Facts, Assumptions and Unknowns: A Decision-Record Checklist

A decision record should let someone understand why an action seemed reasonable at the time. That requires more than a conclusion and a list of links. The reader needs to see what was observed, what was assumed and which questions remained open.

Consider the difference between “the replacement will arrive Friday” and “the supplier’s email says it expects Friday delivery.” The second statement accurately identifies the source and the status of the information. It does not turn an expectation into a verified future event.

ODNI’s Analytic Standards distinguish source information from assumptions and judgments, and address uncertainty and source quality. Those principles are useful context for careful reasoning. The short business record below is BSH’s original explanatory application, not an ODNI form, a government requirement for your business or a claim of compliance. Intelligence Community Directive 203, January 2, 2015, as amended, section D.6.e(1)–(3)

Give each statement a clear role

For this checklist, use four practical labels:

• Observation or reported evidence: what you directly checked, or what an identified source reports. Make the difference visible.
• Assumption: a condition you are temporarily accepting to reason forward. State what changes if it is wrong.
• Unknown: a relevant question the available evidence does not answer.
• Judgment: your conclusion or recommendation based on the evidence and assumptions.

A source can be authentic and still contain a mistake, an estimate or incomplete information. Recording a document’s existence is different from verifying every claim in it.

A fictional decision: repair the workshop printer?

Suppose a small workshop must produce a batch of instruction sheets by Monday. Its printer has stopped feeding paper reliably. The owner is deciding whether to wait for a replacement part or arrange external printing.

A short record could say:

Decision question: Can we rely on an in-house repair for Monday’s print run, or should we reserve an outside option?

Observed evidence: In today’s local check, the printer failed to feed paper on three attempts. This establishes the observed symptom; it does not diagnose the cause.

Reported evidence: The supplier’s dated email estimates Friday delivery for the replacement part. Delivery has not occurred.

Assumption: If the part arrives Friday and resolves the fault, the workshop will have enough time to print and check the sheets before Monday. If either condition fails, the in-house plan is at risk.

Unknowns: Whether that part will resolve the fault; whether an outside printer has capacity; and when the outside booking must be confirmed.

Judgment and next action: Check the outside printer’s capacity and booking deadline before relying solely on the repair. Reconsider after the part arrives and a test print is inspected.

This example does not need a numeric confidence score. It needs a clear distinction between a symptom, a supplier’s estimate, an assumption about recovery time and the owner’s proposed action.

Keep the record short enough to use

For an ordinary business decision, start with these fields:

1. Question and deadline. What must be decided, by whom and when?
2. Evidence. What supports the decision? Record the source, date and relevant scope.
3. Assumptions. Which unverified conditions does the reasoning depend on?
4. Consequential unknowns. Which unanswered questions could change the action?
5. Judgment. What do you recommend, and how does it follow from the record?
6. Alternatives. What other practical option was considered, and why was it not selected?
7. Next verification. What will someone check, who owns that check and what outcome would change the decision?
8. Review point. When, or after which event, will the owner revisit the record?

These are suggested fields, not a claim that every decision needs the same documentation. A routine reversible choice may need a few sentences. A decision with larger consequences may need additional evidence and specialist review.

Update the evidence without rewriting history

When new information arrives, add the result and identify which assumption or unknown it resolves. Retain enough of the earlier record to explain the original decision.

If the supplier later confirms a delay, write that change explicitly. Do not silently replace “estimated Friday delivery” with a new date and leave the original recommendation unexplained. If the repair works, record the test that established the result rather than changing the earlier symptom into a diagnosis that was never verified.

Before closing the note, ask a colleague to point to the evidence behind the recommendation and name the remaining uncertainty. If they cannot, tighten the record.

See the Source Evaluation & Analytic Confidence Toolkit for the scope of the existing BSH product. The checklist and printer example here are original explanatory material, not an excerpt from that toolkit.

Source

ODNI — Intelligence Community Directive 203, Analytic Standards, section D.6.e(1)–(3), supplies the source-attribution, uncertainty and reasoning distinctions discussed above. Its applicability to Intelligence Community elements does not make this business checklist an official intelligence product.