Most transformation offices fail the same way: they slowly turn into a reporting layer. They start as an engine and end as a spreadsheet. The people inside are competent and busy, which is exactly what makes the drift so hard to see.
Collecting status is not driving change
The failure pattern is predictable. The office is asked to give leadership visibility, so it builds templates. The templates need filling, so it chases owners. Chasing becomes the job. Within two quarters the office knows the status of everything and the outcome of nothing. It has become a very expensive way to find out that a project is late.
Visibility is not the deliverable. A program does not move because someone wrote down that it is amber. It moves because a blocker was cleared, a decision was forced, a dependency was resequenced. If the office cannot point to things that finished because it existed, it is overhead wearing a lanyard.
A transformation office should be judged by what it unblocks, not by what it reports. If you removed it tomorrow, would delivery slow down, or would the calendar just get quieter?
A useful office owns friction, not slides
The version that works does fewer things and harder ones. It owns the cross-cutting problems no single project can solve: contested priorities, shared dependencies, the decisions stuck between two executives who will not be in the same room. It runs the operating rhythm so leaders decide on time. It protects the small number of metrics that actually signal progress, and it kills work that no longer earns its place.
That office is smaller than the reporting kind, and far more senior. It needs people who can walk into a stalled situation and change it, not people who can format it.
The test I use is blunt. Reporting offices describe the weather. Useful ones change it.