1. Introduction
Systems development transforms the as-is (existing) system into the to-be (proposed) system.
Requirements determination
- The single most critical step of the whole SDLC.
- Changes are easy and cheap at this stage.
- > 50 % of system failures come from requirements problems.
- OOSAD’s iterative process helps: small batches of requirements are identified and implemented incrementally, and the system evolves.
Business vs. System Requirements
| Business (user) requirements | System requirements | |
|---|---|---|
| When | Analysis | Design |
| Perspective | Businessperson | Developer |
| Focus | What the system does | How it will be implemented |
- No strict boundary — they overlap and are often used interchangeably.
- Requirements are not fixed: they evolve from business capabilities to technical descriptions as the project moves Inception → Elaboration → Construction.
2. Requirements Determination
Definition (Purpose)
Convert the high-level business requirements (from the system request) into detailed requirements, which become the input for models (e.g., UML diagrams).
Definition (Requirement)
A statement of what the system must do or a characteristic it must have. It later evolves into a technical description of how it will be implemented.
Business + IT Collaboration
| Only IT analysts | Only business users |
|---|---|
| May miss real business needs | May just automate existing inefficient processes |
| System solves the wrong problem | Miss opportunities to improve with technology |
✅ Best approach: business + IT collaborate. Users don’t always know exactly what they want → analysts help discover and refine requirements.
Techniques for identifying requirements: interviews, questionnaires, observation, JAD, document analysis (detailed in §6).
Problems in Requirements Determination
- Limited access to the right users → missing, misrepresented, or over-specified requirements.
- Poor specification → unclear, incomplete, not well documented.
- Requirements not known early → understanding improves, new requirements appear, existing ones change or are removed.
- Difficulty in verification & validation — hard to ensure requirements are correct (validation) and properly defined (verification).
3. Types of Requirements
3.1 Functional Requirements
Describe what the system must do or what information it must handle — the core functionality.
Used to build the system models:
- Functional view → what the system does
- Structural view → components (classes, data)
- Behavioral view → how components interact
Examples
- The system should search available inventory. (process)
- The system should generate a report of actual vs. budgeted expenses. (information)
3.2 Non-Functional Requirements
Describe the behavioral properties the system must have — how it should perform (quality attributes: performance, usability, reliability).
- Example: The system should be accessible via a web browser.
- Mainly affect design decisions, not core functionality: database design, UI, hardware/software choices, physical architecture.
- Influence the analysis models only indirectly.
| Type | Concerns | Example |
|---|---|---|
| Operational | Environment the system runs in | OS, hardware, network |
| Performance | Speed, capacity, reliability | Response time, number of users |
| Security | Who can access, under what conditions | Authentication, authorization |
| Cultural & Political | Legal, cultural, organizational constraints | Regulations, company policy |
Wider classification (slide 16):
flowchart TD NF[Non-functional requirements] --> PR[Product] NF --> OR[Organisational] NF --> EX[External] PR --> EF[Efficiency] --> US[Usability] EF --> PE[Performance] EF --> SP[Space] PR --> RE[Reliability] PR --> PO[Portability] OR --> DE[Delivery] OR --> IM[Implementation] OR --> ST[Standards] EX --> IN[Interoperability] EX --> ET[Ethical] EX --> LE[Legislative] --> PV[Privacy] LE --> SA[Safety]
Worked example — classify each requirement (appointment system):
| Requirement | Type |
|---|---|
| Patient makes a new appointment | Functional |
| Office manager prints the daily schedule | Functional |
| System operates in a Windows environment | Non-functional — Operational |
| System stores a new appointment in ≤ 2 seconds | Non-functional — Performance |
| Only doctors can set their availability | Non-functional — Security |
| No special cultural/political requirements | Non-functional — Cultural & Political |
⚠ “Only doctors can set availability” sounds functional, but it restricts who can access a function → security (non-functional).
4. Requirements Definition
Definition (Requirements definition report)
A straightforward text report that:
- Lists all functional and non-functional requirements in outline format.
- Ranks them by importance (High / Medium / Low) or by release (Release 1, 2, 3) — important for incremental delivery.
- Feeds the next workflows: analysis models (functional, structural, behavioral) and design.
- Defines the scope of the system — its MOST important role.
- Serves as a reference to resolve confusion or disagreements in the team.
Example — sample requirements definition (appointment system):
Nonfunctional Requirements
1. Operational
1.1 The system will operate in Windows environment.
1.2 The system should be able to connect to printers wirelessly.
1.3 The system should automatically back up at the end of each day.
2. Performance
2.1 The system will store a new appointment in 2 seconds or less.
2.2 The system will retrieve the daily appointment schedule in 2 seconds or less.
3. Security
3.1 Only doctors can set their availability.
3.2 Only a manager can produce a schedule.
4. Cultural and Political
4.1 No special cultural and political requirements are anticipated.
Functional Requirements
1. Manage Appointments
1.1 Patient makes new appointment.
1.2 Patient changes appointment.
1.3 Patient cancels appointment.
2. Produce Schedule
2.1 Office Manager checks daily schedule.
2.2 Office Manager prints daily schedule.
3. Record Doctor Availability
3.1 Doctor updates schedule.Creating a Requirements Definition
- Determine the types of functional and non-functional requirements for the project.
