The single-system fallacy
Why the instinct to put everyone on one system fails in organisations built from many independent parts, and what tends to work instead.
A central team had produced the official version of a form. It was legally correct, three pages long and written in the language the centre uses about itself. A local leader downloaded it, decided the people in front of them would never get past the first page and rebuilt it on one side of A4. It did the legal job in plainer words, and some of the details were already filled in.
They had tried the official version once and watched people give up early. Nobody could have made them use the long one.
I’ve heard this sentence in more than one boardroom: “Let’s just build one system that everyone uses.” It sounds responsible and efficient, like finally getting a grip on the mess.
I’ve nothing against shared systems. But “one system for everyone” assumes a kind of organisation that plenty of us don’t work in. In organisations like mine it often goes badly. The cost tends to show up a couple of years later, and by then a lot of goodwill has gone too.
What a federated organisation is
A federated organisation is made up of parts that are legally or operationally independent, held together by a shared purpose. The centre can’t tell the rest what to do, so it has to persuade them. The Scouts work this way, with each local group a charity in its own right. So does Citizens Advice, a network of more than 250 independent local charities.
Many organisations are built this way without using the word. With no name for the pattern, each one tends to assume its situation is unique and learns the same hard lessons alone.
Most change advice assumes a corporate hierarchy, where head office decides and the branches comply. Bring that assumption into a federation and it usually goes wrong in one of two ways.
The centre builds a polished single system and finds that nobody was ever obliged to use it, so plenty don’t. Or it pushes hard for adoption. People resent the centre reaching in and taking a decision away from them, and the pushing hardens the local resistance it most needed to avoid. The money has gone either way, and the next proposal starts further back.
A version of that sentence now turns up in conversations about AI: “Let’s pick one AI system and roll it out everywhere.”
A different question to ask
The question I’d rather ask is what’s the smallest shared architecture that lets independent parts work together without asking them to become the same. Answering it means working out what has to be common, and being honest that almost everything else can stay local.
In practice that means:
- a small shared core, such as one identity system
- local freedom around everything else
- standards offered because they make local work easier
- a centre that earns adoption by being useful, again and again.
The clearest public example I know of is the Scouts replacing Compass, the membership system for their adult volunteers. If anything belongs in the shared core, membership does, since it holds training records and permits. Even so, it didn’t reach everyone at once. Ten counties and areas were chosen in 2022 to go first as early adopters. The rest of the UK was promised for early 2024, but testing raised concerns about moving that much data, and the main switch-over happened in late November 2024.
By corporate standards that’s slow. I’d call it a sensible pace for something that has to work for everyone.
Where I’m less sure
I don’t think the federated frame fits every case. Some organisations look federated on the chart but behave as networks, where there isn’t even a centre to do the persuading, and the advice has to change again. I’m still working out where one turns into the other, mostly by writing about it in public.
Where an organisation is federated, though, I’d want any single-system proposal to start with a list of what has to be the same everywhere. It’s usually shorter than the proposal assumes.
Ben
Photograph by Bruno Martins.