Reflections

Transformation can't be delivered. It has to grow.

27 May 2026

What thirty years of observation and practice have taught me about change that actually holds.

The claim that 70% of transformation programmes fail has been repeated so often that it has become background noise. Academics have challenged the figure — it turns out to be an estimate rather than an empirical finding. But the observable reality behind it is harder to dispute: organisations spend enormous sums on change initiatives that do not hold. Consultants leave. The methodology fades. The organisation reverts. A year or two later, a new initiative begins. The question worth asking is not whether this happens, but why.

Two familiar models

For decades, the business world has cycled between two dominant theories of change.

The first focuses on the individual. Train the people, and the organisation will follow. Develop leaders. Run workshops. Certify practitioners. Embed the methodology. The logic is intuitive, but the evidence is sobering. Hermann Ebbinghaus's research on memory — now nearly 140 years old — established that without reinforcement and application in context, the majority of what is learned begins to fade within hours. Training without changed conditions does not produce changed behaviour. It produces temporarily informed people who return to established patterns.

The second theory focuses on structure. Reorganise. Replatform. Restructure the incentives, and the people will adapt. This too contains partial truth. But structural change without corresponding human development tends to produce compliance rather than capability — people who behave differently because the system demands it, not because they understand why, or could sustain it independently.

Both approaches treat change as something that can be delivered: a project with a start date, an end date, and a defined output. What neither fully addresses is this: the two need each other. The development of people and the creation of new structures must happen concurrently, each making the other possible. People need structures that support new behaviour; structures need people who understand why they exist. Neither works without the other, and neither can be sequenced away from the other.

There is a deeper assumption both models share that rarely gets examined: that the right design can be specified in advance and then executed. In predictable systems, this is often true. In unpredictable ones, it almost never is. The structures that actually serve an organisation's development cannot be blueprinted in advance — they have to emerge from experience, adapted incrementally in response to what is actually being learned. The organisation has to discover what it needs as it goes.

Matching the thinking to the challenge

Dave Snowden's Cynefin framework — developed at IBM in the late 1990s and now influential across government, defence, and business — offers a precise diagnosis. Snowden distinguishes between complicated and complex domains. In complicated systems, cause and effect are knowable; expert analysis can identify the right answer. In complex systems, cause and effect are only visible in retrospect; no amount of upfront analysis produces a reliable solution. A clock is complicated: it has hundreds of parts and requires expert knowledge to repair, but cause and effect are fully knowable — a skilled watchmaker can identify exactly why it stopped. Raising a child is complex: the same approach produces different results with different children, the child changes in response to how they are raised, and what worked last year may not work this year. Expertise helps, but does not guarantee outcomes.

Most organisational transformation is complex. And yet we approach it with complicated-domain tools: expert consultants, best practice methodologies, and defined delivery plans. Snowden is explicit about the consequence: in a complex system, setting a fixed future goal and pursuing it linearly produces unintended consequences that cannot be managed. What works instead is what he calls Probe–Sense–Respond — run safe-to-fail experiments, observe what emerges, and adjust continuously.

A common feature of conventional transformation programmes is the assumption that planning, action, and reflection belong in sequence — complete each phase before beginning the next. In complex systems, these happen concurrently. Reflection informs action in progress; action generates questions that reshape the approach. Insisting on a clean linear sequence severs the feedback loops through which the system actually learns.

This is not merely a theoretical distinction. It describes why so many well-designed, well-funded programmes fail despite everyone's best intentions.

A different model

Over 30 years I have been observing an approach that operates on fundamentally different principles. It is not designed in a conference room. It emerges from necessity — from repeated cycles of action and honest reflection on what actually holds under pressure. The understanding accumulates gradually through many iterations. And where these principles take hold, the results prove more durable than those of most planned programmes.

What follows are not prescriptions. They are patterns — features of this approach that I have observed consistently over many years. They are not a complete inventory; other principles will matter in other contexts. But they are the ones that have struck me as consistently significant, and consistently absent from the dominant model.

