When warning isn't the problem

There is a comfortable story we tell about intelligence disasters: the signal was faint, the dots were scattered, no one could reasonably have joined them. It is comfortable because it implies the fix is more – more collection, more coverage, more data. Sometimes that story is true. Often it is not.

In several of the most consequential surprises of the last two decades, the warning existed. It was specific. It was timely. It named what was coming. And it still did not become a decision.

Anatomy of a compound failure

Consider the shape of it. A foreign liaison service passes a detailed warning that names the group, its leader and its intended targets. The warning reaches the head of the domestic service. Days later it is put in writing. On the morning itself, a final, urgent warning arrives. Between that last warning and the first explosion there are roughly two hours – and no standing protocol to trigger, no emergency line to pull.

Every individual link in that chain can be staffed by competent people acting in good faith, and the chain can still break – because the failure is not located in any one person. It is distributed across the seams: analysis anchored to yesterday's threat; dissemination that moves an urgent warning as routine paperwork; agencies that each hold one corner of the picture and never assemble it; an executive tier too fractured to convene. Collection succeeded. Coherence failed.

Why competence isn't enough

The uncomfortable implication is that you cannot train your way out of a structural problem. An analyst who escapes the dominant frame, an officer who escalates with urgency, a duty principal who improvises a protective posture – each can lift the odds. But in a fragmented system the best available individual play still tops out below prevention, because the structure keeps re-imposing the friction the individual is fighting.

This is why "lessons learned" exercises that end in personal accountability so often fail to prevent the next failure. Naming who did not act is not the same as fixing the system that ensured action could not travel. The remedy is structural: a standing escalation duty, a fusion function that joins liaison reporting to open domestic cases by default, an executive tier insulated from political capture, and an analytic cadre sustained well enough to think.

The cascade beyond the event

Fragmentation rarely stays contained to a single event, either. A service built to watch inward, unable to read a shifting threat environment, tends to be equally unprepared for the crises that follow – economic, informational, maritime. The same seams that lost one warning lose the next.

What this means for leaders

If your instinct after a near-miss is to buy more collection, pause. Ask a different question first: can a specific warning entering any node of your enterprise reach the person who must act, in time, by default? If the honest answer is "only if the right individual happens to force it," you do not have a collection problem. You have a coherence problem — and coherence, unlike luck, can be measured and rebuilt.

That is the premise behind the Intelligence Coherence Index: put a number on whether warning becomes decision, before the next warning arrives.