Skip to main content
CBE NEXUS · UNDERGRADUATE INFORMATION SYSTEMS · CHAPTER 3

Systems Analysis and Requirements Engineering

Undergraduate textbook notes · Problem investigation · Requirements · Modelling · Validation

Learning outcomes: Explain systems analysis and the SDLC; investigate organisational problems; identify stakeholders; elicit, model, prioritise and validate requirements; assess feasibility, risk and quality attributes; and plan testing, migration and implementation.

Introduction

Information systems succeed when they address real organisational needs and are usable, secure, reliable and sustainable. Systems analysis provides the disciplined bridge between an organisational problem and a solution that can be designed, tested and operated. Requirements engineering makes stakeholder needs explicit and verifiable.

1. Purpose and scope of systems analysis

Systems analysis is the disciplined investigation of an organisation, problem or opportunity in order to understand needs and determine what an information system must achieve. It connects organisational objectives with people, processes, data, technology and controls. The analyst does not begin by assuming that a new application is the answer. The first task is to understand the situation, identify the problem worth solving and establish whether a system change is justified. The outcome may be a new system, a change to an existing system, a process redesign, a policy change or a decision not to proceed.

A system boundary defines what is included in the investigation and what remains external. The boundary should identify users, organisational units, external organisations, interfaces, data sources and relevant constraints. A narrow boundary can hide the real cause of a problem: an apparently slow application may be affected by manual approval queues, unclear responsibilities or poor-quality source data. A broad boundary can make analysis unmanageable. Analysts should define scope, assumptions, exclusions and dependencies and revise them when evidence changes.

Systems analysis is iterative. Initial understanding is usually incomplete, and stakeholders may disagree about the problem or desired outcome. Analysts progressively refine models, validate interpretations and manage uncertainty. A useful analysis distinguishes observed facts, stakeholder perceptions, assumptions and proposals. This discipline reduces the risk of turning an untested opinion into a system requirement. It also provides a traceable rationale for design and investment decisions.

Figure 1. Systems analysis and delivery relationship
ORGANISATIONAL NEED → PROBLEM INVESTIGATION → REQUIREMENTS
→ FEASIBILITY AND DESIGN → BUILD / CONFIGURE → TEST AND DEPLOY
← FEEDBACK, MEASUREMENT AND CHANGE

2. The systems development life cycle

The systems development life cycle (SDLC) organises the work involved in planning, analysing, designing, building, testing, deploying and maintaining information systems. Common phases include initiation, feasibility, requirements analysis, design, construction or configuration, testing, implementation, operation and retirement. Different methodologies use different names and may combine or repeat phases. The SDLC is a management framework rather than a guarantee that a project will succeed. Its value comes from appropriate decision points, evidence, accountability and feedback.

A traditional sequential lifecycle seeks to complete and approve substantial analysis and design before construction. This can support predictability when requirements are stable and regulatory documentation is important, but it may delay user feedback and make late changes expensive. Iterative and incremental lifecycles deliver a system in smaller parts, enabling earlier learning and adjustment. Agile approaches emphasise collaboration, working increments and response to change, but still require architecture, security, documentation and governance appropriate to the risk.

Lifecycle selection should reflect uncertainty, complexity, criticality, procurement constraints, team capability and user availability. A small internal workflow may be developed incrementally with frequent demonstrations. A safety-critical or highly regulated system may need formal assurance evidence and independent verification. Hybrid approaches are common: an organisation may use formal investment approval and architecture governance while delivering functionality through short iterations. The method should serve the project rather than become an end in itself.

3. Problem definition and root-cause investigation

A problem statement describes a gap between the current situation and a desired outcome. It should identify who is affected, what is happening, where and when it occurs, the evidence available and the consequences. Statements such as 'the organisation needs an app' prescribe a solution before the problem is understood. A stronger statement might describe delays in verifying applications, the proportion of cases affected and the impact on applicants and staff. A clear problem statement provides a basis for evaluating alternative responses.

Root-cause analysis investigates why a problem occurs rather than treating visible symptoms. Techniques include the five whys, cause-and-effect diagrams, fault trees, Pareto analysis and process observation. The five whys asks successive questions about causes, but should not be used mechanically: complex failures often have several interacting causes. A fishbone diagram groups potential causes under categories such as people, methods, machines, materials, measurement and environment. Hypotheses must be checked against evidence.

Analysts should avoid blame-based explanations that attribute failures to individual carelessness without examining system conditions. Workload, unclear instructions, poor interfaces, inadequate training, conflicting incentives and weak controls may contribute. Root causes can also lie outside the immediate organisation, such as supplier delays or unreliable connectivity. Findings should distinguish confirmed causes from plausible hypotheses and identify what additional data or tests are needed.

