Logistics Costing, Performance and Analytics

Introduction and context

Logistics decisions determine how materials, products and information move through a supply network. They also determine a substantial share of an organisation’s total cost and customer experience. A low transport rate may be associated with high inventory, stock-outs, damage or poor service. Conversely, a premium delivery service may be economically justified for a high-margin or strategically important customer. Effective logistics management therefore requires more than reducing individual expenses: it requires understanding total cost, service value and operational performance together.

This chapter develops practical methods for costing and analysing logistics activities. It covers total logistics cost, cost-to-serve, landed cost, activity-based costing (ABC), key performance indicators (KPIs), dashboards, benchmarking, root-cause analysis and three levels of analytics: descriptive, predictive and prescriptive. The emphasis is on decisions that a materials or logistics manager can make using reliable operational and financial data.

Learning objectives

After studying this chapter, you should be able to:

1. Explain why logistics costs must be evaluated as a system rather than as isolated expenses.
2. Calculate and interpret total logistics cost, landed cost and cost-to-serve.
3. Apply activity-based costing to logistics processes.
4. Select useful logistics KPIs and design a balanced performance dashboard.
5. Use benchmarking and root-cause analysis to investigate performance gaps.
6. Distinguish descriptive, predictive and prescriptive analytics.
7. Recognise data-quality, ethical, governance and implementation limitations.

1. The logic of logistics costing

Logistics costing is the measurement and assignment of resources consumed in procurement, transportation, receiving, storage, inventory management, order processing, handling, distribution, returns and related administration. Costs may be classified in several ways.

The same cost can have different classifications depending on the decision. A warehouse lease is fixed in the short run but may be avoidable over a longer planning horizon. Cost analysis must therefore state its time horizon, decision scope and assumptions.

Total logistics cost

A practical total-cost model is:

where:

• TLC = total logistics cost for the selected period;
• C_p = procurement and inbound coordination cost;
• C_t = transportation cost;
• C_h = handling and material movement cost;
• C_w = warehousing cost;
• C_i = inventory carrying cost;
• C_o = order-processing and information cost;
• C_r = returns and reverse-logistics cost;
• C_q = quality, damage, shortage and failure-related cost.

The categories must be defined so that they do not double-count one another. The model is applicable when management is comparing alternatives over the same period and service scope. It may be extended to include taxes, duties, packaging, insurance, obsolescence and carbon costs where relevant.

Inventory carrying cost is commonly estimated as:

where Ā is average inventory value in currency and i is the annual carrying-cost rate expressed as a decimal. The rate may include capital cost, insurance, storage-related risk, deterioration and obsolescence. This is an estimate, not a universal accounting rule; components should be defined and not counted elsewhere.

Worked example: comparing transport alternatives

A company is considering two replenishment policies for one year. Policy A has transport cost of $48,000, warehousing cost of $36,000 and average inventory of $180,000. Policy B has transport cost of $62,000, warehousing cost of $30,000 and average inventory of $120,000. The annual carrying-cost rate is 20%; all other costs are equal.

For Policy A:

TLC_A = 48,000 + 36,000 + (180,000 × 0.20) = $120,000

For Policy B:

TLC_B = 62,000 + 30,000 + (120,000 × 0.20) = $116,000

Policy B costs $14,000 more in transport but reduces total logistics cost by $4,000. The example illustrates why a transport-only decision can be misleading.

2. Landed cost and cost-to-serve

Landed cost

Landed cost is the cost of acquiring an item and bringing it to a defined point of availability, such as a plant, distribution centre or customer location. A general expression is:

where LC is landed cost; P purchase price; F freight; I insurance; D duties; T_x applicable taxes; H handling; C_c compliance and customs costs; and O other directly attributable acquisition costs. The formula applies only when each component is measured on a consistent quantity basis and taxes are treated consistently with the organisation’s accounting policy.

For example, 1,000 units are purchased at $12 each. Freight is $900, insurance $100, duty is 5% of purchase value, and handling is $300. Purchase value is $12,000 and duty is $600. Total landed cost is $13,900, or $13.90 per unit. If a recoverable tax were included in the invoice, it should not necessarily be included in economic landed cost.

