Company twin methodology
Version company-twin-1
Define the company
Identify the company and set the evidence cutoff. Keep the proposed decision separate from the starting state.
Collect evidence
Retain dated company records and their sources. Look for customers, contracts, suppliers, capacity and cash. Record the exact passage for each selected fact.
Record gaps
Label operator statements as supplied assumptions. Leave missing facts unknown. A cited passage supports only what it says.
Freeze the starting state
Save the company, cutoff, sources and passages together. Reuse the same version for each alternative. Refreshing evidence creates a new snapshot.
Calculate the effects
Build a decision-specific graph. Link company and counterparty inputs, apply the action and calculate changes against the current plan. Mark behavioral assumptions and show unresolved effects.
Test accuracy
Freeze forecasts before outcomes are known. Compare the action taken with later records and a simple baseline. Report coverage, missed serious losses and false warnings.