12. Logistics Costing, Analytics and Decision Models

Introduction and Context

International logistics decisions are often presented as choices between freight rates, transit times, carriers, ports, or modes. In practice, however, the financially relevant question is broader: what will the shipment cost the firm from the point of origin to the point at which the goods are available for sale or production? A low ocean rate may be outweighed by port congestion, inventory carrying cost, customs penalties, inland repositioning, or stockouts. Similarly, the cheapest supplier may not offer the lowest total landed cost.

Logistics costing provides the financial structure for these decisions. Analytics converts operational data into evidence about cost, service, risk, and capacity. Decision models help managers compare alternatives when objectives conflict and when future conditions are uncertain. Together, these capabilities support sourcing, routing, inventory positioning, carrier selection, shipment consolidation, mode choice, and network design.

This chapter examines the calculation of freight, duties, inventory cost, and total landed cost; the design and interpretation of logistics performance indicators; forecasting and optimization methods; sensitivity analysis and simulation; and multicriteria decision models. The emphasis is not merely on mathematical technique but on managerial judgment: selecting appropriate assumptions, recognizing uncertainty, and ensuring that a model reflects the commercial and regulatory reality of international trade.

Learning Objectives

After studying this chapter, the student should be able to:

1. Distinguish logistics cost categories and construct a shipment-level cost model.

2. Calculate freight, customs duty, taxes, inventory carrying cost, and total landed cost.

3. Design KPI dashboards that connect financial, operational, service, and risk measures.

4. Apply forecasting, optimization, sensitivity analysis, simulation, and multicriteria decision methods.

5. Interpret model outputs and identify assumptions, limitations, and data-quality problems.

6. Recommend logistics alternatives using transparent and professionally defensible evidence.

The Architecture of International Logistics Cost

A logistics cost model begins by defining the boundary of analysis. The boundary may extend from supplier gate to buyer warehouse, from port of export to distribution centre, or across the entire order-to-delivery process. The selected boundary must be stated because different Incoterms allocate transport, insurance, customs, and risk responsibilities differently. An analysis based on an invoice price under one Incoterm cannot be compared directly with an invoice price under another without normalising the cost elements.

A useful classification is as follows:

• Purchase-related cost: supplier price, tooling, packaging, and origin preparation.
• Transport cost: pre-carriage, export handling, main carriage, terminal charges, fuel or security surcharges, and destination delivery.
• Trade compliance cost: customs duties, import taxes, brokerage, documentation, inspections, and penalties.
• Inventory cost: capital tied up in goods, storage, insurance, obsolescence, shrinkage, and handling.
• Exception and risk cost: demurrage, detention, disruption, claims, expedited freight, stockouts, and quality failures.
• Administrative and systems cost: order processing, transaction fees, data integration, and management overhead.

The distinction between fixed and variable cost is important. A container booking fee may be fixed per shipment, while freight may vary by weight or volume. A distribution centre lease is largely fixed over the short term, whereas handling labour may vary with order lines. Cost behaviour affects the preferred shipment frequency and shipment size.

Basic costing relationships

For a shipment, total landed cost can be expressed as:

TLC = Product cost + Origin logistics cost + International freight + Insurance + Import duties and taxes + Destination logistics cost + Inventory cost + Expected exception cost

The expression is a managerial model rather than a universal accounting formula. Some organisations include only costs required to make goods available for sale; others add post-arrival warehousing, financing, and expected service failure. The model must therefore identify inclusions and exclusions explicitly.

If a shipment contains Q units and the total landed cost is TLC, landed cost per unit is:

Landed cost per unit = TLC / Q

When exchange rates are involved, all values should be converted using a declared rate and date. A sensitivity range is preferable to a single exchange-rate assumption when the purchasing cycle is long.

Freight calculation

Freight may be quoted per container, shipment, kilogram, cubic metre, pallet, chargeable weight, or transport leg. Air freight commonly uses chargeable weight, which is the greater of actual weight and volumetric weight. A simplified volumetric calculation is:

Volumetric weight = Length × Width × Height / Volume divisor

The applicable divisor depends on the carrier and unit system. A cost model should also include minimum charges, origin and destination fees, documentation, security surcharges, terminal handling, and accessorial charges. Comparing only the headline rate creates a systematic understatement of cost.

