Omnichannel, E-Commerce and Last-Mile Networks

Introduction

Distribution networks were traditionally designed around a relatively stable flow: manufacturers supplied regional distribution centres, which replenished stores, and customers travelled to stores to purchase products. E-commerce has transformed this structure. Customers now expect to search across channels, compare prices, order through a website or marketplace, receive products at home, collect them from a store, return them through another channel, and obtain accurate delivery information in real time.

This transformation is not simply a change in sales technology. It changes the physical configuration of the distribution network, inventory ownership, order-management logic, transportation economics, service promises and organisational responsibilities. A store may function simultaneously as a retail outlet, a fulfilment node, a returns centre and a collection point. A marketplace may provide demand access but impose strict service requirements. A micro-fulfilment centre may improve delivery speed while increasing fixed and technology costs.

The central managerial problem is integration: how can stores, warehouses, suppliers, marketplaces, carriers and customers be coordinated as one network rather than as isolated channels? This chapter examines the design and economics of such networks.

Learning Objectives

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

1. Explain the structure and strategic logic of omnichannel distribution.
2. Distinguish stores, fulfilment centres, micro-fulfilment centres, marketplaces and direct-to-customer nodes.
3. Evaluate inventory visibility and order-allocation alternatives.
4. Design click-and-collect, ship-from-store and return flows.
5. Analyse urban logistics, crowdsourced delivery and last-mile service economics.
6. Apply quantitative measures to capacity, delivery cost, profitability and service performance.
7. Assess technology, governance, sustainability and implementation risks.

1. From Multichannel to Omnichannel Distribution

A multichannel retailer operates more than one channel, such as stores, a website, a mobile application and a marketplace. These channels may have separate inventories, prices, customer records and performance measures. An omnichannel network seeks to provide a coordinated customer experience across channels and to use the total network of inventory and fulfilment resources.

The distinction is important:

Omnichannel does not mean that every order must be fulfilled from every node. It means that the network is designed deliberately so that channel boundaries do not unnecessarily restrict customer service or economic optimisation.

Typical omnichannel services include:

• buy online, deliver to home;
• buy online, collect in store (BOPIS or click-and-collect);
• reserve online, pay or collect in store;
• ship-from-store;
• endless aisle, in which a store sells an item not physically available there;
• marketplace fulfilment;
• return online purchases to stores; and
• same-day or scheduled urban delivery.

The network therefore has several simultaneous flows: products, orders, information, payments, returns and customer contacts. Configuration decisions must consider all of them.

2. Network Nodes and Their Strategic Roles

2.1 Regional and national fulfilment centres

Large fulfilment centres provide scale economies in storage, labour, automation and inbound transportation. They generally offer broad assortments and efficient batch picking. Their disadvantages include longer average distance to customers and possible difficulty meeting narrow delivery windows in dense urban markets.

2.2 Retail stores as logistics nodes

Stores provide geographic proximity and existing customer access. They can support click-and-collect, ship-from-store, local delivery and returns. However, store inventory is often less accurate than warehouse inventory, picking interrupts selling activities, and customer traffic creates operational variability. A store should become a fulfilment node only when its inventory accuracy, labour availability, layout and order volume support reliable execution.

2.3 Micro-fulfilment centres

A micro-fulfilment centre (MFC) is a compact, often automated facility located close to demand, frequently inside, beneath or adjacent to a retail site. It is designed for high-frequency, relatively small online orders. MFCs can reduce delivery distance and improve same-day service, but they may have limited assortment, high capital intensity and difficult replenishment requirements.

The economic choice between a regional centre and an MFC depends on order density. If the number of orders within a local area is low, the MFC's fixed cost is difficult to recover. If demand is dense and delivery speed has significant value, local fulfilment may be justified.

2.4 Dark stores and urban hubs

A dark store is a store-like facility dedicated to online fulfilment rather than walk-in sales. An urban hub may consolidate parcels, support bicycle or electric-vehicle delivery, and enable route dispatch. These facilities are useful where curb access, congestion and building density make direct long-haul delivery inefficient.