Landed cost supports supplier comparison, sourcing decisions, quotation analysis and inventory valuation. Managers should identify the point at which the cost is calculated and distinguish estimated landed cost from final invoiced cost.

Cost-to-serve

Cost-to-serve (CTS) measures the resources consumed in serving a particular customer, channel, product family or order segment. It is broader than product margin because it includes service complexity. A useful contribution view is:

CTS may include order lines, order frequency, special packaging, delivery stops, expedited transport, returns, customer support, invoicing complexity, minimum-order failures and promotional handling. It should not be used to penalise customers mechanically; it helps identify service designs and pricing arrangements that are economically sustainable.

A customer may generate high revenue but low profit if it places many small orders, requires urgent delivery and returns a large proportion of goods. Another customer with lower revenue may be more profitable because it orders full pallets on a predictable schedule.

3. Activity-based costing in logistics

Traditional overhead allocation often assigns warehouse or administration overhead using one broad base, such as units sold or sales value. This can distort costs when customers and products consume activities differently. ABC assigns resource costs first to activities and then to cost objects using cost drivers.

The ABC procedure is:

1. Identify activities, such as receiving, put-away, picking, packing, dispatch and returns.
2. Form cost pools for resources consumed by each activity.
3. Select a causal cost driver for each pool.
4. Calculate an activity rate.
5. Assign cost to products, orders or customers according to driver usage.
6. Compare the resulting cost with revenue and service outcomes.

The activity rate is:

Assigned activity cost is:

The driver should have a defensible causal relationship with resource consumption. Receiving cost may use pallets received or receiving lines; picking cost may use order lines; delivery cost may use stops, kilometres or shipment weight. Practical capacity is generally preferable to theoretical capacity because unavoidable unused capacity should be visible rather than hidden in inflated rates.

Worked ABC example

A distribution centre has the following annual cost pools:

Rates are therefore $5 per pallet, $2 per order line and $5 per shipment. Customer X uses 20 pallets, 140 order lines and 30 shipments. Its assigned cost is:

(20 × 5) + (140 × 2) + (30 × 5) = $530

If Customer X’s product margin before logistics is $1,200, its contribution after these activities is $670, before any additional cost pools. The result is more informative than allocating warehouse cost only by sales value.

ABC has limitations. Driver measurement can be expensive, rates may become outdated, and not every cost has a simple causal driver. A hybrid system may use direct tracing for major costs, ABC for high-variation activities and broad allocation for immaterial residual costs.

4. Performance measurement and KPIs

A KPI is a defined measure linked to a decision, objective or control. A good KPI has a clear name, formula, unit, owner, data source, frequency, target and escalation rule. It should also indicate whether higher or lower values are desirable.

OTIF is commonly calculated as:

An order must satisfy both conditions. Counting on-time and in-full separately and averaging the two percentages does not produce OTIF.

Inventory turnover is:

If opening and closing inventory are the only reliable observations, average inventory is often approximated as (Opening + Closing) ÷ 2. The approximation may be poor where seasonality is strong; monthly or daily averages are better.

A dashboard should balance outcome measures and diagnostic measures. OTIF is an outcome KPI, whereas picking accuracy, backlog, labour availability and carrier acceptance may help explain it. Financial, service, process, people, risk and sustainability measures should not be presented as unrelated targets. Improving one measure at the expense of another can create sub-optimisation.

5. Benchmarking and root-cause analysis

Benchmarking compares performance with a reference. Internal benchmarking compares sites or periods; competitive benchmarking compares suitable external organisations; functional benchmarking compares a process with another industry. Comparisons are valid only when definitions, product mix, geography, service promise, accounting treatment and measurement periods are sufficiently comparable.

A useful improvement sequence is:

1. Define the performance gap precisely.
2. Confirm the data and measurement definition.
3. Segment the result by site, product, customer, shift, carrier or lane.
4. Identify where the gap is concentrated.
5. Test likely causes using process evidence.
6. Implement a controlled corrective action.
7. Re-measure and confirm whether the change caused improvement.