For example, assume 1,000 units have a supplier price of $18 each. Origin charges are $700, ocean freight is $3,200, insurance is $180, destination charges are $1,050, and inland delivery is $900. The customs value is defined for this illustration as product value plus freight and insurance:

• Product value = 1,000 × $18 = $18,000
• Customs value = $18,000 + $3,200 + $180 = $21,380
• Duty at 8% = $1,710.40
• Other logistics costs = $700 + $3,200 + $180 + $1,050 + $900 = $6,030
• Total landed cost before inventory cost = $18,000 + $6,030 + $1,710.40 = $25,740.40
• Landed cost per unit = $25.74

Actual customs valuation rules vary by jurisdiction and transaction circumstances. The illustration therefore demonstrates method, not a legal conclusion.

Duties, taxes, and customs valuation

Import duty is usually calculated as:

Duty = Dutiable customs value × Duty rate

The customs value, classification, origin, valuation method, and eligibility for preferential treatment must be verified through applicable law and customs documentation. Tariff classification errors may result in overpayment, delays, audits, or penalties. Preferential trade agreements can reduce duty, but only if origin rules and documentary requirements are satisfied.

Import VAT or similar consumption taxes may be calculated on a base that includes customs value, duty, and certain transport or insurance charges. Whether such tax is a final cost depends on the organisation’s ability to recover it. A non-recoverable tax belongs in landed cost; a recoverable tax may be recorded as a cash-flow and compliance consideration rather than an economic cost. The model must not treat all taxes as interchangeable.

Inventory carrying cost and time

International logistics decisions affect inventory through both quantity and elapsed time. Annual inventory carrying cost is commonly estimated as:

Annual carrying cost = Average inventory value × Carrying-rate percentage

The carrying-rate percentage may include cost of capital, insurance, storage, obsolescence, damage, and administration. For goods in transit, a time-based approximation is:

In-transit inventory cost = Shipment value × Annual carrying rate × Transit days / 365

Suppose the shipment value is $21,380, the annual carrying rate is 20%, and transit and clearance take 42 days. The estimated in-transit cost is:

$21,380 × 0.20 × 42/365 = approximately $492

This amount may appear small for one shipment but becomes material across recurring flows. A slower mode can also require higher cycle stock and safety stock. Thus, the inventory consequence of a mode choice should include both pipeline inventory and service-protection inventory.

Expected cost of risk and service failure

A deterministic model may conceal important exposure. Expected exception cost can be estimated as:

Expected risk cost = Probability of event × Financial impact of event

If a disruption has a 12% probability and would cost $15,000 through expedited transport, lost margin, and administrative recovery, its expected cost is $1,800. This does not predict that the event will occur; it provides a decision value for repeated or portfolio-level analysis. Managers should also consider the size of the worst-case loss, because alternatives with equal expected cost may have very different resilience profiles.

From Cost Tables to Performance Intelligence

Costing explains what was spent; analytics explains why it was spent and what may happen next. A logistics dashboard should connect operational activity with business outcomes. A large number of indicators can weaken attention, so the selection should reflect strategic priorities and decision rights.

Relevant KPIs include:

• Freight cost per unit or per kilogram: useful for comparing lanes after normalising shipment characteristics.
• Total landed cost per unit: combines product, trade, transport, inventory, and relevant exception costs.
• On-time delivery: percentage of shipments delivered within the agreed window.
• Order cycle time: elapsed time from order confirmation to delivery or availability.
• Transit-time variability: dispersion of transit time, not merely its average.
• Customs clearance time: time between arrival and release.
• Documentation accuracy: percentage of documents accepted without correction.
• Container or vehicle utilisation: used capacity relative to available capacity.
• Damage and claims rate: quantity or value of damaged shipments relative to shipments handled.
• Inventory turns and days of supply: indicators of capital productivity.
• Forecast accuracy and bias: measures of demand-planning quality.
• Carbon intensity: emissions per tonne-kilometre, unit, order, or shipment, depending on the decision.

KPI definitions require governance. “On time” must have a defined reference point, tolerance, and data source. A carrier dashboard based on planned departure may show high performance while customers experience late delivery. Likewise, a cost-per-unit KPI may improve because shipment sizes increased, while inventory and obsolescence costs worsened.

A balanced dashboard commonly has four perspectives: financial efficiency, customer service, process reliability, and resilience or sustainability. Drill-down capability should allow managers to move from a network result to country, lane, carrier, port, product, customer, and shipment-level causes.

