Applied Supply Chain Decision-Making Project

Introduction and Context

Supply chain management is ultimately a discipline of informed choice. Managers decide how much inventory to hold, which supplier to use, how products should move, what technology to adopt, and how service commitments can be met at acceptable cost and risk. These decisions are interdependent. A low-cost supplier may increase lead-time variability; a highly responsive transport option may increase emissions and expenditure; and a warehouse automation project may improve accuracy while creating implementation and cybersecurity risks.

This chapter presents an integrated method for completing an applied supply chain decision-making project. The project requires a learner to diagnose a realistic supply chain problem, map the relevant process, analyse evidence, compare alternatives, evaluate cost and service consequences, consider risk and sustainability, and recommend an implementable course of action. The emphasis is not on finding a theoretically perfect answer, but on producing a transparent, evidence-based and measurable decision.

A sound project separates four questions:

1. What is happening in the current process?
2. Why is it happening?
3. What alternatives are feasible?
4. How will the preferred alternative be implemented and evaluated?

Learning Objectives

By the end of this chapter, the learner should be able to:

• define a supply chain problem in measurable terms;
• map a process and identify delays, waste, controls and hand-offs;
• collect, clean and interpret operational data;
• calculate relevant cost, inventory, service and performance measures;
• compare alternatives using cost–service, risk and sustainability criteria;
• distinguish evidence, assumptions and managerial judgement;
• prepare an implementation plan with responsibilities, milestones and controls; and
• communicate a recommendation professionally to decision-makers.

1. Framing the Decision Problem

A project should begin with a decision statement rather than a general complaint. “Warehouse performance is poor” is too broad. A stronger statement is: “During the next six months, the distribution centre must raise order lines picked accurately from 96.0% to at least 98.5% while keeping fulfilment cost per order below $4.20.” This statement identifies the performance gap, target, time horizon and constraints.

A useful problem definition contains:

• scope: products, facilities, suppliers, customers and activities included;
• stakeholders: those affected by, responsible for or able to influence the decision;
• baseline: current performance, measured over a defined period;
• objective: the desired improvement;
• constraints: budget, capacity, regulation, labour, technology or contractual limits; and
• decision date: when action is required.

The project team should distinguish symptoms from causes. Late deliveries may be a symptom of inaccurate inventory records, poor replenishment rules, congestion, supplier unreliability or unrealistic promised dates. A cause-and-effect analysis, such as a fishbone diagram or the “5 Whys” technique, helps prevent premature investment in a solution.

Stakeholder and governance questions

Before analysing data, ask who owns each decision and who may be affected. Procurement, operations, finance, sales, information technology, customers, suppliers and employees may use different definitions of success. Governance should specify data ownership, approval authority, confidentiality requirements, escalation rules and the person accountable for benefits after implementation.

2. Process Mapping as a Diagnostic Tool

A process map represents the sequence of activities, decisions, information flows and physical movements. It makes invisible work visible and supports discussion across functional boundaries. A basic map may show supplier order entry, receiving, inspection, put-away, picking, packing, dispatch and customer delivery. A more advanced value-stream map also records processing time, waiting time, inventory, information triggers and rework.

Use standard symbols consistently:

• rectangles represent activities;
• diamonds represent decisions;
• arrows represent flow;
• documents or system records represent information inputs and outputs; and
• delays, queues and rework should be explicitly labelled.

The team should map the process as it actually operates, not as a procedure manual claims it operates. Observe several transactions, interview operators and reconcile the map with system timestamps. Record exceptions, because exceptions frequently explain most performance failures.

Two related measures are useful. Throughput time is the elapsed time from process entry to exit. Value-added time is the time spent transforming the product or service in a way the customer values. A simple process-efficiency indicator is:

The measure is dimensionless because both numerator and denominator are time quantities. It is meaningful only when the boundaries of the process and the definition of value-added work are consistent. Waiting is not automatically waste: a controlled maturation or legally required inspection may be necessary. Nevertheless, a low ratio signals an opportunity to investigate queues, transport, batching, duplication and approvals.

3. Data Collection and Analysis

Designing the evidence base

A credible project uses several evidence sources where practical:

• enterprise resource planning or warehouse-management records;
• purchase orders, invoices, transport bills and claims;
• cycle counts and physical observations;
• interviews and structured staff questionnaires;
• customer complaints and service reports; and
• supplier performance and market information.

Create a data dictionary before analysis. It should define each field, unit, time period, source, owner and treatment of missing values. For example, “on-time delivery” must specify whether the comparison is with the requested date, confirmed date or promised date, and whether early deliveries count as on time.

