4. Distribution Strategy and Network Design

Introduction and Context

Distribution strategy concerns the deliberate design of the facilities, transportation links, inventory positions, information flows, and managerial policies through which products move from sources to customers. It is not merely a choice of warehouses. It is a strategic decision about how an organisation will balance customer responsiveness, operating cost, asset utilisation, risk, and competitive positioning.

A distribution network may consist of plants, suppliers, ports, regional distribution centres, fulfilment centres, cross-docks, retail outlets, customer locations, and transportation services. The network determines where inventory is held, how far products travel, which orders are consolidated, how quickly customers are served, and how disruption is absorbed. Consequently, network design decisions often have effects lasting many years and involving substantial fixed investment.

The central managerial challenge is that the network cannot simultaneously minimise every relevant cost and maximise every service attribute. A highly centralised system may reduce inventory and facility costs but increase delivery distance. A decentralised system may improve response time but create duplicated inventory and operating capacity. Direct shipping can eliminate intermediate handling, whereas cross-docking can combine consolidation economies with rapid movement. The appropriate design depends on product characteristics, demand geography, service promises, supply uncertainty, regulatory conditions, and the economics of transport.

This chapter develops the concepts and analytical approaches used to design such networks. It links strategic service requirements with facility-location, allocation, gravity, flow, and total-landed-cost models, and then applies these ideas to managerial decisions.

Learning Objectives

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

1. Explain the strategic role and major configurations of distribution networks.

2. Distinguish centralised, decentralised, direct-shipping, cross-docking, and multi-echelon designs.

3. Identify the cost, service, risk, and sustainability trade-offs associated with each configuration.

4. Apply facility-location, allocation, gravity, flow, and total-landed-cost concepts to network decisions.

5. Evaluate network alternatives using quantitative and qualitative criteria.

6. Recommend an appropriate network design for a defined market and service strategy.

7. Recognise implementation risks, governance requirements, and emerging digital capabilities.

Distribution Strategy as a Competitive Choice

Distribution strategy translates the organisation’s value proposition into physical and transactional capabilities. A premium pharmaceutical distributor may require validated temperature control, high traceability, and rapid replenishment. A furniture retailer may accept longer lead times but require economical delivery of bulky items. An online fashion business may prioritise order variety, returns processing, and flexible fulfilment. Thus, network design begins with service segmentation rather than with a search for the cheapest warehouse locations.

Important service dimensions include order-cycle time, delivery reliability, product availability, delivery frequency, order completeness, flexibility, visibility, and returns responsiveness. These dimensions should be specified by customer segment. A uniform service policy can be unnecessarily expensive when customers have different willingness to pay and different operational requirements.

The main network levers are:

• the number and location of facilities;
• the assignment of suppliers and customers to facilities;
• the inventory position at each echelon;
• transportation modes and shipment frequency;
• the use of consolidation, cross-docking, or direct fulfilment;
• ownership and governance of logistics assets; and
• information systems supporting planning and execution.

Network design is strategic, but it should not be isolated from tactical and operational decisions. A network that appears economical under annual average demand may fail during seasonal peaks if labour, docks, fleet capacity, or inventory replenishment are inadequate. Similarly, a location selected without considering transportation schedules and customs processes may produce poor actual service.

Core Network Configurations

Centralised distribution

In a centralised network, inventory is concentrated in one major distribution centre or a small number of facilities. Products may be shipped directly from a plant to the central facility and then to customers or retail locations.

Centralisation can reduce duplicated safety stock because demand is pooled across markets. It may also simplify control, standardise processes, and improve utilisation of material-handling and information systems. The disadvantages are longer outbound distances for remote customers, greater dependence on a limited number of facilities, and possible exposure to regional transport disruption. Centralisation is often suitable for high-value, slow-moving, or highly variable products where inventory pooling is valuable and customers can tolerate moderate lead times.

