How to run a fire drill

What a drill is actually testing

Not whether people can walk downstairs. A drill tests four things: whether everyone heard the alarm, whether they went to the right place, whether you could account for all of them, and how long the whole thing took. The third is the one most workplaces cannot answer honestly, because the list they are checking against was never accurate.

Before the drill

• Decide whether staff are told in advance. An announced drill tests the route; an unannounced one tests the plan. Alternate them.
• Tell your alarm monitoring company and, where relevant, the fire service — so nobody is dispatched.
• Brief the wardens, not the building. They should know their sweep areas and assembly points cold.
• Decide who is timing it, and from what moment. The alarm sounding, not the first person moving.
• Check that the on-site list wardens will use is genuinely current — including visitors and contractors.

On the day

• Sound the alarm and start the clock.
• Wardens sweep their areas, then go to the assembly point — not the other way round.
• At the assembly point, work through the on-site list and mark each person accounted for.
• Record the time when the last person is accounted for. That is the number that matters.
• Report anyone unaccounted for to the incident controller, exactly as you would in a real event.
• Do not send anyone back inside. In a drill it is a bad habit; in a fire it is fatal.

What to measure

• Time from alarm to building clear.
• Time from alarm to everyone accounted for — usually much longer, and usually the surprise.
• How many people on the list had already left the site and were counted as missing.
• How many people were on site but not on the list at all.
• Whether any assembly point was in a bad place — too close, blocked, or unmarked.

Afterwards

Write it down while it is fresh: the date, the times, who was accounted for, what went wrong and what you will change. A drill with no record is an interruption rather than a test. If your sign-in system keeps the roll-call, that record is made for you — including the timestamps, which are the part people forget to note.

The most common failure

Almost every drill that goes badly goes badly for the same reason: the list. A paper visitor book records arrivals but not departures, so it over-counts, and wardens end up hunting for people who went home an hour ago. Meanwhile the contractor on the roof never signed anything and is on no list at all. Fix the list and most of the drill fixes itself.