How to convince the boss good architecture matters

In terms of holding interest, enterprise technical architecture management (ETAM) ranks somewhere between watching paint dry and grass grow, with the additional disadvantage of being intellectually demanding. The less technical your audience, the more ETAM serves as a side-effect-free alternative to Lunesta.

The news gets worse: ETAM is the one subject where IT needs to engage business leaders on a technical level — a conversation about the enterprise technical architecture that’s limited to metaphors and business terms won’t get the job done.

ETAM is strategically vital, an essential aspect of competent IT management, and intellectually challenging and satisfying besides. But is there any easy way to interest business executives in it? Sadly, no.

One alternative is resorting to threats. I once told a client’s CEO he had two choices about this: He could trust me, or I’d explain it to him. He caught a glimpse of my first slide and said, “I think I’d rather trust you.”

It isn’t that the business case is complicated — quite the opposite. But proving it is excruciatingly difficult, even to a technical audience interested in the topic. Making the case for ETAM to someone who would rather have his pancreas eaten by ferrets? Good luck with that.

The business and politics of ETAM

Here’s ETAM’s business case in one sentence: We need to spend more time, effort, and money on everything we do now, so we won’t have to spend a lot more time, effort, and money a few years from now on everything we do then. It’s very much the same as the case for preventive maintenance.

If the business case is that simple, why do most enterprises have a hard time making ETAM happen? Call it the altruism obstacle. ETAM asks everyone in the business to spend more and wait longer now so that someone else in the business will benefit in the future. And it asks the CEO to accept higher costs and smaller profits this year, so profits a few years from now will be better than they otherwise would.

It’s even worse if the company charges back for IT services. If it does, ETAM’s overt costs and deferred benefits aren’t just a matter of image damage (“Why do your projects cost so much?”). It actually hits business budgets.

The ETAM political anticlincher: If it delivers its promised benefits, there will be no way to prove it. Successful prevention is, after all, indistinguishable from absence of risk.

In trying to sell the business on ETAM, you’re best served by following a double-barreled approach: Limit your discussion to the ETAM decisions that are truly business-driven (most aren’t), and present the anti-ETAM political case bluntly, in all its unsavory glory.

Non-business-driven ETAM decisions

That IT must be business-driven is a truism, and like so many other truisms, it’s mostly false. When the subject is enterprise technical architecture, business strategy and tactics have little impact on most of the choices IT has to make. Even the business architecture is often irrelevant.

For example: IT needs a way to manage structured data and the transactions that feed it. As explained last week, that’s the services-level description. That’s followed by choosing the class of technology needed to provide that service. In this case, IT’s alternatives are either an RDBMS, an RDBMS, or an RDBMS. When the subjects are transaction processing and structured data management, the solution is a relational database management system. All the business-driving IT in the world won’t change that.

When the time comes to choose which RDBMS to build the company on, some business information can be handy. How fast the execs plan for the company to grow (measured in transactions per minute, if possible) will help weed out platforms that fail the Goldilocks Test. Though there may very well be more information that would help inform this decision, once you get beyond sizing, you won’t find much connection between what the business is going to do and which RDBMS to buy.

Another example: how to customize COTS (commercial off-the-shelf) software. The ETAM solution: Make use of the built-in configuration tools provided by the application vendor, and if these can’t do the job, write satellite applications that integrate into the COTS application through its defined APIs.

Here’s how not to do this: Modify the vendor’s code. Business plans have no impact on this guidance.

Business-driven ETAM decisions

There are some business futures that truly drive decisions about the enterprise technical architecture. You need to involve senior leadership in all of them. Want an example? The trade-off between independent action and business integration is usually the most important.

Companies that value integration — that consider the enterprise a machine with interlocking gears, pulleys, cams, and levers — need an ERP-oriented enterprise technical architecture. They need, that is, a unifying business suite that serves as a hub for data and business logic, so that when marketing sneezes, supply chain knows to say “gesundheit!”

Companies that value independent action — with a more entrepreneurial form of leadership — want those responsible for a business function, shared service, or line of business to make decisions based on what will make their area more successful, without having to pay too much attention to ripple effects. This sort of company needs a federated architecture built around focused best-of-breed solutions and integrated using some form of EAI (enterprise application integration) system, service bus, or equivalent technology.

Companies that value integration more highly will ask each part of the business to sacrifice functionality if necessary to achieve superior integration. Companies that value independent action will recognize the same trade-off but will make the opposite choice.

Here’s another example: The extent to which the business expects to rely on a mobile, vertical, or (to a lesser extent) virtual workforce. The issue here plays out in the need to support devices other than standard PCs that have one or more monitors plus mice and keyboards — smartphones and tablets, that is. To the extent that need is real, companies will require applications structured to segregate business and integration logic from user-interface logic.

And a third: The more dramatically processing loads vary, whether randomly or seasonally, the more likely it is that a cloud-based solution will prove superior to internally hosted applications.

Handling the political anticase

As mentioned earlier, company politics, especially if silo-driven decision-making dominates, makes enterprise-oriented decisions about the technical architecture nearly impossible. Here’s what IT can do to turn the political situation to its advantage:

• Be blunt and overt: There’s always a regular meeting of the company’s top executives. Introduce the challenge at one of these meetings, and don’t soft-pedal it. Say it more or less as presented above, threaten to explain the engineering behind your assertion to anyone who doesn’t trust you about it, and ask everyone in the room how they propose to overcome this core IT governance challenge.
• Don’t bluster: It’s highly likely the company is already experiencing several warning signs of bad architecture. Point them out, and make it clear they’ll only get worse if the company doesn’t invest in solutions.
• Take advantage of the chargeback system: If you have to deal with chargebacks, corporate should add an architecture subsidy to every software-related project. While you’re almost certainly better off getting rid of the chargeback system altogether, if you can’t, this will eliminate the altruism obstacle.

Finally, make it clear that while the benefits of fixing existing architectural problems can be measured objectively, ETAM is mostly preventive. That makes it like everything else the company does to keep bad things from happening — a good investment whose benefits can’t be measured.

If you’re challenged on that point, you already have the solution: Threaten to explain it.

This story, “How to convince the boss good architecture matters,” was originally published at InfoWorld.com.