Chapter 6 – Sensitivity: measured, not assumed
The Quantitative Schedule Risk Analysis Handbook · Edition 1, August 2026 · by the team behind [Epoch SRA](https://epochsra.com), an Excel add-in for schedule risk analysis · corrections welcome at contact@epochsra.com
Two rankings of the same risks. The first: sorted by likelihood × severity, the register's own arithmetic. The second: sorted by how much each risk actually moves the program finish, measured across thousands of simulated futures. If the two rankings matched, the second would be redundant. In our experience, they do not match – and the differences are where the money is.
Why the rankings diverge
Likelihood × severity is a statement about a risk in isolation. The finish date is a property of the whole network. Between the two sit three amplifiers the register cannot see. Position: a risk feeding a zero-float chain into the integration event transmits every day of impact; the same risk on 200 days of float transmits nothing. Merge structure: a risk on the fifth parallel path into a milestone attacks a merge point already eroded to 59% – its marginal damage is outsized. Correlation with criticality: some tasks are only critical in the bad futures, exactly when risks fire; a static critical path never shows them.
How measurement works
The clean method uses the simulation's own bookkeeping. For each risk, partition the iterations: those where it fired, those where it did not. Compare the finish distributions of the two groups – the difference at the percentile you care about (say P80) is that risk's measured contribution, in days, network effects included. No second model, no assumed elasticities: the same futures, sorted two ways.
Presented as a tornado – bars ordered by contribution – this becomes the most decision-dense chart an SRA produces. It answers the only question a mitigation budget cares about: if we killed this risk, what would we actually buy?
One honesty rule for the method: some risks fire too rarely in a finite run to measure. The honest rendering is "unmeasurable", never zero – zero is a claim, and an unearned one.
The row that wins the meeting
On our demo program, the largest single mover of the P80 finish is a comms-subsystem risk that sits on the critical path in only 19% of futures. Sorted by criticality it is row twenty; a critical-path review never discusses it. Sorted by measured contribution it is row one, at over 130 days of P80 impact. That inversion is not a curiosity – it is the routine product of position and merge structure, and it is invisible to every method short of measurement.
The practical consequence: mitigation money allocated by register score is systematically misallocated. Not because the scores are wrong, but because the scores answer a different question than the one the program finish asks.
--- In practice: before your next mitigation review, write down which risk your team would fund first. Then measure. The comparison of those two answers – gut ranking versus computed contribution – is the fastest credibility test an SRA can pass in front of a skeptical room.