Where does the cost first become avoidable?
Follow a recent piece of work until you find the earliest moment someone could reasonably have prevented the loss. Check what they knew, what authority they had, and what it would have cost to act. That gives you a candidate for intervention. Reconstruct a few other jobs before assuming the same cause explains them.
Consider an illustrative commercial contractor whose crews keep arriving before a space is ready. The schedule shows idle labor. Dispatch hears about it first. The owner could buy better scheduling software. But suppose the project manager already knows the tenant is still in the space and has no clear way to suspend the mobilization. A more accurate calendar would still show a start the crew cannot make.
The testable change is to give someone authority to release a crew once access is confirmed. If access information actually arrives after dispatch, that calls for a different test. The distinction is worth settling before either change is funded.
What should the first investment prove?
Write the decision the pilot needs to inform. A useful test tells you whether a specific change reduced a defined loss under conditions you can describe. If you cannot imagine a result that would cause you to stop, the pilot is already an implementation decision.
For one group of comparable jobs, record late mobilization cancellations, idle crew time, and customer commitments affected. Agree when a release decision must be made and who can override it. Keep the schedule of starts visible. Fewer wasted trips means something different if the team simply delays all starts until it feels safe.
Meadows’s work on intervention points gives leaders a reason to compare changes to information and decision rules with changes to resources. In this example the CEO has two legitimate investments on the table. The purpose of the test is to learn which one the work calls for.
What goes on a one-page decision record?
Record the recurring loss, the decision that can prevent it, two candidate interventions, the owner and test window, and the result that would warrant expansion. Include the cost of running the test and one safeguard for the people who could absorb the burden of the change.
FAQ
Should we start with the biggest cost?
Start where the business can learn something consequential with the evidence and authority it has. A larger loss can justify investigation while a narrower test establishes the mechanism.
What if the pilot results are mixed?
Separate jobs where the proposed cause was present from jobs where it was absent. Keep a safeguard breach visible, even if the main metric improved.
When does the pilot become a budget decision?
When the owner can explain which conditions produced the result, what scaling would cost, and who would own the process in normal operations.
Sources
- Fritjof Capra, Principles of Life, for understanding outcomes through relationships.
- Donella Meadows, Leverage Points, for considering information and rules as places to intervene.