Unit 12: Use a PERT analysis to estimate task durations

I. Orientation: Estimating Duration Under Uncertainty

Program Evaluation and Review Technique (PERT) is a probabilistic project-planning method developed for the United States Navy’s Polaris program (1957–1958). Instead of treating a task duration as a single certain value, PERT combines optimistic, most likely or expected, and pessimistic estimates into a weighted duration. In Microsoft Project’s legacy PERT tools, the middle estimate is called the expected duration.

• Purpose: PERT produces a reasoned duration estimate when activity completion times are uncertain, especially in new, complex, or non-routine projects.
• Three-point estimation:
  • Optimistic estimate (O): The shortest realistic duration under favorable conditions.
  • Most likely or expected estimate (M): The duration anticipated under normal conditions.
  • Pessimistic estimate (P): The longest realistic duration under unfavorable conditions.
• Default weighting convention: Classical PERT assigns weights of 1, 4, and 1 to O, M, and P, respectively.
• Unit consistency: All three estimates must use the same unit, such as working days, hours, or weeks.
• Task-level calculation: PERT estimates each task separately; scheduling software then uses task relationships, calendars, constraints, and resources to calculate project dates.
• Estimate order: A logically consistent set normally satisfies O <= M <= P.
• Uncertainty assumption: The estimates describe a realistic range, not impossible best and worst cases.
• Schedule distinction: A PERT duration is not automatically the elapsed calendar time of the project because tasks may overlap, follow dependencies, or lie outside the critical path.

II. PERT Analysis: Principles and Measures

PERT analysis converts three judgments about an uncertain task into a weighted expected duration and, where required, a numerical measure of variability.

A. About the PERT analysis

The central idea of PERT is that one carefully weighted estimate is more informative than an unsupported single-value guess.

• Expected-duration formula: Classical PERT calculates the weighted mean as:

TE = (O + 4M + P) / 6

• TE = PERT expected task duration.
• O = optimistic task duration.
• M = most likely or expected task duration.
• P = pessimistic task duration.
• 1 + 4 + 1 = 6, so division by 6 converts the weighted total into a weighted average.

• Interpretation of the weights: The coefficient 4 gives the middle estimate four times the influence of either extreme. The result therefore remains close to the normal-condition estimate while still reflecting favorable and unfavorable outcomes.
• Standard deviation: Classical PERT approximates task-duration spread using:

SD = (P - O) / 6

• SD = estimated standard deviation.
• P - O = the estimated duration range.
• A larger SD indicates greater uncertainty.

• Variance: Variance is the square of the standard deviation:

V = [(P - O) / 6]^2

• V = task-duration variance.
• Variances, rather than standard deviations, are added when estimating the uncertainty of an independent sequence of critical-path activities.

• Worked example: Suppose a programming task has O = 4 days, M = 7 days, and P = 16 days.

TE = (4 + 4×7 + 16) / 6
   = 48 / 6
   = 8 days

SD = (16 - 4) / 6
   = 2 days

V = 2^2
  = 4 days^2

The expected duration is 8 days, one day above the most likely estimate because the pessimistic outcome is much farther from M than the optimistic outcome.

• PERT and the critical path: After calculated durations are entered into a network schedule, the critical path is the longest dependency path through the project. Tasks on this path normally have zero or minimal total slack, so their duration changes can affect the project finish date directly.
• Useful applications: PERT is appropriate for research, software development, system integration, prototyping, and other work where historical duration data are limited.
• Limitations: The result depends on the quality of the three inputs. Bias, hidden dependencies, resource shortages, or unrealistic extremes can produce a precise-looking but unreliable estimate.
• Software limitation: Microsoft Project’s PERT calculation primarily produces weighted task durations. It does not remove the need to inspect calendars, constraints, resource assignments, critical tasks, and schedule risk.

III. Default-Weight PERT Analysis

Using default weights means applying the standard 1:4:1 relationship without modifying Microsoft Project’s PERT weighting settings.

A. Perform a PERT analysis by using the default weights for duration estimates

A default-weight analysis requires preparing the schedule, recording three durations for each uncertain task, calculating PERT durations, and examining the resulting schedule.

• Prepare the task network: Create the project, define the working calendar, enter tasks, and establish predecessor relationships before interpreting calculated finish dates. For example, a finish-to-start dependency requires the predecessor to finish before its successor starts.

• Select suitable tasks: Apply three-point estimation to uncertain activities such as “Develop authentication module,” while using known contractual or fixed durations where uncertainty is negligible.