4. Stakeholder identification and analysis

Stakeholders are people or organisations that affect, are affected by, or have an interest in a system. They may include end users, customers, managers, process owners, executives, IT staff, suppliers, regulators, auditors, community representatives and people whose data are processed. Stakeholders have different goals, levels of influence, technical knowledge and exposure to system risks. Identifying only the project sponsor and direct users can omit groups who experience important consequences or control essential resources.

Stakeholder analysis records interests, influence, impact, expectations, concerns and preferred engagement methods. A power-interest matrix can help plan communication, but should not be treated as a substitute for listening to less powerful groups. A receptionist, student, patient or citizen may have little formal authority but possess essential knowledge of service failures. Accessibility and inclusion require deliberate engagement with people who may face language, disability, connectivity or digital-literacy barriers.

Stakeholder engagement is continuous. Analysts may use interviews, workshops, surveys, observation, demonstrations and feedback sessions. The purpose and limits of each activity should be explained. Conflicting stakeholder views should be documented and resolved through explicit priorities, policy decisions or trade-off analysis. Analysts should not promise that every request will be implemented; they should show how requests are evaluated against objectives, constraints, risks and available resources.

5. Requirements engineering overview

Requirements engineering is the systematic process of discovering, analysing, documenting, validating and managing the needs and constraints that a system must satisfy. A requirement states a capability, quality, rule or condition needed by a stakeholder or imposed by the operating environment. Requirements provide the basis for design, development, testing, procurement and acceptance. Poor requirements create rework because teams may build features that are technically correct but do not solve the intended problem.

Requirements are commonly divided into functional and non-functional categories. Functional requirements describe what the system must do, such as register a learner, calculate a fee, approve a request or generate a report. Non-functional requirements describe qualities or constraints such as response time, availability, security, usability, accessibility, interoperability, maintainability and capacity. Both categories are essential. A system that performs the right function but is unavailable during critical periods is not fit for purpose.

Requirements should be necessary, clear, feasible, testable, consistent, traceable and prioritised. Vague terms such as 'fast', 'user-friendly' or 'secure' need measurable interpretation. For example, a response-time requirement should identify the operation, workload, percentile or measurement condition. A security requirement should state the control or outcome expected without prescribing an unsuitable implementation prematurely. Each requirement should have an identifier, rationale, source, owner and acceptance criteria.

6. Requirements elicitation techniques

Elicitation is the activity of discovering requirements from stakeholders, existing systems, documents, regulations, operational data and the wider environment. Interviews allow detailed exploration and follow-up questions. Structured interviews use a consistent set of questions; semi-structured interviews provide a framework while allowing new issues to emerge. Analysts should prepare, listen actively, avoid leading questions and confirm their interpretation at the end.

Workshops bring stakeholders together to map processes, discuss priorities and resolve differences. They can reveal cross-functional dependencies more quickly than separate interviews, but require skilled facilitation so that senior or outspoken participants do not dominate. Observation and contextual inquiry allow analysts to see real work, including informal workarounds and environmental constraints. Document analysis examines forms, policies, reports, manuals, contracts and existing system outputs. Surveys can reach larger populations but may provide less explanatory depth.

Prototypes, storyboards and interface mock-ups help stakeholders react to concrete examples. They are especially useful when users find it difficult to describe an unfamiliar system. Analysts must label prototypes clearly so that users do not mistake a demonstration for a committed feature. Existing logs and analytics can reveal usage patterns, delays and errors, but cannot fully explain motives or unrecorded work. Strong elicitation combines methods and triangulates findings.

7. Interview design and facilitation

An effective requirements interview begins with a clear objective and an understanding of the participant's role. Open questions invite explanation: 'Walk me through how you handle a new application.' Probing questions explore exceptions, decisions and evidence: 'What happens when the supporting document is missing?' Closed questions can confirm details, but an interview dominated by yes-or-no questions may miss important context. Questions should be neutral and avoid implying that a particular solution has already been selected.

Interviewers should distinguish what a participant does, what policy says should happen and what the participant believes ought to happen. These may differ. Asking for a recent example can help turn general descriptions into observable detail. Analysts should explore frequency, volume, timing, error handling, hand-offs, data use and workarounds. Notes should record key statements, unresolved issues and agreed actions, while respecting confidentiality and organisational rules.

At the end of an interview, summarise the main points and invite correction. Findings should be shared in an appropriate form for validation. Sensitive topics may require private interviews rather than group workshops. Analysts should avoid recording personal information that is not needed for the project. Good facilitation builds trust by demonstrating that participants' time is respected and that their contributions are reflected accurately.