Forecasting Demand, Time, and Cost

Forecasting supports capacity booking, inventory positioning, labour planning, and budget control. It is not limited to demand. Transit time, port dwell, fuel price, foreign exchange, and customs workload may also be forecast.

A moving average is appropriate when demand is relatively stable and recent observations deserve emphasis. Exponential smoothing gives greater weight to recent observations and can be extended to trend and seasonality. Regression and causal models are useful where demand depends on price, promotions, economic conditions, or customer activity. Machine-learning methods may identify nonlinear patterns, but they do not remove the need for clean data, explainability, and operational judgment.

Forecast evaluation should distinguish accuracy from bias. Mean absolute error measures average absolute deviation; mean absolute percentage error can be useful but becomes unstable when actual demand is near zero. Forecast bias identifies systematic over- or under-forecasting. For logistics, a forecast that is slightly less accurate but consistently conservative may be preferable to one with severe under-forecasting during peak periods.

Forecasts should generate scenarios rather than a single unquestioned number. A base case, high-demand case, and disruption case can be translated into capacity, inventory, and cost consequences. Forecast error should feed safety-stock calculations and simulation rather than being ignored.

Optimization and Decision Models

Optimization formalises the allocation of scarce resources subject to constraints. A linear programming model may minimise total logistics cost while deciding quantities shipped through plants, ports, modes, and warehouses. A typical objective is:

Minimise total cost = purchase + transport + handling + inventory + fixed facility + penalty costs

Subject to supply capacity, demand fulfilment, lane capacity, warehouse capacity, service requirements, customs restrictions, and non-negativity constraints.

A transportation model assigns quantities from origins to destinations. A transshipment model permits intermediate nodes such as ports or consolidation centres. A mixed-integer model is required when decisions are discrete, such as opening a warehouse, selecting a carrier, or booking a full container. Vehicle-routing models address sequencing and capacity, while inventory models determine replenishment quantities and locations.

Optimization results should not be accepted merely because a solver produced them. Managers must test whether the solution is feasible in practice, whether input data are reliable, and whether constraints have unintentionally excluded important risks. Shadow prices can be informative: they estimate the value of relaxing a binding constraint, such as an additional unit of warehouse capacity or a greater port throughput allowance.

Sensitivity analysis

Sensitivity analysis examines how conclusions change when assumptions vary. Important variables include demand, fuel price, exchange rate, duty rate, freight rate, lead time, carrying rate, and disruption probability. One-way sensitivity changes one variable at a time; scenario analysis changes several related assumptions; break-even analysis identifies the point at which one alternative becomes preferable.

For instance, air freight may be economically justified when the cost of stockout and lost margin exceeds the premium over ocean freight. The break-even condition can be expressed as:

Air premium ≤ Avoided inventory cost + Avoided shortage cost + Other avoided costs

Sensitivity analysis is especially important when a recommendation depends on a narrow cost difference. A solution that is cheapest under one precise forecast but inferior under modest variation is less robust than a slightly more expensive solution with stable performance.

Simulation under uncertainty

Simulation models repeated possible outcomes using probability distributions. A Monte Carlo model might sample demand, transit time, exchange rate, and disruption occurrence, then calculate total cost and service performance for each trial. The result is a distribution rather than a single estimate, allowing managers to examine mean cost, percentile cost, probability of stockout, and downside exposure.

Simulation requires credible distributions and correlations. Demand surges may coincide with capacity shortages; treating these events as independent can understate risk. Simulation also depends on validation: historical results, expert review, and sensitivity tests should be used to assess whether the model behaves plausibly.

Multicriteria Choices in International Logistics

Cost is important but rarely sufficient. Carrier and mode choices may also involve reliability, flexibility, emissions, security, customs capability, visibility, and strategic fit. Multicriteria decision-making makes these trade-offs explicit.

The weighted-scoring method is a practical approach. First, define criteria and measurable scales. Second, assign weights whose sum equals one. Third, score each alternative, preferably using evidence rather than unsupported opinion. Fourth, calculate the weighted total:

Alternative score = Σ (criterion weight × alternative performance score)

Suppose a firm evaluates carriers using cost weight 0.35, reliability 0.30, visibility 0.15, flexibility 0.10, and sustainability 0.10. Scores must be normalised so that a high score has the same meaning across criteria. A cost criterion may be scored inversely, because lower cost is better. The result is not an objective truth; it is a structured expression of preferences.

