Two hours, a scenario, the people who would actually be involved, and a facilitator. The mechanics are simple and the failure rate is high, because most exercises are constructed to be completed rather than to find something.
The reassurance failure
A tabletop where everybody performs well and the plan holds has produced no information. It has consumed two hours and generated confidence, and the confidence is unearned.
This happens because of how such exercises get designed. The scenario is chosen because a response exists for it. The people attending are the ones who know the plan. Nobody is absent. The information arrives in a sensible order. Every one of those choices makes the exercise smoother and makes it useless.
The design question is the opposite one: what would have to be true for this exercise to reveal a gap? Then arrange that.
Design it to break
Remove somebody. Start by announcing that the executive director is on a flight and unreachable for four hours. This single change surfaces more than any scenario detail, because it tests whether authority actually transfers or merely appears to on paper.
Give the information in the wrong order, incomplete, and partly wrong. Real situations do not arrive as a briefing. They arrive as a phone call with three facts, one of which turns out to be mistaken. An exercise that hands out an accurate summary has removed the actual difficulty.
Introduce a second problem at the forty-minute mark. A reporter calls. A family member arrives. A second site reports something. Sequential handling is easy; simultaneous handling is what fails.
Require actual artifacts. Do not let anybody say they would notify the board. Have them write the message, in the room, in five minutes, and read it out. The gap between what people say they would do and what they can produce under time pressure is where most of the learning is.
Scenario selection
Choose something plausible for your organization and uncomfortable rather than something dramatic.
The instinct runs toward fire, flood, and active threat, because those feel like emergencies and there is usually a procedure. More useful, for most organizations, are the situations with no procedure: an allegation against a long-serving employee, a funder withdrawing with sixty days notice, a serious error discovered in work already delivered to a client, a board member speaking publicly against a decision.
These are uncomfortable to rehearse, which is precisely the argument for rehearsing them. They are also considerably more likely than the dramatic versions.
Where the scenario resembles a real situation involving identifiable people, alter it enough that it does not, and say so at the start. A rehearsal that reads as commentary on somebody present will not produce candour.
Running it
The facilitator should not be the senior person in the room. If the executive director runs the exercise, everybody performs for them and the exercise measures performance rather than capability.
Someone should keep a log, both because the log is one of the things being tested and because the record is the output.
State at the outset that finding a gap is the objective and that nothing said here is an assessment of anybody. Then hold to it, because one instance of a discovered gap being treated as a performance failure will convert every future exercise into a demonstration.
The output is a short list
Twenty minutes at the end, and the useful question is not how did we do. It is what did we discover.
What did nobody know. What did we not have. What took longer than it should have. Who did we forget to notify. Where did two people believe somebody else was handling something.
Then three to five specific fixes with a name and a date attached to each. Not a report. An exercise producing a document that nobody actions has produced nothing, and the second exercise will surface the same gaps, which is how organizations conclude that tabletops do not work.
Frequency and scale
Once a year is enough for most organizations, and once is considerably better than never. The gap between an organization that has run one and an organization that has run none is larger than the gap between one and four.
For an organization of a dozen, a compressed version works: forty-five minutes, one scenario, the same design principles. The point is not the duration. It is that the first time somebody asks who calls the funder, the answer is worked out in a room where being wrong is free.
Bodies with statutory obligations should check what their own requirements are, since some districts and licensed operations have specific exercise or drill expectations that this does not substitute for.
Edited by Patrick J. Wolf, PhD