Carrier Comparisons After Adjusting for Lane Mix

A carrier ranking can reverse when the work mix changes

Two carriers may handle different proportions of easy and difficult deliveries. Comparing their overall on-time rates can therefore mix carrier performance with the assignment of work. A distributor deciding how to allocate freight needs to know whether an apparent advantage persists within comparable lanes and service conditions. This research develops a small, reproducible example and a practical method for checking that question.

The analysis does not claim that carriers are interchangeable or that lane adjustment reveals a pure causal effect. Equipment, customer requirements, appointment constraints, shipment size, and selection decisions can remain different within a lane. The purpose is narrower: identify when the aggregate ranking depends materially on the mix of work before turning that ranking into a commercial decision.

Source basis and limits of the analogy

NIST's randomized block design guidance (opens in a new tab) explains how grouping comparable conditions can help account for nuisance factors in experiments. We borrow the comparison principle, not the claim that an observational carrier scorecard is a randomized experiment. The Census Bureau's wholesale survey estimation procedures (opens in a new tab) provide additional context on uncertainty and data errors.

Sources were checked October 2, 2026. All shipment counts below are constructed for explanation. No carrier's actual performance, price, capacity, or suitability is being reported. The proposed standardization is an analytical choice that an operation must evaluate against its own decision and available data.

Reproduce the reversal with two lane groups

Suppose Carrier A handles ten deliveries on an easier lane and one hundred on a more difficult lane. It completes nine of the ten easier deliveries on time and eighty of the hundred difficult deliveries on time. Its combined result is eighty-nine on-time deliveries out of one hundred ten, or approximately eighty point nine percent.

Carrier B handles one hundred easier deliveries and ten difficult deliveries. It completes eighty-five easier deliveries and seven difficult deliveries on time. Its combined result is ninety-two out of one hundred ten, or approximately eighty-three point six percent. The aggregate ranks B above A.

Yet A has the higher observed rate in both lane groups. The overall reversal arises because A performs much more of the difficult work. The example is arithmetic, not evidence that A would necessarily perform better if all future freight were reassigned to it.

Choose a common mix that matches the decision

One simple comparison applies the same lane weights to both carriers. With an illustrative fifty-fifty mix, A's standardized rate is half of ninety plus half of eighty, or eighty-five percent. B's is half of eighty-five plus half of seventy, or seventy-seven point five percent. These numbers describe performance under the selected common weighting of the observed lane rates.

The fifty-fifty choice is not automatically the right one. A manager might use the expected future lane mix, the current combined portfolio, or another approved reference. Publish the weights and their rationale. Choosing weights after seeing which carrier wins would turn a useful adjustment into a tool for justifying a preferred answer.

Keep observed and standardized rates side by side. The observed rate describes the work actually handled. The standardized rate supports a particular like-mix comparison. Neither should replace the other without explanation, especially when the commercial decision concerns an upcoming portfolio that differs from both.

Define comparable groups before calculating rates

A lane name alone may be too broad. Deliveries with strict appointments, special equipment, or different service levels may not be comparable. Choose a small set of grouping variables that reflect the operational question and are recorded consistently. Too few groups leave important differences hidden; too many create sparse cells with unstable rates.

Freeze the grouping rules for the review period. If a shipment changes service level, preserve the accepted commitment and the rule used to classify it. Avoid placing failed shipments into a special difficult category after the outcome is known. That would contaminate the comparison with hindsight.

Reconcile all eligible deliveries, including unresolved proof and canceled movements. A carrier with better documentation should not be penalized merely because its late cases are easier to identify. Report unresolved outcomes by group and test whether they could change the comparison before presenting a firm ranking.

Treat sparse and absent cells as uncertainty

In the example, some rates rely on only ten deliveries. A difference based on small counts may be unstable. Publish numerator and denominator with every rate so readers can see that limitation. Where inferential uncertainty is needed, a qualified analyst should choose an interval appropriate to the sampling or modeling assumptions.

If a carrier has no observations in a lane, there is no observed lane rate to weight. Do not insert zero or copy another carrier's rate. Restrict the comparison to common coverage and disclose what was excluded, or obtain evidence through a carefully designed trial. An adjusted score that conceals missing cells is less defensible than an explicit statement that the comparison cannot yet be made.

Even common coverage can hide selection. A shipper may assign its most urgent freight to one carrier within the same lane. Without recording that assignment logic, a lane-adjusted difference can still reflect the work chosen rather than the carrier's effect.

Connect the analysis to a controlled commercial test

If the comparison materially changes after adjustment, review the assignment policy before changing allocation. Ask which lanes and service conditions the decision will affect, whether each carrier has confirmed capacity, and whether the accepted service definitions match. Historical percentages alone cannot establish future capacity or cost.

A bounded trial can improve evidence if the business can allocate comparable work under an approved design. Define the trial period, eligible shipments, assignment rule, and outcome before starting. Preserve deviations and operational constraints. Do not describe a manager-selected convenience trial as randomized unless random assignment actually occurred.

The transport owner remains responsible for carrier selection and service commitments. Administrative support can prepare the matched shipment groups, retrieve proof, and calculate the approved tables. An analyst reviews the method, while commercial staff evaluate contractual and capacity implications beyond the score.

Preserve the assignment decision as data

Retain the reason a shipment was assigned to a carrier where the organization records it. A capacity refusal, customer requirement, or equipment restriction can explain why a nominally comparable carrier did not receive the same work. These records help distinguish an available alternative from a purely statistical comparison. If the assignment reason is missing, say so; do not infer it from the eventual result.

Before presenting a ranking, have the transport owner inspect several shipments from each comparison group. Ask whether they regard those movements as substitutable for the decision under review. If not, refine the scope and retain the rejected comparison with its reason. This qualitative check cannot remove every bias, but it can catch an analytically tidy grouping that has little operational meaning.

Report the comparison without overstating causality

A useful report includes actual rates, common-mix rates, weights, missing cells, unresolved outcomes, and the key differences that remain uncontrolled. Its conclusion might be that the aggregate advantage disappears under a specified mix. That is more precise than claiming that adjustment proves one carrier is better.

Review sensitivity to reasonable alternative weights. If the ranking changes across plausible future portfolios, the decision may need lane-specific allocation rather than a single company-wide winner. If the ranking remains stable, it is still an observational finding whose limits should remain attached.

For help assembling a comparison that decision-makers can inspect, see freight coordination support and discuss your carrier review. Begin with the portfolio you actually intend to move. The value of this method is exposing the role of work mix, so a carrier decision rests on a relevant comparison rather than on an aggregate percentage with a hidden denominator.