The analytic hierarchy process can be used when criteria are hierarchical and pairwise comparisons are useful. Data-envelopment approaches compare relative efficiency across multiple inputs and outputs. Goal programming is appropriate when management has target levels, such as minimum service and maximum emissions, rather than a single cost objective. The technique should match the decision, data quality, and required level of explanation.

A robust decision process includes a governance review. Weights should be approved by relevant stakeholders, conflicts of interest should be disclosed, and the outcome should be tested by changing weights. If a small change in weights reverses the ranking, the decision is preference-sensitive and should be presented as such.

An Illustrative Case: Choosing a Mode for a Seasonal Import

Background

Northstar Appliances, a fictional importer, sources 4,000 compact appliances from an overseas supplier for a seasonal sales campaign. The goods must be available at the national distribution centre no later than 30 September. Management is considering ocean freight, rail-assisted intermodal transport, and air freight for part of the order.

Situation and problem

Ocean freight has the lowest direct transport cost but a variable lead time. Rail-assisted transport costs more but offers a shorter and more reliable transit window. Air freight is expensive but can protect the campaign if demand materialises late. The commercial team initially recommends ocean freight because its quoted rate is lowest. The logistics team argues that this comparison ignores inventory and shortage consequences.

Analysis

The team defines the decision boundary as supplier gate to distribution-centre availability. It estimates product value, freight, insurance, duty, destination charges, pipeline inventory, and expected shortage cost. It then creates three scenarios:

• Base scenario: forecast demand is achieved and transit follows the expected duration.
• High-demand scenario: demand is 20% above forecast and late delivery causes lost margin.
• Disruption scenario: ocean transit is delayed by 18 days.

A weighted decision model assigns 45% to total economic cost, 25% to delivery reliability, 15% to flexibility, and 15% to emissions. The weights are approved by procurement, sales, finance, and compliance rather than by one department alone.

The analysis shows that ocean is cheapest in the base scenario, but its expected shortage and disruption exposure is high. Rail-assisted transport has a slightly higher expected cost but materially lower delivery variance. A hybrid strategy—most units by ocean, a limited quantity by rail, and an option for emergency air freight—performs best across the scenarios. The hybrid plan also reduces the risk of paying for air freight unnecessarily.

Approach and solution

Northstar books ocean capacity for the core volume, reserves rail capacity for a defined proportion, and establishes an air-freight trigger based on actual demand and vessel status. The trigger is governed by a break-even calculation comparing the air premium with the expected margin protected by avoiding stockout. A dashboard tracks booking confirmation, estimated arrival, customs documentation, inventory coverage, and forecast updates.

Outcome

The case is illustrative rather than a report of an actual company. Its principal outcome is a decision architecture: the firm does not select a mode solely from a rate sheet. It combines landed cost, variability, service exposure, and managerial options. The hybrid plan may not be the lowest-cost outcome in every realised scenario, but it reduces dependence on a single forecast and provides a controlled response to disruption.

Learning points

1. Cost boundaries and Incoterms must be clarified before alternatives are compared.

2. Variability can be more important than average transit time.

3. Scenario and break-even analysis help determine when premium transport is justified.

4. A hybrid solution can outperform a single-mode solution when uncertainty is material.

5. Decision models support judgment; they do not replace cross-functional accountability.

Professional Practice, Governance, and Common Errors

The greatest weakness in logistics costing is often not arithmetic but definition. Common errors include double-counting charges, omitting accessorial fees, using inconsistent currencies, treating recoverable taxes as permanent costs, applying the wrong customs value, and comparing shipments with different utilisation levels. Another frequent error is allocating a container’s cost equally across units when products occupy different volume or weight proportions.

Data governance is essential. Source systems should identify shipment, invoice, purchase order, tariff classification, origin, destination, carrier, and event timestamps consistently. Master-data errors can distort dashboards and optimization models at scale. Access controls are needed because freight rates, supplier prices, and customer information are commercially sensitive.

Ethically, managers must not manipulate KPI definitions to conceal poor performance or select weights that predetermine a preferred supplier without disclosure. Customs information must be accurate, and model outputs must not be used to justify misclassification, undervaluation, evasion of duties, or circumvention of trade controls. Where algorithms rank suppliers or carriers, procurement teams should examine explainability, data bias, and the consequences for smaller providers.

