Order Fulfilment and Customer Service

Introduction and context

Order fulfilment is the coordinated process through which a customer request is converted into a delivered product or service. It begins with order capture and ends with delivery, invoicing, customer acceptance, and, where necessary, returns or after-sales support. Customer service is the policy and operating system that determines how reliably, quickly, completely, and transparently this process is performed.

In contemporary supply chains, fulfilment is no longer merely a warehouse activity. It integrates sales, inventory planning, procurement, transport, information systems, finance, customer support, and reverse logistics. Customers increasingly expect accurate availability information, narrow delivery windows, real-time visibility, simple returns, and consistent experience across stores, websites, marketplaces, and mobile applications. These expectations create a managerial challenge: superior service increases revenue and loyalty, but excessive inventory, expedited transport, and highly fragmented fulfilment can destroy margins.

The central managerial task is therefore not to maximise service at any cost. It is to design a differentiated, economically viable service system aligned with customer value and strategic priorities.

Learning objectives

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

• explain the order-to-delivery cycle and the role of customer service;
• distinguish order fulfilment channels and select appropriate channel designs;
• calculate and interpret important service-level measures;
• apply available-to-promise and capable-to-promise logic;
• design policies for allocation, prioritisation, delivery, and returns;
• evaluate omnichannel fulfilment alternatives;
• identify the role and risks of technology, analytics, and artificial intelligence;
• diagnose fulfilment problems and recommend improvements using a systems perspective.

1. The order-to-delivery system

A typical order cycle contains the following stages:

1. Demand generation and order capture: The customer places an order through a salesperson, electronic data interchange, website, marketplace, store, call centre, or mobile application.
2. Order validation: The organisation verifies price, customer identity, credit status, tax, address, product configuration, and requested delivery date.
3. Availability and promise: The system checks inventory, planned receipts, production capacity, and transportation constraints before confirming what can be supplied and when.
4. Allocation and release: Available units are reserved against the order. The order is released to a warehouse, store, supplier, or production facility.
5. Picking, packing, and dispatch: Items are selected, checked, consolidated, labelled, and handed to a carrier.
6. Transportation and delivery: The shipment moves through the transport network and is delivered within the promised conditions.
7. Confirmation and settlement: Proof of delivery, invoicing, payment, and customer acceptance are completed.
8. After-sales service and returns: Complaints, exchanges, repairs, refunds, and product recovery are managed.

The process is cross-functional. A warehouse may achieve high picking productivity while the overall customer experience remains poor because orders were promised inaccurately or delivered late. Consequently, fulfilment should be governed through end-to-end performance rather than isolated departmental metrics.

Order-cycle time and variability

Order-cycle time is the elapsed time between order receipt and customer delivery. It includes processing, waiting, picking, packing, transport, and delivery. A useful decomposition is:

Here, T_order is total order-cycle time and each component is measured in a consistent time unit. The equation is an additive identity, not a statistical law. It applies when the stages are sequential; if activities occur in parallel, elapsed time is determined by the critical path rather than by simply adding all activity times.

Managers should examine both the mean and variation of cycle time. A mean delivery time of two days is not equivalent to a process that delivers every order in two days and one that delivers some in one day and others in five. Variability increases safety-stock requirements, customer uncertainty, exception handling, and the probability of missing a promise.

2. Customer service as a strategic proposition

Customer service has three dimensions:

• Pre-transaction service: product information, ordering arrangements, credit terms, channel access, and stated delivery policies.
• Transaction service: availability, order accuracy, delivery reliability, completeness, condition, and communication.
• Post-transaction service: installation, technical support, warranty, complaint resolution, returns, repairs, and recycling.

A service policy should define the service offering explicitly. Important policy questions include:

• Which customers and products receive priority?
• What delivery lead time and delivery window will be promised?
• Is the promise based on an order line, order, shipment, or monetary value?
• How are shortages allocated?
• What compensation applies when a commitment is missed?
• Which return reasons are accepted, and who bears the cost?

Service should be segmented rather than standardised blindly. An industrial customer whose production line may stop because of a missing component may justify a higher service target than a low-margin, non-urgent consumer order. Segmentation can use customer profitability, criticality, order frequency, product substitutability, geography, and willingness to pay. It must, however, be transparent and consistent with contractual, competition, consumer-protection, and non-discrimination requirements.

