I’ve lost count of the number of Technical Design Authority meetings I’ve sat in, watching smart people tie themselves in knots over a diagram. We’d debate whether a proposed change conformed to the “target architecture” – a utopian blueprint of a perfectly rationalised, fully integrated, and utterly fictional technology estate.
I remember one team in particular. They had a plan to deliver a new user-facing service in eight weeks. The roadblock? Their plan didn’t “conform”. They were told they had to build it on the new, approved enterprise platform. A platform that wasn’t scheduled to exist for another two years.
And this platform, like so many I’ve seen, was being conceptualised in a vacuum. It was a solution being built without understanding the real-world needs of the services – like this one – that would supposedly use it. The team knew that even if they did wait two years, what was finally delivered probably wouldn’t work as expected.
The service was delayed. The platform never delivered.
This is a scenario I’ve seen play out time and time again. That target architecture wasn’t a map to the future; it was an anchor. It wasn’t solving business challenges; it was creating committees. It was forcing decisions to be made years too early and, in the process, it was killing the one thing we needed most: speed.
The seductive lie of control
This story is not an exception. It’s the norm. I understand the impulse. Modern technology is complex. It’s no longer just the email system. It’s the ERP, the CRM, the Electronic Patient Record, the supply chain logic, and the hundreds of other systems that are the non-negotiable foundation of the entire business. And it’s all in constant flux.
For leaders in large, established organisations, this complexity can be terrifying and can represent an enormous and legitimate business risk. It’s an understandable, even logical, impulse to apply the established playbooks for managing large-scale capital expenditure. They try to manage technology the same way they would manage the construction of a new factory, a hospital or a bridge – disciplines where a detailed, upfront blueprint is critical to success.
This approach is intended to provide strategic control, prevent duplication, and ensure approved tech is secure. Leaders reach for a total inventory, a single picture that promises to provide comprehensive visibility over an estate that is, in reality, too dynamic and complex to fully comprehend in a static document.
This is what drives the cycle.
Leaders hire Enterprise Architects and empower them to build the grand plan. They adopt rigid, certification-heavy frameworks like TOGAF. They burn enormous amounts of time, money, and political capital to design their utopia. They create binders full of diagrams, laminating a pristine “city on a hill” that exists only on paper.
But here’s the hard truth: managing technology isn’t like building a bridge. It’s like gardening. You don’t “finish” it. It requires constant tending, pruning, and adaptation to a changing environment.
The target architecture approach locks in decisions far too early, making it impossible to pivot when, inevitably, circumstances change – whether that’s a new technology, a new cyber threat, or a change in customer expectations.
You can’t blueprint your way out of ‘Deep Digital Geology’
The biggest flaw in the target architecture model is that it ignores the ground it’s supposed to be built on.
Most large organisations aren’t starting with a clean, empty field. They are built on what my colleague Dave Rogers calls “deep digital geology” – decades of accreted technology layers. That mainframe from the 1990s still processing applications. The 20-year-old Oracle database that no one fully understands but everyone relies on. The essential-but-fragile Excel macros built by someone who left a decade ago. The half-completed migration here; the failed replacement there.
This is the reality. And let’s be perfectly clear: this geology is a high-risk problem. That 20-year-old database is an operational time bomb. It must be managed.
But here is the trap: we’ve convinced ourselves that the antidote to this risk is a perfect, utopian architecture. It isn’t. The risk of designing a plan you’ll never deliver is just as high as the risk of the legacy tech itself. It’s just a different flavour of risk – that your perfect plan is a fiction that guarantees you will never actually replace the old tech, all while burning time and money that could have been spent on a real fix.
Your target architecture pretends you can simply sweep this geology away. But the cost and effort to excavate and replace these layers are not just enormous; they are often prohibitively high. The business risk is too high.
And worse? Even if, by some miracle, you achieved it, the sands will have shifted. By the time you arrive, your perfect destination is already obsolete. At best, your target architecture is a monument to wasted effort. At worst, it’s a set of golden handcuffs, actively preventing your teams from innovating.
The alternative isn’t chaos.
The most successful tech-native companies do not have static, target architectures. The alternative to this rigid, top-down control isn’t a ‘Wild West’. It’s about being intentional in a different way.
The goals of enterprise architecture – reducing risk, ensuring security, lowering long-term costs, and enabling interoperability – are not wrong. They are essential. A true Wild West where teams use any tool they want without standards is just as dangerous as a rigid, static plan.
The challenge is that a static, multi-year target architecture is not an effective way to achieve those goals in a fast-moving digital environment.
This doesn’t mean having no vision. There is nothing wrong with having an architectural vision and setting a direction.
But that vision is best expressed as a set of guiding principles and a new operating model, not a detailed map. This is how you achieve those same goals – control, security, and efficiency - but with speed and adaptability.
The way to do this is to:
-
Fund empowered teams, not one-off projects.
The single biggest, most important shift is to stop funding temporary ‘projects,’ ‘system replacements,’ and ‘migrations’. Instead, you must fund permanent, long-lived, and empowered multidisciplinary teams. These teams own a specific service or business capability; they design it, they build it, and, crucially, they maintain it for its entire lifecycle. Because they are permanent and empowered to make decisions about their service, they can make smart, incremental choices, and they are accountable for the real-world results. -
Give them principles, not blueprints.
This is how you guide those empowered teams and establish your vision. Instead of a central committee dictating what platform a team must use, leaders publish clear, testable principles.- A bad rule: “All new projects must use the new Enterprise Data Platform”.
- A good principle: “All services must expose their data via a standard, secure API and be deployable in under 15 minutes”.
This is where architecture becomes an emergent property, guided by shared understanding and real-world results, not a rigid map drawn by people disconnected from the work.
This is what unlocks speed. When decisions are small, reversible, and made by the empowered team, you can make them just-in-time. You don’t need to guess what the world will look like in two years. You can react to the world as it is today. If a decision turns out to be wrong, the team can pivot without asking a central committee for permission to re-draw the master plan.
-
Start ‘untangling,’ not ‘replacing’.
With empowered teams and clear principles, you can finally tackle your digital geology. You don’t do it with a single, massive replacement program. You untangle the spaghetti, one strand at a time. An empowered team will find a pressing user need and build a new, thin, loosely coupled service that meets it. They’ll use proven patterns (like the Strangler Pattern) to put a clean API in front of that messy legacy system. Then they chip away at the old system, moving one piece of functionality at a time into new, modern services. This is iterative, delivers value immediately, and is guided by real-world business needs.
This glacial, committee-driven approach was already failing. In the age of AI, it’s a death sentence. The speed of change is now accelerating exponentially. Organisations are still struggling to adapt to the last generation of cloud and mobile tech, let alone today’s rate of change. You cannot ’target architecture’ your way into this new world. It’s moving too fast. The only way to keep up is to be fast, flexible, and adaptive – the very things your EA process is designed to prevent.
Stop funding expeditions to a mythical, utopian destination. That target architecture you’re pouring money and political capital into? It’s a fantasy. It’s an enormous waste of time and resources, a detailed map to a place that doesn’t exist. It will never become real. It’s a dangerous distraction, a set of golden handcuffs that does nothing to solve your actual legacy problems while actively preventing your teams from innovating.
Your job as a leader isn’t to approve a fictional blueprint. It’s to create empowered, long-lived teams. It’s to get them out of the temporary project hamster wheel and make them permanent owners of their services. And it’s to give them clear principles – the navigational rules they need to move fast, safely, and independently. That is how you navigate – and improve – the real, complex territory you’re on, one valuable step at a time.