In a design review (before build): Run all seven questions against the proposed design. For each question, identify gaps. A gap is not a veto — it is a decision point. Decide explicitly whether the gap is acceptable, and if so, document why.
In an incident (after failure): After stabilisation, run the questions. Q2 always applies. The failure mode you experienced is what happens when Q2’s answer is wrong. Q3 reveals what state was at risk. Q7 reveals how blind you were.
In a system you inherited: Run the questions against the system as it is, not as it was designed. The answers to Q1, Q3, and Q5 reveal what will be problems in the next twelve months.