8. Use cases and user stories

A use case describes how an actor interacts with a system to achieve a goal. It typically identifies the primary actor, trigger, preconditions, main success flow, alternative flows, exceptions and postconditions. For example, an applicant submits an application, the system validates required fields, records the submission and provides a reference. An alternative flow may route an incomplete application to a correction stage. Use cases help define system behaviour from an external viewpoint.

Use-case diagrams show actors and use cases within a system boundary. They provide a high-level view of scope and interactions, but do not normally capture every detail of sequence or business rule. Detailed use-case descriptions provide the steps and alternatives. Analysts should ensure that use cases represent meaningful actor goals rather than every button click. They should also consider actors that are other systems, scheduled processes or external services.

Agile teams often express needs as user stories, commonly using the pattern: 'As a [user], I want [capability], so that [benefit].' A story should have acceptance criteria that define observable completion. INVEST is a useful heuristic: stories should be independent, negotiable, valuable, estimable, small and testable. User stories are not a substitute for detailed rules, data definitions, security requirements or system-wide quality attributes. Those concerns must be captured in an appropriate backlog or specification.

Figure 2. Use-case interaction example
ACTOR: APPLICANT → SUBMIT APPLICATION
SYSTEM: VALIDATE DETAILS → RECORD CASE → ISSUE REFERENCE
ALTERNATIVE: INCOMPLETE → REQUEST CORRECTION

9. Functional requirements and business rules

Functional requirements specify the services and behaviours a system must provide. They may describe data entry, validation, calculations, searches, notifications, approvals, reports, integrations and exception handling. Requirements should state the trigger, input, expected processing and observable result. For example, 'When an authorised finance officer confirms a payment reference, the system shall update the invoice status and record the officer, timestamp and reference' is more testable than 'the system should manage payments'.

Business rules define policies, constraints and decision logic that govern organisational activity. Examples include eligibility criteria, approval thresholds, tax treatment, discount conditions and document retention rules. Rules should be separated from interface wording where possible so they can be reviewed and maintained consistently. A rule catalogue can record its identifier, statement, rationale, source authority, owner, effective date and affected processes.

Analysts should identify rule conflicts, exceptions and effective dates. A policy may change during a project, or different departments may apply inconsistent interpretations. A decision table can make combinations of conditions explicit. The system should handle missing, invalid and contradictory data in a predictable way. Rules that affect legal rights, financial decisions or access to essential services require appropriate domain and compliance review.

10. Non-functional requirements and quality attributes

Non-functional requirements describe how well a system must operate and the constraints under which it must operate. Performance may specify response time, throughput and workload. Availability may specify service hours and acceptable downtime. Reliability concerns consistent operation; recoverability concerns restoration after failure; scalability concerns growth in users or transactions; and maintainability concerns the effort required to correct or enhance the system.

Security requirements may cover authentication, authorisation, encryption, logging, secure development, vulnerability management and incident response. Privacy requirements address purpose limitation, data minimisation, transparency, retention and individual rights where applicable. Accessibility requirements ensure that people with disabilities can use the system, including through keyboard navigation, assistive technology, adequate contrast and understandable content. Usability requirements should be tied to representative users and realistic tasks.

Quality attributes can conflict. Stronger authentication may add friction; extensive logging may increase storage and privacy risk; high availability may increase cost; and broad customisation may reduce maintainability. Analysts should document priorities and trade-offs rather than treating every quality attribute as unlimited. Requirements should specify measurable thresholds and test conditions. A requirement such as 'support 2,000 concurrent users with 95th-percentile page response below two seconds under the agreed workload' is more useful than 'must be scalable and fast'.

11. Requirements modelling and specification

Requirements models provide structured views of system behaviour, information, processes and constraints. Common models include use cases, activity diagrams, data-flow diagrams, entity-relationship diagrams, state-transition diagrams, decision tables and process models. Each model answers a different question. Analysts should select models that clarify the problem rather than produce diagrams simply to satisfy a template. Consistent names and definitions across models reduce ambiguity.

A requirements specification records agreed requirements in a form that stakeholders and technical teams can use. It may include scope, context, functional requirements, quality attributes, interfaces, data requirements, constraints, assumptions, dependencies, acceptance criteria and glossary. The document may be a formal baseline, a set of linked backlog items or a combination. The important features are clarity, controlled change and traceability.

Requirements should be written at an appropriate level of abstraction. Prematurely specifying screen layout or database technology can constrain design before alternatives are assessed. Conversely, a requirement that is too general cannot guide implementation or testing. Analysts should distinguish the required outcome from a proposed solution and identify where a design decision is genuinely constrained by policy, legacy integration or technical standards.

