Assumptions and evidence

An assumption is a statement that must be true, or sufficiently close to true, for the forecast to hold. Typical categories include demand, price, enrollment or utilization, capacity, hiring, wage rates, launch timing, construction, permits, grants, contracts, inflation, financing, and residual value.

For each material assumption record a name, description, unit, base/downside/upside values when applicable, confidence, impact, status, source type, owner, validation action, and evidence links. Confidence describes how well the assumption is known; impact describes how much the decision could change if it is wrong. Low confidence plus high impact is an executive review priority.

Evidence may be a historical record, signed quote, contract, reference, workspace file, market source, expert note, or documented calculation. A link alone is not enough: state what it supports and whether it is current and comparable. Mark an assumption validated only when the responsible person has checked the evidence.

AI-supported drafting

AI can propose a checklist of commonly missed assumptions, extract candidate facts from available sources, challenge internal consistency, and draft validation questions. It cannot know organization-specific amounts without reliable evidence. AI-generated assumptions must remain marked as drafts until a person confirms their meaning, value, timing, confidence, and source.

Connect important line items to the assumptions that drive them. This makes a later change explainable and helps the board distinguish measured facts from management judgment.