When Governance Can No Longer See the Institution

There are moments in an institution when governance is still active, yet strangely disconnected from what the organisation is actually doing. People meet, papers move, controls exist, but the sense of oversight has thinned. Governance is present, but its view of the institution has become narrow.

You notice it in conversations. Someone presents a decision, and the discussion stays on the surface. The numbers are there, the slides are complete, the format is familiar — but the reasoning behind the decision is hard to find. A policy is renewed even though no one remembers the situation it was originally built for. A plan is approved because it fits the template, not because it fits the world the institution now operates in.

Inside the organisation, signals begin to lose clarity.
A metric shows movement, but no one can explain what the movement means.
A process continues, but its purpose is no longer visible.
A document summarises actions, but the intention behind them is missing.
Governance tries to interpret these signals, yet they no longer reveal what they once did.

Where focus shifts away from reality

A common sign is that discussions turn into debates about roles rather than tasks or results.

Who signs?
Who owns?
Who approves?
Who is accountable?

The conversation becomes about the choreography of responsibility instead of the substance of what is being produced.

You see it everywhere:

• A committee spends half its time debating whether a decision belongs to IT, Business, or Risk, while the actual decision remains untouched.
• A project review focuses on whether the RACI is correct, not whether the project is delivering anything meaningful.
• A team argues about who should “own” a KPI, even though no one can explain what the KPI measures.
• A policy update stalls because departments negotiate ownership paragraphs, not because the policy itself needs improvement.

When governance loses sight, roles expand and outputs disappear.
People become guardians of boxes on a chart rather than stewards of actual work.

IT governance examples — where sight is lost fastest

IT governance is often the first place where this narrowing becomes visible.

• An architecture board rejects a solution because it doesn’t fit a diagram drawn five years ago, even though the diagram no longer reflects the system landscape.
• A change advisory board spends more time discussing who should approve a change than what the change actually does.
• A security committee insists on a control that everyone knows is symbolic, but removing it would require rewriting a policy no one wants to touch.
• A data governance council debates ownership of a dataset that no one uses, while critical datasets have no owner at all.

The institution starts to treat governance artefacts as if they are the system.
The real system — the one running in production — becomes secondary.

PMO theatre — governance without sight

PMO theatre is another sign that governance has lost its ability to see.

• Status reports are polished until they say nothing.
• Risks are written in abstract language that avoids naming anything real.
• Milestones are adjusted to match the reporting cycle, mostly the annual or quarterly budget cycle, rather than the work.
• Dashboards show green because no one wants to explain amber.
• Steering committees receive updates that describe activity, not meaningful progress.

The PMO becomes a stage where everyone performs certainty.
The actual work happens somewhere else.

Where bureaucracy thickens

Bureaucracy thrives in the spaces where understanding has faded.

• A simple change request requires eight signatures because no one knows who actually owns the decision.
• A team spends two weeks preparing a report that no one reads, because “it’s required for governance”.
• A policy update takes six months because every department wants to add a paragraph to protect itself.
• A project manager spends more time updating trackers than working on the project.
• A committee asks for “one more version” of a document, even though the content hasn’t changed — only the formatting.
• Real risks are not communicated, trivial risks are because they do not cause controversy

People start spending more time proving that something happened than understanding what happened.
The bureaucracy is not a malfunction.
It is what institutions create when they can no longer see themselves clearly.

The reflection problem

The institution begins to behave like a system looking at its own reflection.
Governance reads the reflection — the templates, the reports, the procedures — and assumes it is reading the institution.
But reflections only show surfaces.
The real activity, the real meaning, the real consequences sit behind them.

Systems don’t fix this.
They work with what is explicit.
If the institution cannot see itself, the systems cannot see it either.
They follow the visible structure, even when the visible structure no longer matches the underlying reality.

The quiet disconnect

Governance believes it is steering.
The institution believes it is being steered.
But they are not looking at the same thing.

Restoring sight

The real work is to restore the institution’s ability to observe itself.
That requires definitions that mean something, rationale that can be retrieved, ownership that is clear, and signals that show structure rather than activity.
It requires documentation that explains instead of records.
It requires governance that sees, not governance that verifies.

Governance only governs when it can see.
When it cannot see, bureaucracy takes over.
And once bureaucracy takes over, the institution begins to move without understanding where it is going.

Institutions that cannot observe themselves cannot steer.
And institutions that cannot steer eventually lose the ability to choose.

Final paragraph to add in AI times

There is one place where this loss of sight becomes unmistakable: the old split between IT and Business. It was never a natural boundary; it was a bureaucratic arrangement that made sense on paper and slowly hardened into an organisational truth. Over time it became a wall. Decisions are framed in terms of who owns a system rather than what the system produces. Governance debates roles instead of outcomes. In the context of AI this divide becomes impossible to maintain, because AI does not live on one side of the wall. It sits inside the institution’s logic, inside its decisions, inside its processes. When an organisation keeps treating technology as something separate from its own reasoning, it cannot see what it is building. And when it cannot see what it is building, it cannot govern it.
We will deepen this in a future essay.