Stress-Test a Plan by Assuming It Already Failed
A stress test is not meant to kill the plan. It is meant to tell you where it breaks before you spend the money.
How to do it
Step 1: Turn the plan into a risk table you can attack
Write the goal, deadline, headcount, and dependencies, then feed them to “Project Risk Register” for a first table with risk, trigger, impact, probability, mitigation, and owner. Flag every cell that has no owner or no date. Vagueness is itself a risk and the easiest to miss, because nothing fails in one place; the plan simply never becomes real.
Step 2: Assume it failed with “Pre-Mortem”
Have it assume the project failed three months from now and list the five most likely reasons, each with one line on the earliest moment that reason could have been noticed. The point is not how bad the failure is, but whether you would see it in time. Anything you would notice too late needs a separate early-warning plan.
Step 3: Audit time and people with “Is the Cost Underestimated?”
For each line, ask: does this estimate include rework, review, waiting, and communication? Compare against the gap between planned and actual in your own past projects, not against a generic multiplier from the internet. With no history at all, split the largest chunk until it can be estimated.
Step 4: Find the cheapest test with “Shortest Path to Disproof”
Ask which single assumption, if false, makes the whole plan unnecessary. Spend the smallest possible amount disproving that one first. Starting with the most certain part feels productive, but it gives you no information for a decision.
Step 5: Return to the risk table and add exit conditions
Go back to the risk table from step one and use “Set Exit Conditions Up Front” to add a check date, a threshold, and a decision-maker to each risk, then write the conditions for pausing, shrinking, or stopping. A risk without a trigger is a slogan. Keep the table short; past one page usually means you are copying a template instead of thinking about your own project.
Prompts used
FAQ
Will a stress test just kill the plan?
The goal is not to kill it but to find where it breaks while changes are still cheap. Most plans survive with a different scope, order, or timeline.
How do I estimate cost with no historical data?
Use the gap between planned and actual in your last two or three projects. Even one data point beats a generic multiplier. With none, split the biggest chunk until it can be estimated.
How detailed should the risk register be?
Each risk needs a trigger signal and a concrete action. If you cannot write those, it is a true but useless sentence, and you are better off leaving it out.