• Open the PERT entry interface: In Microsoft Project editions that provide the legacy PERT Analysis commands, display the PERT Analysis toolbar or open the PERT Entry Sheet. The exact menu or ribbon location depends on the Project version.

• Enter three estimates:

  • Record the shortest credible duration in Optimistic Duration.
  • Record the normal-condition duration in Expected Duration.
  • Record the longest credible duration in Pessimistic Duration.
  • Enter explicit units where necessary, such as 4d, 32h, or 2w.

• Retain default weights: Confirm that the values are:

Optimistic weight  = 1
Expected weight    = 4
Pessimistic weight = 1
Total weight       = 6

• Calculate the estimate: Run Calculate PERT. Project applies the default weighted-average calculation to each populated task:

Calculated Duration = (1O + 4M + 1P) / 6

• Worked example: For an integration task estimated at 6d, 9d, and 18d:

Calculated Duration = (1×6 + 4×9 + 1×18) / 6
                    = 60 / 6
                    = 10d

Project uses approximately 10 working days as the task duration, subject to its duration precision and display settings.

• Inspect scenario views: Legacy PERT tools may provide Optimistic Gantt, Expected Gantt, Pessimistic Gantt, and calculated PERT views. Comparing them reveals how alternative duration assumptions affect task dates and the project finish.
• Recalculate the schedule: When a task duration changes, Project recalculates successor dates according to dependencies, calendars, constraints, and scheduling mode. A ten-day task does not necessarily delay the project by ten days.
• Validate the result: Check for missing estimates, incorrect units, summary-task entries, imposed constraints, and unexpected critical-path changes. Estimates should be obtained from knowledgeable team members and supported by analogous work where available.
• Preserve original information: Calculation can replace the active duration values used by the schedule. Save a baseline or a separate project copy before calculation when the original schedule must remain available for comparison.

IV. Customized-Weight PERT Analysis

Customized weighting changes the relative influence of the optimistic, expected, and pessimistic estimates while preserving the three-point estimation structure.

A. Perform a PERT analysis by changing the way Project weights duration estimates

Changing Project’s PERT weights is appropriate when organizational evidence or risk policy justifies giving one scenario more influence than the classical default.

• General weighted formula: With user-selected weights, the calculation becomes:

TW = (WO×O + WM×M + WP×P) / (WO + WM + WP)

• TW = customized weighted duration.
• WO = optimistic weight.
• WM = expected or most-likely weight.
• WP = pessimistic weight.
• The denominator is the sum of all three weights.

• Open the weighting command: From the legacy PERT Analysis toolbar or related command set, choose Set PERT Weights or PERT Weights, depending on the Project version.
• Enter revised weights: Supply nonnegative relative values for the optimistic, expected, and pessimistic cases. The important property is their ratio; 1:4:1 and 2:8:2 express the same relative weighting.
• Calculate again: After confirming the new values, run Calculate PERT so that Project recomputes task durations. Merely changing the weights does not communicate risk unless the affected durations and schedule are recalculated and reviewed.
• Worked comparison: Consider O = 5d, M = 8d, and P = 17d.

Default 1:4:1
TE = (1×5 + 4×8 + 1×17) / 6
   = 54 / 6
   = 9d

Risk-conscious 1:2:3
TW = (1×5 + 2×8 + 3×17) / 6
   = 72 / 6
   = 12d

Raising the pessimistic weight from 1 to 3 increases the estimate from 9d to 12d, reflecting stronger concern about adverse conditions.

• Paired interpretations:

  1. Optimistic emphasis: A larger WO produces a shorter weighted duration and may suit a demonstrably repeatable process with strong productivity evidence.
  2. Pessimistic emphasis: A larger WP produces a longer weighted duration and may suit high-risk integration, unfamiliar technology, scarce resources, or strict delivery commitments.

• Expected-case emphasis: Increasing WM keeps the result closer to M. For example, 1:6:1 discounts extreme estimates more heavily than the default 1:4:1.

• Consistency requirement: Use a documented organizational rule across comparable tasks. Arbitrarily changing weights task by task can conceal assumptions and make estimates difficult to audit.

• Sensitivity analysis: Compare project finish dates under default and alternative weights. Focus on tasks that become critical, lose slack, or materially alter milestones.

• Caution: Weight adjustment does not fix poor inputs. First revise unrealistic O, M, or P values; then change weights only to represent a defensible probability judgment or risk posture.

• Modern-version consideration: Some current Microsoft Project installations do not expose the legacy PERT toolbar. The same method can be implemented with custom duration fields and a formula, spreadsheet calculation, or approved add-in, while preserving the weighted-average logic and task-duration units.