Separate evidence from assumptions
Label actual observations, estimates and management decisions. Record the source and date of material inputs and make units and time periods consistent.
Compare plausible scenarios
Examine what happens when demand, pricing, cost or implementation timing changes. Scenarios reveal dependencies; they do not promise future outcomes.
Connect the model to a decision
Define which decision the model informs and which thresholds would change the recommendation. Avoid adding complexity that does not improve that decision.
Keep the model reviewable
Use a clear input structure, a version history and simple checks. A colleague should be able to trace a headline result to its underlying assumptions.
General information for discussion. Apply it to your circumstances with the relevant qualified professionals.
Discuss this with Refsnes