Digital Transformation Is Now an Operating Model Question

The adoption paradox

Digital transformation used to begin with access to technology. Cloud infrastructure, advanced analytics, automation and machine learning were expensive capabilities that only some organisations could deploy at scale.

That constraint is disappearing.

AI is moving into the enterprise at remarkable speed. In McKinsey’s 2025 global survey, 88 per cent of respondents said their organisations regularly used AI in at least one business function. Yet only 7 per cent described AI as fully scaled across the organisation.

That gap is more consequential than another debate about whether digital transformation is succeeding or failing. It suggests that the central problem has moved.

For many organisations, access to technology is no longer the principal barrier. Neither is experimentation. Companies have copilots, pilots, analytics platforms, automation programmes and increasingly AI agents. The harder question is why all this technological activity so often leaves the underlying organisation recognisable.

Meetings remain. Approval chains remain. Functions remain fragmented. Budget cycles remain. Decision rights remain ambiguous. Legacy metrics continue rewarding yesterday’s behaviour.

The software changes faster than the system around it.

This is why digital transformation should increasingly be understood not as a technology programme but as an operating-model question: how does the organisation actually create value, allocate authority, combine human and machine capabilities, learn and change?

Recent research on organisations moving through AI transformation reaches a similar conclusion. Value emerges not from one universally superior organisational model, but from deliberate operating-model choices and the ability to execute them.

Buying technology may be getting easier at precisely the moment when transformation is becoming more organisationally demanding.

When technology accelerates the old organisation

One of the most persistent mistakes in digital transformation is to confuse a faster process with a redesigned one.

Add generative AI to an inefficient approval chain and individual tasks may become faster while the approval chain remains. Automate a fragmented customer journey and the organisation may simply move customers through its fragmentation more efficiently. Give managers better analytics but leave decision authority unchanged and information may improve without decisions improving with it.

AI is particularly revealing because it exposes the difference between task productivity and enterprise productivity.

A person may produce a document in half the time. That is useful. But if four layers still need to review it, the end-to-end cycle may barely move. A forecasting system may generate better predictions, but if resource allocation is locked into an annual budget, the organisation may remain unable to act on what it knows.

This is the deeper meaning of the claim that technology acts as an amplifier. It does not necessarily repair the organisational system into which it is introduced. It can make the strengths of that system more powerful — and its weaknesses more visible.

Marco Iansiti and Karim Lakhani anticipated much of this argument in Competing in the Age of AI. Their work focuses not simply on adopting AI but on the new operating architecture created when data, algorithms and networks become embedded in how an organisation creates and delivers value. The strategic change is architectural: processes, learning and scale begin to work differently.

That distinction matters because technology itself is increasingly replicable. Competitors can often access similar foundation models, cloud services and enterprise applications. What is much harder to copy is the accumulated set of organisational choices around them: proprietary workflows, data, governance, talent, customer relationships and the ability to make decisions at speed.

The competitive question shifts from “What technology do we have?” to “What can our organisation now do that it could not do before?”

The unit of transformation is the workflow

Transformation becomes tangible when leaders stop discussing “the organisation” in the abstract and examine how work actually travels through it.

Who receives the information? Who interprets it? Who can decide? Which hand-off creates delay? Which judgement requires a human? Which decision could be automated? Where does accountability sit if an AI system acts? Which metric encourages a manager to preserve a process that the strategy says should disappear?

This is the level at which transformation either becomes operational or remains theatre.

Many organisations are using AI to accelerate existing activities while leaving governance, teams, capabilities and decision structures largely unchanged. The more durable opportunity lies in redesigning workflows and operating models rather than treating AI as a procurement exercise.

These questions are harder than choosing a platform because they expose power, incentives and organisational history.

A redesigned workflow may make a management layer unnecessary. Faster decisions may require executives to surrender approvals they are accustomed to holding. Cross-functional ownership may challenge existing P&Ls. AI may change not only what employees do, but what managers are for.

That is transformation.

Learning is part of the operating model

There is another reason technology programmes stall: organisations often want experimentation without accepting the conditions required for learning.