12. Data requirements and information modelling

Data requirements identify the information a system must collect, use, create, share, retain and protect. Analysts should identify entities, attributes, relationships, identifiers, data sources, validation rules, ownership and retention needs. An entity represents a type of thing about which data are stored, such as Student, Course, Application or Payment. An attribute describes a property, while a relationship describes an association between entities.

Entity-relationship modelling helps represent data structure. A one-to-many relationship may connect one department to many courses; a many-to-many relationship may connect students and courses through an enrolment entity. Primary keys uniquely identify records, while foreign keys link related records. Normalisation reduces avoidable duplication and update anomalies, although analytical and performance needs may sometimes justify carefully controlled denormalisation.

Data quality includes accuracy, completeness, consistency, timeliness, validity and uniqueness. Requirements should specify how data are validated, corrected and governed. Migration from legacy systems needs profiling, cleansing, mapping, deduplication and reconciliation. Sensitive information requires appropriate access, retention and protection. Analysts should not collect data merely because it may be useful in the future; collection should have a defined purpose and lawful basis.

Figure 3. Simple data relationship model
APPLICANT (ApplicantID) 1 ─── * APPLICATION (ApplicationID)
PROGRAMME (ProgrammeID) 1 ─── * APPLICATION
APPLICATION 1 ─── * DOCUMENT (DocumentID)

13. Process and data-flow modelling

Process models describe activities, decisions, events, roles and sequence. Data-flow diagrams focus on how information moves between processes, external entities and data stores. A context diagram presents the system as one process and shows its relationships with external entities. Lower-level diagrams decompose the system into processes and flows. The diagrams should be balanced: inputs and outputs at a parent level should be accounted for in the detailed decomposition.

Data-flow diagrams help identify where data originate, how they are transformed, where they are stored and who receives them. They are useful for analysing interfaces, privacy exposure and missing validation. A data flow should be labelled with meaningful information, not vague terms such as 'data'. A process should transform or use data; a data store should represent retained information; and an external entity should be outside the system boundary.

Process and data-flow models are complementary. A process map may show that an application is checked and approved, while a data-flow model clarifies which records are consulted and what information is passed to the decision maker. Analysts should reconcile names, boundaries and assumptions across models. Model reviews with stakeholders can expose missing activities, duplicated capture and unclear ownership.

14. State modelling and decision logic

A state model describes the states an entity can occupy and the events or conditions that cause transitions. An application might move through Draft, Submitted, Under Review, Awaiting Information, Approved, Rejected and Withdrawn. A state-transition model helps clarify which actions are valid at each stage, who may perform them and what records must be created. It prevents contradictory status changes and supports consistent reporting.

Decision logic determines how the system responds to conditions. Decision tables are useful when several conditions combine to determine an action; decision trees show branching sequences. Analysts should define how missing information, boundary values and conflicting conditions are handled. For example, an eligibility rule should state whether an applicant exactly at a threshold qualifies and how equivalent evidence is assessed.

State and decision models are especially important in systems that manage cases, applications, orders, claims, payments or regulated approvals. They support test design because each state and transition can be checked. They also help identify security requirements: a user may be allowed to view a case but not approve it, or may only change a record before a particular transition.

15. Feasibility analysis and business cases

Feasibility analysis evaluates whether a proposed response is viable and worthwhile. Technical feasibility examines whether technology, skills, integration and infrastructure can support it. Economic feasibility compares costs and benefits. Operational feasibility considers whether people and processes can adopt the change. Legal and regulatory feasibility considers obligations and permissions. Schedule feasibility assesses whether delivery can occur within required time constraints. Organisational and ethical considerations may also be material.

A business case explains the problem, options, recommended approach, expected benefits, costs, risks, dependencies and measures of success. Costs include acquisition, configuration, development, integration, migration, training, support, security, maintenance and eventual replacement. Benefits may include reduced processing time, fewer errors, improved access, better decisions, compliance and service quality. Benefits should have owners and measurement plans rather than remain optimistic statements.

Options analysis should include more than build versus buy. Alternatives may include process simplification, configuration of an existing platform, shared services, outsourcing, incremental enhancement or no change. Net present value and payback period can support financial evaluation, but estimates are uncertain and should be tested through sensitivity analysis. A positive financial result does not automatically justify a solution that creates unacceptable safety, privacy or fairness risks.

Figure 4. Feasibility dimensions
TECHNICAL · ECONOMIC · OPERATIONAL
LEGAL · SCHEDULE · ETHICAL
ALL INFORM THE INVESTMENT DECISION

16. Requirements prioritisation and negotiation

Projects operate under constraints of time, budget, skills and risk. Requirements prioritisation helps decide what must be delivered first and what can be deferred. MoSCoW classifies requirements as Must have, Should have, Could have and Won't have for the current release. The categories need explicit definitions; if every stakeholder labels every request as essential, prioritisation has failed. Value, urgency, risk reduction, dependency and implementation effort should inform decisions.