Models also have limitations. Historical data may not represent a new trade lane. Average costs can hide tail events. A linear model may ignore congestion thresholds. A forecast may fail after a structural market change. Simulation can create false confidence when probabilities are invented rather than estimated. Professional practice therefore requires validation, periodic recalibration, and explicit communication of uncertainty.

Digital Analytics and Emerging Practice

Transport-management systems, enterprise-resource-planning platforms, customs systems, warehouse systems, and visibility platforms can provide the data foundation for costing and decision models. Application programming interfaces and event standards allow milestone data to flow into control towers and dashboards. Robotic process automation can reconcile invoices, identify duplicate charges, and flag missing documents.

Artificial intelligence can assist with demand forecasting, anomaly detection, estimated-time-of-arrival prediction, and dynamic risk alerts. Its value depends on data quality and human oversight. A machine-learning model that predicts delay from historical port events may perform poorly when a new geopolitical or regulatory event occurs. Managers should retain escalation rules, monitor model drift, and distinguish a prediction from a decision.

Digital twins and scenario engines increasingly allow firms to test network changes before implementation. Sustainability analytics can incorporate emissions into routing and supplier decisions, although boundaries, emission factors, and data quality must be transparent. The emerging professional standard is therefore not simply automation, but auditable, explainable, and decision-relevant analytics.

Best-Practice Decision Sequence

A disciplined logistics-cost decision normally follows this sequence:

1. Define the decision, time horizon, cost boundary, Incoterm, and service requirement.

2. Collect and validate product, trade, transport, inventory, and event data.

3. Build a transparent baseline landed-cost model.

4. Separate fixed, variable, recoverable, avoidable, and risk-related costs.

5. Develop realistic alternatives and constraints.

6. Forecast demand and operational conditions using scenarios.

7. Apply optimization or multicriteria methods where appropriate.

8. Test sensitivity, break-even points, and downside exposure.

9. Review compliance, ethics, data governance, and implementation feasibility.

10. Monitor actual performance against the model and update assumptions.

The sequence prevents a technically sophisticated model from becoming disconnected from commercial reality. It also creates an audit trail that can be explained to finance, operations, senior management, auditors, and regulators.

Chapter Summary

International logistics costing extends beyond freight rates. A meaningful total landed cost model includes product value, origin and destination charges, freight, insurance, duties, relevant taxes, inventory carrying cost, and expected exception costs within a clearly defined boundary. Customs valuation, tariff classification, Incoterms, exchange rates, and tax recoverability must be handled carefully.

Analytics links cost to service, reliability, capacity, inventory, sustainability, and risk. Forecasting supports planning, while optimization allocates resources subject to operational constraints. Sensitivity analysis reveals fragile assumptions; simulation represents uncertain outcomes; and multicriteria methods make trade-offs explicit when cost is not the only objective. These methods are valuable only when data are governed, assumptions are visible, results are validated, and professional judgment remains accountable.

Key Terms

Total landed cost: The complete cost of acquiring and making goods available at a defined destination.

Customs value: The value used by customs authorities to assess import duty.

Dutiable value: The portion of value to which a customs duty rate is applied.

Incoterm: A standard trade term allocating cost, responsibility, and risk between buyer and seller.

Chargeable weight: The weight used for freight pricing, often the greater of actual and volumetric weight.

Accessorial charge: An additional transport or handling charge outside the basic freight rate.

Inventory carrying cost: The cost of holding inventory, including capital, storage, insurance, and obsolescence.

Pipeline inventory: Goods that are in transit between origin and destination.

Expected risk cost: Probability of an event multiplied by its financial impact.

KPI: A key performance indicator used to monitor performance against objectives.

Forecast bias: A systematic tendency for forecasts to exceed or fall below actual outcomes.

Linear programming: An optimization method with linear objectives and constraints.

Mixed-integer optimization: Optimization involving continuous variables and discrete yes-or-no decisions.

Sensitivity analysis: Examination of how results change when assumptions vary.

Break-even analysis: Determination of the point at which competing alternatives have equal economic value.

Monte Carlo simulation: Repeated random sampling used to model uncertain outcomes.

Multicriteria decision-making: Evaluation of alternatives against multiple weighted criteria.

Shadow price: The estimated value of relaxing a binding model constraint.

Model governance: Processes for controlling assumptions, data, validation, access, and use of analytical models.

