Running a crisis simulation your team won't sleepwalk through

A crisis simulation is a timed exercise in which a response team works a realistic incident in real time, producing actual statements and decisions against a clock rather than discussing what they would do.

A simulation earns its cost by producing failures in a room where they are free. If your last exercise ended with the team broadly succeeding, it did not test anything — it rehearsed a comfortable version of an event that will not be comfortable.

The difference between a useful simulation and a pleasant one is almost entirely in the design.

Build the scenario against your actual weakness

The instinct is to pick a dramatic scenario. Resist it. The useful scenario is the plausible one you are least prepared for, which is rarely the most spectacular.

Three questions find it:

• What is the thing you would least like a journalist to ask about, that is true?
• Which part of the organisation would you least like to be the source of an incident, because you have the least visibility into it?
• Which spokesperson would you least like to be unavailable?

Build around those answers. A scenario that makes the room slightly uncomfortable during planning is the right one. If designing it is entirely comfortable, you have chosen a scenario you are already prepared for.

Injects: the timing is the exercise

An inject is new information arriving mid-exercise. Injects are where the pressure comes from, and their timing matters more than their content.

The single most instructive inject is the contradiction: something the team accepted as fact in the first twenty minutes turns out to be wrong, ideally just after they have used it in a draft statement. This tests the thing that most often causes real damage — an organisation publicly committed to a fact it had not verified.

A workable ninety-minute sequence:

• T+0 — the initial report. Deliberately incomplete and slightly ambiguous.
• T+15 — a journalist calls with a deadline. Now there is a clock the team did not set.
• T+30 — a staff member posts something on social media. Tests internal comms, which is almost always the neglected channel.
• T+45 — the contradiction. A fact from T+0 was wrong.
• T+60 — the named spokesperson becomes unavailable.
• T+75 — a second outlet arrives with a document the team has not seen.

Do not send the schedule in advance. A team that knows an inject is coming at T+45 is not being tested on anything.

Demand real outputs

The difference between a tabletop discussion and a simulation is that a simulation produces artefacts. Talking about what you would say is not the exercise; writing it, under a clock, and having it reviewed, is.

Insist on: a holding statement drafted and approved to whatever standard your plan requires, an actual decision on who speaks, a written internal message to staff, and a response to the journalist's specific question. Set the same deadlines the real event would impose and hold them. When a deadline passes with nothing produced, let it pass and record it — that is a finding, and a valuable one.

The four failure modes worth designing for

• Nobody escalates. Everyone waits for someone more senior to declare it a crisis. Test by starting with something ambiguous enough that declaring it requires a judgement call.
• Committee drafting. Four people editing one statement, forty minutes gone, nothing published. Test by imposing a deadline shorter than a committee can meet.
• Internal last. Staff learn from the news. Test with the social media inject — it exposes this immediately.
• Premature certainty. The team states a cause before it is known. Test with the contradiction inject.

Each of these is common, expensive and entirely preventable once a team has experienced it. That is precisely why simulating them is worth the afternoon.

Facilitate without rescuing

The hardest discipline is letting things go badly. When a room is floundering, the natural instinct is to offer a hint or accept a half-finished statement as adequate. Doing so converts the exercise into a demonstration.

If the deadline passes with nothing published, the deadline passes with nothing published. Note the time. That number is the most useful output of the day, and it is the one that changes behaviour — a team that has watched itself take seventy minutes to produce a holding statement does not need to be persuaded that pre-cleared templates are worth having.

A simulation where nothing goes wrong is a presentation with extra steps.

What to measure

Debriefs that ask how people felt produce agreement and no change. Measure instead:

• Time to first external statement from the initial report. The single most useful number.
• Time to first internal message. Usually much worse than people expect.
• How many drafts existed simultaneously at any point. More than one is a process finding.
• Whether the contradiction was caught before it reached a published draft.
• How long the decision on who speaks took, and whether it was ever formally made.

Then change the plan. An exercise that produces findings and no amendments to the plan has been a training event rather than a test. Most findings translate directly: pre-cleared templates for slow drafting, a lower escalation threshold for late escalation, a named deputy for spokesperson unavailability.

Individual spokesperson performance is a separate exercise — see media training — but a simulation will usually tell you which person needs it.