Part III – The Signals You See Before Crisis Begins

Part III | Signals Before Crisis

Parts I and II focused on what happens once the crisis has already begun – how behaviour forms under pressure, and how leadership tone becomes the operating environment in real time.

This piece steps back slightly. A crisis doesn’t begin when something breaks. It begins when the system can no longer absorb what’s happening.

Many of the same patterns are already visible much earlier – before anything has failed, and before it would be recognised as a crisis at all.

On the face of it, exercises often look like they’ve worked. The right people are in the room, the process that has been developed appears to hold, and the system recovers within an acceptable window. Nothing particularly dramatic happens, and nothing obvious presents itself as a cause for concern. By the end, there is usually a sense that, broadly speaking, things did what they were supposed to do – at least on the surface.

That is part of the problem. These moments are experienced as evidence that the system works, not as signals of how it actually behaves.

In one exercise, the team assumed the trading platform would be restored before the reporting system. It wasn’t written down anywhere – it was simply taken for granted. When someone raised it forty minutes in, the room paused. People looked at each other, and it took twelve minutes to resolve something that should have taken two.

By the end, the exercise was considered successful.

Near-misses can leave a similar impression. Something happens, it’s contained, the impact is limited, and afterwards there is often a quiet sense of reassurance that the organisation absorbed the shock, that the controls held, and that resilience is, in some way, working.

And yet, if you sit in enough of these exercises, or stay close enough to near-misses as they unfold, a different picture begins to emerge. Nothing clearly breaks, and there is rarely a single moment you can point to and say “this is the problem”. What begins to appear is more subtle than that. Not a single point of failure, but small moments where the system hesitates, where something assumed becomes something that has to be worked out in real time. The exercise continues, but not quite as expected, and the difference is easy to absorb as part of the flow rather than something worth stopping for.

They are signals – small, often ambiguous – of how the system actually behaves when it starts to come under strain.

It shows less how it was designed to work, and more how it really holds together.

That, in many ways, is the point. Things usually do move on. The exercise continues, the near-miss remains contained, and the overall outcome still appears acceptable. But it continues slightly differently from how it was expected to, and, if you are paying attention, that difference starts to matter. Because things still appear to have worked, that difference is rarely experienced as a problem.

One of the first signals sits in how quickly people begin to adapt – and more importantly, what that adaptation is doing.

Workarounds do not wait for the formal process to fail. They tend to emerge much earlier, often as soon as the system starts to come under strain. Someone bridges a gap between two teams, someone else picks up a piece of coordination that does not clearly belong anywhere, and what looked structured on paper becomes something more fluid in practice.

The system continues to function, but part of what is sustaining it is no longer what was designed.

People are stepping in to hold together something that was assumed to be in place, and the work of recovery begins to rely on that intervention earlier than expected.

It still works – just not in the way it was assumed.

The next signal is less visible, but just as telling – how time begins to behave differently. Nothing is completely blocked, and there may be no single point of failure, but progress does not move at the pace everyone expected. Steps that looked straightforward on paper take longer once they are actually attempted. Access has to be checked, then rechecked. Sequencing becomes less obvious once dependencies begin interacting with one another rather than sitting neatly on a diagram. It is not always experienced as a problem in the room, but it often carries a certain weight – a sense that things are proving more difficult to execute than they were to describe.

And eventually, the signal becomes harder to ignore. The picture never quite settles. Updates are given, but they do not always align perfectly. Different parts of the organisation carry slightly different versions of the situation, and decisions made with confidence at one moment are adjusted as new information comes in.

What begins to shift here is not just understanding, but the sense of shared reality. The situation continues to evolve, but not always from the same starting point. From the outside, it may still look controlled. From the inside, however, that control depends on continual adjustment, interpretation, and correction – and confidence becomes something that is actively constructed, rather than simply held.

Even the language in the room rarely reflects this directly. Progress is emphasised, reassurance comes naturally, and the overall tone remains steady even as small signals begin to suggest that things are not unfolding quite as expected. That does not usually feel misleading. It feels normal, which is precisely why it can be so easy to accept.

None of these moments, taken individually, are enough to trigger serious concern. When the outcome remains acceptable, the system is experienced as having worked, and the signals that emerged along the way are absorbed into that success rather than interpreted as something separate from it.

The service is still running, or it has been recovered. The exercise completes. The near-miss does not escalate. And so whatever felt slightly off fades into the background – not ignored exactly, but never fully understood for what it is.

What matters, though, is that these moments are not created by the exercise or by the near-miss. They are already there. What changes is the condition the system is operating under. There is less time, less separation between teams, and less room to pause and reset. Assumptions that usually sit quietly in the background are drawn into the open, not because anyone deliberately went looking for them, but because the system starts to rely on them more directly. What becomes visible under pressure is not something new. It is the same system, only with less room to hide.

Which means those earlier signals were never hypothetical to begin with.

They were already showing you the system as it is – just in conditions where it was still manageable.

The same patterns appear in real incidents, only more forcefully. The dependency that caused a moment of hesitation becomes critical. The workaround that quietly kept things moving becomes essential to recovery. The delay that merely stretched the timeline begins to threaten it. The slight misalignment in understanding turns into confusion that has to be actively managed. Nothing fundamental has changed in the nature of the signal. It has simply become harder to ignore.

That is why some incidents still feel surprising, even in organisations that test regularly. It is not that the signals were absent. More often, they were encountered earlier in exercises and near-misses, but in forms that were easier to absorb, easier to accommodate, and easier to explain away. The system continued to function, the outcome remained within tolerance, and confidence, understandably, held. The underlying conditions, however, often remained largely untouched.

Exercises and near-misses are therefore often treated as confirmation that things are working and, in many cases, they are. But they are also something else. They are moments in which the system begins to reveal what it actually depends on: where the edges are less clear than assumed, where coordination relies more heavily on people than design, and where recovery is possible, but not quite as straightforward as it appears.

They are not failures, but nor are they clean confirmations.

Which means the difference between an exercise and a crisis is often not the presence of fragility, but whether it can still be absorbed into a successful outcome – or whether it can no longer be ignored.