- Use gathering techniques (interviews, observation…) to collect details.
- Work with users to validate, refine and prioritize.
- Keep refining throughout analysis.
⚠ Avoid scope creep. Don’t keep adding requirements — the system keeps growing and the project never finishes. A useful but out-of-scope requirement goes on a list of future enhancements. Managing scope is one of the hardest parts of project management.
5. Requirements Analysis Strategies
| Strategy | Core question | Change size |
|---|---|---|
| Problem Analysis | What’s wrong now? How would users fix it? | Small (efficiency, usability) |
| Root Cause Analysis | Why does the problem exist? | Small–moderate |
| Duration Analysis | Where does the time go? | Moderate |
| Activity-Based Costing | Where does the money go? | Moderate |
| Informal Benchmarking | How do others do it? | Moderate |
| Outcome Analysis | What does the customer really want? | Large |
| Technology Analysis | How can new tech add value? | Large |
| Activity Elimination | Can we do it without this step? | Large |
5.1 Problem Analysis
The most straightforward and most used technique. Ask users what the current issues are and how they’d solve them. Improves efficiency and usability.
- Examples: more space to type the customer’s address; a new report that doesn’t exist yet.
5.2 Root Cause Analysis
Focuses on why the problem exists (the cause, not just a fix).
- List and prioritize problems.
- Identify their root causes.
- Design solutions for the causes.
Problem analysis → WHAT is wrong. Root cause analysis → WHY it is wrong.
5.3 Duration Analysis
- Measure the time of each step.
- Sum them → total actual work time.
- Compare with the total end-to-end process time.
- Big gap → the process is broken. Fix by combining steps or running steps in parallel.
Worked example — mortgage approval
flowchart LR A[Data entry] --> B[Credit check] --> C[Title search] --> D[Appraisal] --> E[Approval]
- End-to-end time: 30 days
- Actual work across all steps: ≈ 8 hours (≈ 1 working day)
- Conclusion: the problem is waiting/hand-offs, not the work → improve process flow (e.g., run credit check, title search and appraisal in parallel).
5.4 Activity-Based Costing
Like duration analysis but measures cost. Find the most expensive activities and focus improvement there.
- Direct costs: labor (), materials.
- Indirect costs: rent, depreciation, other overhead.
5.5 Informal Benchmarking
Study how other successful organizations perform similar processes to learn best practices — e.g., visit another company as a customer.
5.6 Outcome Analysis
“What does the customer really want?” — the obvious answer isn’t always the real need.
Example — car insurance
- ❌ Assumed need: fast payment.
- ✅ Real need: a repaired car.
- Opportunity: repair the car directly or partner with authorized repair shops.
5.7 Technology Analysis
Identify relevant new technologies (AI, automation, cloud…), evaluate how they apply to business processes and what benefits they bring.
⚠ Don’t use technology for its own sake — only when it adds value and supports business goals.
5.8 Activity Elimination
A force-fit exercise: try to remove each step and ask “can we do this without it?” Goal: simplify the process, remove unnecessary work.
6. Requirements Gathering Techniques
Purpose: identify all requirements early (late ones are harder to incorporate) and build trust/support with users.
Who: all key stakeholders — managers, employees, customers, suppliers.
⚠ No single technique is enough — use a combination.
6.1 Interviews
The most popular technique: to know something, ask users directly.
flowchart LR A[1. Select interviewees] --> B[2. Design questions] --> C[3. Prepare] --> D[4. Conduct] --> E[5. Post-interview follow-up]
- Select interviewees — choose participants, build an interview schedule.
- Design questions
| Type | Purpose | Examples |
|---|---|---|
| Closed-ended | Specific answers | How many telephone orders are received per day? How do customers place orders? |
| Open-ended | Detailed answers, room to elaborate | What do you think about the current system? What problems do you face daily? |
| Probing | Follow-up | Why? Can you give me an example? Can you explain that in more detail? |
- Prepare — structured (planned questions) or unstructured (flexible discussion).
- Conduct — top-down (general → specific) or bottom-up (specific → big picture).
flowchart TD H["High-level (very general):<br/>How can order processing be improved?"] --> M["Medium-level:<br/>How can we reduce the number of times customers return items?"] --> L["Low-level (very specific):<br/>How can we reduce errors in order processing (e.g., shipping wrong products)?"]
Top-down reads the triangle downward; bottom-up reads it upward.
- Follow-up — document results in an interview report (person interviewed, interviewer, purpose, summary, open items, detailed notes).