Reading current reality before prescribing solutions. Most transformation programmes begin from a picture of the target state — what good looks like, usually drawn from the consultant's experience elsewhere. The current state is mapped briefly, then set aside as an obstacle to be overcome on the way to the destination. What is lost in this framing is the specific conditions, history, and dynamics of the actual organisation in front of you. Peter Senge, in The Fifth Discipline, put it plainly: the energy for genuine change comes from the gap between a clear vision and an honest picture of current reality — but only if the current reality picture is accurate. When it is replaced by an externally imposed target state that has not been earned through actual diagnosis, that gap closes artificially, and the source of energy for change closes with it. The question worth learning to ask is not 'how do we reach the target state?' but 'given what is actually here — the capabilities, the relationships, the history, the constraints — what is the most fitting next step?' A consultant who arrives with the answer already written has already stopped seeing. And an organisation that learns to read its own reality clearly, and to ask that question for itself, has already changed in the way that matters most. Reading reality, however, is not only about where the organisation is. It is equally about the direction it is trying to move. The gap between the two — understood honestly — generates the learning questions that give purpose to inquiry: not 'what can we conclude from what happened?' but 'what do we most need to understand next, given where we are heading?' This is what gives the cycle of inquiry, action, and reflection its direction rather than simply its rhythm.

The pattern repeats in recognisable ways. Some time ago, a client set out to define what a great leader looks like in their context — theoretically, through workshops and frameworks — and then commissioned a programme to develop those leaders from scratch. A question that had not been asked, which to me was surprising, was: surely there are great leaders here already. What makes them such? Have we spoken to them? To their teams? In almost every case, the answer is no. The organisation's own existing capability goes unread because the starting assumption is deficit.

Systematising and disseminating learning, not just generating it. Most organisations treat learning as a byproduct — something that happens to individuals or teams in the course of a project, then dissipates when the team moves on, or the consultant departs. This model treats it as a primary output requiring deliberate infrastructure: mechanisms to gather what is being discovered at the front line, reflect on it collectively, distil the principles from the particular, and make them available in forms others can actually build on. The distinction matters at scale. A genuine disposition toward learning remains local if nothing carries what is learned to those who need it. Etienne Wenger's research on communities of practice found exactly this: learning that travels through networks of practitioners — carried by people who move between contexts, accumulated in shared language over time — produces a qualitatively different kind of capability than learning that stays inside teams or projects. The infrastructure this requires is not complicated, but it demands intentionality: regular spaces for collective reflection, processes for capturing what those reflections produce, and channels through which the resulting insights reach the people who need them.

The consulting model works against this almost by design. The practitioner gathers the learning — through interviews, observation, workshops — synthesises it into recommendations, and delivers the output. The organisation receives a conclusion; the capacity to generate it again independently is not transferred. An organisation that has learned to systematise its own learning has, in this sense, already begun to dissolve its dependence on that arrangement. Over time, a further possibility opens. Accumulated learning, carefully organised, begins to cohere into something more than a record of what worked: a developing conceptual framework — a working set of insights about why things work, and under what conditions. This is qualitatively different from documentation. It guides future action and is itself refined as experience accumulates. Held as insight rather than fixed theory, it frees practitioners to adapt intelligently to conditions that have not yet been encountered, without forcing novel situations into familiar categories.

What this requires in practice is more deliberate than it sounds. Learning at the front line does not systematise itself — it has to be actively captured. That means being present where experience is actually accumulating, not waiting for it to be reported upward. It means describing what is happening with enough accuracy and specificity that patterns begin to emerge when many such descriptions are placed alongside each other. As those descriptions multiply, analysis becomes possible: what conditions are producing what results, what is transferable and what is context-specific, what principles can be distilled from the particular. It is that analysis — not raw experience — that feeds the next cycle of inquiry and action. Without this deliberate work, learning remains trapped in the person or team that generated it and never reaches those who could build on it.

What distinguishes this from a knowledge management function or an L&D department is the direction of flow. In most organisations, learning is designed centrally and pushed outward. In this model, it arises at the front line — from experience, from reflection on action, from practitioners comparing notes — and travels in all directions: upward to inform strategy, laterally to practitioners in other contexts, and back again to those who generated it, enriched by what others have learned elsewhere. Leadership's role here is not to manage the learning but to tend it: to create the spaces in which genuine reflection can happen, to ensure that what is discovered in one place reaches those who need it elsewhere, and to cultivate the relationships through which understanding flows. You do not own an ecosystem. You create the conditions for it to thrive.

