Workbench guidance

Use the modules from left to right for the first build. Return to assumptions, line items, and scenarios whenever evidence changes. Save named versions at meaningful review points so later reports can be tied to the exact case seen by decision makers.

Data-entry conventions

• Use one concept per line item and names a non-author can recognize.
• Store cash as positive amounts; direction determines whether it enters or leaves the model.
• Use month numbers consistently: the first modeled month is month 0.
• Prefer drivers such as units × rate when they explain the amount better than a fixed total.
• Use manual monthly values only where the profile cannot be expressed clearly by frequency and growth.
• Disable obsolete items instead of reusing them for a different concept when traceability matters.
• Link high-value lines to assumptions or evidence and explain material overrides.

Quality checklist

• Decision question, owner, horizon, currency, and reserve are explicit.
• Income and all startup, operating, financing, and contingency costs are considered.
• High-impact assumptions have owners and validation actions.
• Base and downside cases are credible and materially different.
• The cash low point occurs within the chosen horizon or the horizon is extended.
• Funding sources and approval conditions cover any reserve shortfall.
• NPV, ROI, payback, and IRR are interpreted with their limitations.
• Recorded recommendation, calculated signal, rationale, risks, and conditions are internally consistent.
• AI-generated content is disclosed and reviewed by a person.
• The board report is generated from the approved saved version.

When the data cannot support a responsible choice, defer is a valid outcome. Specify the evidence or milestone needed to return to the decision rather than presenting false precision.