Emergency response flowchart: decision tree template
A decision tree for the first minutes of a workplace emergency: who is at risk, is it safe to approach, evacuate or shelter, and who gives the all-clear.
How it works
Rename the lanes to your decision rights
Replace first person on scene, warden or first aider, incident controller and emergency services with the roles you actually staff. Keep the lanes to who answers each question, not who does the work — that is what makes it a decision tree. If nobody holds the incident controller role at night, at weekends or on a lone-working shift, the empty lane is the finding, and naming a deputy is the fix.
Write your own safe-to-approach rule
Open the Is the area safe to approach? decision and replace the generic list with the hazards you hold: the substances on site, the plant that stays energised, the areas that need breathing apparatus, the spaces that need a permit. This is the decision that produces second casualties, so it should be answerable by someone with no time to think and no training in the specific hazard.
Set the evacuate versus shelter criteria for your site
Decide in advance which events send people out and which keep them in, and write it next to the decision. Evacuation assumes the escape routes are safer than staying; shelter in place assumes the opposite, which is usually an external plume, severe weather or a security threat on the approach. Name the refuge, the assembly points and the alternative assembly point for when the primary one is downwind.
Fix the roll-call source and who holds it
Point the Is everyone accounted for? step at a list that includes contractors, visitors, drivers and anyone working alone, and say who carries it to the assembly point. A roll call taken from a staff list is how a visitor is left unaccounted for, and a roll call held only on a server inside the building is no roll call at all.
Name who gives the all-clear and who may not
State on the reoccupation decision who is authorised to release the site. Where the emergency services have attended, re-entry is normally their call rather than the site's, and a silent alarm panel is not an all-clear. Add the Site not released branch to your plan properly: where people go, who tells them, and who speaks to families and to the media.
Add your reporting duties, then keep the chart approved and exercised
Every branch ends in a report, including false alarms and shelters that were stood down. Record what has to be captured — times of discovery, alarm, call, evacuation and all-clear, the roll-call result, injuries and near misses — and the external reports you owe a regulator or insurer, whose scope and timescales vary by jurisdiction. Then put the chart through approval, date it, and re-run it after every drill and every real event.
Frequently asked questions
How is an emergency response flowchart different from an emergency response process map?
A process map answers "what happens next, and who does it". It moves through stages and role lanes, and it is the right shape for the hours and days after an event: containment, communication, investigation, corrective action, closure. A decision tree answers "which option do we choose, and who is authorised to choose it". Its spine is a chain of questions, its branch labels are the answers, and its paths end in different outcomes rather than one. Emergencies need both, but not on the same page: this chart covers the minutes when the response is still being decided, and the end-to-end sequence that follows is a separate map.
When should you evacuate and when should you shelter in place?
Evacuate when the hazard is inside and the escape routes are usable — fire, smoke, internal release, structural damage. Shelter in place when the outside is the more dangerous side: an external gas or chemical plume, severe weather, or a security threat on the approach route. The failure mode is deciding it on the spot without criteria, which is how people are sent out into the plume they were being protected from. Write the trigger conditions next to the decision, decide it once, announce it once, and make sure every warden hears the same instruction.
Should employees tackle a fire or a spill themselves?
Only when all of a small number of conditions hold at once: the hazard is small and not spreading, the person is trained for that specific hazard, the correct equipment is within reach, and there is a clear escape route behind them. Failing any one of them makes the answer no, and the correct response is to withdraw, secure a cordon and call the emergency services. On-site intervention is a permission that has to be granted in advance to named, trained people, not a judgement made by whoever arrives first.
Who gives the all-clear after an emergency?
Not the alarm panel, and not the person who is keenest to get back to work. Where the emergency services have attended, releasing the site is normally the decision of the officer in charge, with the incident controller then standing the site down. This chart draws that as its own decision with two outcomes, because "site not released" is a real state that needs planning for — somewhere for people to go, someone to tell them, and a decision about the rest of the shift — and treating it as an afterthought is why crowds drift back to the doors.
Does an incident report have to be filed if nothing happened?
Yes. Every branch of this chart ends in a report, including the false alarm, the small fire put out with an extinguisher and the shelter that was stood down after twenty minutes. The report is what turns a response into evidence: it fixes the times of discovery, alarm, call, evacuation and all-clear, the roll-call result, and the near misses nobody would otherwise record. Some of those events also carry an external reporting duty to a regulator or insurer, and the scope and timescales for that differ by jurisdiction, so state on your own version which events trigger which report.