2.5 Marketplaces and drop-ship suppliers

Marketplaces expand assortment without requiring the focal retailer to own all inventory. The seller or supplier may fulfil the order directly, a practice known as drop shipping. The benefit is lower inventory investment; the risks include variable service quality, fragmented data, inconsistent packaging and weaker control over returns and customer experience.

2.6 Direct-to-customer networks

Direct-to-customer (DTC) networks remove or reduce intermediary dependence. They can improve customer data and brand control, but they require capabilities in demand planning, fulfilment, parcel contracting, customer service, tax compliance and reverse logistics.

3. Inventory Visibility and the Order Promise

Omnichannel performance depends on a reliable answer to four questions:

1. What inventory exists?
2. Where is it located?
3. Is it available for the promised customer order?
4. When can it be delivered or collected?

A basic inventory record can be represented as:

where ATP_i is available-to-promise inventory for item i, OHI_i is on-hand inventory, SR_i is scheduled and reliable inbound replenishment, SO_i is inventory committed to existing sales orders, and R_i is protected or reserved inventory. All terms must be measured in the same units and time horizon.

This is a planning expression, not a guarantee. It assumes that on-hand inventory is physically usable, inventory records are accurate, inbound supply will arrive as planned, and reservations are enforced. In practice, retailers often distinguish:

• physical inventory: units recorded at a node;
• available inventory: units not damaged, quarantined or otherwise restricted;
• safety stock: inventory protected against uncertainty;
• sellable inventory: inventory meeting quality and regulatory requirements; and
• promiseable inventory: inventory that can satisfy a specific order by a stated deadline.

Inventory visibility requires integration among enterprise resource planning, warehouse-management, order-management, point-of-sale, transport-management and marketplace systems. Real-time visibility is valuable only when the underlying transactions are timely and accurate. A sophisticated dashboard cannot compensate for unrecorded damage, delayed receipts or unprocessed returns.

Inventory accuracy

Inventory accuracy may be measured as:

The numerator and denominator must use the same counting basis. A retailer should also measure accuracy at the stock-keeping-unit and location level, because aggregate accuracy can conceal severe errors in popular products.

4. Order Allocation and Fulfilment Logic

An order-management system (OMS) determines which node should fulfil each order. The decision should balance service, cost, inventory risk and operational feasibility.

A useful conceptual objective is:

subject to node capacity, inventory availability, promised delivery time, labour constraints, carrier availability and regulatory restrictions. This is a conceptual optimisation statement; the exact objective and constraints depend on the organisation's strategy.

4.1 Common allocation rules

Nearest-node allocation reduces transportation distance but can consume scarce local stock or create expensive split shipments.

Lowest-cost allocation selects the node with the lowest estimated fulfilment and transport cost. It may be unsuitable when a premium customer promise requires speed.

Inventory-balancing allocation protects inventory at constrained nodes and directs orders toward locations with surplus stock.

Service-priority allocation reserves fast-moving or strategically important stock for high-value, urgent or contractual orders.

Margin-aware allocation considers contribution margin after fulfilment, shipping, discounts, payment fees and expected returns. Revenue alone is an inadequate allocation criterion.

A general cost comparison for node j can be expressed as:

where P_j is picking and packing cost, L_j is labour or handling cost, T_j is transport cost, H_j is incremental inventory-handling cost, and E_j is an expected penalty or risk cost. Each term must be calculated on the same order basis and currency. The node is feasible only if it has the required stock and can meet the promised service time.

4.2 Split orders

A split order is fulfilled from more than one node. Splitting can increase product availability, but it often increases packaging, handling, transport and customer-contact costs. The OMS should compare the incremental benefit of fulfilling immediately with the cost of splitting. Consolidation may be preferable where the customer accepts a later delivery.

4.3 Substitution and partial fulfilment

For grocery and other replenishment categories, substitution rules may permit an alternative product. Substitution requires customer consent, product compatibility, price treatment and food or safety controls. Partial fulfilment should be communicated clearly because it affects customer satisfaction and subsequent delivery economics.