3. Available-to-promise and capable-to-promise

Available-to-promise

Available-to-promise (ATP) is the quantity of uncommitted supply that can be promised to customers at a particular time. For a replenishment period, a basic ATP relationship is:

where ATP_t is available-to-promise quantity in period t, S_t is usable on-hand inventory at the start of the period, R_t is confirmed replenishment arriving in the period, and C_t is existing committed demand due in the period. All quantities must use the same unit of measure and time bucket.

This simple expression assumes that inventory is usable, receipts arrive as planned, demand commitments are correctly recorded, and no safety-stock or allocation restriction must be protected. In practice, a firm may reserve safety stock, exclude quality-hold inventory, or apply customer-specific allocation rules. A more operational expression is:

where B_t represents inventory deliberately blocked or protected for safety, strategic customers, regulatory reasons, or other policy purposes.

ATP should be calculated across time buckets, because promising a quantity today may consume supply needed for a later order. A reservation should be made immediately after a promise is accepted; otherwise two sales channels may promise the same unit.

Capable-to-promise

Capable-to-promise (CTP) extends ATP by considering production or service capacity. It is appropriate when the requested product is not in stock but can be manufactured, configured, assembled, or delivered through available capacity. CTP requires information about material availability, labour, machine capacity, tooling, production sequencing, and transport lead time. ATP asks, “Is the item available?” CTP asks, “Can the item be made and delivered by the requested date?”

CTP is more computationally demanding because a promise may alter the feasible schedule for other orders. It should be connected to finite-capacity planning rather than to an unconstrained master schedule.

Worked example: ATP

A distribution centre has 420 units of usable stock at the beginning of week 3. A confirmed receipt of 180 units is due during the week. Existing commitments due in week 3 total 470 units, and management protects 50 units as safety stock. Therefore:

ATP₃ = 420 + 180 − 470 − 50 = 80 units.

The system can promise 80 units in week 3, subject to order allocation rules and confirmation that the receipt arrives before the promised deliveries. Promising 100 units would either consume protected stock or create a shortage.

4. Fulfilment channels and network choices

Major fulfilment channels

The appropriate channel depends on demand density, product characteristics, order size, margin, delivery expectations, inventory ownership, and information quality. Centralisation can reduce inventory pooling and operating cost, while decentralisation can reduce transport time and improve responsiveness. The correct design is a network trade-off, not a universal rule.

Allocation and prioritisation

When supply is scarce, allocation rules determine which customers receive product. Common rules include first-come-first-served, proportional allocation, contractual priority, revenue or margin priority, criticality, and strategic-account allocation. Each rule has consequences. First-come-first-served appears neutral but may disadvantage customers in slower channels. Margin-based allocation may improve economic performance but damage strategic relationships or raise fairness concerns.

A robust policy states the trigger for allocation, the decision authority, the information required, and the communication procedure. Substitution rules should also be defined: whether an equivalent product may be shipped, whether customer approval is required, and how price differences are handled.

5. Measuring service performance

A measurement system should combine reliability, speed, completeness, flexibility, experience, and cost. No single metric captures fulfilment quality.

Fill rate

The order-line fill rate measures the proportion of ordered lines supplied completely from available stock:

It is appropriate when each line is treated as one service event. A unit fill rate instead weights lines by quantity:

These measures can produce different results. Supplying 98 of 100 units may look strong by unit fill rate, while failing two critical order lines may be unacceptable to the customer.

On-time delivery and OTIF

On-time delivery is commonly calculated as:

The exact definition must specify the timestamp, permitted tolerance, treatment of customer-requested delays, and whether the original or revised promise is used. On-time in-full (OTIF) requires both conditions to be satisfied for the same order or delivery:

OTIF is generally stricter than reporting on-time and in-full separately. A delivery that is punctual but short is not OTIF.

Perfect order and customer experience

A perfect order may require correct product, correct quantity, no damage, accurate documentation, and on-time delivery. Organisations must define the denominator and event logic carefully. Multiplying independent component percentages can be misleading unless independence and consistent populations are justified.

Other useful indicators include order accuracy, promise accuracy, complaint rate, first-contact resolution, average response time, return cycle time, cost per order, and customer effort. Financial measures such as contribution margin after fulfilment cost help prevent service improvement from becoming economically destructive.

A balanced dashboard should show:

Metrics must be disaggregated by customer, product, channel, region, carrier, and fulfilment node. An aggregate 95% OTIF may conceal severe failure for a strategically important segment.

6. Returns and reverse logistics

Returns arise from defects, damage, wrong item, over-ordering, buyer remorse, warranty claims, rental termination, product upgrades, or regulatory recall. Reverse logistics includes authorisation, collection, transportation, inspection, disposition, refurbishment, resale, recycling, and disposal.