Decentralised distribution

A decentralised network places inventory in several regional or local facilities. It is designed to bring stock physically closer to demand. This can shorten delivery times, support frequent replenishment, and reduce the final-mile distance.

The principal cost is duplication. Each facility may need safety stock, personnel, equipment, information-system capability, and management overhead. Forecast errors at individual locations can also lead to simultaneous overstocks and stockouts. Decentralisation is attractive when customer density is geographically dispersed, delivery time is strategically important, products are bulky or low-value relative to transport cost, or local regulations make regional presence necessary.

Direct shipping

Under direct shipping, an order moves from the supplier or plant directly to the customer without passing through a stocking distribution centre. It may be used for make-to-order products, large equipment, drop-shipped e-commerce orders, or products with limited demand at each destination.

Direct shipping can reduce handling, shorten the physical route, and avoid facility inventory. However, it may produce small shipments, poor vehicle utilisation, complex order tracking, and inconsistent customer experience when multiple suppliers are involved. The decision should therefore compare the complete door-to-door cost, not merely the absence of warehouse cost.

Cross-docking

Cross-docking transfers incoming goods directly to outbound vehicles or staging lanes with little or no storage. In a retail application, suppliers may deliver labelled cases that are sorted by store and rapidly consolidated for onward delivery. In parcel networks, shipments may be scanned, sorted, and dispatched within a short dwell period.

Cross-docking is effective when demand is relatively predictable, product flow is synchronised, shipment information is accurate, and inbound and outbound schedules are coordinated. It reduces storage and may improve vehicle utilisation. It is not simply a warehouse without racks: it requires disciplined appointment management, barcoding or equivalent identification, rapid exception handling, and reliable transportation capacity. Unplanned arrivals, poor labelling, variable quantities, or demand uncertainty can convert a cross-dock into a congested temporary warehouse.

Multi-echelon distribution

A multi-echelon network contains two or more inventory and distribution levels, such as plant–national distribution centre–regional distribution centre–customer. Echelons can provide economies of scale upstream while preserving local responsiveness downstream.

The design question is not only where facilities should be located but also which inventory and decision responsibilities belong at each level. National facilities may hold slow-moving or high-value products, while regional facilities hold fast-moving stock. This differentiation can reduce total inventory compared with independently stocking every item at every location. Nevertheless, multi-echelon systems add handling, replenishment dependencies, planning complexity, and risks of inventory imbalance.

Terminology for Network Design

A facility is a physical node where products are produced, stored, consolidated, sorted, or transferred. A distribution centre normally holds and processes inventory, whereas a cross-dock primarily transfers goods. An echelon is a level in a distribution hierarchy. A node is any relevant point in a network; an arc is a connection between nodes over which flow occurs.

Demand points are customer or market locations requiring supply. Allocation assigns demand points to facilities or suppliers. Throughput is the quantity processed by a facility in a period. Lead time is the elapsed time between order release and receipt. Safety stock protects against demand and replenishment uncertainty. The service radius is the geographic or time-based area a facility can serve within a defined service standard.

Fixed cost is incurred when a facility is opened or operated, largely independent of short-term volume. Variable cost changes with flow or activity. Capacity is the maximum feasible output or handling volume under stated conditions. Total landed cost is the complete cost of making a product available at the destination, including purchase or production, transport, duties, handling, inventory carrying, order processing, and relevant failure or compliance costs.

Analytical Models and Their Interpretation

Facility-location models

Facility-location models determine which facilities should operate and where they should be positioned. A simplified capacitated model can be expressed as:

Minimise:

`Σj Fj yj + ΣiΣj cij xij`

subject to:

`Σj xij = Di` for each demand point i

`Σi xij ≤ Kj yj` for each candidate facility j

`xij ≥ 0; yj ∈ {0,1}`

Here, `Fj` is the fixed cost of opening facility j, `Di` is demand at customer i, `cij` is the cost of serving i from j, `Kj` is facility capacity, `xij` is the quantity allocated from j to i, and `yj` indicates whether facility j is opened. Real models may include service-time constraints, multiple products, transport modes, inventory costs, taxes, carbon emissions, and facility-opening limits.

