Institutions Rarely Collapse. They Drift.

The most dangerous period for an institution is when it still looks successful. Performance reduces curiosity. Returns delay correction. By the time the structural conditions for failure become visible, they have usually been in place for years.‍

I. The Myth of the Sudden Failure

When an institution fails, the instinct is to locate a cause. A bad hire. A market shock. A governance lapse that became public. This framing is convenient because it is containable. It implies that failure is exceptional, identifiable, andcorrectable.

Most institutional failures are not events. They are processes. A reporting line that never quite worked. A risk appetite that crept beyond what the governance framework was designed to hold. A board that received the correct information and asked the wrong questions. The failure was assembled slowly, across decisions that each seemed defensible at the time.

What appears to be a moment is actually a conclusion.‍

II. What Drift Actually Looks Like

Drift moves through institutions in patterns that are recognisable in retrospect and almost invisible in real time.

The first is tolerance creep. A standard gets set. An exception gets made once, for reasons that seemed adequate. The exception is not reversed. It becomes the new floor. Another exception follows from that floor. Over time, the original standard exists only in documentation. The institution is operating to a different standard entirely, one that was never formally adopted and cannot be located in any single decision.

The second is misalignment absorption. Most institutions contain tensions between functions, between performance pressure and structural health, between what gets reported and what is actually happening. Absorbed tension does not disappear. It accumulates and surfaces later, at greater cost, often when the institution is least equipped to handle it.

The third is accountability migration. In drifting institutions, responsibility for difficult decisions moves upward or outward. Problems get escalated, reviewed, considered, and parked. The review process becomes a holding mechanism. The problem compounds while the process continues to look functional.

Strong performance is often the camouflage under which drift accelerates. Success reduces curiosity.‍

III. The Role of Tolerated Misalignment

Misalignment between stated values and operational behaviour is one of the earliest signals of institutional drift. It is also the most consistently underweighted because it is cheap in the early stages. No immediate consequences. Performance metrics may look strong. The misalignment falls below the threshold for formal concern, so it does not trigger it. It simply persists.

Each period of tolerated misalignment makes it more embedded. People adjust their behaviour to the actual operating reality rather than the stated one. When a stress event arrives, the institution discovers it has been running on a different set of principles than the ones it believed it held.

Drift accelerates further when reputation becomes part of institutional identity. Organisations that have come to think of themselves as well-governed are the hardest to correct. The self-image absorbs the signal before it reaches the people who need to act on it.

I have seen this in institutions with strong documentation, active boards, and capable leadership. The documentation described a well-governed institution. The operating reality had moved. The gap was the risk.‍

IV. Why Governance Fails to Catch Drift

Governance frameworks are designed to catch events. They are considerably less effective at catching processes.

A threshold breach triggers a review. A single quarter of underperformance triggers scrutiny. These mechanisms work well for what they were designed to detect.

The most sophisticated drift happens through the formal amendment of the very thresholds designed to detect it. The risk limit is revised to accommodate a position already held. A reporting format gets simplified in ways that reduce visibility of emerging concentrations. A committee meets less frequently because the agenda has become routine. None of these changes would fail a governance audit. Together, they represent a material weakening of the institution's ability to recognise problems as they form.

Oversight rarely fails because it is absent. It fails because it becomes socially aligned with the behaviour it was meant to challenge. Most governance failures are social, not technical.‍

V. What Small Decisions Actually Signal

How a minor policy exception gets handled tells you more about the actual governance culture than any framework document.

The question is not whether exceptions happen. The question is what happens around them. Is the exception documented? Does it get reversed, or does it quietly become precedent?

The discipline required to document a minor exception, to flag a slight misalignment, to raise a question the room would prefer not to discuss, is the same discipline that prevents drift. Not a different capability. The same one, exercised at lower visibility.

The institutions most resistant to drift are often the least celebrated. They do not produce dramatic governance recoveries because they rarely require them.‍

VI. Recovery Is Harder Than Prevention

Once drift is embedded, recovery is a different order of problem. Prevention requires a willingness to create friction at low cost. Recovery requires confronting a gap that the institution has spent considerable time not confronting.

Structural resistance protects drift. The people best positioned to diagnose the problem are often the ones with the most substantial incentive not to. External reference points matter precisely because the internal system has already adjusted to what it is meant to correct.

When recovery happens, it is usually faster than expected. What was required was not reconstruction but realignment. The institution's structural health was better than its operating reality suggested. The drift had not destroyed the foundation. It had simply obscured it.

Institutions rarely fail without warning. They fail after the warning has been reclassified as a tolerance.