Where does enterprise-architecture sit?

This one’s a follow-up to a useful comment today by David Deighton to my ‘TOGAF Certified‘ entry back in August. David asks:

However I’m a little concerned about the idea that Enterprise Architecture is somehow bigger than IT Architecture. Who are we kidding here? Just try going in to an organization and telling them, from within the IT department, you’re going to re-architect their business processes and see what kind of reception you get. Some of us have to put in some effort to try to avoid giving the impression that IT are trying to take over the world.

Talk about scope creep.

In one sense, yes, agreed, there’ll be problems on this approach if we start “from within the IT department”. We’d have exactly the outcome David describes – in fact I’ve seen exactly that happening in practice, too many times for comfort.

But the whole point of what I’ve been saying over the last few months is that enterprise-architecture doesn’t belong in IT. We should not be starting from “within the IT department”, because EA shouldn’t be under their aegis in the first place. IT-architectures (plural) are subsets of enterprise-wide IT-architecture, which is often mistakenly described as ‘enterprise-architecture’ but is in reality just one more subset of real enterprise-architecture, which covers the whole of the enterprise – not just the IT components.

Let’s say that again more clearly:

  • enterprise-architecture covers the whole of the enterprise
  • IT-architecture covers the IT-specific aspects of the enterprise
  • IT-architecture is a subset of enterprise-architecture – not the whole of it

The proper place for an enterprise-architecture unit is under an organisation-wide governance / change-management facility reporting directly to board level. Its role, in essence, is to manage the enterprise’s knowledge of its structure, and of itself as an entity, a functioning whole. It must operate above any sector-specific architecture functions such as IT-architecture, since it must be able to track processes, for example, across all implementations (IT, human, machine or mixed) and independent of any implementation – otherwise true full-scope process reengineering is impossible.

The most appropriate group to initiate a true enterprise-architecture facility is a whole-of-enterprise business-transformation team. That works: I know it works, I’ve seen it work, and well. But time and again I’ve seen that the least appropriate group to do true enterprise-architecture is the IT ‘enterprise-architecture’ unit: in every organisation I’ve dealt with, the IT EA team has fought tooth-and-nail to sabotage any attempt to broaden the scope to a true enterprise-wide level, or to address any issues at all beyond low-level and relatively-short-term information-technology concerns.

Enterprise-wide architecture isn’t ‘scope creep’: enterprise-wide is the natural and necessary scope for true enterprise-architecture. Any less scope isn’t enterprise-architecture – it’s just a subset pretending to be the whole scope, with all of the problems that that mistake immediately entails.

So whilst David is entirely correct about what happens if we do try to start “from within the IT department”, his description is still too IT-centric. “Some of us have to put in some effort to try to avoid giving the impression that IT are trying to take over the world”, says David: but that effort (and yes, I know it well) is only necessary because business is becoming less and less willing to tolerate the way that most IT departments are too ‘busy’, too self-centred or too damned stupid to even bother to listen to what the business wants, prancing around with an arrogant “we’re IT so we know what’s best for you” attitude. When that attitude dominates, yes, enterprise-architecture will always look like “IT trying to take over the world” – because EA must eventually have that broader scope in order to function at all.

So the real problem here is to get IT to let go of enterprise-architecture, because it should never have been under their control in the first place. How we get that to happen is going to be the interesting problem of the next few years, I suspect… but unless and until we do manage to wrest it from their cold dead hands, real enterprise-architecture ain’t goin’ nowhere, folks.

Over to you?

Leave a Reply

Your email address will not be published. Required fields are marked *