The model’s value lies in revealing trade-offs rather than producing an unquestionable answer. Results depend on demand forecasts, cost assumptions, capacity estimates, and the geographic resolution of the analysis. Sensitivity testing is therefore essential.

Centre-of-gravity analysis

The centre-of-gravity method provides an initial approximation of a facility location. If customer coordinates are `(xi, yi)` and demand weights are `wi`, the weighted location is:

`X = Σ(wi xi) / Σwi` and `Y = Σ(wi yi) / Σwi`.

The method is useful for screening locations where transport cost is approximately proportional to distance and routes are not severely constrained. It does not account adequately for highways, terrain, congestion, land cost, labour availability, taxes, service-time requirements, or network economies. The calculated point is therefore a starting point for evaluation, not a final site decision.

Allocation and transportation-flow models

Allocation analysis determines which facility supplies which demand point. A transportation model minimises shipping cost subject to supply and demand constraints. Flow models extend this logic across several stages and can represent plants, ports, distribution centres, and customers simultaneously. They are especially useful when capacity, mode choice, transshipment, or multiple products influence the result.

A flow solution should be reviewed for operational feasibility. A mathematically attractive allocation may require an unrealistic number of small deliveries, cross-border movements with uncertain clearance, or a transport mode unavailable in the required lane.

Total-landed-cost analysis

Total-landed-cost analysis prevents narrow decisions based only on purchase price or line-haul rate. A general expression is:

`TLC = acquisition cost + inbound transport + duties and taxes + receiving and handling + storage + inventory carrying cost + outbound transport + order processing + expected shortage, damage, and return cost`.

For network design, it may also include facility fixed cost, packaging changes, carbon cost, quality assurance, and the financial impact of longer lead times. The relevant unit may be a case, pallet, order, kilogram, cubic metre, or customer delivery. The unit should match the decision.

Service-cost frontiers and scenario analysis

A network should be evaluated on a service-cost frontier. One alternative may minimise annual cost, another may provide faster response, and a third may offer greater resilience. Scenario analysis tests performance under different demand volumes, fuel prices, product mixes, disruptions, and service policies. Useful scenarios include expected demand, high growth, peak season, facility outage, border delay, supplier failure, and major transport-rate changes.

A Practical Network-Design Process

The process begins by defining the strategic question. “How many warehouses should we open?” is less useful than “What network can provide two-day delivery to priority customers at an acceptable total cost and risk?” The organisation should then establish the planning horizon, product scope, demand zones, service segments, and constraints.

The next stage is data preparation. Required data commonly include historical and forecast demand, customer coordinates, order profiles, product dimensions and values, current facility costs and capacities, transport rates, supplier locations, inventory policies, labour conditions, taxes, and lead-time performance. Data quality must be assessed before modelling. Inconsistent demand units, missing returns, and averages that hide peaks can materially distort outcomes.

Candidate network configurations are then developed. These may include one central facility, several regional facilities, direct shipment for selected products, cross-docking for predictable retail flows, or a hybrid multi-echelon design. Quantitative models estimate cost, capacity, distance, inventory, and service performance. Managers should combine model outputs with qualitative assessment of labour markets, community impact, regulatory exposure, supplier relationships, and organisational capability.

The preferred design is tested through sensitivity and risk analysis. A robust solution may not be the absolute lowest-cost solution; it may be the alternative that performs acceptably across a broad range of conditions. Implementation should then proceed through a transition plan covering inventory migration, systems configuration, transport procurement, workforce, customer communication, performance baselines, and contingency arrangements.

Illustrative Applied Case: Meridian Home Appliances

Background

Meridian Home Appliances, an illustrative mid-sized manufacturer, sells refrigerators, washing machines, and small appliances through retailers and an online channel. Its existing network consists of one central distribution centre located near the manufacturing plant. The facility serves customers nationwide.