Review Questions

Short Answer Questions

1. What elements should normally be considered in a total landed cost model?

2. Explain the difference between actual weight and chargeable weight.

3. Why must an Incoterm be identified before comparing supplier prices?

4. What is forecast bias, and why does it matter in logistics planning?

5. Define expected risk cost and state its basic formula.

Long Answer Questions

1. Explain how customs valuation, duty, taxes, and tax recoverability influence landed cost.

2. Discuss the design of a balanced international logistics KPI dashboard.

3. Compare optimization, sensitivity analysis, and simulation as decision-support methods.

4. Explain how inventory carrying cost changes the comparison between ocean, rail, and air freight.

5. Discuss the governance, ethical, and professional risks associated with logistics analytics.

Analytical / Application Questions

1. A shipment has a product value of $30,000, freight of $4,000, insurance of $250, destination charges of $1,400, inland delivery of $1,000, and a duty rate of 6% applied to product value plus freight and insurance. Calculate landed cost before inventory cost and landed cost per unit if the shipment contains 1,500 units.

2. A slow transport option costs $2,500 less than a faster option but adds 20 days of transit. The shipment value is $40,000 and the annual carrying rate is 18%. Calculate the additional pipeline inventory cost and explain what other factors should be evaluated before choosing the slow option.

3. Design a weighted multicriteria model for selecting an international carrier. Select at least five criteria, assign weights, explain how each criterion would be measured, and describe how you would test whether the ranking is sensitive to the weights.

22. Multiple Choice Questions with Answers

Question 1

Which expression best represents total landed cost in the chapter’s framework?

A. Product cost plus headline freight rate only

B. Product cost, logistics, trade, inventory, and relevant exception costs within a defined boundary

C. Customs duty plus warehouse rent only

D. Invoice value excluding transport and insurance

Correct Answer: B. Product cost, logistics, trade, inventory, and relevant exception costs within a defined boundary

Question 2

A shipment has a customs value of $20,000 and a duty rate of 7%. What is the duty?

A. $140

B. $700

C. $1,400

D. $14,000

Correct Answer: C. $1,400

Question 3

Why can a lower freight rate fail to produce the lowest economic outcome?

A. Freight rates never affect landed cost

B. Lower rates always create higher customs duties

C. Inventory, delay, shortage, and exception costs may offset the rate advantage

D. Freight rates are unrelated to shipment size

Correct Answer: C. Inventory, delay, shortage, and exception costs may offset the rate advantage

Question 4

Which measure is most directly concerned with systematic over- or under-forecasting?

A. Forecast bias

B. Container utilisation

C. Customs clearance time

D. Transit-time variability

Correct Answer: A. Forecast bias

Question 5

What is the principal purpose of sensitivity analysis?

A. To eliminate all uncertainty from a decision

B. To examine how results change when assumptions vary

C. To replace customs classification

D. To calculate only historical freight cost

Correct Answer: B. To examine how results change when assumptions vary

Question 6

Which situation most clearly requires mixed-integer optimization?

A. Calculating duty on a shipment

B. Computing average transit time

C. Choosing whether to open a warehouse while allocating shipment quantities

D. Converting currencies using a stated exchange rate

Correct Answer: C. Choosing whether to open a warehouse while allocating shipment quantities

Question 7

A disruption has a 10% probability and a financial impact of $25,000. What is its expected risk cost?

A. $250

B. $2,500

C. $15,000

D. $250,000

Correct Answer: B. $2,500

Question 8

In a weighted-scoring model, why must criteria be normalised carefully?

A. To ensure every alternative receives the same score

B. To make different measurement scales comparable

C. To remove the need for managerial judgment

D. To guarantee that the cheapest option wins

Correct Answer: B. To make different measurement scales comparable

Question 9

What is a major limitation of a simulation model?

A. It cannot represent uncertainty

B. It always produces a single deterministic answer

C. Its results are only as credible as its assumptions and probability distributions

D. It cannot be used for inventory decisions

Correct Answer: C. Its results are only as credible as its assumptions and probability distributions

Question 10

Which practice best supports ethical logistics analytics?

A. Changing KPI definitions after poor performance occurs

B. Using unsupported weights to predetermine a supplier ranking

C. Maintaining transparent assumptions, accurate trade data, and auditable model governance

D. Ignoring recoverable taxes in every situation

Correct Answer: C. Maintaining transparent assumptions, accurate trade data, and auditable model governance