Build a carrier scorecard that separates acceptance, execution and evidence

A carrier performance scorecard should explain where service breaks down. A single on-time percentage cannot show whether a carrier declines difficult work, accepts more than it can place, delivers late, or completes the movement but supplies evidence too slowly.

Start with separate populations and traceable events. Procurement needs to know what was offered, what the carrier accepted, what it executed and what evidence it returned. Those counts should connect without forcing every metric to use the same denominator.

Follow an illustrative cohort through the operation

Suppose a shipper reviews 100 unique, eligible load offers sent to one carrier during a defined period. Each offer has a valid response deadline and required service specification. Duplicate transmissions are removed.

The carrier accepts 60 offers and declines or times out on 40. All 60 accepted loads have reached their pickup commitment by the reporting cut-off. Vehicles were placed on time for 55 and late for five.

For delivery, 50 of the 60 accepted loads are complete: 45 completed on time and five completed late. Five additional loads are still open after their delivery commitment, and the remaining five have future delivery commitments. Valid delivery evidence has arrived for 40 of the 50 completed loads.

The delivery-due denominator includes the 50 completed loads and the five overdue open loads. It excludes the five whose delivery commitments are still in the future. Those five are pending exposure, not current service failures.

The carrier can therefore show 90% on-time delivery on completed work while providing a materially different picture across all work already due. Acceptance at 60% adds another question: which offers were refused, and why?

Keep offer and load identities separate

A load can be offered to more than one carrier, sequentially or under an agreed procurement process. Count the chosen unit explicitly. A carrier's response rate may use unique eligible offers to that carrier, while a shipper's coverage metric may use unique loads requiring transport.

A retransmitted message is not automatically a new commercial offer. A revised offer with a different pickup date or equipment requirement may need a new version. Preserve those links so retries do not inflate the denominator and amendments do not erase an earlier response.

Oracle documents distinct tender acceptance statuses in its transport workflow. Those statuses can help identify response states, but the organization still needs to define its metric population and treatment of withdrawn or superseded offers. Oracle Transportation Management 26C — Tender Acceptance Status

Capture refusal and timeout reasons. An offer withdrawn by the shipper before the response deadline should not silently become a carrier rejection. An out-of-scope offer should be visible as a procurement or master-data problem rather than counted as ordinary eligible work.

Measure commitments without rewriting history

Store the original accepted pickup and delivery commitments, the actual events and any subsequently approved revisions. A new ETA is not necessarily a new contractual commitment.

Define whether “on time” means arrival, check-in, start of unloading or completion. Include the agreed tolerance and time zone. If appointment access is controlled by a consignee, preserve the evidence needed to distinguish arrival from access delay without automatically assigning liability.

Freeze a reporting cut-off and retain open overdue work. A completed-only metric can be useful for studying executed trips, but label it as such and show the overdue count beside it. Recomputing after late trips close should not erase what the team knew at the earlier review.

A record entered late also needs both event time and receipt time. Backdated data may correct a physical-service measure while still revealing a visibility problem. Treat the correction as an auditable change, not a reason to collapse the two measures.

Separate physical completion from evidence quality

In the worked example, the 50 completed deliveries have independently verified completion events under the shipper's defined rule. Only 40 have the required valid evidence package. The ten missing packages are evidence gaps, not ten additional undelivered loads.

If the operation cannot verify completion without that package, it needs a different state and denominator. It should not copy this example's assumption into its scorecard.

Define what makes evidence valid: correct shipment or consignment association, required acknowledgement, legible information and handling of partial or disputed deliveries. A file uploaded against the wrong load does not satisfy the requirement.

GS1 EPCIS provides standards for exchanging visibility event information. It can support structured event records, but it does not define the shipper's contractual proof requirements or prove that an uploaded document is valid. GS1 — EPCIS and CBV artefacts

Evidence timeliness also needs its own deadline. The 80% in the table measures completeness at a cut-off. It cannot be called “on-time POD” unless each completed load's evidence deadline and submission history are evaluated.

Compare like work before changing allocation

Break the offer and execution populations into meaningful lane, equipment and service groups. A carrier may accept mostly easy urban work while declining remote destinations. Another may complete a harder mix and look worse in an unadjusted aggregate.

Show volumes with percentages. Two late deliveries out of ten and 20 out of 100 share a percentage but provide different amounts of evidence. Retain approved exclusions and their reasons, then show both the original population and the adjusted calculation where an adjustment is justified.

Avoid a weighted score that lets strong document uploads offset repeated inability to cover an essential lane. Mandatory service or capacity requirements can remain explicit gates. If a composite score is used, publish the weights, population rules and sensitivity to small samples.

Review the cause before imposing a remedy. Low acceptance may require a different specification or capacity agreement. Poor placement may require operational escalation. Late evidence may require a document handoff change. Each finding should lead to an owner and an action that addresses the observed failure.

Make the scorecard reproducible

The useful scorecard lets procurement and operations trace a percentage back to individual offers, commitments and events. Its purpose is a defensible allocation conversation, with pending work and missing evidence still visible.

Next actions

1. Define unique offer and load identities, eligibility rules and the reporting cut-off.
2. Calculate acceptance, placement, delivery-due performance and evidence completion from their own populations.
3. Reconcile exclusions, revised commitments and overdue open work before comparing carriers.
4. Review results by comparable lane and service mix, then assign a specific corrective action to each material gap.