The Evacuation Drill Measures Attendance. It Should Measure Readiness.

Most organisations run an evacuation drill at the required frequency and record that it happened. The record usually says the date, the duration, and that the site was cleared. It is enough to satisfy an obligation and almost useless as preparation, because it answers a question nobody will ask during a real emergency.

During an actual incident, the incident commander is not asking how many people reached the assembly point. They are asking who has not — because that list determines whether anyone goes back into the building, and who goes where. A headcount answers the first question. Only a named list answers the second, and those are very different operational positions.

This guide covers why manual roll call struggles for reasons unrelated to the competence of whoever holds the clipboard, what a drill should actually measure, and how to make each evacuation drill comparable to the last one so the programme improves rather than merely repeats.

The clipboard was accurate this morning. The site was not.

Manual roll call fails in a specific and predictable way, and it has nothing to do with the person performing it. The list they are working from was compiled at the start of the shift. The population of the site changed continuously afterwards.

Contractors arrived at ten. Two people left early for a medical appointment. A delivery driver is on the loading dock. A visitor is in a meeting room on the second floor, signed in at reception in a book that is now inside the building nobody may re-enter. An agency worker was added to the night shift by phone. None of them appear on the printed list, and at least one of them is unaccounted for in a way nobody can distinguish from a missing employee.

There is a second failure that compounds the first. Manual roll call is sequential — names are read, hands are raised, a marshal ticks a box. For two hundred people that takes time, and the time is spent confirming the overwhelming majority who are obviously safe and standing in front of you, before you learn anything about the handful who are not.

Starting from a live on-site count inverts that. The people who arrived are already known, so the display shows the exception rather than the roll. Effort goes immediately to the four names outstanding rather than to verifying the hundred and ninety-six who never needed verifying.

Six measurements that make a drill worth running

An evacuation drill that produces only a date and a duration cannot be compared with the previous one in any useful way. These six can, and together they describe readiness rather than attendance.

The third row deserves emphasis because it is the one most often conflated. A site can be physically clear several minutes before anyone can state with confidence that it is clear. During a real incident, that interval is exactly when someone decides whether to send a search team back inside — and every minute of uncertainty is a minute of pressure to make that decision without information.

Design the drill to fail somewhere

An evacuation drill everyone passes teaches nothing. The purpose is to find the weakness before an emergency does, which means the design should deliberately include conditions your plan has not been tested against.

One caution on realism: a drill that causes genuine alarm, or that stops safety-critical processes without preparation, creates hazards of its own. The aim is to test the plan under awkward conditions, not to simulate an actual emergency. Where a scenario cannot be run safely, walk it through as a tabletop exercise instead and record it as such.

The debrief is where the value is, and it is usually skipped

Most evacuation drill exercises end when the site is declared clear. That is roughly halfway. The findings — a route that took too long, a group that went to the wrong assembly point, a marshal who could not be reached — are worth nothing unless each one becomes a corrective action with an owner and a date, exactly as an incident finding would.

This is also the moment to compare against the previous drill, which is only possible if both were recorded the same way. A drill that produced a paragraph of narrative cannot be compared with one that produced a spreadsheet. Recording drills the same way as live events solves this and has a second benefit: when a real evacuation eventually occurs, its record sits in the same series, so you can see whether practice predicted performance.

The evidence question follows naturally. An inspector asking about emergency preparedness does not want a note that a drill occurred — they want the muster history, the time to accountability, who was unaccounted for and what was done about it. That is the seventh stage of the safety lifecycle arriving as a by-product, and it is covered in more depth in the EHS audit checklist.