Software Project Management and Estimation
Domain 7 tests planning, estimation, scheduling and progress tracking. Expect arithmetic: PERT, float on a network diagram, or earned value indices that you must interpret rather than just compute.
Estimation techniques
Analogy compares with a similar past project. Expert judgment, often via Delphi to converge anonymously, suits novel work. Bottom-up estimation decomposes a work breakdown structure into tasks and rolls up; top-down allocates from a total. Three-point PERT uses E = (O + 4M + P) / 6 with standard deviation approximately (P - O) / 6. Story points measure relative size against a team-specific velocity, while ideal days measure effort; points are not comparable across teams. The cone of uncertainty narrows as the project progresses. COCOMO derives effort from size in KLOC or function points adjusted by cost drivers.
Scheduling and tracking
A WBS decomposes deliverables. The critical path is the longest dependency path through the network; its tasks have zero float, so any delay there delays the project end. Float equals LS - ES, equivalently LF - EF. Earned value: SV = EV - PV, CV = EV - AC, SPI = EV / PV, CPI = EV / AC; an SPI or CPI below 1 means behind schedule or over budget. Risk management cycles through identification, analysis as probability times impact, response planning (avoid, mitigate, transfer, accept), and monitoring, all recorded with owners and triggers in a risk register.
Worked example
A task has optimistic 4, most likely 6 and pessimistic 14 days. PERT expected duration = (4 + 4x6 + 14) / 6 = 42 / 6 = 7 days, and sigma is about (14 - 4) / 6 = 1.67 days. Reasoning: weighting the most likely value four times pulls the mean away from the pessimistic tail, while sigma quantifies uncertainty for contingency sizing. Because this task lies on the critical path with zero float, a one-day slip moves the project end by one day, so the response is a schedule buffer or resource action, not just a register entry.
Traps that cost marks
- Adding people to a late project; Brooks's law says communication overhead grows roughly with the square of team size.
- Assuming the critical path never changes after re-estimation or a slip.
- Reading SPI above 1 as unambiguously good; it can reflect under-scoping or early progress claims.
- Padding every task and then adding a project buffer, double counting contingency.
- Comparing velocity across teams as if the units were universal.