A returns policy balances customer convenience against fraud, processing cost, recovery value, and environmental impact. Important design choices include return period, proof of purchase, free versus paid return, collection versus drop-off, condition requirements, refund timing, and disposition authority.

A return should be classified at receipt. Possible dispositions are:

• return to stock as new;
• repackage and sell as open-box;
• refurbish or repair;
• harvest components;
• return to supplier under agreement;
• recycle;
• dispose safely.

The economically appropriate disposition depends on recovery value, processing cost, remaining product life, legal requirements, and customer disclosure obligations. Data from returns should be fed back to design, packaging, forecasting, supplier quality, and product information. A high return rate is not only a logistics problem; it may signal inaccurate descriptions, poor fit, weak quality, or inappropriate commercial incentives.

7. Omnichannel logistics

Multichannel operations provide several channels; omnichannel operations coordinate them so that the customer experiences one integrated proposition. Examples include ordering online and collecting in store, returning a marketplace purchase through a store, viewing store inventory online, or receiving a split shipment from several nodes.

Omnichannel design requires a common product catalogue, customer and order identity, inventory visibility, pricing and promotion governance, payment integration, and consistent returns rules. Inventory visibility must include reservations, safety stock, goods in transit, damaged stock, and the time required to pick the item. “Available” should mean available for the selected fulfilment option, not merely physically present somewhere in the network.

Common omnichannel models include ship-from-store, buy-online-pick-up-in-store, curbside collection, endless aisle, and dark stores. Their benefits include shorter delivery distance and better use of distributed stock. Their risks include inaccurate store inventory, labour conflicts between shoppers and store operations, fragmented orders, higher handling cost, and customer confusion about responsibility.

A useful decision framework evaluates each option against total cost, promised lead time, inventory accuracy, capacity, customer convenience, returns complexity, and resilience. The lowest transport distance does not necessarily produce the lowest total cost if store picking is slow or inaccurate.

8. Technology, data, and artificial intelligence

Order management systems coordinate order capture, inventory allocation, payment, fulfilment, shipment, and exception handling. Warehouse management systems control locations, waves, labour, picking, packing, and dispatch. Transport management systems support carrier selection, routing, tendering, tracking, and freight settlement. Application programming interfaces connect internal systems with carriers, marketplaces, suppliers, and customers.

Artificial intelligence can support demand sensing, ETA prediction, customer-contact classification, delivery-slot optimisation, fraud detection, return-risk analysis, and warehouse labour planning. Its value depends on data quality and operational adoption. An algorithm trained on historical allocation decisions may reproduce systematic bias. A delivery promise model that ignores carrier capacity may be accurate in normal conditions but fail during peaks.

Governance should include data ownership, access controls, audit trails, model validation, human override, cybersecurity, privacy protection, and monitoring for drift. Customers should not be denied remedies solely because an opaque model classified an order as suspicious. Personal data should be collected only for legitimate purposes, retained appropriately, and protected under applicable privacy law.

9. Detailed applied case: Northstar Home Appliances

Background

Northstar Home Appliances sells refrigerators and small kitchen appliances through its website, 65 stores, and two online marketplaces. It promises two-day delivery for major cities and store collection within four hours. The company uses a central warehouse, store inventory, and supplier drop-shipping.

Problem

Sales grew rapidly, but customer complaints increased. Reported on-time delivery was 93%, yet only 81% of orders were delivered complete and on time. Website availability was also unreliable: customers could order products shown as available, but cancellation rates rose during promotional periods. Stores complained that online picking disrupted in-store service. Returns increased after a new product-description format was introduced.

Analysis

An investigation separated performance by channel and examined the original promise rather than revised dates. It found four main causes:

1. store inventory records were updated only several times per day;
2. the order-management system did not subtract reservations from displayed availability;
3. marketplace orders received no capacity reservation during peak periods;
4. product dimensions and installation requirements were incomplete, causing unsuitable purchases and failed deliveries.

The company also discovered that its 93% on-time metric excluded orders cancelled before dispatch. This made the published figure appear better than the customer’s actual experience.

Approach and solution

Northstar introduced a unified available-to-promise view. Store stock was classified as sellable, reserved, damaged, display, or unavailable. Inventory feeds were updated continuously for high-volume products and at shorter intervals for the remaining range. A protected quantity was retained for store demand and service replacements.

The company created channel-specific fulfilment rules. Small appliances could be picked in stores when labour capacity was available; refrigerators were generally fulfilled from the central warehouse because of handling requirements. During promotions, the system used finite dispatch capacity and stopped promising two-day delivery when the capacity threshold was reached.

