Requirements Gathering for IT Projects: Techniques, Documentation and Traceability
Studies of failed projects repeatedly list unclear, incomplete or changing requirements among the main causes. A requirement is a condition or capability that a product, service or result must meet to satisfy a need. Gathering them well is part business analysis, part negotiation and part detective work. Even when your organisation has a dedicated business analyst, the project manager must make sure requirements work happens and is of good quality.
Types of Requirements
| Type | Describes | Example |
|---|---|---|
| Business requirements | High-level goals of the organisation | Reduce order-processing cost by 20 per cent |
| Stakeholder requirements | Needs of particular groups | Sales managers need a daily view of open quotes |
| Functional requirements | What the system must do | The system shall allow a user to reset a password by email |
| Non-functional requirements | How well the system must perform | Pages shall load within two seconds for 95 per cent of requests |
| Transition requirements | What is needed to move to the new solution | Migrate five years of customer data; train 200 users |
| Constraints | Limits on the solution | Must run on the company’s existing cloud provider |
Non-functional requirements, covering performance, security, availability, usability and compliance, are the ones most often missed, and the ones that most often cause expensive surprises at the end.
Elicitation Techniques
Interviews
One-to-one conversations draw out detail and sensitive views. Prepare open questions and ask “why” to reach the underlying need.
Workshops
Facilitated group sessions that resolve conflicting views quickly. They work best with clear agendas and decision makers present.
Surveys and questionnaires
Useful for gathering input from many users, but poorly suited to uncovering nuanced needs.
Observation (job shadowing)
Watching users work reveals tasks they forget to mention or have learned to work around.
Document analysis
Reviewing existing procedures, reports, system documentation and incident logs shows how things actually work today.
Prototyping and wireframes
Showing a mock-up draws out reactions that abstract descriptions cannot. It is especially valuable in IT because people often cannot describe what they want until they see something.
User stories and use cases
Structured ways to express requirements from the user’s viewpoint. See User Stories in Scrum for the Agile format.
Writing Good Requirements
A good requirement is:
- Clear: understood the same way by everyone.
- Testable: you can prove it was met.
- Complete: contains everything needed to understand it.
- Consistent: does not contradict others.
- Feasible: can be achieved within constraints.
- Necessary: traces to a real need.
- Prioritised: ranked for importance.
Compare these two versions:
- Weak: “The system should be fast and user friendly.”
- Strong: “A first-time user shall be able to submit a leave request in under two minutes without training, and search results shall display within three seconds for 95 per cent of queries with 500 concurrent users.”
The strong version can be tested. The weak version invites argument at acceptance.
Prioritisation: MoSCoW
MoSCoW sorts requirements into Must have (the project fails without them), Should have (important but not vital), Could have (desirable if time allows) and Won’t have this time (agreed exclusions). A common guide is to keep must-haves to about 60 per cent of effort, leaving room for the unexpected. The “Won’t have” list is valuable: it records explicitly what you decided not to do.
Documentation and Traceability
Requirements are typically recorded in a requirements document, a backlog or a requirements management tool. Whatever the format, give each item a unique ID, owner, priority and status. A requirements traceability matrix links each requirement forward to design elements, code or configuration and test cases, and backward to the business need.
| Req ID | Requirement | Source | Priority | Design ref | Test case | Status |
|---|---|---|---|---|---|---|
| R-014 | User can reset password by email | Support manager | Must | D-07 | T-031, T-032 | Tested |
| R-015 | Admin can export audit log as CSV | Compliance | Should | D-09 | T-040 | In build |
| R-016 | Dark mode theme | Marketing | Could | — | — | Deferred |
Traceability reveals gaps (a requirement with no test) and unnecessary work (a feature with no requirement), and makes the impact of change visible.
Requirements in Agile Projects
Agile teams still gather requirements, but progressively. The Product Owner maintains a prioritised backlog that is refined continuously, and detail is added just before a story is built. Acceptance criteria play the role of testable requirements; see Acceptance Criteria in Scrum and Product Backlog Explained.
Getting Sign-Off
Agree the baseline of requirements with the sponsor and key stakeholders. Sign-off does not mean requirements never change; it means changes pass through the process described in Scope Creep and Change Control. Seek agreement in writing and keep a record of who approved what.
Common Mistakes
- Gathering requirements from only one group, for example managers but not users.
- Recording solutions (“add a dropdown”) in place of needs (“choose from approved options”).
- Ignoring non-functional requirements until testing.
- Failing to prioritise, so everything becomes “critical”.
- Treating sign-off as a signature ritual rather than real agreement.
- Capture business, functional, non-functional and transition requirements.
- Combine techniques: interviews, workshops, observation and prototypes.
- Write requirements that are testable and prioritised, for example with MoSCoW.
- A traceability matrix connects needs to design and tests, and exposes change impact.