Weighted scoring can compare requirements against agreed criteria. For example, stakeholders may score customer value, legal necessity, operational risk, strategic alignment and cost. Scores should support discussion rather than create false precision. A low-scoring requirement may still be mandatory because of law or safety. Dependencies matter: a seemingly low-value data capability may be necessary before a high-value feature can work.

Negotiation should make trade-offs transparent. Analysts can present options, consequences and evidence, then record decisions and unresolved disagreements. Scope changes should be assessed for impact on cost, schedule, architecture, testing, training and controls. A change-control process prevents informal additions from undermining delivery. In iterative projects, the backlog is continuously refined, but changes still need ownership and impact awareness.

17. Prototyping and user-centred design

Prototyping creates an early representation of a proposed system so that assumptions can be tested. Low-fidelity prototypes may be sketches, paper screens or simple wireframes. High-fidelity prototypes may simulate navigation and interaction. Throwaway prototypes are used to learn and then discarded; evolutionary prototypes are progressively refined into a working solution. The choice depends on the question being tested and the risk of confusing a demonstration with a production-ready product.

User-centred design places users, tasks and environments at the centre of design decisions. It begins by understanding context of use, specifies user needs, develops design alternatives and evaluates them with representative users. Usability testing observes people attempting realistic tasks and identifies errors, confusion and unnecessary effort. The sample should include users with different abilities, devices, languages and levels of digital confidence where relevant.

Accessibility should be built into requirements and evaluated throughout design. Semantic structure, keyboard operation, visible focus, captions, labels, readable language and sufficient contrast support a wider range of users. Design should also account for low bandwidth, small screens and interruptions. User feedback is evidence, but design teams must interpret it alongside policy, security, technical constraints and the needs of users who cannot attend testing sessions.

18. Architecture and interface requirements

System architecture describes the major components of a system, their responsibilities, relationships and deployment environment. Requirements analysis should identify architectural constraints such as existing platforms, approved technologies, data residency, integration standards, scalability, resilience and security. Analysts do not need to make every architecture decision, but they must expose requirements that materially influence architecture.

Interfaces connect a system to users, devices, internal applications, external services and data providers. Interface requirements should define the purpose, data exchanged, format, timing, protocol, authentication, error handling, retry behaviour and ownership. An integration that works under normal conditions may fail when a partner is unavailable or sends duplicate messages. Requirements should include observable behaviour for timeouts, partial failure and recovery.

Interoperability depends on shared semantics as well as technical connectivity. Two systems may exchange a date or status code but interpret it differently. Data contracts, schemas, versioning and change notification help manage these dependencies. Analysts should identify who owns each interface and how changes are tested. Integration testing should cover normal flows, invalid data, security failures, load and recovery.

19. Security, privacy and ethical requirements

Security requirements protect confidentiality, integrity and availability. They may address identity verification, authentication, authorisation, least privilege, encryption, secure configuration, audit logging, vulnerability handling, backup and incident response. Threat modelling identifies assets, actors, attack paths and possible harms. Security should be considered during requirements and architecture, not added only before launch.

Privacy requirements govern how personal information is collected, used, shared, retained and disposed of. Analysts should identify the purpose for each data item, minimise collection, define access, establish retention rules and explain transparency obligations. Privacy impact assessment may be required where processing is likely to create significant risks. Legal requirements vary by jurisdiction and sector, so qualified advice should be obtained when necessary.

Ethical analysis considers fairness, transparency, autonomy, accessibility, accountability and the distribution of benefits and harms. Automated scoring or eligibility decisions may disadvantage groups if data or rules are biased. Requirements may need explainability, human review, appeal routes, monitoring and restrictions on secondary use. A system can comply with minimum technical standards and still produce unacceptable outcomes; ethical review should examine real-world effects.

20. Testing requirements and acceptance criteria

Requirements must be testable. Acceptance criteria define the conditions under which a stakeholder will consider a requirement satisfied. They may specify input conditions, expected output, timing, permissions, error handling and evidence. Test cases should trace back to requirements so that coverage can be assessed. If a requirement cannot be tested, it may be ambiguous or may need a measurable proxy.

Testing occurs at several levels. Unit testing checks small components; integration testing checks interactions; system testing evaluates the complete solution; acceptance testing checks whether the system meets user and business needs. Non-functional testing may cover performance, security, accessibility, usability, compatibility, reliability and recovery. Independent testing can provide assurance for high-risk systems.

