The Three Missing Theories of Enterprise Architecture: Absorptive Capacity, Organisational Learning, and Why Frameworks Alone Cannot Close the Strategic Gap
The complaint about business and enterprise architecture has hardened into a familiar pattern.
Failed digital transformations. Debates about three-level versus five-level capability maps. Frustration that value streams “don’t translate into outcomes.” Practitioners asking, increasingly often, whether the discipline still has a strategic role.
I think the diagnosis is wrong, but not in the way most defenders of the discipline argue.
The problem isn’t that BA/EA frameworks are immature. TOGAF, BIZBOK, ArchiMate: the apparatus is sophisticated. The problem is that the discipline emerged on the precipice of the technology domain, where systems are deterministic. Data has a schema. Integrations have endpoints. Infrastructure has capacity. You can model it cleanly because it behaves cleanly.
Strategic enterprise change does not behave cleanly.
This is not the first time I’ve made this argument. In Why Enterprise Architecture Still Struggles, I named the symptoms. In Dynamic Capabilities: The Missing Link in Enterprise Architecture, I proposed Teece’s framework as part of the answer. This piece extends both. The constructive answer is not one missing theory. It is three. And the three together require a paradigm BA/EA has historically refused to adopt.
The Diagnostic Question, Revisited
The earlier argument was that EA struggles because it builds models in tools that have no operational job. The artefacts are produced for a planning cycle, then put down. The business changes; the artefacts do not. Within eighteen months the architecture is fiction.
That diagnostic is right. But it is incomplete.
Even an EA function that produces operationally-living artefacts (the Palantir-Foundry-style architecture I wrote about in Operational Architecture: Why Palantir Is Enterprise Architecture That Actually Worked) still encounters the strategic-execution gap. The ontology can be correct on a Tuesday morning and the strategic transformation can still fail.
Why?
Because the architecture, however operationally alive, is still working at a level that the harder constraints on strategic change do not respect. Three constraints in particular. Each comes from established academic literature. None of them sit inside the architecture frameworks. And without them, the discipline is asking artefacts to do work that artefacts cannot do.
I. The First Missing Theory: Absorptive Capacity Has a Window
In 1990, Wesley Cohen and Daniel Levinthal published the foundational paper on absorptive capacity: an organisation’s capacity to recognise the value of new external information, assimilate it, and apply it to commercial ends. They argued that this capacity is itself a learned capability, built through prior investments in research and development. In 2002, Shaker Zahra and Gerard George decomposed it further: potential absorptive capacity (acquisition and assimilation) and realised absorptive capacity (transformation and exploitation). The distinction mattered because organisations can have one without the other.
What is rarely emphasised in practitioner discussions of absorptive capacity is its temporal dynamic.
Absorptive capacity has a window. Use it, and it strengthens. Neglect it, and it atrophies. By the time a strategic opportunity arrives (a new market, a disruptive technology, a competitive shock), the window may have already closed. The organisation can be told what to do. It cannot do it. The learning required to absorb the change is no longer available within the time the strategy demands.
This is what I see, again and again, in transformation programmes that “fail despite excellent architecture.” The future-state diagrams are pristine. The capability heat-maps are colour-coded with appropriate care. The roadmap has dependencies tracked in Gantt detail. And the organisation cannot move. Not because anyone is resisting. Because the absorptive infrastructure that would have made the change executable was not built when it could have been built. The window closed years before the architecture exercise began.
BA/EA tools document the destination. They do not extend the window during which the journey is possible.
This is a structural blind spot. The discipline has no theory of organisational readiness over time. It has no construct for the use-it-or-lose-it dynamic. It treats the enterprise as if its capacity to change were available on demand, when in fact that capacity is a depreciating asset that must be continuously exercised.
II. The Second Missing Theory: Organisational Learning Runs on a Different Clock
If absorptive capacity is the window for change, organisational learning is the clock.
Chris Argyris, in his work on action science, distinguished single-loop learning (correcting behaviour without changing the underlying mental model) from double-loop learning, which revises the model itself. Most “transformation” failures are single-loop change. New structures imposed on unchanged theories-in-use. New language pinned over unchanged decision habits. The structures decay back into the prior pattern within months because the model that produced the prior pattern was never engaged.
Mary Crossan, Henry Lane and Roderick White added the temporal layer. Their 4I framework describes how learning flows through four interconnected processes: individual intuition, group interpretation, organisational integration, and institutional embedding. Each layer has its own pace. Intuition is fast. An individual sees something and grasps it in minutes. Interpretation is slower: a group must build shared language, which takes weeks. Integration is slower still: organisational practices must align across functions, which takes months. Institutionalisation is slowest of all, with culture, norms, and identity shifting over years.
A capability map produced in three weeks cannot enforce institutionalisation that takes three years.
This is the structural mismatch nobody in BA/EA discourse names plainly. The architecture is produced at planning speed. The organisation learns at organisational-learning speed. The frameworks assume the enterprise can be redesigned in the time it takes to draw the diagrams. The literature on organisational learning has been telling us, for forty years, that this assumption is false.
When practitioners ask why digital transformation programmes fail to translate into operational change, this is much of the answer. The architecture has moved through interpretation in months. The organisation needs years to integrate and institutionalise. Without explicit design for that pacing, without a learning clock built into the architecture process, the artefacts run ahead of the organisation and become irrelevant before the organisation catches up.
