HKSM

A data-driven assessment that models how identified risks and general uncertainty could change project cost and schedule, producing probabilities, ranges, and guidance for reserves and responses.

Purpose & When to Use

• Estimate the likelihood of meeting key dates and budgets using numbers, not just ratings.
• Size cost and schedule contingency based on modeled uncertainty.
• Identify which risks and assumptions drive the most variance.
• Compare alternatives or response strategies using expected value and scenarios.
• Use after qualitative risk prioritization, and on projects that are large, complex, or have tight targets or high uncertainty.
• Repeat at major milestones or after significant scope, estimate, or risk changes.

Mini Flow (How It’s Done)

• Confirm inputs and data quality: risk register with well-defined risks, schedule network, cost estimates, assumptions, and constraints.
• Build the model: add uncertainty to durations and costs (e.g., three-point estimates) and include discrete risk events with probabilities and impacts.
• Select distributions and dependencies: choose suitable probability distributions and model correlations where variables move together.
• Run analysis: use simulation for cost and/or schedule, perform sensitivity checks (e.g., tornado charts), and use decision trees or expected value where choices exist.
• Interpret results: review percentiles (e.g., P50, P80), ranges, probability of meeting targets, and the top risk drivers.
• Decide and update: propose contingency reserves, refine risk responses, adjust targets if needed, and update the risk report and risk register.
• Communicate and iterate: explain assumptions and confidence levels, then rerun when plans or risks change.

Quality & Acceptance Checklist

• Scope, schedule, and cost models reflect the current baseline and logic ties are valid.
• Key threats and opportunities are modeled, with clear probabilities and impacts.
• Uncertainty ranges for major estimates are justified and sourced.
• Chosen distributions and correlations are reasonable and documented.
• Outputs include percentiles, probability of meeting key targets, and a list of main risk drivers.
• Recommended contingency is traceable to analysis results, not arbitrary padding.
• Assumptions, data quality limits, and scenario choices are recorded.
• Stakeholders reviewed results and understand confidence levels and implications.

Common Mistakes & Exam Traps

• Confusing qualitative with quantitative analysis; quantitative uses numeric models and probabilities, not just risk ratings.
• Assuming it is required on every project; it is used when the benefit outweighs the effort and data is sufficient.
• Ignoring opportunities; include positive impacts as well as threats in the model.
• Leaving out correlations; dependencies between variables can materially change ranges.
• Using point estimates only; always include uncertainty ranges for key costs and durations.
• Misreading percentiles; P80 is not a guarantee, it is a confidence level under stated assumptions.
• Double counting reserves; contingency comes from analysis for known-unknowns, while management reserve is separate for unknown-unknowns.
• Running simulations on a weak schedule; missing logic ties or unrealistic calendars produce misleading results.

PMP Example Question

Your sponsor asks for the probability of finishing within the approved budget and a data-backed recommendation for cost contingency. You have already prioritized risks using likelihood and impact scores. What should you do next?

1. Add a fixed percentage contingency based on organizational policy.
2. Perform quantitative risk analysis using a cost model with uncertainties and discrete risks.
3. Update the risk register with more qualitative categories.
4. Request management reserve to cover unknown risks.

Correct Answer: B — Perform quantitative risk analysis using a cost model with uncertainties and discrete risks.

Explanation: The sponsor wants probabilities and defensible contingency. This requires quantitative analysis using a numerical model, not qualitative scoring or arbitrary percentages.