Planning and Analysis
Decompose, estimate, sequence, baseline
Planning converts scope into executable work. Analysis converts vague needs into structured, testable statements. Both are judged on traceability: can you follow a thread from an objective to a requirement to a task to a test?
Concepts to hold
- Work breakdown. Decompose until each element can be estimated and assigned. The 100% rule (children sum to the parent) and mutual exclusivity (no overlap between siblings) are the two tests.
- Estimating techniques. Analogous (similar past work, fast, weak on differences), parametric (rate × quantity), three-point (optimistic, most likely, pessimistic), and bottom-up (accurate, expensive).
- Dependencies and sequencing. Finish-to-start is the default; check for external dependencies that no plan can compress.
- Prioritisation. MoSCoW, scoring models, or value versus effort. Prioritisation without a stated criterion is just opinion.
- Baselines and change control. A baseline is the approved reference against which variance is measured. Change control protects the baseline; it does not forbid change.
Traps
Adding contingency at every level and again at the top. Using analogous estimates where the comparison project differs materially. Assuming the critical path is the longest list of tasks rather than the longest dependency chain. Treating a priority ranking as a commitment of sequence.
Rule of thumb: three-point estimate
Expected = (Optimistic + 4 × Most likely + Pessimistic) ÷ 6. Use it to show that a spread is not the same as a single number.
Worked illustration
A task is estimated at 4 days optimistic, 6 most likely, 14 pessimistic. (4 + 24 + 14) ÷ 6 = 7 days expected. The pessimistic tail matters: contingency is justified by the spread, not by a general percentage added on top.