5. Click-and-Collect and Store-Based Fulfilment

Click-and-collect changes the last-mile problem by transferring part of the journey to the customer. It can reduce home-delivery cost and failed deliveries, but it introduces collection-point capacity, staging, identification and customer-waiting requirements.

A typical process is:

1. The customer places an order and selects a collection location and time window.
2. The OMS reserves inventory.
3. The store or fulfilment centre picks and checks the order.
4. The order is staged in a labelled, secure location.
5. The customer receives a readiness notification.
6. Identity or order verification occurs at collection.
7. Uncollected orders are returned to stock or managed under a defined exception policy.

Important performance measures include readiness accuracy, average collection waiting time, percentage collected within the reserved window, staging-space utilisation and uncollected-order rate.

Ship-from-store can improve delivery speed and use existing inventory, but it may reduce shelf availability, disrupt store labour and produce inconsistent packing. A store fulfilment model should specify cut-off times, picking priorities, exception handling and responsibility for inventory discrepancies.

6. Last-Mile Network Economics

The last mile is often the most expensive and operationally variable part of e-commerce fulfilment. Cost is influenced by stop density, distance, parcel size, delivery windows, failed attempts, parking, congestion, building access, customer availability and reverse logistics.

A simple average cost per successful delivery is:

The denominator must exclude failed attempts if the purpose is to reveal the cost of completed deliveries. If management wants the cost of the entire service process, total delivery attempts and reattempt costs should also be reported.

For a delivery route, stop density can be represented as:

If stops are measured in stops and distance in kilometres, the unit is stops per kilometre. Higher density generally reduces distance per stop, but the relationship is not unlimited: dense areas may generate congestion, parking delays and difficult building access.

Worked example: delivery contribution

An online order has a selling price of £80. Product cost is £48, payment and platform fees are £4, picking and packing cost is £3, outbound delivery is £7, and expected return processing cost is £2. The contribution after these order-related costs is:

If a same-day premium service adds £5 of delivery cost and the customer pays only £2 extra, contribution becomes:

The faster service may still be strategically justified if it increases conversion, retention or basket size, but it is not economically superior on this order alone. The calculation excludes fixed facility costs, taxes and broader customer-lifetime effects.

Failed deliveries and delivery windows

Narrow time windows increase customer value but reduce route flexibility. A failed delivery creates additional cost and may damage customer trust. Options include secure lockers, collection points, delivery instructions, appointment pricing, recipient choice and dynamic re-routing. The network should segment service promises rather than offer the most expensive promise universally.

7. Urban Logistics and Crowdsourced Delivery

Urban logistics involves dense demand, constrained road space, emissions concerns, limited loading areas and complex buildings. Common solutions include consolidation hubs, parcel lockers, cargo bicycles, electric vans, scheduled delivery windows and off-peak replenishment.

Crowdsourced delivery uses independent individuals or flexible fleets to perform deliveries, usually through a digital platform. It can provide capacity during peaks and reduce the need for dedicated vehicles. However, service consistency, worker classification, insurance, safeguarding, data protection and customer complaint handling require careful governance.

A sound decision framework compares crowdsourced capacity with contracted or owned capacity on:

• cost per completed stop;
• delivery reliability;
• available capacity during peaks;
• geographic coverage;
• data and control;
• safety and legal exposure; and
• carbon and social impact.

Crowdsourcing is not automatically cheaper. Payment for the driver, platform fees, incentives, failed attempts, customer support and insurance must be included.

8. Returns and the Reverse Omnichannel Network

Returns are a design feature, not an afterthought. Customers may return an online purchase by parcel carrier, to a store, through a locker or directly to a supplier. The network must determine whether the item is resalable, repairable, recyclable or disposable.

A return decision may follow this sequence:

1. Verify order and item identity.
2. Inspect condition and safety status.
3. Classify as new, open-box, repairable, defective or unsellable.
4. Route to the economically and environmentally appropriate destination.
5. Refund or replace according to policy.
6. Record the disposition and update inventory.

The net recovery value of a returned item can be approximated as:

This expression is useful for comparison, but it omits customer-retention benefits and reputational effects. A return policy should therefore be assessed both transactionally and strategically.

