Centralisation Myths?

Centralisation is everywhere.

• Call centres are points of centralisation.
• IT is constantly trying to pull everything back to the data centre.
• Many business functions are centralised.
• Organisations outsource so they can get the benefits of centralisation.

Why? There are normally two reasons and they are the usual business reasons – cost and quality. Functions done centrally are supposed to be more cost effective than those done in a distributed fashion. Centralisation allows specialisation which should lead to an increase in quality.

If there is a quality problem, or a need to reduce costs the first mechanism that people turn to it centralisation.

The increase in quality and the reduction in cost is regarded as a certainty.

It’s almost become a business mantra – “I must centralise”.

I am personally becoming increasingly skeptical about centralisation.

I feel that I should declare my experience here. Throughout my nearly 20 years in IT I have been involved in a number of centralisation activities; centralisation to a data centre on a site, then to a data centre in a country, then to global data centres. I have also been involved in IT help desks. When I was a new graduate I would man the help desk for two afternoons a week. I watched on as the help desk went from supporting a site, then to supporting a number of sites within a country, then a country, eventually it supported a number of different customers across different time zones.

I’m not doubting that these activities reduced costs, though none of them gained the cost reductions people were hoping for. There was also a change in quality but it not dramatic and not on every measure.

So why am I skeptical?

My main area of skepticism is caused by one word – change. These centralised entities are terrible at responding to change.

They naturally become highly integrated within themselves

The help desk naturally consolidates to a single set of systems. That is how costs are reduced after all. The consolidation of the systems creates cost reduction and increases the quality. That is until something comes along which drives change. Lots of small systems, each running independently can change when they need to change. There doesn’t need to be a huge requirement to change. When a huge integrated system exists change becomes more and more difficult. Where change is difficult change will either stop, or the cost of change will be dramatic.

People forget that change is inevitable.

The same is also true for IT systems. Changing a single system that does a single role is far easier than one large system that handles lots of roles. I’m sure that some people believe that if the cost of changing one system with one role is X then the cost of changing one system with lots of roles (Y) is less than X * Y. In my experience it’s more than X * Y it’s more like X * 1.5Y.

Because change is difficult it happens rarely. When it does occur the change ends up being massive, it’s normally not possible to change a single entity. The system has coalesced and for one thing to change, lots of things have to change. It’s become a chain reaction. Between the massive changes, though, the quality of services is constantly decreasing as the service delivered becomes further and further from the service required.

I’m not suggesting that we throw out the baby with the bath water here. What I am advocating is that we approach consolidation in a more pragmatic manner. Rather than blindly following the centralisation mantra we should evaluate the centralisation option knowing that change is inevitable and plan for it. In planning for it we may discover that centralization isn’t actually the correct option.