A recommendation an architect has not personally tested is an opinion with a logo on it, and a governance body can tell the difference.
That is the whole argument, and in an estate where six agencies could decline any standard put in front of them, it was not a philosophical position — it was the only thing that worked. An agency free to say no will say no to a preference. It is much harder to say no to something that has already been run against your own conditions and did what it was said to do.
The worked example
Microsoft Intune is the clearest one. Piloted against the organization’s own conditions rather than against the manufacturer’s material; deployed alongside desktop engineering rather than over the top of them; carried through change governance so it did not disrupt production; and handed over with the team who would operate it trained on it.
Four steps, and the last two are the ones most evaluations skip. A pilot that ends in a recommendation has done half the job. A pilot that ends with somebody else able to run the thing has done all of it.
The same pattern, eight times over
It ran across mobility and endpoint management, network orchestration, wireless survey and propagation tooling, building automation and lighting control, video and acoustic detection platforms, enterprise integration and, latterly, AI capability.
The tooling has nothing in common. The method does: state the constraints before looking at solutions, build the thing out rather than judge it on a datasheet, run it where it will actually have to live, and keep the verification with the engineer rather than delegating it to the vendor’s demonstration.
What it costs, and why it is worth it
It is slower. It happens on evenings and weekends as often as not, because the estate does not stop needing to work while something is being proved against it. And a good proportion of it ends in a no.
That last part is the point rather than the waste. Knowing what an organization should not buy, and being able to show the evidence for it, is the same discipline as knowing what it should.
Outcome
Recommendations a governance body of independently governed agencies would accept, because the evidence was first-hand.