6.2 Joint Application Development (JAD)
A group meeting of users and analysts led by a facilitator, with a clear agenda; in person or online (e-JAD, can be anonymous).
| Participants | Count |
|---|---|
| Users | 10–20 |
| Scribes (note-takers) | 1–2 |
| Facilitator | 1 |
- Purpose: collect requirements quickly and collaboratively; encourage discussion.
- ⚠ Challenges: people may not speak freely (e.g., in front of the boss), may dominate, or participate unequally → e-JAD reduces these.
- 💡 Success factors: careful planning (users bring documents/manuals), clear rules (respect, one speaker at a time), skilled facilitator.
6.3 Questionnaires
A set of written questions (paper or electronic).
- When: many people, different locations, need facts and opinions, external users (customers, vendors).
- Question types: closed-ended (most common), open-ended.
- ⚠ Challenges: low response rates (< 50 % paper/email, < 30 % web); no immediate clarification → questions must be very clear.
Good questionnaire design
- Begin with non-threatening, interesting questions.
- Group items into logical sections.
- Don’t put important items at the very end.
- Don’t crowd a page; avoid abbreviations.
- Avoid biased or suggestive items.
- Number questions; pretest to find confusing ones.
- Provide anonymity.
6.4 Document Analysis
Purpose: understand the as-is system.
- Review: technical documentation; user documents (forms, reports, policy manuals).
- Look for: user modifications (added fields, notes) and unused elements (empty fields, unread reports).
- Insight: documents show the formal system; real usage may differ.
Worked example — vet clinic Patient Information Card
| Observation on the filled form | Implication for the to-be system |
|---|---|
| ”Name” field holds the owner’s name, not the pet’s | Relabel as Owner’s Name |
| Staff wrote the animal’s type and birth date next to pet’s name | Add pet type and date of birth fields |
| Phone number written without area code | Make the area-code field explicit |
⚠ Warning signs: users create their own forms or combine multiple reports → the system needs improvement.
6.5 Observation
Watching how work is actually performed.
- Why: see reality, not just what people say; users forget or misreport; validates interview/JAD/questionnaire data.
- ⚠ Challenge: people behave more carefully when observed (Hawthorne-like effect).
- 💡 Best practices: keep a low profile, don’t interrupt or influence, observe daily, weekly and monthly activities.
7. Selecting the Appropriate Technique
Comparison factors: depth, breadth, integration of information, user involvement, cost.
| Interviews | JAD | Questionnaires | Document Analysis | Observation | |
|---|---|---|---|---|---|
| Depth | High | High | Medium | Low | High |
| Breadth | Low | Medium | High | High | Low |
| Integration | Low | High | Low | Low | Low |
| User involvement | Medium | High | Low | Low | Low |
| Cost | Medium | Low–Medium | Low | Low | Low–Medium |
(Standard textbook comparison; the slide highlights the key conclusions below.)
- JAD → best for integration.
- Questionnaires → best for large-scale (broad) data.
- Interviews → best for deep understanding.
- Document analysis & observation → require little training.
- JAD sessions → can be very challenging to run.
8. Alternative Techniques
| Technique | Idea |
|---|---|
| Throwaway prototyping | Small model to test ideas |
| Use cases | How users interact with the system — main technique in UML & UP, captures functional requirements |
| Role-playing CRC cards with use-case scenarios | Define classes, responsibilities, interactions → helps build functional, structural, behavioral models |
| Concept mapping | Visualize relationships between requirements |
| Story cards & task lists | Short user needs (Agile) |
Example — concept map of the appointment requirements:
flowchart TD R[Requirements] -->|include| NF[Nonfunctional] R -->|include| F[Functional] NF -->|include| OP[Operational] NF -->|include| SEC[Security] NF -->|include| PERF[Performance] NF -->|include| CP[Cultural & Political] OP --> W[Operate in Windows] OP --> B[Backup schedule daily] SEC --> D1[Only doctors set availability] SEC --> M1[Only managers produce schedule] PERF --> S2[Store appt in ≤ 2 s] F -->|include| MA[Manage Appointments] F -->|include| PS[Produce Schedule] F -->|include| RA[Record Doctor Availability] MA --> MK[Make] & CH[Change] & CA[Cancel] D1 -. impacts .-> RA M1 -. impacts .-> PS S2 -. impacts .-> MK
9. The System Proposal
Definition
The final document combining everything from planning and analysis.
| # | Section | Content |
|---|---|---|
| 1 | Table of Contents | |
| 2 | Executive Summary | Short overview so a busy manager can decide which sections to read |
| 3 | System Request | Revised: business need, project goals |
| 4 | Workplan | Revised tasks, timeline, resources |
| 5 | Feasibility Analysis | Revised: technical, economic, organizational |
| 6 | Requirements Definition | Functional & non-functional requirements |
| 7 | Functional Model | Activity diagrams, use-case descriptions, use-case diagram |
| 8 | Structural Models | CRC cards, class diagram, object diagrams |
| 9 | Behavioral Models | Sequence/communication diagrams, state machines, CRUDE matrix |
| 10 | Appendices | Survey results, interview data, industry reports… |