Root-cause tools include the Pareto chart, process mapping, five-whys analysis, fishbone diagram and failure mode and effects analysis (FMEA). Five-whys is not a command to ask exactly five questions; it is a disciplined way to move from a symptom to controllable causes. A fishbone structure may examine people, methods, machines, materials, measurement and environment.

For example, late deliveries may be caused by poor picking sequence rather than carrier performance. Segmenting by dispatch time and route may reveal that late orders are concentrated in a particular shift. Corrective action could then involve wave-release rules, slotting or staffing instead of changing carriers.

6. Analytics for logistics decisions

Descriptive analytics

Descriptive analytics answers “What happened?” It uses reports, scorecards, trend lines, averages, medians, frequency distributions and Pareto analysis. A mean is:

where xᵢ is an observation and n is the number of observations. The median may better represent delivery time when a few extreme delays exist. Descriptive analysis should show variation, not just a single average.

Predictive analytics

Predictive analytics answers “What is likely to happen?” Examples include forecasting demand, estimating late-delivery risk, predicting equipment failure and identifying likely stock-outs. A simple moving-average forecast is:

where F_t+1 is the next-period forecast, D_t is observed demand in period t, and k is the number of periods. The method assumes recent observations are informative and gives equal weight to the selected periods. It is unsuitable when strong trend or seasonality is ignored.

Forecast error can be measured by mean absolute error:

MAE is expressed in the same units as demand and treats errors proportionally to their absolute size. Forecasts support decisions; they do not eliminate uncertainty.

Prescriptive analytics

Prescriptive analytics answers “What should we do?” It evaluates alternatives subject to constraints. Examples include selecting carriers, assigning orders to warehouses, setting delivery routes and choosing inventory policies. A simple decision model can minimise total cost:

subject to demand, capacity and service constraints, where xᵢⱼ is the quantity shipped from source i to destination j and cᵢⱼ is the unit cost. Real models may add fixed costs, minimum quantities, lead-time restrictions, emissions and risk. An optimisation result is useful only when costs, constraints and data reflect operational reality.

Data governance and technology

A logistics analytics system commonly combines an enterprise resource planning system, warehouse-management system, transport-management system, barcode or RFID data, telematics and business-intelligence dashboards. Artificial intelligence can support anomaly detection, ETA prediction, document classification and demand forecasting. However, automation should not replace managerial review where data are incomplete or decisions affect employment, customer access or safety.

Data governance requires common definitions, ownership, access controls, audit trails, retention rules and protection of personal or commercially sensitive information. A dashboard with precise-looking numbers can still be misleading if timestamps, master data or event definitions are inconsistent.

7. Illustrative case: improving cost-to-serve at Meridian Components

The following case is fictional and illustrative.

Background

Meridian Components supplies industrial parts to three customer segments: contract manufacturers, regional distributors and emergency-repair customers. Revenue was growing, but logistics cost per order and delivery complaints were increasing. Management initially proposed changing to a lower-rate carrier.

Problem and analysis

The logistics team built an ABC model with receiving, picking, packing, dispatch and returns pools. It also calculated cost-to-serve by segment. Emergency-repair customers generated 18% of revenue but 41% of logistics activity because they placed small orders, required same-day dispatch and frequently used premium transport. Regional distributors had moderate revenue but high returns caused by mixed-pallet labelling errors.

A dashboard showed OTIF of 91%, but the result concealed variation: contract manufacturers achieved 97%, distributors 93% and emergency-repair customers 76%. A Pareto analysis showed that most late orders arose from a small number of late order releases and stock-location errors. Carrier performance was responsible for fewer failures than expected.

Approach and solution

Meridian introduced three service policies. Contract manufacturers received scheduled consolidated deliveries. Distributors received standard delivery windows and improved pallet labelling. Emergency customers were offered a premium service with transparent surcharge rules and a cut-off time. Warehouse slotting was changed for fast-moving repair items, and order-release controls were added.

The company continued to monitor total cost, not only freight rate. It used separate dashboards for customer service, warehouse process, cost-to-serve and exceptions. A monthly review checked whether the new policies shifted costs or damaged service elsewhere.