BA/EA has no theory of pacing. It has no methodology for sequencing interventions across the four learning layers. It treats organisational change as if it were an engineering schedule.
III. The Third Missing Theory: Dynamic Capabilities Live Below the Architecture
I covered the dynamic capabilities argument in the earlier piece. Briefly recapped: David Teece’s framework locates strategic adaptive capacity in three dynamic capabilities: sensing opportunities and threats, seizing them through committed action, and transforming the organisation to capture sustained value.
What I want to emphasise here, building on the earlier piece, is where these capabilities live.
They live at the micro-foundations level. In the routines, decision rights, and behavioural patterns of individuals and teams. Not in the structural diagrams. Not in the capability maps. Not in the value streams. The architecture frameworks operate at the level of the organisational entity (boxes, flows, layers) but the capabilities themselves are constituted at a level beneath that, in the daily decisions and habitual practices of people doing the work.
This means BA/EA can support sensing reasonably well. Visibility, inventory, structural clarity: these are real contributions to an organisation’s capacity to see what is happening. But seizing (committing the organisation under uncertainty to a particular direction) happens in decision rights, governance habits, and the quality of executive judgement under pressure. And transforming (continuously reconfiguring the organisation to capture value from the seized opportunity) happens in routines, capabilities, and the everyday choices of teams.
Architecture frameworks do not reach these levels.
This is why digital transformations “fail” even when the architecture is sound. The architecture is not the binding constraint. The binding constraint is the organisation’s absorptive capacity, learning velocity, and dynamic-capability development at the micro-foundations level. The diagram is downstream of the dynamics it claims to govern.
The Synthesis: An Engineering Paradigm on a Sociotechnical Reality
Read the three theories together and a deeper pattern surfaces.
Absorptive capacity gives us the window during which strategic change is possible. Organisational learning gives us the clock on which the change unfolds. Dynamic capabilities tells us where the change actually happens: at the micro-foundations level, beneath the architecture artefacts.
None of these are properties of the architecture. They are properties of the organisation as a sociotechnical system.
BA/EA has been operating with engineering metaphors on sociotechnical realities. The frameworks are sophisticated; the theoretical foundation is thin. We model the enterprise as a designable artefact because the IT systems the discipline originally addressed were designable artefacts. The paradigm came along when the discipline expanded upward into business architecture. Capabilities became boxes. Value streams became flows. The enterprise became a redrawable diagram.
The three missing theories are not three more constructs to add to the framework. They are signals that the framework’s underlying paradigm is mis-fit to the problem.
This is why incremental improvements to the architecture toolkit do not close the gap. Better capability maps cannot recover the absorptive-capacity window that closed five years ago. More precise value streams cannot compress the organisational-learning clock from years into weeks. Tighter governance cannot reach the micro-foundations where dynamic capabilities are built or fail. The toolkit is operating at a layer above where the binding constraints actually live.
What a Paradigm-Different Architecture Practice Looks Like
If this is right, the constructive move is not to keep refining the architecture artefacts. It is to make absorptive capacity, organisational learning, and dynamic capabilities the primary objects of the discipline, with architecture artefacts becoming derivative: useful where they support these primary concerns, expendable where they do not.
In practice this means three changes.
First, the architecture practice acquires a temporal dimension. The question is no longer just what the future state is. It is also: how wide is the absorptive-capacity window for this change, and what does the organisation need to invest now to keep the window open later? Architecture work becomes longitudinal, not snapshot.
Second, the architecture practice acquires a learning-clock literacy. The question is no longer just what the structural target is. It is also: which of the four learning layers (intuition, interpretation, integration, institutionalisation) does each intervention engage, and what is the realistic pacing for each? Architecture deliverables get sequenced not by structural dependency but by learning-cycle dependency.
Third, the architecture practice extends down into the micro-foundations. Capability maps describe the structural target. The architecture work describes the routines, decision rights, and behavioural patterns through which the target is actually constituted. The Palantir-Foundry pattern of operational architecture I wrote about earlier is one model for this (architecture that lives in the operational layer because it has an operational job) but the principle generalises beyond the technology layer.
This is a different discipline. It overlaps substantially with what consultancies in the strategy-execution space (firms like BTS Group, the simulation-and-capability-building consultancies) have been doing for years, often in territory where BA/EA has been notably absent. The work is real. The methodology exists. The discipline has just been letting it happen in adjacent professions rather than claiming the territory.
The Provocation
The question I keep returning to, when I hear the discipline-in-crisis posts on LinkedIn, is not the one most practitioners are asking.
It is not how do we map capabilities better?
It is what would the discipline look like if it took absorptive capacity, organisational learning, and dynamic capabilities as primary, and architecture artefacts as derivative?
I suspect we’d stop arguing about how many levels the capability map should have. I suspect the discipline would stop fighting for legitimacy. And I suspect the digital transformations that currently “fail despite excellent architecture” would start succeeding, not because the architecture got better, but because the discipline finally engaged with the dynamics that determine whether strategic change is even possible.
The frameworks have always claimed to bridge business and technology. The next move is to recognise the bridge runs in the wrong direction. Strategic enterprise change is led by the organisational dynamics. Architecture catches up.