P50 vs P80: how to set a defensible completion date

Every programme baseline states a single completion date. That date is a fiction, not a lie, a fiction. It is one draw from a distribution of possible outcomes, and quoting it alone tells you nothing about how likely it is. Quantitative Schedule Risk Analysis (QSRA) replaces the single date with the distribution itself.

What does a Quantitative Schedule Risk Analysis actually do?

QSRA runs a Monte Carlo simulation: thousands of iterations of the schedule, each one sampling activity durations and risk events from defined ranges and probabilities, with correlations applied so that related risks move together. The output is an exceedance curve, the probability of finishing by any given date.

• P50, a 50% chance of finishing on or before this date. This is your working programme: the target the team plans to.
• P80, an 80% chance. This is your contractual commitment and the basis for contingency.
• P90, used for high-consequence milestones where the cost of being late is severe.

What is the most common mistake teams make with a QSRA?

Most risk registers fail not in the mathematics but in the governance. The simulation runs, the exceedance curve gets a slide, and nothing about the contract or the contingency changes. The curve becomes decoration.

Bind each percentile to an action. P50 drives the working programme. P80 drives the contractual commitment and the contingency-drawdown rules. The gap between them is the risk budget: owned, spent, and reported like money.

When the gap between P50 and P80 is eight weeks, that eight weeks is the schedule contingency. It is not a buffer to be quietly absorbed; it is a budget with an owner, drawn down against named risks, and reported every period like any other cost.

Which risk drivers should a QSRA model?

A QSRA is only as good as its inputs. The programmes that get value from it model design readiness, regulatory approvals and interface risk as first-class drivers. Not as optimistic assumptions baked into deterministic durations. Those are where real delay originates, and where the distribution's tail comes from.

How does an exceedance curve become the programme's operating manual?

Done well, the exceedance curve stops being a slide and becomes the programme's operating manual: it says which date to work to, which date to commit to, how much contingency exists, and what has to go wrong to consume it. Our Risk Intelligence engine runs three-dimensional Monte Carlo across time, cost and performance, and our planning & controls team turns the output into contingency and contractual decisions that hold.