A couple of recent conversations have really caused this theme to spike in my head. In my experience, I’ve seen successful enterprise architecture and I’ve seen unsuccessful enterprise architecture. While many may put the blame on a failure to define what architecture is, I think that’s wrong. I think a recipe of failure is a lack of understanding of what the enterprise is. By that, I mean a lack of understanding of what capabilities should be managed at an enterprise level and what capabilities should not. There’s no uniform right or wrong approach, it is highly dependent on your company’s operating model as articulated in the book “Enterprise Architecture as Strategy.” In that book, the authors even state that at one extreme, a completely diversified company may have no enterprise architecture at all.
Once you understand what “enterprise” is, you have to set up the organization, processes, and governance to support that. To illustrate this point, however, let’s look at the differences today between two commonly “shared” organizations: HR and IT. Most big organizations have a centralized HR department. While I’ve never looked at the funding model for HR, my guess is that there is some overhead tacked onto every employee and every contractor that winds up paying the costs of the HR department. It’s a shared cost. One organization is normally not able to throw extra money in the pot and fund extra recruiters or benefits managers for their organization. Contrast this with IT where that’s exactly what happens. While IT may be centralized, it’s funding model is not. This results in IT organizations whose structure mirrors the business silos, and whose actions are completely controlled by the funding provided from each of those silos. What’s the point in having a central organization? That central organization really can’t dictate priorities or approaches, because someone else controls the purse strings. The right way of handling it is where the IT leader meets with the business leaders, they jointly agree on IT priorities and budget, and then the centralized IT department is funded based on those priorities with the discretion it needs to manage it from within. I’ve been at an organization that made that transition, and it was a definite improvement in the working environment, in my opinion.
Now the really interesting thing is getting that initial definition of the enterprise. The more I think about this, the more I realize that this definition is the job of Enterprise Architecture. We must architect the enterprise, meaning we come up with the proposal for what things should be enterprise and what things are best left to sub-domains. But wait, this sounds like a task, not a team. Initially, it is a task. Ongoing, however, the refinement of that definition and adjustment based on changes in the business and its climate will occur, and that is the job of the team. On top of that, there’s the enforcement aspect of making sure everyone continues to play by the rules. As an example, think about architecting the enterprise of the United States of America. The initial “architecture” creating the federated governing structure, with some capabilities provided at a federal, or enterprise, level (e.g. military forces) and others left to the discretion of the states. Since that time, the government has continued, in part, to tweak and refine what things are handled federally and what things are handled locally. The approach must always be monitored for effectiveness, responsiveness, cost, etc. Imagine, now, a different world where that structure doesn’t exist, and every cross-cutting topic is brought before a collection of state governors. Debate would occur on every single item, decisions would be extremely slow, and people with the power ($) to do so, simply would, even when it may not be in the best interest of the country as a whole. Take another scenario in recent press like immigration and border security. If the federal government says, “we must secure our borders” but doesn’t create a centralized approach for doing it, what happens? States with money may do a good job, states with others will not. One state’s approach may be completely different, making it difficult for legal immigrants to travel from state to state because the rules are so different. In other words, while all agreed that it was an enterprise goal, it was not managed as an enterprise capability. This is what happens in IT departments around the globe every single day.
My advice to my fellow EA practitioners: if you’re struggling, focus on defining the enterprise first getting buy off from senior executives on what items must be managed at an enterprise level, and then guide the transition to that approach.