Schedule risk analysis that changes decisions, not just decorates reports.
A completion date without a confidence level is an opinion. Quantitative schedule risk analysis (QSRA) turns a deterministic programme into a distribution — P50, P80, P90 completion positions — so contingency, commitments, and acceleration decisions rest on evidence. Done badly, QSRA is theatre; done properly, it changes what the project decides.
QSRA built on a programme worth analysing
Monte Carlo on a broken logic network produces confident nonsense. We validate the programme first (the same integrity checks as our audits), then build uncertainty ranges and discrete risk events with the people who own the work — not from a generic risk register.
Where it earns its keep
At tender: pricing time-risk deliberately across expected, accelerated, and shortest-viable scenarios. At baseline: setting contingency and interim milestone confidence. In delivery: testing recovery options for probability of success before money is spent on them.
Outputs decision-makers use
Confidence curves on completion and key milestones, tornado charts identifying which risks actually drive the tail, and plain-language findings: what to buy down, what to carry, and what the schedule can genuinely commit to.
Common questions
What is the difference between QSRA and a risk register?
A risk register lists what might happen; a QSRA quantifies what those risks and everyday estimate uncertainty do to your completion date, through the programme's actual logic. The register is an input — the confidence curve is the answer.
When should a QSRA be run?
At tender for scenario pricing, at baseline for contingency setting, and at major change or recovery decisions. Annual set-piece QSRAs that inform nothing are the theatre version — we decline those politely.
What tools do you use for schedule risk analysis?
Industry-standard Monte Carlo tooling over Primavera P6 (and MS Project where required). The tool matters less than the ranges: we build them from the delivery team's evidence, not defaults.