HKSM

Key Points

• Expresses variability using ranges, percentiles, confidence levels, or probability distributions.
• Supports estimating, forecasting, and quantitative risk analysis to set realistic reserves and commitments.
• Depends on clear assumptions, calibrated expert input, and relevant historical data to reduce bias.
• Common visuals include tornado charts, S-curves, and fan charts to communicate uncertainty and sensitivity.
• Can incorporate correlations and scenario sets to reflect dependencies and external conditions.
• Aim is better decision-making and transparency, not false precision.

Purpose

Make uncertainty visible so stakeholders understand possible ranges of outcomes and their likelihood. This enables balanced trade-offs, credible schedules and budgets, and informed decisions about reserves and risk responses.

Facilitation Steps

• Define the scope of analysis and list variables with meaningful uncertainty (e.g., key activities, major cost drivers, benefits).
• Agree on common terms for ranges and percentiles (e.g., P10 low, P50 most likely, P90 high) and how to treat risk events versus variability.
• Gather reference data and context (historical results, benchmarks, prior estimates, risk register entries).
• Elicit three-point estimates or ranges from subject matter experts for each variable, capturing rationale and assumptions.
• Select suitable distribution shapes based on data and expert judgment (triangular, beta, normal, discrete) and document the choice.
• Identify dependencies and correlations among variables; define correlation coefficients where material.
• Visualize the results (tornado charts, S-curves, fan charts) and review with stakeholders to validate realism.
• Record assumptions, agree on contingencies or buffers, and plan follow-up to refine inputs as new information emerges.

Inputs Needed

• Scope baseline, WBS or backlog items, and key milestones.
• Historical performance data, benchmarks, and reference class information.
• Risk register with identified threats and opportunities.
• Expert judgment from people closest to the work.
• Estimation models, productivity rates, and cost or duration drivers.
• Constraints, assumptions, and dependency maps.
• Tooling for visualization and, if applicable, simulation.

Outputs Produced

• Documented ranges or three-point estimates for selected variables and chosen distribution types.
• Correlation assumptions and any scenario definitions.
• Basis-of-estimate entries and updates to the assumptions log.
• Visuals such as tornado charts, S-curves, or fan charts for stakeholder communication.
• Simulation-ready inputs and summary statistics (e.g., P50, P80 outcomes).
• Recommendations for contingency reserves, buffers, and targeted risk responses.

Tips

• Calibrate experts using seed questions or historical comparisons to counter optimism bias.
• Use consistent percentile definitions for low, most likely, and high values across the team.
• Start simple (ranges, triangular distributions), then add complexity only if it changes decisions.
• Avoid false precision; round to meaningful units and highlight confidence levels.
• Check for double-counting between inherent variability and explicit risk events.
• Model correlations for major drivers; independent assumptions can mislead.
• Revisit representations as the project learns and conditions change.

Example

A team identifies three activities that drive schedule risk. For each, they provide three-point estimates: Task A 8-12-20 days, Task B 5-9-14 days, Task C 10-15-24 days. They choose triangular distributions, set a 0.5 correlation between A and C due to shared resources, and produce an S-curve showing total duration P50 = 42 days and P80 = 46 days. The sponsor approves a 4-day schedule buffer based on the P80 outcome.

Pitfalls

• Using single-point estimates disguised as ranges that are unrealistically narrow.
• Ignoring correlations, leading to overly optimistic aggregate results.
• Choosing distribution shapes without data or rationale.
• Building black-box models that stakeholders do not understand or trust.
• Failing to separate variability from discrete risk events, causing double-counting or gaps.
• Not updating assumptions and ranges as new information becomes available.

PMP Example Question

During a risk workshop, the team needs to capture uncertainty for key cost items in a way that can feed a quantitative analysis and be clearly explained to stakeholders. What should the project manager ask the team to provide?

1. A single number for each cost item based on expert judgment.
2. Three-point estimates with agreed percentiles and brief assumptions for each item.
3. A fixed contingency percentage applied uniformly across all items.
4. Only qualitative risk ratings for the highest-cost items.

Correct Answer: B — Three-point estimates with agreed percentiles and brief assumptions for each item.

Explanation: Three-point estimates and clear assumptions enable probability distributions and transparent communication, supporting quantitative risk analysis. Single-point values, blanket contingencies, or purely qualitative ratings do not adequately represent uncertainty.