Give forecast assumptions an evidence review date

A model can contain today's actuals and last quarter's unsupported assumptions. Updating the transaction feed does not reconfirm an equipment quotation, hiring expectation, implementation slot or customer launch date. Review the inputs that support the next decision, especially where their validity depends on a condition that can change before approval.

ICAEW's spreadsheet principles address input quality, documented assumptions and appropriate testing. An assumption review date is a practical control built on those ideas. It records when the evidence should be checked again, rather than implying that a professional body prescribes one universal expiration period for every forecast input.

Distinguish the date information was received from the period it covers and any stated validity limit. A recently forwarded quotation can still be old. A current estimate may apply only to a particular specification, volume or delivery window. Store those conditions beside the amount used in the model.

For a supplier quotation, use its actual validity terms and the next purchase decision. For an operating estimate, use a relevant change such as a revised specification or staffing pattern. For a management judgment with no formal expiry, choose a review date based on its sensitivity and the pace at which its supporting evidence can change.

Suppose a hypothetical quote is valid for 30 days, but management expects to approve the purchase on day 45. The current model can show the quoted amount as a reference, while flagging that the approval will need a renewed price and delivery confirmation. It should not silently treat the earlier quote as an available offer at the later decision date.

A trigger can arrive before the scheduled review. A customer changes scope, a key employee leaves or an operating test fails. Have the assumption owner notify finance when the condition changes, rather than waiting for a monthly process to discover that the model's premise no longer holds.

Keep the last supported value visible and identify the evidence gap. If a reasonable range can be defended, test that range and show whether it changes the recommendation. If no credible replacement estimate exists, state which conclusion remains conditional. Replacing an expired quote with an arbitrary increase creates a new unsupported assumption.

Prioritize inputs by decision sensitivity. A stale minor travel estimate may have little effect on the preferred option, while an uncertain implementation date can change the entire benefit period. Ask the reviewer to explain the consequence of the uncertainty, not merely count overdue assumption records.

Do not delete an assumption because its evidence needs renewal. Preserve the previous value, source and review outcome. A changed price and a changed delivery promise may move different parts of the model, and the history helps explain why the approved case differs from an earlier proposal.

Before the commitment, review the small set of assumptions on which approval depends. Record who reconfirmed each one, what evidence they used and whether the model changed. An owner's acknowledgement is useful only if it addresses the actual condition, such as the available capacity or current commercial terms.

If management proceeds with unresolved evidence, document the chosen assumption and the action required if it proves wrong. That may involve a conditional purchase, a staged commitment or a later decision. Finance should show the consequences of those choices without implying that a model can establish contractual rights.

At handover, include the evidence location, owner, next review date and event triggers for material assumptions. The next preparer should be able to distinguish a confirmed input from a provisional judgment. This makes the forecast maintainable after the person who assembled the original evidence has moved on.