The pattern is familiar to anyone who has run a multi-site estate. A programme lands, performance improves for two quarters, and then the numbers drift back toward where they started. Nobody sabotaged it. The organisation simply returned to the routine it actually runs on.
Transformation does not fail at design stage. It fails at the point where a new way of working has to survive a short-staffed Saturday.
The three failure modes
First, the model is documented but not deployable. A target operating model that requires a manager to read forty pages will not be used at 7am during a delivery.
Second, accountability is described rather than assigned. If a measure has no single named owner at site level, it belongs to nobody.
Third, the management layer was never built. Programmes routinely invest in process and systems while assuming the general manager population can absorb the change. That assumption is the most expensive one in the plan.
What holds instead
Change holds when the new routine is simpler than the old one, when it is measured weekly rather than quarterly, and when the people running it were involved in designing it.
The practical test is straightforward: if the consultant left tomorrow, could the estate run the model on Monday? If the answer requires a caveat, the programme is not finished.