Acceptance should not be based solely on a successful demonstration. Tests need representative data, realistic workloads, negative cases, boundary conditions and exception paths. Defects should be classified by severity and impact, assigned and retested after correction. Stakeholders should understand any known limitations and formally accept residual risks through the organisation's governance process.

21. Requirements traceability and change management

Requirements traceability links each requirement to its source, rationale, design elements, implementation components, test cases and acceptance evidence. A traceability matrix helps identify requirements that have no tests, design elements with no approved need or changes that affect many parts of a system. Traceability is particularly important in complex, regulated or safety-sensitive projects.

Requirements change because stakeholders learn, policies evolve, threats emerge, suppliers alter interfaces or assumptions prove false. Change is not inherently a failure; unmanaged change is the risk. A change request should describe the proposed change, reason, urgency, affected requirements and expected benefit. Impact analysis considers architecture, data, cost, schedule, risk, training, migration and support.

An approved baseline provides a reference for assessing change. In iterative development, the baseline may be a release plan or a prioritised backlog rather than a frozen document. The team should still record decisions and maintain version history. Stakeholders need to know which requirements are current, which are superseded and which remain unresolved. Clear configuration management prevents teams from implementing outdated specifications.

Figure 5. Requirements traceability
STAKEHOLDER NEED → REQUIREMENT → DESIGN ELEMENT
→ IMPLEMENTED COMPONENT → TEST CASE → ACCEPTANCE EVIDENCE

22. Legacy systems and requirements for migration

Legacy systems may contain valuable business rules, historical data and integrations even when their technology is difficult to maintain. Analysis should identify current functionality, dependencies, data quality, operating costs, security exposure, user workarounds and knowledge held by staff. Replacing a legacy system without understanding its hidden role can disrupt important services.

Migration requirements address data extraction, mapping, cleansing, transformation, loading, reconciliation, cutover and rollback. Data owners should define authoritative sources and rules for duplicates, missing values and historical records. Trial migrations reveal problems before the final move. Reconciliation checks record counts, totals, relationships and selected samples. Sensitive information must be protected throughout migration, including temporary files and test environments.

Migration strategy may be a single cutover, phased transition, parallel operation or gradual replacement of components. Each option has different risks. Parallel operation can improve confidence but increase workload and create reconciliation challenges. Phased migration can reduce the size of each change but require temporary integration. The chosen approach should include user communication, training, support, contingency and a clear decision point for rollback.

23. Vendor selection and procurement requirements

When acquiring a system or service, requirements form the basis of procurement and evaluation. A request for proposal should describe outcomes, mandatory capabilities, quality attributes, interfaces, security, data handling, service levels, implementation expectations and evaluation criteria. Requirements should be technology-neutral where possible so suppliers can propose alternatives, while remaining specific enough to compare responses fairly.

Supplier evaluation may use demonstrations, scripted scenarios, proof-of-concept exercises, reference checks, security assessments and total-cost analysis. Marketing claims should be verified against evidence. Analysts should examine configuration limits, data export, integration costs, upgrade practices, support arrangements, subcontractors, service continuity and exit provisions. A product that meets functional needs may still create unacceptable vendor lock-in or data portability risk.

Contracts and service-level agreements should define responsibilities, availability, response and resolution targets, incident notification, data ownership, security obligations, audit rights, change management and termination assistance. Procurement decisions should involve technical, legal, financial, operational and security specialists. The cheapest initial price may not produce the lowest whole-life cost or best organisational outcome.

24. Implementation, deployment and organisational readiness

Implementation converts a tested solution into an operational service. Readiness includes infrastructure, configuration, data, access, support, procedures, training, communications, controls and business continuity. A go-live checklist should assign owners and evidence for each critical condition. Deployment should be planned around business calendars, user availability, transaction volumes and the consequences of interruption.

Common deployment approaches include direct cutover, phased deployment, pilot deployment and parallel running. Direct cutover is fast but exposes the organisation to concentrated risk. Phased deployment limits the scope of each transition but may require temporary interfaces. Pilot deployment tests the solution with a controlled group. Parallel running compares old and new processes but can increase cost and confusion. Selection depends on criticality, reversibility and operational constraints.

After launch, hypercare provides enhanced support while users adapt and defects are resolved. Monitoring should cover system health, process performance, user adoption, security events, customer outcomes and unresolved issues. A project should not be declared successful merely because the software is live. Benefits realisation requires comparing actual results with the business case and assigning responsibility for corrective action.

25. Integrated case study: university admissions information system

A university receives applications through email, paper forms and a web portal. Staff re-enter applicant details into separate systems, departments use different eligibility spreadsheets and applicants cannot reliably track progress. The project team first defines the problem in terms of incomplete applications, processing delay, inconsistent decisions and avoidable enquiries. It maps the end-to-end journey from enquiry through submission, verification, academic assessment, decision and enrolment.

