Estimating Duration and Resources
Estimation approaches
Analogous estimating uses a similar past project; fast and cheap, low accuracy. Parametric uses a rate (e.g. 12 screens at 6 hours each); accurate when the rate is validated. Bottom-up sums work-package estimates; most accurate, most effort. Three-point estimating handles uncertainty: PERT = (O + 4M + P) / 6, and triangular = (O + M + P) / 3. PERT weights the most likely value.
Duration versus effort
Effort is person-hours; duration is elapsed time. Duration = effort / number of resources (adjusted for productivity and availability). Adding people rarely halves duration - Brooks's law and the crashing trade-off apply. Resource calendars (holidays, part-time staff) and productivity factors (learning curve, multitasking) change elapsed time. Contingency reserves cover known-unknowns; management reserve covers unknown-unknowns.
Traps in questions
- Reporting effort as duration (40 hours of work is not 40 hours elapsed if two people share it).
- Using the triangular formula when the question says PERT.
- Assuming linear speed-up from adding staff.
- Confusing contingency (in the estimate) with management reserve (outside the baseline).
Rule of thumb
Estimate at the work-package level, record the basis of estimate, and state the confidence range. If the range is wider than ±25%, decompose further.
Worked example: A task has O=4, M=6, P=14 days. PERT = (4 + 24 + 14)/6 = 7 days. Triangular = (4+6+14)/3 = 8 days. If two people share the work and productivity drops 10%, duration is not simply halved; apply the factor to effort first.