9. Technology, Data and Artificial Intelligence

Core omnichannel technologies include:

• order-management systems;
• warehouse- and transportation-management systems;
• distributed order management;
• application programming interfaces and event streaming;
• barcode, RFID and computer-vision identification;
• geocoding, route optimisation and delivery-exception platforms; and
• customer-facing tracking and notification systems.

Artificial intelligence can support demand forecasting, dynamic slot pricing, fraud detection, product substitution, route prediction and customer-service automation. Its value depends on data quality, explainability and operational integration. A forecasting model that ignores promotions, stockouts or channel substitution may produce systematically biased demand estimates.

Governance should address data minimisation, cybersecurity, access controls, model monitoring, auditability and human escalation. Customer location and delivery information may be personal data and must be processed in accordance with applicable privacy law. Automated allocation should not produce unjustified discrimination in service availability or pricing.

10. Illustrative Case: Meridian Home and Living

Background

Meridian Home and Living is an illustrative national retailer selling furniture accessories and household products through 80 stores, one national fulfilment centre and two online marketplaces. Home delivery is available in two to five days. Stores have frequent inventory discrepancies, and online orders are often split between the national centre and stores.

Problem

The company wants to introduce next-day delivery in three metropolitan areas. Management is considering ship-from-store, a micro-fulfilment centre and a crowdsourced delivery contract. The objective is to improve service without destroying contribution margin.

Analysis

The national centre has low unit handling cost and good inventory accuracy, but the average metropolitan delivery distance is high. Selected stores are close to customers, yet their inventory accuracy is only 91%, compared with 99% at the national centre. Store labour is already constrained during weekends. A local MFC would improve availability and speed but would require substantial fixed investment and would carry only the top 35% of online-selling items.

The company therefore segments products and orders. Fast-moving, compact products in metropolitan areas are eligible for MFC fulfilment. Long-tail products remain at the national centre. Stores fulfil only orders for which a cycle count confirms inventory and for which the store has spare labour. Crowdsourced delivery is used for same-day peak periods, while contracted carriers handle scheduled next-day routes.

The OMS applies the following sequence: first test the promised delivery time; then test inventory and capacity; then estimate total fulfilment cost; finally select the node with the lowest feasible cost. Customers are offered a later consolidated delivery when a split shipment would exceed a defined cost threshold.

Outcome

In the pilot, the retailer reports improved metropolitan delivery speed and fewer long-distance parcel movements. However, the MFC requires a broader assortment than originally planned, and store fulfilment is restricted to selected locations because of labour conflicts. The pilot also reveals that returns to stores create unplanned staging work. Meridian therefore adds return-capacity planning to its OMS and measures contribution after returns rather than measuring delivery speed alone.

Learning points

1. A node should be evaluated as part of the complete network, not by delivery speed alone.
2. Inventory accuracy is a prerequisite for a credible customer promise.
3. Different product segments may require different fulfilment architectures.
4. Flexible delivery capacity is useful during peaks but requires governance.
5. Returns and labour capacity can change the economics of an apparently successful pilot.

11. Performance Measurement and Strategic Trade-offs

A balanced omnichannel dashboard should include service, financial, operational, inventory, customer and sustainability measures.

Metrics must be interpreted together. High vehicle utilisation may increase route duration and reduce punctuality. High inventory availability may result from excessive stock. A low delivery cost may reflect poor service or hidden failed attempts. Managers should use contribution after variable fulfilment costs and assess fixed-cost recovery separately.

A useful strategic principle is service differentiation. Customers and products should not all receive the same promise. Premium same-day service may be offered where demand density, willingness to pay and operational feasibility justify it. Standard delivery, lockers or collection points may be more appropriate for price-sensitive or flexible customers.

12. Challenges, Ethics and Regulation

Omnichannel networks create several risks:

• inaccurate promises caused by weak master data;
• labour stress and unsafe productivity targets;
• privacy breaches through location and purchase data;
• cyberattacks on connected fulfilment systems;
• unfair treatment of gig workers;
• congestion and emissions from excessive delivery attempts;
• packaging waste and difficult-to-recycle materials;
• misleading delivery claims; and
• exclusion of customers without digital access or convenient collection facilities.