Stakeholder analysis includes applicants, guardians where appropriate, admissions officers, academic departments, finance, IT, records staff, accessibility specialists and regulators. Elicitation uses interviews, workshops, observation, document analysis and review of application data. Functional requirements cover account creation, application submission, document upload, eligibility checks, status tracking, decision recording and notifications. Non-functional requirements cover privacy, accessibility, response time, auditability, availability and data retention.

Data modelling defines applicants, programmes, applications, qualifications, documents, decisions and payments with clear relationships and identifiers. Use cases describe applicant and staff interactions; state models define valid application transitions. Acceptance criteria test complete and incomplete submissions, duplicate records, permission boundaries and exception handling. Feasibility analysis compares configuring an existing platform with custom development. A pilot is tested with representative applicants and staff, followed by a phased rollout, support and benefits review. The case illustrates how analysis turns a broad request for a portal into a controlled set of organisational and technical requirements.

26. Professional practice, documentation and analyst skills

Systems analysts combine communication, critical thinking, domain understanding, modelling, facilitation, data literacy and technical awareness. They translate between people who understand organisational work and teams who build or configure technology. They must be curious without being careless, precise without becoming inflexible and neutral enough to investigate competing explanations. Domain knowledge is valuable, but assumptions should still be tested.

Professional documentation should be accurate, readable, version-controlled and appropriate to its audience. Models need titles, scope, legends, definitions and ownership. Meeting records should distinguish decisions from discussion. Requirements should have stable identifiers and clear acceptance criteria. Analysts should maintain a decision log, issue log, assumptions register and traceability information where the project's complexity warrants them.

Professional conduct includes confidentiality, honest reporting, respect for stakeholder differences, disclosure of conflicts of interest and awareness of competence limits. Analysts should escalate material safety, privacy or security concerns rather than conceal them to protect a schedule. They should avoid collecting unnecessary personal data and should communicate uncertainty clearly. The quality of analysis depends not only on technique but also on integrity and responsible judgement.

27. Emerging approaches: cloud, APIs, AI and continuous delivery

Cloud services change how systems are acquired, scaled and operated. Requirements should address service availability, data location, identity integration, backup, monitoring, portability, shared-responsibility boundaries and exit arrangements. Cloud adoption does not transfer every security or compliance responsibility to the provider. The organisation remains accountable for its data, access decisions, configuration and appropriate use of the service.

APIs enable systems to exchange data and invoke capabilities. API requirements should specify resources, operations, authentication, authorisation, rate limits, versioning, error responses, documentation and monitoring. Continuous integration and delivery automate parts of build, test and release. These practices can improve feedback and consistency, but require reliable automated tests, configuration control, secure pipelines and rollback capability.

AI-enabled systems introduce requirements for data provenance, model performance, human oversight, explainability appropriate to the context, monitoring for drift, bias evaluation and incident response. Analysts should identify where AI output is advisory and where it influences consequential decisions. Users need ways to challenge errors and obtain human review where appropriate. The system should record relevant decisions and model versions without collecting excessive data.

28. Synthesis: from organisational need to accepted system

Systems analysis begins with organisational purpose and a well-defined problem. Analysts identify stakeholders, investigate current processes, examine causes and evaluate alternative responses. Requirements engineering then converts validated needs into functional capabilities, quality attributes, data rules, interfaces, constraints and acceptance criteria. Models make complex behaviour and information relationships easier to inspect and communicate.

Feasibility and business-case analysis test whether a solution is technically, economically, operationally, legally and ethically viable. Prioritisation and negotiation establish a manageable scope. Prototypes and user-centred evaluation test assumptions. Traceability and change management preserve alignment as requirements evolve. Security, privacy, accessibility, resilience and compliance should influence the system from the beginning.

Implementation is successful only when the system is adopted and produces intended outcomes. Testing, migration, deployment planning, training, support and monitoring all contribute. Benefits must be measured against a baseline, and lessons should inform further improvement. The analyst's central responsibility is to maintain a defensible connection between the real organisational need, the system delivered and the outcomes experienced by people.

Structured university examination questions and model answers

Strong examination responses define the relevant concept, explain its purpose, apply it to an example and evaluate limitations or risks.

1. Define systems analysis.

Model answer: The disciplined investigation of an organisation, problem or opportunity to understand needs and determine what an information system must achieve.

2. What is the SDLC?

Model answer: A framework organising system work through stages such as initiation, analysis, design, construction, testing, implementation and maintenance.

3. Why should analysts avoid solution-first problem statements?

Model answer: They can lock the project into an unsuitable technology before the underlying causes and organisational need are understood.

