Owner decision
Which business problem is the first failed layer rather than the loudest symptom?
Treat the earliest condition that still explains the repeated symptom as a constraint hypothesis, not a proven root cause. Try to disprove it with a bounded test; if the symptom persists after the condition is removed, reject the hypothesis and move one layer earlier.
Use this now: Trace symptom, repeated event, immediate cause, prior failed layer, decision owner, and evidence that the layer changed.
Owner worksheet
Constraint falsification chain
| Check | Write down |
|---|---|
| Presenting symptom | State the repeated visible problem, its unit, frequency, time window, and business consequence. |
| Failed interventions | List what was already changed, when, by whom, and the result that shows the symptom persisted. |
| Upstream shared constraint | Name the earliest capacity, authority, demand, cash, information, or handoff condition shared by the failed interventions. |
| First falsifiable proof | Define the observation or small test that would disprove this constraint hypothesis before a larger fix is funded. |
| Owner and decision | Assign the test owner, due date, pass/fail threshold, and the action released by each result. |
Close the decision: Fix the constraint only after the falsification test survives; otherwise move upstream and test the next shared cause.
Worked example
Illustrative symptom-to-falsified-constraint trace
Illustrative only: Illustrative operating example only; replace inputs with verified business evidence and do not present the example as a universal benchmark or achieved result.
| Step | Illustrative input | Replace with your evidence | Completed test or status |
|---|---|---|---|
| Presenting symptom | Illustrative late-delivery rate rose from 8 of 100 orders to 19 of 100 in one declared month. | Order cohort, promised date, delivered date, and exclusion rule. | The symptom has a denominator and window. |
| Failed interventions | Overtime and a new dashboard were added; late deliveries persisted in the illustrative trace. | Change dates, affected orders, and before/after cohort. | Repeated action did not establish the producing condition. |
| First hypothesis | Initial hypothesis: poor sales forecasting causes late delivery because promised mix differs from planned capacity. | Forecast versions and capacity plan. | This is labeled a hypothesis, not a discovered root cause. |
| Disconfirming test | For two weeks, freeze accepted mix to the plan; in the illustrative completed case late orders remain concentrated after production release. | Frozen-mix cohort and release timestamps. | Result rejects forecasting as the first failed layer for this case. |
| Next constraint and owner | Test release sequencing with operations owner, due date, pass/fail condition, and no further overtime until the test resolves. | Sequencing record and decision receipt. | Decision: reject the first hypothesis and test the earlier observable release condition. |
Decision produced: Do not call the loud symptom or first plausible explanation the root cause; reject the forecasting hypothesis from the illustrative result and test release sequencing next.