Data should be checked for duplicate transactions, impossible dates, inconsistent product codes, unit conversions and missing observations. Cleaning must not silently remove inconvenient results. Every exclusion or imputation should be documented.

Core operational measures

The appropriate measures depend on the decision. Common indicators include:

Turnover is normally expressed as times per year. Average inventory may be estimated as the mean of beginning and ending inventory, although monthly averages are preferable where demand is seasonal. A high turnover is not automatically good: it may indicate efficient use of working capital or dangerously low stock.

For a period, inventory flow can be reconciled as:

All terms must use the same units and period. Adjustments include damage, theft, write-offs and counting corrections.

For a stable average demand rate, cycle stock associated with a replenishment order quantity can be approximated by:

where Q is the order quantity, assuming replenishment is received in full and demand is reasonably continuous. This is not total safety stock and should not be used when stockouts, variable deliveries or staged receipts dominate the situation.

If annual demand is D, ordering cost per order is S, and annual holding cost per unit is H, the basic economic order quantity is:

The model assumes known and relatively constant demand, constant lead time, a fixed ordering cost, a linear holding cost, instantaneous replenishment, no quantity discounts and no stockouts. It is a benchmark, not an automatic policy. Supplier minimums, capacity, perishability and service requirements may make another quantity preferable.

Service and variability

A project should examine averages and variation. Two suppliers can have the same average lead time but very different reliability. Useful descriptive statistics include the range, standard deviation and coefficient of variation:

The coefficient is unit-free and is useful for comparing variability across items with different means, provided the mean is positive and not close to zero. Small samples and seasonal data require cautious interpretation. Correlation does not prove causation; for example, high overtime and late orders may both be consequences of a third factor such as an unexpected promotion.

4. Cost–Service Trade-offs

Supply chain decisions should use relevant total cost rather than a single visible price. A practical total-cost model may include purchase, ordering, transport, handling, inventory carrying, quality, stockout, administration, disposal and implementation costs. Avoid double counting: for example, a transport surcharge already included in an invoice should not be added again.

For an alternative j, a decision model can be expressed as:

The terms must be measured over the same time horizon and in the same currency. Costs that do not change between alternatives are not relevant to ranking them, although they may remain important for budgeting.

Service can be represented by fill rate, order cycle time, on-time delivery, accuracy, availability or customer retention. A weighted score may support a structured comparison:

where w_i is the weight assigned to criterion i, r_ij is the rating of alternative j on that criterion, and the weights sum to 1. Ratings must be defined consistently, such as 1–5, and sensitivity analysis should test whether a small change in weights changes the recommendation. A score does not replace managerial judgement or financial analysis.

Worked example: comparing replenishment alternatives

A retailer requires 12,000 units annually. Alternative A has a unit purchase cost of $10, transport and handling cost of $1.20 per unit, and expected annual stockout cost of $8,000. Alternative B costs $10.40 per unit, transport and handling cost of $0.90 per unit, and expected stockout cost of $2,000. Annual implementation costs are $1,000 for A and $3,000 for B. Ignoring other costs, the annual totals are:

• A = (12,000 × $10) + (12,000 × $1.20) + $8,000 + $1,000 = $141,400.
• B = (12,000 × $10.40) + (12,000 × $0.90) + $2,000 + $3,000 = $138,800.

B is cheaper by $2,600 annually and has the lower expected stockout cost. However, the result depends on the estimates. If B requires a major system change that causes temporary disruption, or if its supplier has significant political risk, the decision should not be based on this calculation alone.

5. Risk, Resilience and Sustainability

Risk analysis asks what may go wrong, how likely it is, how severe the consequence may be, and how effectively the organisation can detect or respond. A risk register commonly includes risk description, cause, consequence, likelihood, impact, existing controls, owner, treatment and contingency.

A simple prioritisation index is:

This ordinal score is a screening tool, not a probability-based financial forecast. A rating of 4 × 4 does not necessarily mean twice the risk of 2 × 4. For material decisions, estimate financial exposure or scenario outcomes where reliable data exist.

Consider single sourcing, supplier financial health, geopolitical disruption, cyberattack, transport interruption, labour shortages, product safety, regulatory failure and extreme weather. Resilience options include dual sourcing, strategic buffers, alternative routes, supplier development, postponement, contingency contracts and shared visibility. Each option has cost and may create new risks.

Sustainability should be integrated rather than appended to the analysis. Relevant dimensions include:

• environmental: energy, emissions, packaging, waste, water and land use;
• social: worker safety, wages, labour rights, community impact and modern slavery risk; and
• economic: long-term viability, affordability and responsible value creation.