Governance should define accountability for customer promises, carrier performance, marketplace sellers, data access, product safety and returns. Contracts should specify service-level measures, evidence requirements, liability, insurance, data ownership and escalation procedures. Environmental assessment should include failed deliveries, returns, packaging and vehicle repositioning, not only the direct outbound journey.

13. Best-Practice Design Principles

1. Design the network around total customer journeys, including returns.
2. Establish a single order and inventory truth while retaining source-system accountability.
3. Promise only what inventory, labour and transport capacity can reliably support.
4. Use stores selectively as fulfilment nodes rather than assuming every store is suitable.
5. Segment products and customers by urgency, value, size, margin and demand density.
6. Compare fulfilment alternatives using total cost and contribution, not transport cost alone.
7. Treat click-and-collect as an operational process requiring staging and capacity design.
8. Build exception management into the OMS from the beginning.
9. Pilot MFCs and crowdsourced delivery in areas with sufficient order density.
10. Measure delivery, returns, customer experience, labour and environmental consequences together.
11. Apply human oversight to AI-supported decisions.
12. Review service promises periodically as demand, regulations and urban conditions change.

Chapter Summary

Omnichannel distribution integrates stores, fulfilment centres, marketplaces, suppliers, carriers and customers into a coordinated network. Its defining challenge is not merely selling through several channels but allocating inventory and orders across nodes while maintaining reliable promises and acceptable economics.

Regional centres provide scale and assortment; stores provide proximity but require strong inventory and labour controls; micro-fulfilment centres support dense urban demand but involve fixed cost and assortment limitations. Inventory visibility depends on accurate transactions and a common order-management architecture. Order allocation must balance service, cost, capacity and inventory risk. Click-and-collect reduces home-delivery requirements but creates staging and collection tasks. Last-mile economics are shaped by stop density, failed attempts, delivery windows, congestion and returns.

Urban hubs, lockers, electric vehicles and crowdsourced delivery can improve flexibility, but they require careful economic, legal and ethical evaluation. AI and digital platforms can enhance forecasting and routing only when governed through high-quality data, privacy protection and human accountability. The appropriate network is therefore a segmented, measurable and adaptable system rather than a universal fastest-delivery model.

Key Terms

Omnichannel distribution: Coordinated management of multiple customer and fulfilment channels as one network.

Multichannel: Operation through several channels that may remain operationally separate.

Order-management system: System that captures, allocates, monitors and coordinates customer orders.

Available-to-promise: Inventory that can credibly be committed to a specified customer requirement.

Ship-from-store: Fulfilment of an online order from a retail store.

Click-and-collect: Online ordering followed by customer collection at a designated location.

Micro-fulfilment centre: Compact fulfilment facility located close to concentrated demand.

Dark store: Store-like facility dedicated primarily to online fulfilment.

Marketplace: Digital platform connecting customers with multiple sellers.

Drop shipping: Supplier or seller fulfilment directly to the customer.

Last mile: Final movement of an order from a fulfilment node to the customer or collection point.

Stop density: Number of delivery stops relative to route distance.

Split shipment: An order dispatched from more than one fulfilment node.

Inventory accuracy: Degree to which recorded inventory corresponds to actual usable inventory.

Reverse logistics: Movement and management of returns, repairs, recovery and disposal.

Crowdsourced delivery: Delivery performed through flexible, platform-coordinated external capacity.

Service promise: Stated commitment concerning availability, delivery or collection timing.

Contribution margin: Revenue remaining after specified variable costs.

Distributed order management: Coordinated allocation and execution of orders across multiple nodes.

Urban logistics: Planning and control of goods movement within densely populated areas.

Review Questions

Short Answer Questions

1. Distinguish multichannel distribution from omnichannel distribution.
2. What operational conditions make a retail store suitable for ship-from-store fulfilment?
3. Define available-to-promise inventory and identify two assumptions behind its calculation.
4. Why can click-and-collect reduce last-mile cost while increasing store complexity?
5. Identify four factors that influence the cost of a last-mile delivery.

