Process Design, Analysis and Improvement

Introduction and context

Operations are performed through processes: linked activities that transform inputs into outputs for a customer or user. A process may be physical, administrative, digital, or a combination of these. Examples include receiving materials, approving a loan, scheduling a patient, assembling a product, processing an online order, and resolving a service complaint.

Process design establishes how work should be organised. Process analysis evaluates how work actually occurs, using evidence about time, quality, cost, capacity, and customer experience. Process improvement changes the process so that it delivers greater value with less waste and more consistent performance. These activities are central to operations management because strategy is realised through processes. A low-cost strategy requires efficient and stable processes; a quality strategy requires capable and controlled processes; a responsive strategy requires short lead times and flexible processes.

A process should not be assessed only by asking whether individual tasks are busy. The more important questions are whether the complete process creates customer value, whether work flows smoothly, where delays occur, and whether the process can meet demand reliably. Effective analysis therefore combines visual mapping, quantitative measures, customer requirements, employee knowledge, and disciplined improvement methods.

Learning objectives

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

1. Define and distinguish process design, process analysis, process improvement, and process redesign.
2. Map processes using flowcharts, SIPOC diagrams, and value-stream maps.
3. Identify bottlenecks and evaluate cycle time, throughput, capacity, utilisation, and lead time.
4. Apply Little’s Law appropriately and interpret its operational implications.
5. Select an appropriate process type for volume, variety, flexibility, and customer requirements.
6. Explain business process redesign and distinguish it from incremental improvement.
7. Develop standard work and use it to improve consistency, training, and control.
8. Evaluate the technological, human, ethical, and managerial consequences of process change.

1. The process perspective

A process is a sequence or network of activities that converts inputs into outputs. It has a purpose, identifiable customers, resources, controls, and performance measures. Processes may cross departmental boundaries; consequently, improving one department in isolation can worsen total performance. For example, purchasing may reduce purchase price by ordering large batches, while stores and production experience higher inventory and longer handling times.

Four concepts should be distinguished:

• Process design: deciding the structure, sequence, resources, technology, roles, and controls required to deliver an intended outcome.
• Process analysis: studying the current process to understand activities, relationships, performance, causes of variation, and constraints.
• Process improvement: making focused changes to remove waste, reduce variation, improve quality, or increase flow.
• Business process redesign (BPR): fundamentally rethinking and restructuring a process, often enabled by information technology, rather than making only local adjustments.

A process has several types of work. Value-adding work changes a product or service in a way the customer requires and would be willing to pay for. Business-value-adding work may be required by regulation, risk control, or the organisation even though the customer does not directly value it. Non-value-adding work consumes resources without contributing to the required outcome, such as avoidable waiting, duplication, searching, correction, or unnecessary movement.

The distinction is not always simple. A quality inspection may not add physical value, but it can be necessary until the process is sufficiently capable. The long-term improvement objective is often to build quality into the process so that inspection and rework are reduced.

2. Process mapping tools

Flowcharts

A flowchart represents activities and decisions in sequence. Common conventions include rectangles for activities, diamonds for decisions, arrows for direction, and terminators for start and end points. A useful flowchart identifies:

• the process boundary;
• the person or system responsible for each activity;
• inputs and outputs;
• decision rules and alternative paths;
• rework loops and handoffs;
• delays, queues, and control points.

A cross-functional flowchart or swimlane map places activities in lanes representing departments, roles, or systems. It makes handoffs visible, which is particularly useful in administrative and service processes. Mapping should be based on observation, records, interviews, and system data. A map that describes the official procedure but not actual practice is a design document, not a reliable current-state analysis.

SIPOC analysis

SIPOC stands for Suppliers, Inputs, Process, Outputs, and Customers. It is a high-level map used at the beginning of analysis to establish scope and shared understanding.

SIPOC is deliberately less detailed than a flowchart. It is valuable for defining boundaries, identifying stakeholders, and preventing premature concentration on individual tasks. It can also expose unclear requirements: an output is meaningful only if a customer and a specification can be identified.

Value-stream mapping

Value-stream mapping (VSM) examines the flow of material and information across a complete value stream. It distinguishes processing time from waiting time and usually records inventory, queue sizes, information signals, staffing, uptime, and production or service demand. A current-state map describes existing performance; a future-state map proposes a better flow.

