Bottom line
An assumption register is a short list of uncertain claims that a decision depends on. Give each important claim an owner, an evidence check, and a review trigger. The goal is not to remove all uncertainty, but to avoid treating a convenient guess as an established fact.
Extract the claims beneath the recommendation.
Read the proposed plan and ask what must be true for it to work. Examples include customers having a particular problem, the team being able to deliver within a deadline, or a supplier providing a needed capability. Write each assumption as a specific statement that could turn out to be wrong, rather than a vague topic such as “market demand.”
Distinguish facts, estimates, and unknowns.
An approved budget is a fact about authorization, not proof that the project will finish within it. A forecast is an estimate even when it is presented in a detailed spreadsheet. An unanswered question should remain unknown. Record where each claim came from and when it was checked. Do not upgrade an AI-generated explanation to evidence simply because it is plausible.
Prioritize impact and uncertainty separately.
Ask how much the decision would change if the assumption failed, then ask how strong the current evidence is. A high-impact claim with little support deserves attention before a low-impact detail with the same uncertainty. Simple labels such as high, medium, and low can be enough; a complex scoring model can hide subjective judgments behind arithmetic.
More on Decision Science — every week.
Join 2,400+ leaders getting the decision intelligence briefing.
Choose an evidence check with an owner.
For a hypothetical expansion into a new customer segment, an important assumption might be that buyers can approve a pilot without a lengthy procurement process. Assign someone to interview relevant buyers and inspect actual procurement requirements. Define what result would support or challenge the assumption. Set a date so “we will validate this later” does not become indefinite.
Attach a response to failure.
Decide whether a failed assumption would lead to a smaller scope, an alternative option, a pause, or a new approval. Some assumptions can be monitored during execution; others must be checked before resources are committed. Match that timing to the consequences. A cheap reversible trial can tolerate uncertainty that a binding long-term contract cannot.
Review the register at meaningful moments.
Revisit it before approval, when a major milestone is reached, and when evidence changes. Close entries that have been resolved and keep the outcome alongside the original claim. This record helps future teams understand why a decision made sense at the time without rewriting history to make the result look inevitable.
Research context
In a McKinsey survey, 72% of senior executives said bad strategic decisions were as frequent as good ones or were the prevailing norm.
McKinsey & Company: Untangling your organization's decision making (2017)This independently published statistic provides context for the topic; it does not substantiate every claim in this essay.
More on Decision Science, weekly.
One essay. Every week. No noise.