Outcome

After three months, the company reported lower cost per order, improved emergency-service reliability and fewer distributor returns. The largest improvement came from reducing order complexity and picking errors, not from selecting a cheaper carrier. The case demonstrates that segmentation, ABC, KPI analysis and root-cause investigation should precede a major logistics change.

Learning points

• High revenue does not guarantee high contribution.
• Service promises should reflect customer value and operational cost.
• KPI averages can hide segment-level problems.
• Cost-to-serve analysis is most useful when connected to redesigned service choices.
• Corrective action should target verified causes rather than visible symptoms.

8. Challenges, limitations and professional implications

Logistics costing involves managerial judgement. Allocation choices can influence supplier selection, customer profitability and performance evaluation. Managers should disclose assumptions, avoid arbitrary allocations and test sensitivity to uncertain inputs. A reported “lowest cost” may depend on excluding risk, working capital, failure cost or unused capacity.

Performance targets can create dysfunctional behaviour. A warehouse may maximise pick rate by sacrificing accuracy; a transport team may reduce cost by consolidating shipments and missing delivery promises. Balanced measures, exception review and cross-functional ownership reduce these risks.

Ethically, organisations should protect worker privacy when using location or productivity data, avoid discriminatory algorithmic decisions, and explain automated decisions that materially affect employees or customers. Sustainability metrics should define system boundaries and avoid claiming reductions that result only from shifting emissions to another stage of the supply chain. Regulatory requirements may apply to customs records, product traceability, dangerous goods, financial reporting, data protection and environmental disclosures.

Best practice is to begin with a small decision and a reliable dataset. Define the decision, map the process, establish a baseline, calculate relevant total cost, test service impacts, pilot the change and review results. Sophisticated analytics cannot compensate for poor master data or unclear process ownership.

Chapter Summary

Logistics costing evaluates the resources used to move, store and manage materials and products. Total logistics cost prevents narrow decisions based only on freight or warehouse expense. Landed cost supports sourcing and acquisition decisions, while cost-to-serve reveals the economic effect of customer and order complexity. Activity-based costing improves visibility by assigning activity costs through causal drivers.

Effective performance management combines service, cost, asset, process, quality, risk and sustainability KPIs. Dashboards should use clear definitions and show both outcomes and diagnostic drivers. Benchmarking is meaningful only when comparisons are comparable. Root-cause analysis turns performance gaps into targeted corrective action.

Descriptive analytics explains past performance, predictive analytics estimates likely future outcomes, and prescriptive analytics evaluates actions under constraints. Technology and AI can improve visibility and forecasting, but governance, data quality, human oversight and ethical safeguards remain essential. The central professional principle is to optimise the logistics system and customer value, not an isolated metric.

Key Terms

Total logistics cost: The combined cost of logistics activities within a defined scope and period.

Landed cost: Acquisition cost plus directly attributable costs of bringing an item to a defined destination.

Cost-to-serve: Cost of activities required to serve a customer, channel, product or order segment.

Activity-based costing: A costing method that assigns resource costs through activities and causal cost drivers.

Cost pool: A group of costs associated with a defined activity.

Cost driver: A measurable factor that causes or explains activity-resource consumption.

Carrying cost: Cost of holding inventory, including capital, insurance, storage risk and obsolescence components.

OTIF: Percentage of eligible orders delivered on time and in full.

KPI: A defined performance measure linked to an objective or decision.

Dashboard: A visual display of selected performance measures and exceptions.

Benchmarking: Comparison with an internal, external or functional reference.

Root cause: An underlying controllable cause of a problem.

Pareto analysis: Prioritisation method that identifies the relatively few causes producing much of an effect.

Descriptive analytics: Analysis of historical or current performance.

Predictive analytics: Analysis estimating future outcomes or probabilities.

Prescriptive analytics: Analysis recommending actions subject to objectives and constraints.

MAE: Mean absolute error, the average absolute difference between actual and forecast values.

Data governance: Policies and controls for data quality, ownership, access, use and protection.