VSM is particularly useful in Lean operations because it connects local activities to end-to-end lead time. It helps identify excessive work-in-process, batch production, repeated scheduling, long changeovers, rework, and disconnected information. However, VSM should not be treated as a universal solution. It may require adaptation for highly variable professional services, project work, or processes where the “product” is intangible.

A sound mapping sequence is:

1. Identify the customer and the value proposition.
2. Define the start and end points.
3. Observe the process and record actual activities.
4. Record demand, processing time, waiting time, defects, queues, and handoffs.
5. Validate the map with process participants.
6. Identify constraints and improvement opportunities.
7. Design and test a future state.

3. Quantitative process analysis

Visual maps show structure; measures show performance. Measures must be defined consistently. A “cycle time” measured from the first operation to completion is not comparable with a cycle time measured for one workstation only.

Cycle time, lead time, throughput, and capacity

Cycle time is the elapsed time between completion of successive units at a process or work centre under stated operating conditions. It may also be used to mean the time required to process one unit; the definition must therefore be stated.

Lead time is the total elapsed time from the customer’s request or order entry to delivery. It commonly includes processing, transport, waiting, inspection, queueing, and rework.

Throughput is the rate at which completed units leave a process. If N units are completed during an observation period T, then:

where N is measured in units and T in time, giving units per time. The observation period should be long enough to represent normal conditions.

If a resource completes one unit every c time units and has m parallel, identical resources, an idealised capacity is:

This assumes the resources are available, adequately staffed, supplied, and capable of producing acceptable output. Breaks, setup, downtime, quality losses, and variability reduce practical capacity.

Takt time expresses the production pace required to meet customer demand:

Takt time is a demand-based requirement, not the actual processing time. If a shift has 420 net available minutes and demand is 140 units, takt time is 3 minutes per unit. A process with a stable effective cycle time greater than 3 minutes cannot meet demand without additional capacity, work sharing, overtime, or demand adjustment.

Bottleneck analysis

A bottleneck is a resource, policy, or activity that constrains the throughput of the entire system. In a serial process, the bottleneck is commonly the stage with the lowest effective capacity. For example, suppose three stages have capacities of 30, 18, and 24 units per hour. The theoretical process capacity is 18 units per hour, subject to demand and other constraints.

A bottleneck may be physical, such as a machine, or managerial, such as an approval rule or an information-system limit. It may move when demand, product mix, staffing, or operating conditions change. Analysis should therefore examine:

• effective capacity rather than nominal capacity;
• demand by product or service family;
• setup and changeover losses;
• downtime and quality losses;
• queue growth before the constraint;
• starvation of the constraint due to upstream shortages;
• blocking after the constraint because downstream capacity is unavailable.

A useful improvement sequence is to identify the constraint, exploit existing capacity, subordinate other activities to it, elevate capacity where justified, and then reassess because the constraint may move. Increasing output at a non-bottleneck does not necessarily increase total throughput and may only increase inventory.

Utilisation and variability

Utilisation is the proportion of available capacity that is being used:

It is often expressed as a percentage. High utilisation can appear attractive, but a process operating close to 100% utilisation has little protection against variability. Random arrivals, variable processing times, absenteeism, failures, and urgent work can cause queues to increase sharply. Managers should distinguish productive use from congestion: a worker may be busy while customer lead time becomes unacceptable.

Little’s Law

Little’s Law connects average work-in-process, throughput, and average flow time:

where L is average number of units in the system (work-in-process), λ is the average throughput rate in units per time, and W is average time in the system. Therefore:

The law applies to a stable process over a sufficiently long period: average input and output rates are approximately equal, the system is not continually accumulating work, and the measures refer to the same system boundary and time period. It does not require constant arrival or processing times, but averages must be compatible. Units must be consistent: if L is 60 orders and λ is 15 orders per hour, W is 4 hours.

Worked example. A repair centre completes an average of 12 repairs per day and has an average of 36 repairs in the system, including waiting and repair. By Little’s Law:

If management reduces work-in-process to 24 repairs while maintaining throughput of 12 repairs per day, average flow time becomes 2 days. This is not a guarantee that every repair takes two days; it is an average relationship. If output falls because work is removed without addressing the constraint, the result may not improve.

4. Process selection and operating characteristics

Process selection aligns the process structure with volume, variety, flexibility, technology, and customer contact. Traditional classifications include:

Service processes have related dimensions: customer contact, customisation, labour intensity, and degree of front-office interaction. A hospital emergency department, a legal consultancy, and a self-service parcel locker require different process designs even when all deliver “service”.