Situation and problem

The centralised network has low inventory duplication and simple management, but delivery performance is weak in distant regions. Large appliances generate high transport cost because of their volume, while small appliances are ordered frequently and can be shipped economically in parcels. Demand has grown unevenly, and retailers in two regions now require scheduled next-day replenishment. Management is considering a second regional distribution centre, direct shipment for large appliances, and cross-docking for selected retail orders.

Analysis

A demand-zone analysis first separates large and small appliances and identifies service requirements by customer segment. The centre-of-gravity calculation suggests a location near the largest regional demand cluster, but the candidate is screened against motorway access, labour availability, flood exposure, and lease cost. A capacitated facility-location model compares the current centralised network with a two-echelon alternative.

The analysis shows that placing all products in the regional facility would duplicate too much slow-moving inventory. A multi-echelon design is therefore evaluated: the central facility retains slow-moving and high-value models, while the regional facility holds fast-moving small appliances and selected large-appliance stock. Large appliances with irregular demand are shipped directly from the plant or central facility to customers. Retail replenishment for predictable mixed orders is routed through a cross-dock.

The model includes outbound transport, facility cost, inventory carrying cost, expected stockout cost, and delivery failure cost. It also tests peak-season demand and a temporary closure of either facility. The proposed hybrid network has slightly higher fixed cost but improves regional service and reduces the distance of frequent small-appliance deliveries. Direct shipment avoids unnecessary handling of bulky units, while cross-docking supports retailer schedules without requiring extensive storage.

Approach and outcome

Meridian adopts the hybrid design in stages. It first introduces direct shipping for selected large appliances, then pilots cross-docking with two retail customers, and finally opens the regional facility after validating demand and transport capacity. The company uses common inventory visibility across the central and regional nodes and establishes rules for product allocation during shortages.

The case illustrates that network design should be product- and segment-specific. A single configuration for every item would either over-centralise fast-moving products or over-decentralise slow-moving products.

Learning points

The important lessons are the value of segmentation, the need to include inventory and service costs in facility decisions, the distinction between a cross-dock and a storage warehouse, and the importance of phased implementation. Resilience testing also changes the decision: redundancy has economic value when service failure is costly.

Managerial, Ethical, and Governance Considerations

Network design affects employment, land use, traffic, emissions, customer access, and supplier relationships. A location decision should therefore consider more than financial return. Environmental assessment may include transport emissions, building energy, packaging, refrigeration, and reverse logistics. Consolidation can reduce vehicle kilometres, but a decentralised network may shorten final-mile routes; the result must be measured rather than assumed.

Governance is essential when logistics is outsourced. Contracts should define service levels, capacity commitments, data ownership, audit rights, safety responsibilities, subcontracting rules, business continuity, and procedures for disputes. Regulatory requirements may concern customs, dangerous goods, food safety, pharmaceuticals, labour, privacy, and competition. Ethical practice includes safe working conditions, truthful service reporting, fair treatment of carriers and warehouse employees, and avoidance of discriminatory access decisions.

Common analytical mistakes include relying on average demand, ignoring peak capacity, excluding returns and damages, treating distance as a perfect proxy for transport cost, assuming all facilities have equivalent labour productivity, and selecting a location before defining the service promise. Another frequent error is optimising one function—such as warehouse cost—while shifting cost to transport, inventory, customer service, or emissions.

Digital Tools, Analytics, and Emerging Developments

Modern network design uses geographic information systems, transport-management data, warehouse-management data, enterprise resource planning records, and optimisation software. GIS tools display demand density, travel-time zones, infrastructure, hazards, and candidate sites. Digital twins can simulate facilities and flows before physical investment. Machine-learning methods can improve demand forecasts and identify patterns in delivery failure, although forecasts remain uncertain and require managerial review.

