In most engagements, the diagnosis surprises nobody. People know. They have known for two years which process is broken, which team is understaffed, which project should have been stopped. What they do not have is a place to say it without consequence.
The consultant's role is then less that of an investigator than of a clerk of the court: to collect what is already known, to make it sayable, and to take public responsibility for its wording. That assumes one simple rule — the finding is about mechanisms, never about people.
The phrase that defuses
"This organisation produces this result" is always worth more than "this team is failing". It is not a precaution of language: it is a more accurate statement. A badly cut structure turns competent people into slow people. Naming the mechanism makes the problem repairable; naming the people only makes it painful.
A diagnosis that leaves no way out is not a diagnosis, it is a verdict.
That leaves the question of courage. Softening things until they are inaudible is the symmetrical failure: you leave with a satisfied client and an unchanged situation. The good readout is the one where somebody goes quiet for a moment, then asks where to begin.