The volume–variety relationship is a guide, not a rigid law. Flexible automation, modular design, postponement, digital configuration, and advanced planning can allow higher variety at greater volume. The choice involves trade-offs. Highly standardised processes can lower cost and variation but may reduce flexibility. Highly flexible processes can serve diverse needs but may require more skill, coordination, and capacity protection.

5. Business process redesign

Business process redesign asks whether the existing process should exist in its current form at all. Instead of automating every existing step, redesign challenges assumptions about sequence, ownership, approvals, information, and customer interaction. Typical redesign principles include:

• organise around outcomes rather than departmental tasks;
• place decisions where the work is performed, within appropriate controls;
• capture information once and make it available to authorised users;
• remove unnecessary handoffs and approvals;
• combine related activities;
• use digital workflow, self-service, analytics, or automation where they improve the outcome;
• provide a single accountable process owner;
• design controls into the process rather than relying only on end-stage inspection.

BPR can produce major gains but carries substantial risk. A technically efficient process may create exclusion for customers without digital access, reduce employee discretion, introduce cyber risk, or embed poor rules at greater speed. Redesign should therefore include customer research, employee participation, control assessment, pilot testing, and benefits measurement. Incremental improvement is preferable when the process is fundamentally sound and risk, cost, or regulatory constraints make radical change inappropriate.

6. Standard work

Standard work is the best currently known, safe, and repeatable method for performing an activity under specified conditions. It is not intended to prevent all improvement; rather, it establishes a baseline from which improvement can be measured.

A standard-work document normally specifies:

• the purpose and scope of the task;
• sequence of steps;
• expected time or pace;
• required materials, tools, and information;
• quality and safety points;
• responsibility and escalation rules;
• visual aids where useful;
• revision date, owner, and approval status.

Standard work differs from a vague policy. It should be observable and usable at the workplace. Employees should be trained, allowed to identify abnormalities, and involved in revising the standard. A standard that cannot be followed because materials, systems, or time are unavailable is not an effective control.

Standard work supports quality, onboarding, cross-training, workload balancing, auditability, and continuous improvement. It must also allow controlled responses to exceptions. In healthcare, aviation, laboratories, and other high-risk settings, standardisation should be combined with professional judgement, human-factor design, and escalation rather than used as a reason to suppress legitimate warnings.

7. A disciplined improvement method

A practical improvement project can follow the DMAIC logic: Define the problem and customer requirement; Measure current performance; Analyse causes and constraints; Improve the process; Control the gains. Other approaches, including PDCA, Kaizen, Lean, and Six Sigma, may be appropriate. The method should match the problem rather than become a ritual.

A robust project includes:

1. Problem definition: specify the gap, location, customer impact, time period, and business significance.
2. Baseline: establish current lead time, throughput, defect rate, queue size, cost, and variation.
3. Process map: validate the actual process and identify boundaries and handoffs.
4. Root-cause analysis: use evidence, stratification, observation, the five whys, cause-and-effect analysis, or statistical methods as appropriate.
5. Countermeasure design: address causes, not merely symptoms; consider capacity, quality, people, technology, and risk.
6. Pilot and evaluation: test changes on a controlled scale and compare with the baseline.
7. Control and learning: update standard work, train users, monitor measures, and review unintended effects.

A process measure should be balanced. Speed alone can encourage errors; utilisation alone can encourage queues; cost alone can damage quality or resilience. A useful performance set may include customer outcome, first-time-right quality, flow time, throughput, employee workload, safety, and environmental impact.

8. Illustrative case: reducing order fulfilment lead time

Background. This fictional case concerns Northstar Components, a distributor serving industrial customers. Customers complained that orders requiring credit approval took five working days, although the physical picking and packing work required less than one hour.

Problem. A SIPOC workshop identified sales staff, the customer, the credit team, warehouse staff, and the courier as suppliers or stakeholders. The current-state flowchart showed repeated data entry, an email handoff to credit, a queue for supervisor approval, and a second data check in the warehouse. The average process contained 50 open orders and completed 10 orders per day.

Analysis. Applying Little’s Law to the defined order process:

Observation showed that only 0.75 hours per order were active processing time. The principal problem was not worker effort but waiting, batching, and unclear approval rules. Credit staff processed requests in two daily batches. Warehouse staff could not see whether an order was awaiting information or approval.