Real-time visibility from telematics, shipment tracking, and electronic proof of delivery supports dynamic allocation and exception management. Automated storage, robotics, sortation, and goods-to-person systems can increase throughput, but they also require capital, maintenance, technical skills, and suitable product profiles. Artificial intelligence can assist scenario generation and route or inventory recommendations; governance should require explainability, data quality controls, cybersecurity, and human accountability for significant decisions.

Current network design is also influenced by omnichannel fulfilment, urban micro-fulfilment, parcel locker systems, returns networks, nearshoring, geopolitical uncertainty, and carbon-accounting requirements. These developments favour flexible and modular networks rather than designs optimised only for stable historical patterns.

Best-Practice Principles

Effective network design generally follows these principles:

1. Begin with customer service segments and strategic objectives, not with facility count.

2. Model total landed cost, including inventory, failures, returns, and external impacts where material.

3. Use demand scenarios and sensitivity analysis rather than a single forecast.

4. Match configuration to product characteristics, order profile, and geographic demand.

5. Validate mathematical results against operational, legal, labour, and environmental realities.

6. Design visibility and governance into the network from the beginning.

7. Preserve flexibility through scalable capacity, qualified alternatives, and phased investment.

8. Establish performance measures covering cost, service, resilience, safety, and sustainability.

Chapter Summary

Distribution strategy determines how a firm positions facilities, inventory, transportation, and information to meet customer requirements. Centralised networks pool inventory and simplify control, while decentralised networks improve proximity and responsiveness at the cost of duplication. Direct shipping reduces intermediate handling, cross-docking enables rapid transfer with minimal storage, and multi-echelon networks combine upstream scale with downstream service.

Facility-location, allocation, transportation-flow, gravity, and total-landed-cost models provide structured support for network decisions. Their outputs depend on data and assumptions, so scenario testing and managerial judgement are essential. A sound design evaluates cost, service, capacity, resilience, sustainability, governance, and implementation feasibility together. Digital visibility, optimisation, simulation, automation, and AI increasingly support these decisions, but they do not eliminate uncertainty or accountability.

Key Terms

Facility location: Selection of physical sites from which logistics activities are performed.

Centralised network: A network concentrating inventory in one or a few distribution facilities.

Decentralised network: A network distributing inventory across several regional or local facilities.

Direct shipping: Movement from supplier or plant directly to the customer without an intermediate stocking facility.

Cross-docking: Rapid transfer of inbound goods to outbound transport with little or no storage.

Multi-echelon network: A network containing multiple hierarchical levels of inventory or distribution.

Echelon: A distinct level in a logistics distribution hierarchy.

Demand point: A customer, market, or geographic location requiring supply.

Allocation: Assignment of demand or flow to a supply facility or source.

Throughput: Quantity processed by a facility during a specified period.

Capacity: Maximum feasible processing, storage, or flow volume under stated conditions.

Safety stock: Inventory held to protect against demand or replenishment uncertainty.

Centre of gravity: A demand-weighted geographic approximation for a facility location.

Flow model: A mathematical representation of movement through connected network nodes and arcs.

Fixed cost: Cost associated with establishing or operating capacity, largely independent of short-term volume.

Variable cost: Cost that changes with volume, distance, activity, or quantity handled.

Total landed cost: Complete cost of making a product available at its destination.

Service radius: Geographic or time-based area that a facility can serve to a specified standard.

Network resilience: Ability of a distribution network to absorb disruption and recover acceptable performance.

Digital twin: A digital representation used to simulate and analyse a physical logistics system.

