slip on a shipment expected to arrive in three days.
The scoring model therefore considers ETA variance in relation to the shipment's total planned lead time. A useful representation is:
relative ETA variance = ETA movement / planned lead time
This does not mean that a long-lead shipment can never be urgent. It means that the same number of delayed days should not automatically receive the same priority across different shipment profiles. The model can also use the current date, the original ETA, and the current ETA to distinguish a change that is still distant from a change that has already consumed most of the available delivery window.
Carrier history qualifies the current ETA
A revised ETA is not equally informative for every carrier. Some carriers regularly update their estimates and then deliver within the revised window. Others often revise an estimate again before the shipment arrives.
Tenoro reads historical accuracy on revised ETAs by carrier_id. This produces a carrier reliability index based on past shipments and their milestone events. The index is not a general assessment of carrier quality. It answers a narrower question: how much confidence should an operations team place in this carrier's revised arrival estimates?
For the severity calculation, the carrier signal is represented on a 0 to 10 scale. A carrier with strong historical accuracy contributes less concern to the priority score. A carrier whose revised ETAs have often been missed contributes more. This lets the scoring model distinguish between a current delay that may be stable and one that may compound.
Commitment proximity connects delay to impact
The third signal is order commitment proximity. A shipment can be late without creating an immediate customer or operational problem if there is still sufficient time before the delivery commitment. Conversely, a small ETA movement can become critical when several orders depend on the shipment in the next 24 or 48 hours.
The model considers how close the next commitment is to the shipment's current ETA and how many orders are at risk in that window. This makes the score sensitive to timing, not just delay size. An exception with three delivery commitments approaching may deserve attention before a larger slip that has no commitments in the affected period.
What data the scoring model reads
Severity scoring depends on connecting transport events with order commitments. Tenoro reads the relevant fields from transportation and order systems rather than evaluating a shipment record in isolation.
| Signal | Primary data | What it contributes |
|---|---|---|
| ETA variance | eta_original, eta_current, milestone events |
The size and timing of the estimate movement, relative to planned lead time |
| Carrier reliability history | carrier_id, historical shipments, revised ETA outcomes |
How dependable the carrier's revised ETAs have been |
| Order commitment proximity | order_commit_date, orders_at_risk |
How close and how material the dependent delivery commitments are |
Milestone events add context to the ETA fields. A shipment that has departed, reached a handoff point, or stopped updating may require different interpretation from one with a recent, consistent event sequence. The score is still based on the exception that is visible now, but the event history helps avoid treating every ETA change as identical.
How the three signals become one priority score
Each signal can first be normalized to an illustrative 0 to 10 score. The scoring model then applies weights and combines the results into a priority score. The exact weighting can reflect the operational purpose of the queue, but the important property is interaction.
For example:
- ETA variance score: 0 to 10
- Carrier reliability concern score: 0 to 10
- Commitment proximity score: 0 to 10
An illustrative formulation could be:
priority score = weighted ETA variance + weighted carrier concern + weighted commitment proximity
This is not a prediction of a future delay. It is a ranking of the severity of an exception already present in the data. The current ETA has moved, a milestone condition has triggered, or another defined exception is already visible. The score determines where that exception appears in the priority queue.
The interaction matters more than any single input. A moderate ETA variance combined with weak carrier reliability and a near delivery commitment can outrank a larger ETA variance when the carrier is historically accurate and no commitments fall within the slip window.
Example: the smaller slip that ranks first
Consider a logistics manager at an online retailer who runs daily exception triage. Two open shipping exceptions are waiting for review.
| Exception | ETA movement | Carrier history | Commitments | Result |
|---|---|---|---|---|
| Shipment A | Five-day slip | High reliability on revised ETAs | No delivery commitments in the window | Lower priority |
| Shipment B | Two-day slip | Low reliability on revised ETAs | Three customer commitments in the next 48 hours | Higher priority |
Shipment A has the larger absolute delay. However, its carrier usually delivers within revised ETAs, and no immediate delivery commitment depends on it. Shipment B has a smaller slip, but the current estimate is less trustworthy and three customer commitments are close. Its combined priority score is therefore higher.
This ranking gives the logistics manager a reason to investigate Shipment B first. The next action might include checking milestone events, contacting the carrier, reviewing substitute inventory, or assessing which orders need a customer service response. The score does not decide the action. It helps place the limited attention of the operations team where the timing and potential impact are greatest.
Where the model needs operational judgment
Carrier reliability history is only meaningful when there is enough data behind it. For a new carrier, a rarely used lane, or a carrier with very few comparable shipments, the reliability index is shallow. Treating that thin history as a strong signal would create false confidence.
In those cases, Tenoro weights the carrier signal less. Operations teams can see when a score relies on limited carrier history and apply their own judgment. They may know about a lane change, a seasonal constraint, a newly opened facility, or a service arrangement that is not represented in past shipment outcomes.
The boundary is important: the model does not predict whether a shipment will be delayed. It scores the severity of exceptions that have already occurred, using the available ETA data, carrier history, milestone events, and order commitments at the time of calculation.
Teams can trust the priority score to make the tradeoff between delay size, ETA credibility, and commitment timing explicit. They still need to review thin-data cases and operational context that the source systems do not capture. That combination keeps the queue technically grounded without turning a score into a substitute for judgment.