Carbon data must state the boundary and unit, such as kilograms of carbon-dioxide equivalent per shipment or per product. Avoid comparing alternatives using one option’s direct emissions and another’s full life-cycle emissions. Supplier claims should be verified through documentation, audits or credible certification where feasible.

Ethical decision-making requires attention to affected groups, not only the organisation’s financial return. A cheaper supplier may rely on unsafe work, forced labour or unlawful disposal. Procurement teams should apply supplier codes of conduct, contractual rights to audit, grievance channels and escalation procedures. Data protection and competition law may also apply when sharing customer, supplier or pricing data.

6. Technology and AI in Decision-Making

Technology can improve visibility and control through barcode scanning, radio-frequency identification, warehouse-management systems, transport-management systems, electronic data interchange, cloud analytics and supplier portals. The correct question is not “Which technology is newest?” but “Which information or process failure will it correct, at what total cost and with what controls?”

AI may support demand forecasting, anomaly detection, route planning, document classification and predictive maintenance. Its output should be treated as decision support unless governance explicitly permits automated action. Managers must check data quality, model drift, explainability, bias, cybersecurity, privacy and human accountability. A forecast that is statistically accurate on average may still fail during promotions, disruptions or structural market change.

Good practice includes role-based access, audit trails, backups, change control, user training, validation samples and a manual fallback. Never upload confidential customer, employee or supplier information to an unapproved public tool. Technology benefits should be measured against a baseline, including adoption and process compliance rather than only system installation.

7. Integrated Project Method

A diploma-level project can follow the sequence below.

Stage 1: Charter the project

Prepare a one-page charter stating the problem, objective, scope, sponsor, team, deliverables, assumptions, constraints and completion date. Agree the baseline period and success measures.

Stage 2: Map the current state

Observe the process, interview participants, record physical and information flows, quantify queues and identify failure points. Validate the map with process owners.

Stage 3: Establish the baseline

Calculate selected cost, service, time, quality, risk and sustainability measures. Show numerator, denominator, unit, period and data source for every KPI.

Stage 4: Diagnose causes

Use Pareto analysis, stratification by product or supplier, 5 Whys, cause-and-effect analysis and comparison of planned versus actual performance. Do not confuse a frequent cause with the most expensive cause.

Stage 5: Generate alternatives

Develop at least two feasible alternatives, including a “do nothing” or control option where appropriate. Describe required resources, dependencies, benefits, disadvantages and implementation risks.

Stage 6: Evaluate alternatives

Use total relevant cost, service impact, capacity, risk, sustainability, legal requirements and strategic fit. State assumptions and conduct sensitivity analysis on uncertain variables such as demand, fuel price, failure rate or implementation duration.

Stage 7: Recommend and justify

The recommendation should answer: why this option, compared with what, based on which evidence, under what assumptions, and with what safeguards? State conditions that would cause the recommendation to be revisited.

Stage 8: Plan implementation

Translate the recommendation into actions, owners, resources, milestones, dependencies, training, communication, controls and contingency arrangements. Use a pilot where uncertainty or operational risk is high.

Stage 9: Measure benefits

Define baseline, target, data source, reporting frequency, owner and review date. Benefits should be validated after implementation, not merely claimed in the business case.

A concise implementation plan may be presented as follows:

8. Illustrative Case: Northstar Home Appliances

This is a fictional illustrative case.

Background and problem

Northstar distributes small kitchen appliances from one regional warehouse to 180 retailers. Over the previous quarter, on-time delivery fell from 94% to 86%, while order accuracy declined from 98.2% to 95.5%. Expedite freight increased and customer complaints doubled. Management initially proposed purchasing a larger fleet of delivery vehicles.

Analysis

The project team mapped order release, picking, packing, staging and dispatch. The map showed that sales orders were released in large batches twice daily. Fast-moving and slow-moving products were stored together, and inventory adjustments were entered manually at the end of each shift. Orders frequently waited in staging because the dispatch cut-off was not visible to pickers.

Data from 12 weeks showed that 62% of late orders contained at least one inventory discrepancy or missed pick. The average picking time was 18 minutes per order, but average order throughput time was 9.4 hours. A process walk revealed that only about 1.1 hours represented direct picking, packing and checking activity. The remainder was waiting, batching and rework.

The team considered three alternatives:

• A: purchase additional vehicles;
• B: introduce location discipline, barcode scanning, smaller release batches and a visible dispatch schedule;
• C: outsource all regional distribution to a third-party logistics provider.

