Start with consequence, not row count
- Do not treat every matrix line as equal. Prioritise sequences that affect evacuation, smoke control, suppression, access, lifts, monitoring or operator decisions.
- Group repeated low-risk outputs so witness time is reserved for complex dependencies.
Test the dependency chain
- For each scenario, identify the initiating event, fire panel logic, interfaces, receiving systems, feedback, operator action and reset path.
- A pass at the fire panel is not a pass for the building-wide response.
Control prerequisites and temporary conditions
- Record isolations, overrides, temporary software, disabled outputs and incomplete systems before the test begins.
- State whether the observed result is representative of the final operational condition.
Capture evidence that can be audited
- Use a unique scenario reference, expected result, observed result, timestamps, logs, witness names, defects and retest link.
- Photographs are supporting evidence, not a substitute for a structured record.
Classify deviations consistently
- Separate safety-critical failure, acceptance blocker, partial response, evidence weakness and minor observation.
- Avoid closing a technical issue solely because an administrative status changed.
Close the loop
- Retest against the same expected outcome after correction.
- Update the final matrix, scripts, defect register and handover record so they tell the same story.
Use proportionately. This guide supports project preparation; the applicable fire strategy, specification, legislation and standards determine the actual project requirements.