4. Name four root-cause analysis techniques.

Model answer: Five whys, cause-and-effect diagrams, Pareto analysis, fault trees and direct process observation.

5. Who are stakeholders?

Model answer: People or organisations that affect, are affected by, or have an interest in a system.

6. Define requirements engineering.

Model answer: The process of discovering, analysing, documenting, validating and managing system needs and constraints.

7. Distinguish functional and non-functional requirements.

Model answer: Functional requirements specify system behaviours; non-functional requirements specify qualities and constraints such as security, performance and accessibility.

8. Give four requirements elicitation methods.

Model answer: Interviews, workshops, observation, document analysis, surveys, prototyping and system-log analysis.

9. What is a use case?

Model answer: A description of how an actor interacts with a system to achieve a goal, including main and alternative flows.

10. Explain a user story.

Model answer: A concise statement of a user capability and its benefit, usually supported by testable acceptance criteria.

11. What makes a requirement testable?

Model answer: It has clear conditions and observable outcomes that can be verified through inspection, demonstration, analysis or testing.

12. State four non-functional quality attributes.

Model answer: Performance, availability, reliability, security, usability, accessibility, scalability and maintainability.

13. What is an entity-relationship model?

Model answer: A representation of data entities, their attributes and the relationships among them.

14. Differentiate a primary key and a foreign key.

Model answer: A primary key uniquely identifies a record; a foreign key references a key in another table to represent a relationship.

15. What is a context diagram?

Model answer: A high-level data-flow diagram showing the system as one process and its interactions with external entities.

16. What is state modelling used for?

Model answer: It represents valid states of an entity and the events or conditions that cause transitions between states.

17. Identify five dimensions of feasibility.

Model answer: Technical, economic, operational, legal and schedule feasibility.

18. Explain MoSCoW prioritisation.

Model answer: It categorises requirements as Must have, Should have, Could have and Won't have for the current release.

19. Distinguish throwaway and evolutionary prototypes.

Model answer: A throwaway prototype is discarded after learning; an evolutionary prototype is progressively refined into the solution.

20. Why are interface requirements important?

Model answer: They define data exchange, timing, security, error handling and compatibility between systems and participants.

21. What is requirements traceability?

Model answer: The linking of requirements to sources, design, implementation, tests and acceptance evidence.

22. Name four data migration activities.

Model answer: Extraction, mapping, cleansing, transformation, loading, reconciliation and validation.

23. What should a supplier evaluation examine beyond feature lists?

Model answer: Security, integration, total cost, support, data portability, upgrade practices, continuity and exit arrangements.

24. Compare direct cutover and phased deployment.

Model answer: Direct cutover replaces the old system at once; phased deployment introduces the new system in controlled stages.

25. Why is user acceptance testing necessary?

Model answer: It verifies that the solution supports real tasks and satisfies agreed business and user acceptance criteria.

26. Give three examples of ethical system requirements.

Model answer: Fairness monitoring, transparency, human review, appeal routes, accessibility and limits on secondary data use.

27. What is a business case?

Model answer: A structured justification describing the problem, options, recommended approach, costs, benefits, risks and success measures.

28. Why should benefits be measured after go-live?

Model answer: To determine whether expected outcomes occurred, identify unintended effects and assign corrective action.

29. What is the purpose of a decision log?

Model answer: To preserve the rationale, owner and date of important project decisions for traceability and future review.

30. Explain the analyst's role in system development.

Model answer: The analyst connects organisational needs to system requirements, models processes and data, facilitates agreement and maintains traceability.

Chapter synthesis and revision guide

Systems analysis investigates the real organisational problem before selecting a solution. It defines boundaries, identifies stakeholders, examines processes and root causes, and evaluates options. Requirements engineering discovers and manages functional capabilities, non-functional qualities, data, interfaces, rules and constraints. Elicitation methods include interviews, workshops, observation, document analysis, surveys, prototypes and operational data.

Models such as use cases, process diagrams, data-flow diagrams, entity-relationship models and state models make requirements understandable and testable. Feasibility, business cases and prioritisation support investment decisions. Security, privacy, accessibility, ethics and resilience must shape requirements from the outset. Traceability, change control, testing, migration, deployment and benefits measurement preserve alignment from initial need to operational outcome.

Revision sequence: SDLC → problem definition → root causes → stakeholders → requirements lifecycle → elicitation → interviews → use cases and stories → functional rules → quality attributes → specifications → data models → process/data flows → state logic → feasibility → prioritisation → prototypes → architecture/interfaces → security/privacy → testing → traceability → migration → procurement → implementation → case study → professional practice → emerging technologies.

Official Education References

Use these official organisations for curriculum guidance, examinations and further learning information.