Review Questions

Short Answer Questions

1. Why can reducing transport cost increase total logistics cost?
2. Define landed cost and identify four possible components.
3. What is a cost driver in activity-based costing?
4. Why should a logistics dashboard include diagnostic KPIs?
5. Distinguish descriptive analytics from predictive analytics.

Long Answer Questions

1. Explain the structure and managerial value of a total logistics cost model.
2. Describe the steps for applying activity-based costing in a distribution centre.
3. Discuss how cost-to-serve can influence customer segmentation and service design.
4. Explain the conditions required for meaningful logistics benchmarking.
5. Evaluate the benefits and limitations of using AI in logistics analytics.

Analytical / Application Questions

1. A firm can choose a policy with annual freight of $70,000, warehousing of $25,000 and average inventory of $100,000, or a policy with freight of $50,000, warehousing of $34,000 and average inventory of $160,000. Using a 15% carrying-cost rate, calculate total logistics cost for each and recommend a policy.
2. A warehouse has picking cost of $96,000 and 48,000 practical order lines. Customer A uses 180 lines and Customer B uses 60 lines. Calculate the picking activity rate and each customer’s assigned picking cost. Explain why this allocation may be more useful than allocating by sales value.
3. OTIF is falling although average transport cost is stable. Design a root-cause investigation using segmentation, at least three diagnostic KPIs and one corrective-action test.

Multiple Choice Questions with Answers

Question 1

Which statement best explains the purpose of total logistics cost analysis?

A. It considers only the lowest freight quotation.

B. It evaluates relevant logistics costs together, including possible trade-offs.

C. It replaces all financial accounting reports.

D. It measures warehouse labour productivity only.

Correct Answer: B. It evaluates relevant logistics costs together, including possible trade-offs.

Question 2

A shipment has a purchase value of $20,000, freight of $1,200, duty of $800 and handling of $300. What is its landed cost before any excluded taxes?

A. $20,800

B. $21,200

C. $22,000

D. $22,300

Correct Answer: D. $22,300

Question 3

Which is the most appropriate cost driver for a picking activity cost pool?

A. Number of customer complaints only

B. Number of order lines picked

C. Annual sales revenue of the company

D. Supplier payment terms

Correct Answer: B. Number of order lines picked

Question 4

An order is counted as OTIF when it is:

A. Delivered early, regardless of quantity.

B. Delivered on time and in full.

C. Delivered in full, regardless of the promised date.

D. Dispatched on the promised date.

Correct Answer: B. Delivered on time and in full.

Question 5

A warehouse has a picking cost pool of $120,000 and practical capacity of 60,000 order lines. What is the activity rate?

A. $0.50 per line

B. $2.00 per line

C. $5.00 per line

D. $72.00 per line

Correct Answer: B. $2.00 per line

Question 6

Which analytical approach primarily answers the question “What is likely to happen?”

A. Descriptive analytics

B. Predictive analytics

C. Prescriptive analytics

D. Historical accounting

Correct Answer: B. Predictive analytics

Question 7

If actual demand is 110 units and the forecast is 100 units, what is the absolute forecast error?

A. 10 units

B. −10 units

C. 100 units

D. 210 units

Correct Answer: A. 10 units

Question 8

What is the main purpose of root-cause analysis?

A. To assign blame immediately.

B. To identify underlying causes that can be addressed.

C. To remove all performance targets.

D. To increase the number of dashboard measures.

Correct Answer: B. To identify underlying causes that can be addressed.

Question 9

Which situation most clearly indicates a cost-to-serve problem?

A. A customer places predictable full-pallet orders.

B. A customer has low revenue and no logistics activity.

C. A high-revenue customer requires many urgent small orders and frequent returns.

D. A supplier provides standard electronic documents.

Correct Answer: C. A high-revenue customer requires many urgent small orders and frequent returns.

Question 10

Which governance practice is most important for trustworthy logistics dashboards?

A. Adding as many metrics as possible.

B. Using common definitions, data ownership and access controls.

C. Removing all human review.

D. Reporting only favourable results.

Correct Answer: B. Using common definitions, data ownership and access controls.