Approach. The company introduced a single digital order record, risk-based approval thresholds, an immediate electronic notification, and a visual queue showing order status. Low-risk orders were auto-approved within defined limits; exceptions were routed to a credit specialist. The warehouse standard work was revised to include a single verification step and an escalation rule for missing information. Employees tested the new process during a two-week pilot.

Outcome. Average work-in-process fell to 24 orders and average throughput remained 10 orders per day. Little’s Law indicated:

The measured average lead time fell to approximately 2.4 days. The result was not attributed to automation alone: it depended on clearer rules, reduced batching, visible status information, and standard work. The company also monitored credit errors, customer complaints, employee workload, and orders incorrectly auto-approved.

Learning points. First, an end-to-end boundary revealed that departmental productivity was not the same as customer responsiveness. Second, reducing queues required changing the process logic, not merely asking employees to work faster. Third, digital controls required governance, exception handling, and monitoring. Finally, the improvement remained provisional until performance was sustained under changing demand.

9. Technology, AI, and governance

Process technology includes workflow systems, robotics, sensors, enterprise software, process mining, digital twins, and artificial intelligence. Process-mining tools can reconstruct actual process paths from event logs and reveal rework, delays, and compliance deviations. AI may classify requests, predict demand, identify anomalies, recommend schedules, or support agents. These tools are valuable when the underlying data are accurate and the process objective is clear.

Technology does not remove the need for process understanding. Automating an unnecessary approval can increase cost; an AI model trained on biased historical decisions can reproduce discrimination; a sensor system can create surveillance concerns; and a dashboard can encourage gaming of measures. Governance should address data quality, privacy, cybersecurity, explainability, human review, access rights, retention, audit trails, and accountability for decisions.

Operational redesign also has human and ethical implications. Consultation, training, accessibility, fair workload allocation, health and safety, and job-transition support should be considered. Regulatory requirements may apply to records, financial controls, personal data, medical services, product safety, or environmental reporting. Compliance should be designed into the process without creating unnecessary bureaucracy.

Challenges and limitations

Process analysis has several limitations. Maps can become too detailed to use or too general to diagnose. Average measures can hide peaks, variation, rework, and differences between customer segments. Little’s Law describes an observed relationship but does not identify causes. Capacity calculations can be misleading when demand is seasonal, resources are shared, or output quality varies. Bottlenecks can move after an intervention. Standard work can become obsolete, and redesign can fail through weak adoption rather than poor technical logic.

Managers should therefore triangulate evidence: combine maps, direct observation, transaction data, employee knowledge, customer feedback, financial information, and risk assessment. Improvement should be evaluated over an appropriate period and include unintended consequences. A faster process that increases defects, employee injury, exclusion, or environmental damage is not necessarily a better process.

Chapter Summary

Process design determines how resources, activities, information, technology, and responsibilities are configured to deliver an outcome. Process analysis makes the current system visible through flowcharts, SIPOC, and value-stream maps. Quantitative measures such as cycle time, lead time, throughput, capacity, utilisation, and takt time provide evidence about performance. Bottleneck analysis focuses attention on the constraint that limits system output, while Little’s Law links average work-in-process, throughput, and flow time under stable conditions.

Process selection should match volume, variety, flexibility, and customer contact. Business process redesign challenges the logic and structure of an existing process, whereas incremental improvement makes focused changes. Standard work establishes a safe and repeatable baseline while enabling controlled improvement. Successful improvement combines customer requirements, employee involvement, reliable data, pilot testing, balanced measures, and governance. Technology and AI can improve visibility and automation, but they also introduce risks involving bias, privacy, cybersecurity, accountability, and exclusion.

Key Terms

Bottleneck: The resource, activity, or rule that constrains total process throughput.

Business process redesign: Fundamental rethinking and restructuring of an existing business process.

Capacity: The maximum feasible output of a process under specified conditions.

Cycle time: The elapsed time between successive completed units or the processing time for a defined activity, as explicitly specified.

Flowchart: A graphical representation of activities, decisions, and sequence.

Flow time: Total time a unit spends in a defined process, including waiting and processing.

Lead time: Elapsed time from customer request or order to delivery.

Little’s Law: The relationship L = λW linking average work-in-process, throughput rate, and average flow time.

Process: A set of linked activities that transforms inputs into outputs.

