Test Planning and Design
Test Planning and Design covers how you decide what to test, how much, and with which techniques. CAST expects you to know a test plan's components, measurable entry and exit criteria, risk-based prioritisation, and the black-box techniques that turn requirements into concrete test cases.
What a test plan contains
Following IEEE 829 and ISO/IEC/IEEE 29119, a plan states scope and objectives, test items, the test approach, entry and exit criteria, deliverables, environment needs, schedule, roles, risks and contingencies, and suspension/resumption criteria. The plan is a living control document, updated as risk and scope change - not a one-off artefact written to satisfy an audit.
Design techniques
- Equivalence partitioning: one representative per class, covering valid and invalid classes.
- Boundary value analysis: for the range 1–100, test 0, 1, 2, 99, 100, 101 to catch off-by-one faults at both edges.
- Decision tables: combinations of conditions mapped to actions, collapsing to distinct rules.
- State transition: states, events, and transitions, including invalid transitions.
- Use case and pairwise: reduce combinatorial explosion while keeping interaction coverage.
Entry and exit criteria
Entry criteria might be: environment available, build smoke-tested, test cases reviewed and approved. Exit criteria might be: planned cases executed, a defined pass threshold met, and no open critical or high-severity defects. Criteria must be measurable; "testing complete" is not a criterion.
Worked example
An age field accepts 18 to 65 inclusive. How many values does three-point boundary value analysis give?
Reasoning: valid boundaries are 18 and 65. The invalid values immediately outside are 17 and 66. Three-point BVA also takes the value just inside each boundary: 19 and 64. That yields six values - 17, 18, 19, 64, 65, 66 - covering both valid edges and both invalid edges. Testing only 18 and 65 would miss the classic defect where the code accepts 17. A mid-range value such as 41 belongs to equivalence partitioning, not boundary analysis.
Traps that cost marks
- Testing only valid equivalence classes and skipping negative paths.
- Confusing boundary value analysis with equivalence partitioning.
- Writing exit criteria as a feeling rather than a metric.
- Omitting invalid state transitions from a state-transition suite.
- Treating the test plan as a static document that is never revised.