A Fire Drill Is Not the Same Thing as a Rehearsal

Almost every site we assess runs evacuation drills. Far fewer learn anything from them, because the drill is built to succeed rather than to find out what does not work.

The drill everyone runs

It is announced in advance. It happens mid-morning on a weekday, in good weather, with a full day shift present and no visitors on site. Everyone walks out of the door they came in through, stands in the car park, and goes back to work. The record says the building was cleared in four minutes and the drill is filed.

What has been proven is that people who know a drill is coming can walk out of a building. Nobody doubted that. The assumptions that would actually fail in a fire were never touched.

What a rehearsal tests instead

• A blocked route. Close the main exit before you start. The stairwell everyone uses is the one most likely to be full of smoke, and the alternative is the one nobody has walked.
• The wrong shift. Run one at night, or on a Friday, or during a shift handover. Fewer people, different supervisors, and the person who knows where the assembly point is has gone home.
• People who are not staff. Contractors, drivers waiting at the gate, a visitor in a meeting room. They are on your site and in your headcount whether or not anyone briefed them.
• Roll call that actually resolves. Not counting heads, but naming who is missing and knowing within two minutes whether they left early, are on leave, or are still inside.
• Anyone who cannot use stairs. If the plan is that somebody will help, name the somebody and check they were on shift.

The findings are the point

A rehearsal that goes badly is worth more than a drill that goes perfectly. A fire door propped open with a pallet, a muster point in the path of the prevailing wind, a warden who has left the company, an alarm that cannot be heard over plant noise in one corner of the workshop — none of these appear in a well-behaved drill, and all of them are ordinary.

Write the findings down as actions with owners and dates, in the same way you would treat an audit finding. A drill report that says only evacuation completed in four minutes is a record that something happened, not evidence that the arrangements work.

How often

Twice a year is a common minimum, but frequency matters less than variety. Two thoughtful rehearsals that each test something different will tell you more than four identical walk-outs. Change one assumption each time, and keep a note of which ones you have not tested yet.