George Westerman’s work on digital transformation has long distinguished technological capability from the leadership capability required to use it. In the AI era, the tension remains: the difficult part is not simply AI adoption, but leading change at the speed of technology inside organisations designed for stability.

An adaptable organisation therefore needs more than pilots. It needs a learning system.

That means knowing what can be tested safely, what evidence will determine whether an experiment continues, how resources move towards promising initiatives and — just as importantly — how weak initiatives get stopped.

Amy Edmondson adds an essential human dimension. Her research on psychological safety and intelligent failure is sometimes simplified into permission to fail. The more useful distinction is between failures that generate knowledge under uncertainty and failures that arise from preventable mistakes. Her work argues for candour and disciplined experimentation, not lowered standards.

For transformation leaders, that distinction is crucial.

A culture in which every unsuccessful pilot threatens a career will learn too slowly. But a culture that celebrates experimentation without rigorous stopping rules can generate permanent pilot portfolios, exhausted employees and little enterprise value.

The operating model therefore needs to answer not only how work gets done, but how the organisation learns what should change next.

Transformation without an end date

This changes the meaning of successful digital transformation.

The objective is not to reach a future state in which an organisation has finally “become digital”. Digital capability is already woven into ordinary business. AI will make the distinction even less useful.

The more durable objective is to develop an organisation that can repeatedly reconfigure itself as technologies, markets and customer expectations change.

Rita McGrath’s work on transient competitive advantage provides a strategic lens for this. When advantages become less permanent, strategy cannot depend solely on defending an established position. Organisations need mechanisms for sensing change, testing options and reallocating resources as conditions evolve.

Seen through this lens, transformation is not an exceptional programme layered on top of the business. It is a capability of the business.

This does not mean permanent restructuring. Constant organisational upheaval can destroy focus just as effectively as inertia. The challenge is more precise: leaders must distinguish what should remain stable from what must be capable of changing quickly.

Purpose may remain stable while products change. Strategic guardrails may remain stable while teams gain more autonomy. Standards for risk may remain stable while individual approvals disappear. Customer promises may remain stable while the operating system behind them is rebuilt.

The mature transformation question is no longer “When will this be finished?” It is: What have we built that will make the next change easier?

What this changes for leaders

For boards and executive teams, one practical consequence follows: technology activity is an increasingly weak proxy for transformation progress.

Numbers of AI licences, cloud migrations, pilots or automated tasks can measure implementation. They do not necessarily measure organisational change.

A stronger review begins with business outcomes and works backwards. Has decision time fallen? Has a customer journey materially changed? Has a workflow lost unnecessary hand-offs? Has accountability become clearer? Can the organisation redirect capital and talent faster when evidence changes? Are employees using AI to redesign work, rather than merely producing the same outputs more quickly?

This also changes ownership. Digital transformation cannot be delegated entirely to IT, a Chief Digital Officer or an AI team because the decisions that matter eventually reach strategy, P&L ownership, organisation design, risk, talent and leadership behaviour.

Technology leaders remain essential. But the transformation becomes real only when business leaders are willing to change the system in which the technology operates.

Four lenses, four different stage conversations

A keynote on digital transformation should no longer be selected simply because a speaker can explain emerging technology. The more useful question is what the audience must understand or reconsider.

Transformation capability

George Westerman

Why the difficult part of AI remains organisational transformation rather than the technology itself. ↗

Operating architecture

Marco Iansiti

How AI changes the architecture, scale and learning model of the firm. ↗

Learning organisation

Amy Edmondson

How psychological safety, candour and intelligent failure enable disciplined learning. ↗

Continuous reinvention

Rita McGrath

How transient advantage changes experimentation, resource allocation and strategic adaptation. ↗

These are related conversations, but they are not interchangeable. The subject is no longer simply “digital transformation”. The real brief is the organisational question underneath it.

For another practical starting point, explore Scurati’s perspectives on Digital Transformation, AI, Change Management and Leadership & Corporate Culture.