Process analysis: Systematic examination of process structure and performance.

Process design: The planned configuration of activities, resources, roles, technology, and controls.

Process selection: Choosing a process type suited to volume, variety, flexibility, and customer requirements.

SIPOC: A high-level view of Suppliers, Inputs, Process, Outputs, and Customers.

Standard work: The best currently known, safe, repeatable method for performing a task.

Takt time: Net available production time divided by customer demand.

Throughput: The rate at which completed units leave a process.

Utilisation: The proportion of available capacity or time being used.

Value-adding activity: Work that changes the output in a way required and valued by the customer.

Value-stream mapping: Mapping of material and information flow across an end-to-end value stream.

Work-in-process: Units currently within a process but not yet completed.

Review Questions

Short Answer Questions

1. Distinguish process design from process improvement.
2. What information does a SIPOC analysis provide?
3. Define a bottleneck and explain why it may move.
4. State Little’s Law and define each variable.
5. Why is standard work important for continuous improvement?

Long Answer Questions

1. Explain how flowcharts, SIPOC, and value-stream mapping differ and describe when each is most useful.
2. Discuss the relationship between throughput, capacity, utilisation, and bottlenecks.
3. Compare project, job-shop, batch, repetitive, and continuous-flow processes.
4. Explain the principles, benefits, and risks of business process redesign.
5. Describe a complete DMAIC-based process improvement project, including measurement and control.

Analytical / Application Questions

1. A service process has an average of 80 cases in the system and completes 20 cases per day. Calculate average flow time using Little’s Law and state the assumptions required for interpretation.
2. Three serial work centres have effective capacities of 40, 25, and 32 units per hour. Identify the bottleneck, determine theoretical process capacity, and recommend two questions that should be investigated before adding resources.
3. Design a SIPOC and current-state improvement plan for a university student-enrolment process. Identify at least three likely forms of waste, two performance measures, and one ethical or accessibility risk.

Multiple Choice Questions with Answers

Question 1

Which tool provides a high-level view of Suppliers, Inputs, Process, Outputs, and Customers?

A. SIPOC

B. Control chart

C. Bill of materials

D. Capacity schedule

Correct Answer: A. SIPOC

Question 2

A process completes 15 units per hour and has an average of 45 units in the system. What is the average flow time?

A. 2 hours

B. 3 hours

C. 30 hours

D. 675 hours

Correct Answer: B. 3 hours

Question 3

In a serial process with effective capacities of 30, 18, and 24 units per hour, which stage is the bottleneck?

A. The first stage

B. The second stage

C. The third stage

D. All stages equally

Correct Answer: B. The second stage

Question 4

What does takt time primarily represent?

A. The actual repair time for one unit

B. The required production pace based on customer demand

C. The average queue length

D. The maximum machine speed

Correct Answer: B. The required production pace based on customer demand

Question 5

Which statement best describes value-stream mapping?

A. It maps only the organisation’s reporting structure

B. It examines material and information flow across an end-to-end value stream

C. It calculates employee salaries

D. It replaces all process measurements

Correct Answer: B. It examines material and information flow across an end-to-end value stream

Question 6

If average work-in-process is 24 orders and throughput is 10 orders per day, what is average flow time?

A. 0.42 day

B. 2.4 days

C. 14 days

D. 240 days

Correct Answer: B. 2.4 days

Question 7

What is the primary purpose of standard work?

A. To eliminate all employee judgement

B. To establish a safe, repeatable baseline for performance and improvement

C. To maximise inventory

D. To avoid measuring process performance

Correct Answer: B. To establish a safe, repeatable baseline for performance and improvement

Question 8

Which process type is generally associated with very high volume and very low variety?

A. Project

B. Job shop

C. Continuous flow

D. Custom consultancy

Correct Answer: C. Continuous flow

Question 9

Which statement about Little’s Law is correct?

A. It applies only when processing times are constant

B. It relates average work-in-process, throughput, and average flow time in a stable system

C. It identifies the root cause of every bottleneck

D. It states that utilisation must always equal 100 percent

Correct Answer: B. It relates average work-in-process, throughput, and average flow time in a stable system

Question 10

A major governance risk when using AI in process decisions is:

A. The existence of any process map

B. The possibility of biased data producing unfair decisions

C. The use of standard work

D. The measurement of throughput

Correct Answer: B. The possibility of biased data producing unfair decisions