Northstar redesigned performance reporting around original promises. It reported fill rate, cancellation rate, OTIF, return reason, and cost per fulfilled order by channel. A cross-functional weekly meeting reviewed exceptions rather than blaming individual departments.

Finally, product pages displayed accurate dimensions, delivery access requirements, installation options, and return conditions. Customers could select a delivery slot only when both inventory and carrier capacity were available.

Outcome

After three months, the illustrative pilot reduced cancellations by 38%, increased OTIF from 81% to 91%, and reduced avoidable returns by 22%. Store labour cost per online order increased slightly, but total fulfilment cost declined because failed deliveries and emergency shipments fell. The improvement did not result from one technology purchase. It resulted from consistent definitions, reliable inventory data, capacity-aware promises, channel governance, and feedback from returns.

Learning points

• A high on-time figure can conceal poor completeness or selective measurement.
• ATP must include reservations and policy-protected stock.
• Omnichannel inventory is useful only when its status and location are trustworthy.
• Promise dates should reflect capacity, not only physical stock.
• Returns data can reveal commercial and product-information failures.
• Service metrics should be connected to cost and customer consequences.

10. Challenges, limitations, and managerial implications

The first challenge is the service-cost trade-off. Higher safety stock, more fulfilment nodes, premium transport, and generous returns policies can improve experience but reduce profitability. Managers should evaluate incremental service benefits against total cost-to-serve and customer lifetime value.

The second challenge is demand volatility. Promotions, weather, disruptions, viral products, and competitor actions make historical averages unreliable. Scenario planning, capacity buffers, postponement, and clear promise throttling can reduce the damage.

The third challenge is data inconsistency. Different systems may use different product identifiers, time zones, customer addresses, and definitions of “shipped” or “delivered.” Data governance is therefore an operational requirement, not merely an information-technology concern.

The fourth challenge is sustainability. Faster delivery and individual parcel shipments may increase emissions and packaging. Consolidation, collection points, realistic delivery windows, right-sized packaging, electric vehicles, and transparent carbon accounting can improve the balance. Sustainability claims should be evidence-based and should include returns and failed deliveries.

The fifth challenge is resilience. A highly efficient network may be vulnerable to carrier failure, cyberattack, labour disruption, extreme weather, or supplier concentration. Dual sourcing, alternate nodes, manual fallback procedures, and tested continuity plans should be evaluated according to the criticality of the service.

Best practice is to manage fulfilment as a closed learning loop: sense demand and capacity, promise realistically, execute reliably, measure the customer outcome, analyse exceptions and returns, and redesign the process. Service excellence is thus a capability built through policy, process, people, data, and continuous improvement.

Chapter Summary

Order fulfilment converts customer demand into delivered value through order capture, validation, availability promising, allocation, warehouse execution, transportation, delivery, settlement, and after-sales support. Customer service includes pre-transaction, transaction, and post-transaction activities and should be segmented according to customer value and operational criticality.

ATP calculates uncommitted supply, while CTP considers production or service capacity. Fill rate, on-time delivery, OTIF, perfect-order performance, customer effort, cost, and sustainability measures provide complementary views of performance. Definitions and denominators must be explicit, and metrics should be analysed by customer, product, channel, carrier, and location.

Fulfilment channels include central warehouses, stores, click-and-collect, drop-shipping, marketplaces, and field stock. Omnichannel logistics integrates these channels but requires accurate inventory, common data, capacity-aware promises, and coordinated returns. Reverse logistics is both a cost-control activity and a source of product and customer insight. Technology and AI can improve visibility and decision-making, but require governance, privacy, cybersecurity, human oversight, and bias control.

The managerial objective is not maximum service regardless of cost. It is reliable, transparent, economically sustainable service designed around customer needs and supported by end-to-end process discipline.

Key Terms

Available-to-promise (ATP): Uncommitted supply that can be promised to customers.

Capable-to-promise (CTP): A promise based on materials, production or service capacity, and delivery feasibility.

Order cycle: The elapsed process from order receipt to customer delivery.

Fill rate: The proportion of ordered lines or units supplied from available stock.

On-time delivery: Delivery by the agreed date or within the agreed window.

OTIF: On-time in-full performance.

Perfect order: An order meeting defined requirements for accuracy, completeness, condition, documentation, and timing.

Promise accuracy: The extent to which actual performance matches the commitment made at order acceptance.

Allocation: The process of distributing scarce supply among competing demands.

