Read the plan as a system
Four workstreams share people, data, controls, and a vendor. That coupling is where this plan gets interesting.
The board sees faster claims and service. Frontline teams see harder cases, review delays, and untracked work.
Compose the weather
Select any combination. Three or more stressors reveal compound behavior that isolated stressors miss.
Simulate the plan
640 seeded local trials, six quarters each. The distributions describe this model under stated assumptions.
No run yet. The seeded strategy is ready to test.
Model assumptions and what the numbers mean
Each trial draws deterministic, correlated operational variation across six quarters. Data defects, review load, vendor reliability, metric incentives, trust, and shadow work feed one another. Specific stressor pairs add non-linear effects. Mitigations begin only after their stated delay.
Indices run from 0–100 and are comparative severity scales, not observed rates. Dollar and month values are illustrative planning quantities, not a valuation or schedule forecast. The 10th–90th percentile band and synthetic threshold frequency show sensitivity to assumptions. No empirical calibration is claimed.
Reference thresholds: customer-harm index ≥60; adoption below 50%; control load ≥90. “Failure frequency” means trials crossing any reference threshold, not probability of real-world failure.
Trace the causal map
Select a mechanism, node, or link. Line color names the relationship; the explanation follows the active path.
Inspect seed inventory: people, systems, programs, controls
Replay the failure path
A representative synthetic trial, selected by proximity to the median outcome. It is one coherent path, not an inevitable future.
Run a scenario to reveal decision points, weak signals, and interventions.
Choose the counter-moves
You cannot buy every fix. Capacity is the binding constraint when governance is already overloaded.
A selected move takes effect after its delay. Side effects and new risks remain in the model.
Cross-examine the plan
Six parties read the same state through different incentives. Their disagreements are part of the evidence.
Metric corruption lab
A single headline KPI changes what gets optimized and what disappears from the dashboard.
Write the uncomfortable memo
The pre-mortem updates with scenario, metric, and portfolio. Claims are labeled by provenance.
Local download only. The file includes the model version and seed so the synthetic run is reproducible.