Long Answer Questions

1. Explain the strategic roles of regional fulfilment centres, stores, micro-fulfilment centres and marketplaces in an integrated distribution network.
2. Discuss how an order-management system can balance service promises, fulfilment cost, inventory risk and node capacity.
3. Evaluate the advantages and limitations of using stores as omnichannel fulfilment nodes.
4. Explain why returns should be designed as part of the original distribution network rather than managed as an exception.
5. Assess the governance, ethical and regulatory issues associated with crowdsourced delivery and AI-supported allocation.

Analytical / Application Questions

1. A retailer records 1,200 units on hand, 300 units of reliable inbound supply, 450 units committed to existing orders and 100 units reserved as safety stock. Calculate available-to-promise inventory and state the assumptions required.
2. Compare a national fulfilment centre, a micro-fulfilment centre and ship-from-store for a metropolitan next-day service. Develop a decision matrix using cost, speed, assortment, inventory accuracy, labour and scalability.
3. Design an implementation plan for a retailer introducing click-and-collect, including technology integration, store processes, customer communication, performance measures and exception handling.

Multiple Choice Questions with Answers

Question 1

Which characteristic most clearly distinguishes omnichannel distribution from multichannel distribution?

A. Use of more than one sales channel

B. Coordination of inventory, orders and customer experience across channels

C. Exclusive use of physical stores

D. Elimination of marketplace sellers

Correct Answer: B. Coordination of inventory, orders and customer experience across channels

Question 2

A retailer has 1,200 units on hand, 300 units of reliable inbound supply, 450 units committed to orders and 100 units reserved. What is its available-to-promise inventory?

A. 650 units

B. 750 units

C. 950 units

D. 1,050 units

Correct Answer: A. 650 units

Question 3

Which is a principal advantage of a micro-fulfilment centre?

A. Unlimited product assortment

B. Elimination of all fixed costs

C. Proximity to dense customer demand

D. No requirement for replenishment

Correct Answer: C. Proximity to dense customer demand

Question 4

What is a major operational risk of ship-from-store?

A. Stores cannot hold physical inventory

B. Store picking may conflict with selling activities and labour capacity

C. It always requires international transport

D. It prevents online returns

Correct Answer: B. Store picking may conflict with selling activities and labour capacity

Question 5

Which cost is included in the illustrative total fulfilment cost expression Cⱼ = Pⱼ + Lⱼ + Tⱼ + Hⱼ + Eⱼ?

A. Only product selling price

B. Picking, labour, transport, handling and expected risk or penalty cost

C. Only warehouse rent

D. Only customer satisfaction

Correct Answer: B. Picking, labour, transport, handling and expected risk or penalty cost

Question 6

What does stop density measure?

A. Orders per employee

B. Stops per unit of route distance

C. Inventory units per warehouse

D. Returns per customer

Correct Answer: B. Stops per unit of route distance

Question 7

Why can split shipments reduce omnichannel profitability?

A. They always increase product availability to zero

B. They may increase packaging, handling and transport costs

C. They eliminate the need for an order-management system

D. They prevent all customer communication

Correct Answer: B. They may increase packaging, handling and transport costs

Question 8

Which practice best supports reliable inventory visibility?

A. Using delayed manual updates only

B. Integrating point-of-sale, warehouse, order and transport data with accurate transactions

C. Ignoring damaged inventory

D. Allocating every order to the nearest node regardless of stock

Correct Answer: B. Integrating point-of-sale, warehouse, order and transport data with accurate transactions

Question 9

Which issue is particularly relevant to crowdsourced delivery governance?

A. Whether products can be manufactured

B. Worker classification, insurance, safety and service consistency

C. Whether stores can display merchandise

D. Whether regional warehouses need roofs

Correct Answer: B. Worker classification, insurance, safety and service consistency

Question 10

A return has a resale value of £30, processing cost of £8 and additional transport cost of £5. What is its net recovery value?

A. £13

B. £17

C. £22

D. £43

Correct Answer: B. £17