6
Operating hypotheses
Operations Lab · Evidence-weighted triage
Decide what to test first.
Structure the six most common explanations for an operating problem, mark what the evidence actually supports, and leave with a falsifiable first test—not a software-generated claim that the root cause has been found.
24
Evidence inputs
1
Prioritized test sequence
No signup
Evidence-weighted triage
Explicit optional submission
Working data
Assessment selections are evaluated in the current page. Nothing is submitted unless the visitor explicitly sends the optional scenario and contact form.
Decision boundary
This self-assessment ranks hypotheses and proposes a falsifier; it does not diagnose a root cause or replace direct observation and source-system evidence.
Interactive evidence workspace
Build the evidence pattern before naming the constraint.
Rate the strength of the evidence, mark the observations already supported, and use the resulting sequence to design the smallest useful falsification test.
01 · Build the evidence pattern
Rate the strength of evidence already available, then mark only the observations you can support. The output is a test-priority map—not a probability or root-cause verdict.
02 · Prioritized test sequenceEvidence required
Highest-priority hypothesisNo hypothesis ranked
Add evidence strength or a supported observation to create the first test sequence.
- 1Demand signal0
- 2Effective capacity0
- 3Quality and rework0
- 4Sequencing and flow0
- 5Decision rights0
- 6Measurement integrity0
Published scoring: 40% stated evidence strength + 60% supported observations. The score ranks what to test first; it does not estimate causal probability. An engagement validates the hypothesis against operating data before action.
Published scoring
40% stated evidence strength + 60% supported observations.
Each supported observation contributes exactly 20 points. The score ranks what to test first; it does not estimate causal probability. An engagement validates the hypothesis against operating data before action.
01
Demand signal
How strong is the operating evidence that inbound demand changed outside its normal range?
Supported observations
• The change is visible in raw inbound counts
• It begins at a nameable trigger or date
• The increase remains after seasonality is removed
First test
Build an indexed inbound trend around the trigger and compare it with the same period last year.
What would weaken it
Backlog grows while normalized inbound remains inside its historical range.
Decision this test unlocks
Whether intake controls or demand-shaping belong in the recovery plan.
02
Effective capacity
How strong is the evidence that rated capacity is not translating into available output?
Supported observations
• Actual output persistently trails rated capacity
• Absence, turnover, or downtime explains part of the loss
• Surge labor is already required to hold service
First test
Measure effective capacity by workflow stage and shift, then reconcile the result to the planning assumption.
What would weaken it
Effective output matches plan even during the periods when the queue grows.
Decision this test unlocks
Whether the intervention requires added capacity or removal of a different constraint.
03
Quality and rework
How strong is the evidence that apparent throughput includes work being done more than once?
Supported observations
• First-pass output is materially below reported output
• Rework rate increased before the queue worsened
• Defects cluster around a specific step or cohort
First test
Separate first-pass yield from gross throughput and build a defect Pareto by workflow step.
What would weaken it
First-pass yield is stable and rework consumes an immaterial share of capacity.
Decision this test unlocks
Whether quality containment produces more capacity than staffing or automation.
04
Sequencing and flow
How strong is the evidence that work is entering the right system but moving in the wrong order?
Supported observations
• Aging work exceeds its service target
• Priority rules differ by team or shift
• Work-in-process accumulates at a specific handoff
First test
Map age cohorts and work-in-process across each handoff, then compare actual sequencing with the stated rule.
What would weaken it
Age cohorts move proportionally and no handoff accumulates work faster than the others.
Decision this test unlocks
Whether queue discipline and flow controls can recover service without adding capacity.
05
Decision rights
How strong is the evidence that the operating problem changes faster than the organization can respond?
Supported observations
• Frontline reallocation requires higher approval
• Escalation time exceeds the problem’s timescale
• The same decision is repeatedly reopened
First test
Time the escalation path from detection to action and identify the first approval that does not change the risk.
What would weaken it
Operators can act inside the required timescale and escalation rarely delays recovery.
Decision this test unlocks
Which decisions should move closer to the work and which controls must remain centralized.
06
Measurement integrity
How strong is the evidence that the reported signal is materially different from operating reality?
Supported observations
• A definition, exemption, or classification rule changed
• Raw counts disagree with the management metric
• Frontline experience contradicts the reported trend
First test
Reconcile a sample period from raw events through every adjustment used in the management metric.
What would weaken it
Raw events, adjusted reporting, and frontline observations reconcile within an acceptable tolerance.
Decision this test unlocks
Whether leadership can use the current signal or must repair measurement before allocating resources.
How to use the output
Use the sequence to create a falsifiable operating test.
01
Challenge the evidence
Ask which observation is measured, which is inferred, and which would be contested by the frontline or finance.
02
Run the falsifier
Test the leading hypothesis against the condition that would weaken it before allocating resources.
03
Model the decision
If the hypothesis survives, use the Recovery Model to quantify the capacity or timing consequence.