Open any organisation's last three fire audit reports side by side and you will usually find the same two or three items in all of them. That is not a discipline problem. It is a structural one, and it has four specific causes.
A fire safety finding recurs when any one of four elements is missing: a named owner, a target date driven by severity, a recorded root cause, and an independent verification before closure. Findings that have all four close in days. Findings missing any one of them reappear in the next audit — and the most commonly missing element is the root cause, which is why the correction gets done and the cause does not.
Three audit reports, one blocked escape route in all three. Each time it was recorded. Each time somebody moved the obstruction. Each time it came back, because nothing in the process ever asked why pallets keep being staged in that corridor.
This is the most common failure in fire compliance, and it is almost entirely structural. The people involved are not careless. The process simply has no step that forces the loop to close.
A finding recorded as belonging to "maintenance" or "the safety department" belongs to nobody. Departments do not walk to corridors; people do.
Every finding needs a named individual attached at the moment it is raised. Not to assign blame — to make it visible on somebody's list. The moment a finding has a name against it, it starts appearing in that person's alerts, their manager's escalations and the ageing report. Without one, it exists only in a document nobody re-reads.
Test your own system: pick any open finding from your last audit and ask who owns it. If the answer is a department rather than a person, you have found the reason it is still open.
A finding without a date has no way of being late, and something that cannot be late is never urgent.
But the more subtle failure is a date set by availability rather than severity. If the blocked fire exit and the faded signage board both get "next month", the system has quietly declared them equally important. They are not. Target dates should come from a severity rule applied consistently, so a serious finding cannot be scheduled behind a cosmetic one because the cosmetic one was easier to plan.
| Severity | Example | Typical target |
|---|---|---|
| Critical | Blocked escape route, dead alarm panel | Same day |
| High | Extinguisher below pressure, fire door wedged | Within 7 days |
| Medium | Refill overdue, detector due retest | Within 30 days |
| Low | Faded signage, minor housekeeping | Next cycle |
This is the one that actually drives recurrence. Someone moved the pallet. Nobody wrote down why the pallet was there.
The distinction matters: a correction fixes the instance, a corrective action addresses the cause. Moving the obstruction is a correction. Discovering that inbound stock is staged in that corridor every Tuesday because the receiving bay is undersized — and changing the staging point — is a corrective action.
Recording the cause has a second benefit that only appears over time. Once causes are captured against devices and areas, repeats become visible as a pattern rather than as unrelated notes in four separate reports. A device that has failed the same way twice in twelve months is telling you something the individual findings never could.
If the person who did the work is the person who closes the finding, the record proves only that they believe it is fixed.
Independent verification does not need to be heavy. In practice it is the safety officer walking to the device, scanning its tag, confirming the condition and closing the item — a two-minute task. What it produces is a closure that survives a question, because a second named person confirmed it at the device.
Most organisations that start measuring closure time fixate on the average. The average is the least interesting number in the report.
What matters is the tail: the findings still open past thirty days. In practice these cluster in two places — one area that everybody quietly avoids, and one external vendor whose site visits are slow. Both are solvable, but only once they are visible as a tail rather than buried in an average that looks acceptable.
A closure-time average of four days can comfortably hide two findings that have been open since March. Report the buckets, not the mean.
None of that requires software. All of it is easier with a system that enforces the four elements automatically — which is what NC and CAPA management in Fast Fire Audit does: a finding cannot be raised without an owner, cannot be dated outside the severity rule, and cannot be closed by the person who fixed it.
Raised at the device, owned, escalated and closed only after somebody else verifies it. Live on the call.
