Performance indicators can conceal delivered service

A corridor scorecard should make a decision easier to challenge. It should show whether the declared user service is being delivered, which mechanism threatens it and what action the evidence supports. A page of local utilisation, gross tonnage and average dwell can look comprehensive while failing to answer those questions. The designer should begin with the service boundary and a small set of diagnostic relationships, then retain the underlying definitions so that the reader can reconstruct each measure.

Delivered quantity needs an acceptance basis. Gross handling can include recirculation, rejected cargo or movement before final release. A local transfer may count progress that has not reached the customer boundary. The headline should state the quantity basis and event that enters its numerator. A separate local indicator can remain useful for operations, provided it is labelled as such. The scorecard should not force all actors to use one number when their legitimate purposes differ, but it should explain how their numbers reconcile.

Timeliness should describe the commitment and population. An on-time measure needs a defined due event and treatment of revised promises, cancellations and unfinished movements. If the schedule is repeatedly moved after a delay, apparent punctuality can improve without the user receiving a better service. Preserve original and revised commitments so that the report can distinguish planning adaptation from fulfilment. A punctuality claim without that basis invites a favourable interpretation that the evidence may not support.

A diagnostic indicator should also have an owner and response. A growing eligible queue can trigger dispatch coordination; missing release information can trigger a documentary action; loss of reserve can trigger a recovery plan. A metric with no decision consequence is at risk of becoming reporting furniture. The committee should ask what the organisation will do when the indicator changes and who can legitimately make that choice.

An average duration can be useful for a defined population, but it does not describe every user's experience. A small group of severe delays can matter commercially even when the central tendency changes little. Conversely, a change in cargo mix can alter the average without any process becoming worse. The report should retain meaningful case classes and show the evidence relevant to the service commitment. It should avoid publishing tail measures that have not been supported by a suitable observation universe.

Unfinished service should remain visible at the reporting cutoff. Completed movements tell one story; cargo still waiting tells another. A scorecard that reports only completed cases can improve when the most difficult work remains unresolved. The dashboard should show the unfinished population and its states, with durations treated as incomplete. It should also preserve cancellations and diversion where they matter to the user outcome rather than quietly remove them from the performance universe.

Period comparisons require an edition bridge. A new timestamp, route boundary, cargo eligibility rule or sensor coverage can change a metric without changing the underlying service. The report should record that change and decide whether the periods remain comparable. A trend line that joins incompatible definitions can create a false performance story. Where a bridge is not feasible, show a break and explain the new basis instead of prioritising visual continuity over evidential continuity.

The scorecard should separate capability from use. Low use of a spare stage can reflect a binding neighbour, demand limitation or a deliberate recovery reserve. High use can reflect efficient service or a fragile absence of slack. Neither percentage interprets itself. The annual app's utilisation is a ratio within its fictional capacity balance, while an operational scorecard may use productive time, duty availability or resource occupancy. The labels and definitions should make that distinction clear before comparisons are made.

A broad logistics index and a corridor operating measure answer different questions. A perception-based index can describe respondents' assessments of a national logistics environment under its methodology. A site record can describe movements within a declared facility or route universe. Neither should be substituted for the other. A favourable national context does not establish a particular stage's capacity, and a local delay episode does not determine an entire country's logistics performance.

Methodology changes further limit comparison. The source cards retained for the historical and redesigned World Bank logistics indicators describe different evidence bases. The paper does not treat them as an unchanged time series. It also imports no country score, ranking or site capacity. Their useful role here is to demonstrate why a scorecard needs an evidence-object definition and an edition bridge. A new indicator can be valuable precisely because it measures something different.

A corridor committee can use external context alongside internal evidence if the roles are clear. Context can suggest questions about institutional dependencies or service uncertainty. Internal records can test the mechanism within the actual service boundary. The decision should state which source supports which proposition. An external label should not supply an empirical parameter merely because the project team lacks local observations, nor should a local model result be presented as validation of the wider index.

The same caution applies to benchmarking across ports or routes. Cargo mix, boundaries, timestamps, operating conditions and acceptance rules can differ. A comparison can still be informative if those differences are declared and the decision consequence is appropriately narrow. The team should resist a single ranking that conceals incompatible bases. An original comparison table showing scope and limitations can be more useful than a league table whose apparent precision exceeds its evidence.

The scorecard can pair a service outcome with the mechanism that explains it. Accepted delivery should be accompanied by eligible demand and unfinished commitments. End-to-end duration should be accompanied by state intervals and missingness. Reserve readiness should be accompanied by the protected interruption and restoration plan. This structure gives the reader both the result and the conditions needed to interpret it, reducing the temptation to treat a favourable local measure as a complete system judgement.

Test a scorecard revision with a hypothetical queue moved from the gate to an appointment hold upstream. The terminal's gate-to-receipt measure improves because its start event now occurs later in the user's journey. The end-to-end accepted-delivery measure may remain unchanged. Add the held consignments to the same service cohort and retain the original planned arrival, its revisions and actual receipt. If the local indicator still improves while the cohort indicator does not, report an internal process improvement with displaced waiting, rather than a total service gain. This adversarial comparison makes the measure's start event operationally consequential and exposes the effect of changing where eligible demand becomes visible.

A challenge column can identify what would make the current interpretation wrong. If low delivered flow is attributed to rail capability, the scorecard should ask whether demand, cargo eligibility or destination receipt explains it instead. If a shorter delay is presented as an intervention benefit, identify the concurrent changes that could account for it. The challenge need not paralyse action. It directs the next observation or trial and prevents a dashboard from becoming a one-way presentation of management's preferred explanation.

A scorecard can carry a reconciliation disposition instead of forcing competing reports into a single total. When production release rises while accepted delivery is flat, reconcile the difference into in-transit eligible cargo, unfinished receipt, rejection, loss or unresolved identity. When both quantities are flat but completion time improves, retain that time benefit separately from throughput. When a denominator shrinks because users withdrew, show withdrawn demand rather than celebrating utilisation. These branches distinguish a service change from a reporting change and locate the next investigation. The fictional capacity balance supplies no observations for this reconciliation; its contribution is the structural bound against which an actual cohort account can be compared.