This guide covers an ordinary project-planning draft. Use the appropriate qualified process for safety, medical, legal, security, or other specialist assessments. The fictional case below concerns delivery of printed exhibition labels and makes no safety or compliance judgment.
Give the assessment an objective and a time boundary
The original teaching case has a narrow objective: receive 120 printed labels by November 6 for a November 8 exhibition setup. Its review date is October 24, 2026. These are authored records, not observations from a real project.
| Record | Supplied information |
|---|---|
| R1: supplier note | Printing starts after artwork approval |
| R2: artwork log | Approval is still pending |
| R3: delivery record | No committed delivery date is recorded |
| R4: team suggestion | Ask whether a smaller first delivery is possible |
The current issue is pending artwork approval. A possible future event is labels arriving after November 6. The consequence for this objective is that the planned setup may lack the complete label set.
Keep the chain conditional: pending approval may delay the start of printing, which could affect delivery. The records do not prove the deadline will be missed. They also do not prove that a smaller first delivery is available.
Distinguish identification from scoring
The UK government's Orange Book treats risk in relation to objectives and emphasizes the quality of information and uncertainty in assessment. Its government framework is contextual guidance here, not a certification of this example.
A useful draft risk statement is:
Because artwork approval is pending and printing starts after approval, the labels may arrive after November 6, leaving the November 8 setup without its complete label set. Delivery commitment and production timing remain unconfirmed.
The statement shows the mechanism and the missing evidence. “High likelihood, severe impact” would add unsupported judgments unless the team supplies a method and a basis for applying it.
Once the team has definitions and supported assessments, the risk-matrix guide covers how to combine declared ratings. That is a later task than identifying the scenario.
Copy a source-to-scenario prompt
OpenAI's prompting guidance, checked October 24, 2026, supports giving the task its context and keeping unsupported information visible. This prompt applies that approach to the fictional records:
Draft a project risk scenario for a team review, using supplied facts.
Objective: receive 120 printed labels by November 6 for November 8 setup.
Review date: October 24, 2026.
R1: printing starts after artwork approval.
R2: artwork approval remains pending.
R3: no committed delivery date is recorded.
R4: the team suggested asking about a smaller first delivery.
Separate current conditions/issues, an uncertain future event, and the
consequence for the stated objective. Link each part to its source.
Write one concise conditional risk statement. Identify missing evidence
and draft questions that would help assess the scenario.
Treat R4 as a proposed inquiry, not an agreed supplier arrangement or
an effective control. Do not invent lead times, probabilities, ratings,
loss amounts, owner acceptance, or completed actions. Do not imply the
deadline has already been missed. Return a draft for review only.
Test the causal chain with the people who know the work
Ask the artwork approver about the actual approval state. Ask the supplier, through the authorized contact, what production and delivery timing it can commit to after approval. Those answers can change the scenario; ChatGPT's rewritten sentence cannot supply them.
If the supplier confirms the labels can arrive by November 6 under stated conditions, preserve those conditions and reconsider the scenario with the team's assessment method. If November 6 passes without delivery, record the observed issue and its actual consequences. Retain the earlier risk statement as history.
End the review with the specific questions, responsible contacts, and next review date agreed by the team. Keep the maintained risk record linked to the evidence that changes it.