Alternative A addressed transport capacity but did not address inventory accuracy or internal waiting. Alternative C offered scalability but involved contract, data-integration and service-transition risk. Alternative B required scanner rental, location relabelling, training and process redesign, but could be piloted within one zone.

Evaluation and recommendation

The team recommended B as the first intervention. It had the lowest implementation complexity, targeted the dominant causes and preserved the option of later vehicle or 3PL changes. The recommendation was conditional: if the pilot failed to achieve the target, the steering group would review transport capacity and external fulfilment.

The pilot set targets of 98.0% order accuracy, 92% on-time dispatch and a 20% reduction in average throughput time within six weeks. It included daily cycle counts for high-value items, barcode confirmation at pick and pack, release windows every two hours, and a visual exception board. Customer and employee data were restricted to authorised users, and no productivity target was introduced without a safety review.

Outcome and learning points

After six weeks, the pilot reached 98.4% order accuracy and 94% on-time dispatch. Throughput time fell to 6.8 hours, and expedite freight decreased. The figures were not treated as proof of permanent success: the team continued monitoring after the pilot and compared results with the previous baseline while controlling for seasonal demand.

The case demonstrates five principles. First, a visible symptom may have several causes. Second, process mapping can expose waiting and information failures. Third, the least expensive intervention is not always best, but an intervention should target verified causes. Fourth, a pilot reduces uncertainty. Fifth, a recommendation is incomplete without ownership, controls and measurable outcomes.

9. Common Challenges and Limitations

Applied projects often face incomplete data, inconsistent definitions, confirmation bias, limited time and resistance from affected employees. Teams may overvalue precise-looking numbers whose assumptions are weak. They may also select only measures that support a preferred solution.

To improve rigour, triangulate sources, document assumptions, show ranges where estimates are uncertain, invite challenge from process owners and report adverse findings. Avoid false precision: a forecast of 97.36% service is not credible if the underlying sample is small or the measure is inconsistently recorded. A project may recommend further data collection rather than a high-cost decision when uncertainty is material.

Correlation, averages and model outputs do not eliminate judgement. External conditions can change after implementation, and benefits may be displaced from one department to another. For example, reducing warehouse labour may increase transport waiting or customer-service workload. Evaluate the end-to-end system and revisit the decision as evidence changes.

Best-Practice Checklist

A professional project should:

• define the decision and boundary clearly;
• use a validated current-state process map;
• identify data sources, units, periods and owners;
• separate facts, assumptions and estimates;
• compare at least two feasible alternatives;
• include total relevant cost and service consequences;
• assess operational, ethical, environmental and technology risks;
• conduct sensitivity or scenario analysis;
• assign implementation ownership and controls; and
• specify baseline, target, review date and benefit owner for every major outcome.

Chapter Summary

Applied supply chain decision-making integrates process understanding, quantitative evidence and responsible managerial judgement. A strong project begins with a measurable problem statement and a validated map of the current process. Data must be defined, cleaned and interpreted with attention to units, variation and limitations. Cost analysis should use relevant total cost, while service measures should reflect customer and operational requirements. Alternatives must also be assessed for resilience, sustainability, ethics, technology governance and implementation feasibility. The preferred option should be supported by explicit assumptions, sensitivity analysis and a practical implementation plan. Finally, benefits must be measured against a baseline and assigned to accountable owners.

Key Terms

Applied decision-making: Using evidence and structured judgement to choose among feasible supply chain actions.

Baseline: The measured starting condition against which improvement is assessed.

Cause-and-effect analysis: A structured method for linking observed outcomes to possible contributing causes.

Cycle stock: Inventory held to meet demand between replenishments.

Data dictionary: A record defining data fields, units, sources and treatment rules.

Economic order quantity: A benchmark order size balancing ordering and holding costs under stated assumptions.

Exception: A transaction or condition that does not follow the normal process.

Fill rate: The proportion of demand supplied immediately from available stock.

Implementation risk: The possibility that a planned change will fail, be delayed or create adverse effects.

Key performance indicator: A defined measure used to monitor performance against an objective.

Lead time: Elapsed time between initiating and receiving or completing an activity.

Process map: A visual representation of activities, decisions, flows and delays.

Relevant cost: A cost that changes between decision alternatives over the evaluation horizon.

Risk register: A controlled record of risks, causes, effects, ratings, owners and treatments.

Sensitivity analysis: Testing how a conclusion changes when assumptions or inputs vary.

Service level: A measure of the supply chain’s ability to meet specified customer requirements.

Stakeholder: A person or group affected by, responsible for or able to influence a decision.

Sustainability: Managing environmental, social and economic effects over the long term.