Drop-shipping: Supplier shipment directly to the end customer on behalf of another seller.

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

Omnichannel logistics: Coordinated fulfilment across multiple integrated customer channels.

Reverse logistics: Movement and management of products from customers back through the supply chain.

Disposition: The decision regarding a returned product’s next use or final treatment.

Cost-to-serve: The total cost of serving a specified customer, order, product, or channel.

Order-management system: Technology coordinating order capture, allocation, fulfilment, and status.

Exception management: Identification and resolution of events that threaten normal fulfilment.

Service segmentation: Designing differentiated service policies for distinct customer or product groups.

Review Questions

Short Answer Questions

1. What are the principal stages in an order-to-delivery cycle?
2. Distinguish ATP from CTP.
3. Why can unit fill rate and order-line fill rate produce different conclusions?
4. What conditions must be satisfied for a delivery to be OTIF?
5. Identify three possible dispositions for returned products.

Long Answer Questions

1. Explain how customer service can be segmented and discuss the risks of applying one service policy to all customers.
2. Describe the information and governance requirements for an effective omnichannel fulfilment system.
3. Discuss the economic and operational trade-offs between centralised and decentralised fulfilment.
4. Explain how returns data can improve product design, supplier management, and customer experience.
5. Evaluate the role of artificial intelligence in order fulfilment, including its governance risks.

Analytical / Application Questions

1. A firm has 600 usable units, 250 confirmed receipts, 520 existing commitments, and 80 units protected for safety stock. Calculate ATP and explain whether a new order for 300 units can be promised.
2. A retailer reports 96% on-time delivery, 94% in-full delivery, and 91% OTIF. Explain why OTIF is lower and propose two diagnostic analyses.
3. Design a fulfilment policy for a company selling urgent hospital components and routine consumer accessories. Specify service segmentation, allocation priorities, promise logic, and suitable performance measures.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes ATP from CTP?

A. ATP considers only transport cost, whereas CTP considers only inventory cost.

B. ATP considers uncommitted supply, whereas CTP also considers production or service capacity.

C. ATP applies only to returns, whereas CTP applies only to deliveries.

D. ATP is a customer-satisfaction measure, whereas CTP is a financial measure.

Correct Answer: B. ATP considers uncommitted supply, whereas CTP also considers production or service capacity.

Question 2

A firm has 420 usable units, 180 units arriving, 470 committed units, and 50 protected units. What is ATP?

A. 80 units

B. 130 units

C. 550 units

D. 1,120 units

Correct Answer: A. 80 units

Question 3

Which condition is required for an order to be counted as OTIF?

A. It is shipped early, regardless of quantity.

B. It arrives on time and is complete.

C. It is invoiced before dispatch.

D. It is delivered with premium transport.

Correct Answer: B. It arrives on time and is complete.

Question 4

Which channel normally involves the supplier shipping directly to the end customer?

A. Click-and-collect

B. Store fulfilment

C. Drop-shipping

D. Central warehouse picking

Correct Answer: C. Drop-shipping

Question 5

Why should ATP subtract reservations?

A. To increase reported inventory turnover

B. To prevent the same physical supply being promised to multiple customers

C. To eliminate all safety stock

D. To calculate transport emissions

Correct Answer: B. To prevent the same physical supply being promised to multiple customers

Question 6

Which is a post-transaction customer-service activity?

A. Product catalogue design

B. Credit approval before ordering

C. Warranty and returns support

D. Initial market segmentation

Correct Answer: C. Warranty and returns support

Question 7

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

A. It makes local inventory impossible to use.

B. It can disrupt store operations and expose inaccurate store inventory.

C. It always increases delivery distance.

D. It eliminates the need for order-management systems.

Correct Answer: B. It can disrupt store operations and expose inaccurate store inventory.

Question 8

Which measure requires both timely delivery and complete delivery for the same delivery event?

A. Unit fill rate

B. On-time delivery rate

C. OTIF

D. Complaint rate

Correct Answer: C. OTIF

Question 9

Which activity is part of reverse logistics?

A. Forecasting initial demand only

B. Inspecting and deciding the disposition of returned goods

C. Creating a sales commission plan

D. Setting a production wage rate

Correct Answer: B. Inspecting and deciding the disposition of returned goods

Question 10

Why is a single aggregate service metric potentially misleading?

A. It always contains too many variables.

B. It may conceal poor performance for a particular customer, product, or channel.

C. It cannot be calculated mathematically.

D. It measures only warehouse temperature.

Correct Answer: B. It may conceal poor performance for a particular customer, product, or channel.