Chapter 4 – Is your schedule analyzable?
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
Before running any simulation, ask a harder question: is your schedule a network, or a drawing of one?
The distinction is simple to test. In a network, dates are consequences: task B starts when task A finishes, so changing A moves B. In a drawing, dates are decorations: bars placed by hand where someone wanted them, connected by few or no links. Both look identical on a Gantt chart. Only one can carry a risk analysis.
The test
Pick a task in the middle of your plan. Delay its finish by a month. If nothing downstream moves, you have found painted bars. Repeat with an early task feeding your main integration milestone. If the milestone holds its date regardless, the milestone is not scheduled – it is asserted.
We have run this test on real programs. One 50-task plan we imported had exactly one task on the critical path – a level-of-effort management bar spanning the whole program – while every technical task floated with hundreds of days of slack. Reviews sat at the project start date, disconnected from the phases they were meant to gate. The plan had been maintained for years. It was a drawing.
Why simulation amplifies the problem
A Monte Carlo run on a painted schedule executes without complaint and produces precise percentiles. They mean nothing. Risk propagates through links; where there are no links, a stretched task stretches alone and the program finish never feels it. The result is a suspiciously tight distribution and false comfort, delivered with three decimal places.
This is why schedule quality checks belong before probability, not after. The discipline has names: the [DCMA 14-Point Assessment](/guide.html#dcma-14-point), the [GAO Schedule Assessment Guide](/guide.html#gao-schedule-guide), NASA's schedule management practices. Their shared core is a handful of questions any tool or analyst can ask: Do tasks have predecessors and successors? Is float plausible, or are hundreds of days of slack hiding missing logic? Are hard date constraints doing a link's job? Do relationship types and lags reflect real dependencies, or workarounds?
What the checks buy you
Not compliance – information. A missing-predecessor count tells you which fraction of your plan is connected to reality. A float distribution with a long tail of 500-day slack tells you where logic was never entered. A constraint doing a link's work tells you a dependency exists in someone's head but not in the file. Each finding is repairable in minutes in your scheduling tool; the checks are a repair map, not a verdict.
An honest import report belongs in the same category. When a schedule moves between tools, some constructs may not translate. A tool that lists what it could not carry – by name, with the reason – is giving you the same gift as the quality checks: the difference between what everyone assumes the plan contains and what it actually contains. In our experience that difference, seen once, changes how a team maintains its schedule permanently.
--- In practice: run the delay test on your current plan today – one mid-plan task, plus one month. Then count tasks with more than 200 days of float. The first number tells you if the network exists; the second tells you how much of it is missing.