Having established that the three categories differ, the practical question is how to tell which one you are in while it is happening, when the information is incomplete and everybody is looking at you.
Classification by feeling is the default and it fails in a patterned way. Situations that are frightening but survivable get overclassified. Situations that are quiet and existential get underclassified. Both errors come from reading the emotional temperature rather than the threat.
Five types, five clocks
Crises are not interchangeable. The type determines how fast it moves, who has to be involved, and what the first correct action is.
Human. An injury, a death, a serious allegation. The fastest clock of the five, and the one where the first hour is dominated by obligations to people rather than by managing the situation. Everything institutional waits.
Operational. The plant is down, the system failed, the facility is unusable. Fast, visible, and usually the best understood, because operational failure is what organizations rehearse.
Financial. A covenant, a receivable, a funding source. Slower than it appears from outside, and characterised by a specific danger: the situation is knowable in advance and frequently is known by two or three people who are hoping it resolves.
Legal or regulatory. Runs on somebody else’s timetable, which cannot be accelerated and must not be ignored. The main failure here is treating a long clock as no clock.
Reputational. The most misunderstood. It moves fastest at the beginning, is largely settled within days, and then persists for years in a form that cannot be corrected. The window for shaping it closes before most organizations have finished deciding whether to speak.
Serious situations are frequently two or three at once, and the sequence matters: human obligations first, always, regardless of what the institutional exposure looks like.
The slow crisis
The version that does the most damage never gets declared, because it never presents as an event.
Enrolment declines three percent a year for six years. Reserves are drawn down slightly each cycle to cover an operating gap. Two experienced people leave and are replaced with less experienced ones, then two more. Each year is defensible in isolation, each is discussed at the appropriate meeting, and the organization is a different and considerably weaker thing than it was.
Nothing in the ordinary machinery detects this, because the machinery reports on the period rather than on the trajectory. A board sees this year against budget, not this year against six years ago.
The correction is mechanical: look at the five-year line rather than the annual variance, on a small number of measures, at a fixed interval. Reserves as months of operating cost. Headcount and average tenure. Whatever the organization’s core volume measure is. Presented as a trend, the slow crisis becomes visible immediately and cannot be discussed away.
What precedes it
Most crises are preceded by signals that were available and were not acted upon, and the reasons they were not are structural rather than a matter of anybody being asleep.
The signals arrive as anomalies rather than as warnings: a near miss with no consequence, a customer complaint that gets resolved, a resignation with an unusual reason given, a number that is off and gets explained. Each has a plausible explanation attached, and each explanation is offered by somebody with an interest in the matter being minor.
The organizational reason they are suppressed is covered elsewhere in this Library: raising something costs the person who raises it, and containment is rewarded more reliably than escalation.
Which means the practical instrument is not better detection. It is a channel where somebody can say I am not sure this is anything, but, and be thanked for it whether or not it turns out to be something.
Declaring it
Settle in advance who may declare, and set the threshold low enough that being wrong is cheap.
The useful formulation is a defined initial response rather than a general alarm. Declaration means the named people are called, one person takes the log, and nobody speaks externally until the situation is understood. That is a proportionate response to a possible crisis and costs almost nothing if it turns out to be a problem.
An organization that has never had a false declaration has set the threshold too high. The first genuine one will therefore be recognised late, by which time the decisions that mattered have already been taken by people who did not know what they were in.
Edited by Patrick J. Wolf, PhD