Working Tools
Risk tools and worksheets
Three spreadsheets that do the arithmetic described in the Risk and Uncertainty articles. Each carries live formulas rather than baked-in numbers, a legend naming the cells to edit, and one filled example row. No sign up, no email, no gate.
They are deliberately small. Each answers one question an executive has actually asked, using data the organization already holds, in about an afternoon.
Do our stated confidence levels mean anything?
Forecast Calibration Scorer
Enter the ranges you published and the outcomes that actually arrived. It returns your observed coverage against your stated confidence, separates a centring fault from a width fault, and computes both the re-centring uplift and the widening factor your ranges need.
Because an organization that has never scored its own forecasts has no evidence that its P80 is a P80. This is the cheapest real test of risk work that exists, and it runs on files you already have.
Is the risk profile deteriorating before the forecast admits it?
Contingency Drawdown Tracker
Plots contingency actually remaining against the rate at which work is actually being completed, and flags divergence. One input row per reporting period.
Contingency is consumed quietly by the people doing the work while the forecast is defended in meetings. The divergence shows up here first, and it is difficult to present optimistically because either the money is gone or it is not.
Which of our register entries can actually be modelled?
Risk Register to Model Worksheet
Takes a register row and tests it against what a simulation requires: a checkable trigger, a probability strictly between zero and one, and three ordered impact points. Tells you specifically what is missing on each entry that fails.
Most registers cannot be modelled without rewriting, and the rewrite is usually blocked on entries that have already occurred or were never written as events. This finds them in one pass.
Before you use them
These are instruments, not answers. The calibration scorer will tell you your coverage was 55 percent when you claimed 80, and it will not tell you which of your estimators is responsible or why. The drawdown tracker will show you a divergence and will not explain it. That interpretation is the work, and it is the part worth having a conversation about.
The widening factor in the calibration scorer assumes a roughly normal underlying distribution. Cost outcomes are right skewed, so treat the factor it returns as a floor rather than a final answer. And twenty observations is enough to detect a large fault while being nowhere near enough to calibrate finely: keep scoring toward a hundred before trusting the number.
Practice in the workbooks is anchored in the GAO Cost Estimating and Assessment Guide (GAO-20-195G) and ISO 31000:2018, with the calibration method resting on Lichtenstein, Fischhoff and Phillips (1982) and Brier (1950). Each workbook names its sources on the sheet.