Review Questions

Short Answer Questions

1. Why is distribution network design considered a strategic decision?

2. State two benefits and two limitations of centralised distribution.

3. What operational conditions are necessary for successful cross-docking?

4. Define total landed cost and identify four of its components.

5. What is the purpose of a centre-of-gravity calculation?

Long Answer Questions

1. Compare centralised, decentralised, direct-shipping, cross-docking, and multi-echelon networks.

2. Explain how a capacitated facility-location model supports distribution network decisions.

3. Discuss why service segmentation should precede facility-location analysis.

4. Describe a practical process for designing and implementing a new distribution network.

5. Explain the managerial, ethical, environmental, and governance issues in network design.

Analytical / Application Questions

1. A firm serves dispersed customers with high-value, slow-moving products and highly variable demand. Recommend a network configuration and justify it using inventory, service, and transport considerations.

2. Using the Meridian Home Appliances case, explain why a hybrid network is more appropriate than either complete centralisation or complete decentralisation.

3. A proposed facility has the lowest modelled transport cost but is exposed to flooding and has limited skilled labour. How should the decision team evaluate whether to select the site?

22. Multiple Choice Questions with Answers

Question 1

Which characteristic most strongly supports a centralised distribution network?

A. Highly dispersed demand for bulky, low-value products

B. High demand variability where inventory pooling is valuable

C. A requirement for immediate local delivery to every customer

D. Severe restrictions on long-distance transportation

Correct Answer: B. High demand variability where inventory pooling is valuable

Question 2

What is the defining operational feature of cross-docking?

A. Long-term storage of products near customers

B. Direct delivery from a supplier to an individual customer

C. Rapid transfer from inbound to outbound transport with minimal storage

D. Decentralised safety-stock positioning at every retail outlet

Correct Answer: C. Rapid transfer from inbound to outbound transport with minimal storage

Question 3

In a capacitated facility-location model, the binary variable for a candidate facility generally indicates whether the facility is:

A. Assigned a transport mode

B. Opened or operated

C. Fully stocked with safety inventory

D. Allocated to a single customer

Correct Answer: B. Opened or operated

Question 4

A centre-of-gravity calculation is primarily used to:

A. Estimate a demand-weighted geographic starting point for a facility

B. Determine the exact labour requirement of a warehouse

C. Calculate customs duties on imported goods

D. Measure the reliability of a transport carrier

Correct Answer: A. Estimate a demand-weighted geographic starting point for a facility

Question 5

Which cost is most likely to be omitted in a narrow purchase-price comparison but included in total landed cost?

A. Product colour

B. Inbound transport and duties

C. Employee job title

D. Supplier advertising expenditure

Correct Answer: B. Inbound transport and duties

Question 6

A company holds slow-moving items nationally but fast-moving items regionally. This is an example of:

A. Pure direct shipping

B. Product-differentiated multi-echelon positioning

C. Uncontrolled decentralisation

D. Single-echelon cross-docking

Correct Answer: B. Product-differentiated multi-echelon positioning

Question 7

Why can a mathematically optimal allocation be operationally infeasible?

A. Models never use demand data

B. It may require unavailable transport capacity or unrealistic delivery patterns

C. Optimisation models cannot represent facility capacity

D. Allocation models always minimise inventory instead of cost

Correct Answer: B. It may require unavailable transport capacity or unrealistic delivery patterns

Question 8

Which situation most directly favours direct shipping?

A. Predictable, high-volume flows requiring store consolidation

B. Make-to-order or bulky products with irregular customer demand

C. Products requiring extensive regional safety stock

D. Small, frequent orders requiring route consolidation

Correct Answer: B. Make-to-order or bulky products with irregular customer demand

Question 9

What is a major risk of decentralising inventory without careful analysis?

A. Elimination of all delivery capability

B. Duplication of safety stock and inconsistent inventory balances

C. Removal of all facility fixed costs

D. Inability to serve local customers

Correct Answer: B. Duplication of safety stock and inconsistent inventory balances

Question 10

Which practice best improves the robustness of a network-design decision?

A. Using only the most optimistic demand forecast

B. Selecting the lowest-rent site without testing disruption

C. Testing demand, cost, capacity, and disruption scenarios

D. Optimising warehouse cost while excluding transport cost

Correct Answer: C. Testing demand, cost, capacity, and disruption scenarios