Total throughput time: Elapsed time from entry to completion of a process.

Value-added time: Time spent performing work that transforms an output in a way valued by the customer.

Review Questions

5 Short Answer Questions

1. What elements should be included in a supply chain problem statement?
2. Why should a process map represent the process as actually performed?
3. Distinguish relevant cost from total accounting cost.
4. State two limitations of a risk priority score based on ordinal ratings.
5. Why is a baseline necessary for benefits measurement?

5 Long Answer Questions

1. Explain how process mapping, data analysis and root-cause analysis can be integrated in a supply chain project.
2. Discuss the assumptions and limitations of the economic order quantity model.
3. Explain how cost, service, resilience and sustainability criteria can be balanced when comparing alternatives.
4. Describe a suitable governance framework for using AI in supply chain decision support.
5. Develop an implementation approach for a warehouse accuracy improvement project, including stakeholders, controls and KPIs.

3 Analytical / Application Questions

1. A supplier offers a 6% price reduction but has twice the average lead-time variability and a higher minimum order quantity. Design an analysis to determine whether the offer should be accepted.
2. A distribution centre has an average throughput time of 10 hours, of which 1.5 hours is value-added. Calculate process efficiency and propose three investigations that could improve it.
3. Construct a comparison matrix for two supply chain alternatives using cost, service, risk, sustainability and implementation feasibility. State reasonable weights, assumptions and the conditions under which your recommendation would change.

Multiple Choice Questions with Answers

Question 1

Which statement is the strongest supply chain problem definition?

A. Warehouse performance is poor.

B. Employees need to work faster.

C. Raise order accuracy from 96.0% to 98.5% within six months while keeping fulfilment cost below $4.20 per order.

D. Purchase new warehouse technology.

Correct Answer: C. Raise order accuracy from 96.0% to 98.5% within six months while keeping fulfilment cost below $4.20 per order.

Question 2

What does a process map primarily help a project team identify?

A. Only supplier prices

B. Activities, decisions, information flows, delays and rework

C. Only employee satisfaction

D. The organisation’s annual profit

Correct Answer: B. Activities, decisions, information flows, delays and rework

Question 3

A process has 2 hours of value-added time and 10 hours of total throughput time. What is its process efficiency?

A. 5%

B. 20%

C. 50%

D. 80%

Correct Answer: B. 20%

Question 4

Which is an assumption of the basic EOQ model?

A. Demand is completely unknown

B. Replenishment is received instantaneously and in full

C. Stockouts are required

D. Holding cost is zero

Correct Answer: B. Replenishment is received instantaneously and in full

Question 5

Which formula correctly represents inventory turnover?

A. Average inventory ÷ annual sales revenue

B. Annual cost of goods sold ÷ average inventory at cost

C. Annual demand × ordering cost

D. Ending inventory − beginning inventory

Correct Answer: B. Annual cost of goods sold ÷ average inventory at cost

Question 6

Why should a project use total relevant cost rather than purchase price alone?

A. Purchase price is never recorded.

B. Other costs such as transport, holding, quality and stockouts may differ between alternatives.

C. Total relevant cost ignores implementation.

D. Purchase price is always the smallest cost.

Correct Answer: B. Other costs such as transport, holding, quality and stockouts may differ between alternatives.

Question 7

What is the main limitation of a likelihood-rating multiplied by impact-rating risk score?

A. It cannot list risks.

B. Ordinal scores do not necessarily represent proportional probabilities or financial exposure.

C. It requires no managerial judgement.

D. It measures sustainability automatically.

Correct Answer: B. Ordinal scores do not necessarily represent proportional probabilities or financial exposure.

Question 8

Which practice best supports responsible use of AI in supply chain decisions?

A. Allowing an unverified model to make all decisions automatically

B. Uploading confidential data to any public tool

C. Using validation, audit trails, access controls and human accountability

D. Ignoring model drift after deployment

Correct Answer: C. Using validation, audit trails, access controls and human accountability

Question 9

What is the purpose of sensitivity analysis?

A. To remove all uncertainty from a decision

B. To test how a conclusion changes when assumptions or inputs vary

C. To replace process observation

D. To guarantee the cheapest option wins

Correct Answer: B. To test how a conclusion changes when assumptions or inputs vary

Question 10

Which implementation measure most directly verifies whether an improvement has been achieved?

A. The project team’s opinion that the change looks successful

B. The number of meetings held

C. Comparison of a defined post-implementation KPI with its baseline and target

D. The number of pages in the business case

Correct Answer: C. Comparison of a defined post-implementation KPI with its baseline and target