Learning as a permanent mode, not a project phase. The cycle of inquiry, action, and reflection repeats without end. Knowledge is not transferred from expert to recipient; it is generated through action and then systematised. Reginald Revans, the British physicist who pioneered action learning in the 1940s, made the point plainly: an organisation's rate of learning must at least equal its rate of change. This is not a phase with a completion date. It is a disposition — an orientation toward one's own practice that, once genuinely cultivated, does not stop when the programme ends. Underlying this is a recognition that genuine capability cannot be reduced to the mastery of a set of tasks, however well designed. What develops when the model is working is judgment — the capacity to read novel conditions and generate appropriate responses to situations that no training programme anticipated.

Understanding over compliance. Training that produces compliance succeeds on its own terms: behaviour changes while the system demands it. But compliance is fragile. Remove the external pressure — the reminders, the metrics, the visible accountability — and behaviour often reverts. Understanding produces something more durable. When people understand why a different approach is better — not in the abstract, but through having worked through it in practice and encountered its logic from the inside — they act from that understanding rather than from management. They adapt when conditions change, because the principle is internalised rather than the procedure. Most training programmes optimise for the transmission of knowledge and the rehearsal of skills. Neither of these is understanding. Understanding develops through application, through encountering resistance, and through honest reflection on why something that seemed clear in a workshop behaves differently in actual work.

Accompaniment over delivery. The word is chosen carefully. Accompaniment is not coaching, mentoring, or supervision. It is closer to two colleagues — one with more direct experience in a specific area of practice — learning together in the course of real work. Someone walks alongside you as you develop. If you stumble, someone is nearby. Mistakes are not managed — they are expected, and learned from. Confidence is built incrementally, through real work, not simulated exercises. This is not soft. Amy Edmondson's research at Harvard, corroborated by Google's Project Aristotle, demonstrates that psychological safety — the belief that mistakes will not be punished — is the strongest predictor of team learning behaviour, and learning behaviour is what drives sustained performance. And crucially: those who are accompanied, in time, become those who accompany others. The capacity becomes self-propagating — no longer dependent on the original external actor to sustain it. Nor does the system that supports this remain static. The structures, roles, and learning processes that enable it at greater scale emerge through the same cycle — reading what is actually happening, acting, reflecting, and adapting. New complexity is not designed in advance; it grows from experience.

What makes this work is the posture of the person accompanying. They arrive without a pre-formed assessment — with genuine curiosity about what the person is actually experiencing, not with a diagnosis already in hand. They are present in the actual work, not reviewing it after the fact. Rather than advising, they help the person see what they are doing and draw their own conclusions from it. And they are genuinely changed by the encounter: if they leave having learned nothing, something has gone wrong. Think of a senior practitioner sitting with a colleague through difficult, unfamiliar work — not to assess, not to debrief afterward, but to be present as things unfold and to reflect together on what was discovered. Both leave with something they did not have before. The accompanier's task is not to teach but to create the conditions in which learning can happen — a fundamentally different orientation to knowledge and to people.

Dissolving the expert–client boundary. In this model, everyone is simultaneously a learner and a contributor. The person accompanying you is also being developed. Knowledge flows in both directions. This inverts the standard consulting relationship — in which the expert arrives with answers and the client receives them — and replaces it with something more honest about how understanding actually develops. It also dissolves a subtle but consequential hierarchy: the assumption that development is something the more capable person does to the less capable one.

Protagonists, not recipients. This is perhaps the most important distinction, and the one most absent from conventional change management thinking. The goal is not to produce people who comply with a new system, or who have become competent technicians of a methodology. It is to develop people who have become confident protagonists of their own development and their organisation's — capable of identifying problems, generating responses, and sustaining the process of learning without external prompting. The test of whether change has genuinely occurred is not whether behaviour changed while you were present. It is whether something has been created that the people inside it feel compelled to sustain and develop of their own accord. If people have to be reminded to keep going, the work is not finished. If they cannot imagine stopping, it is. A related test: are they the experimenters, or are they the experiment? An organisation running its own tests, learning from them, and adjusting has already changed in the way that matters. One that is the subject of someone else's intervention has not.

But what about results?

The economic objection arrives quickly: boards want results in quarters, not years. Long-term thinking is a luxury. The commercial model that most organisations inhabit does not have patience built into it.

The accompaniment model is not inherently slower in the ways that matter. What it replaces is not speed — it replaces repeated failure. The cost of a transformation programme that does not hold, followed by another that does not hold, followed by another, is not short-term efficiency. It is a large and recurring waste. The relevant question is not whether organisations can afford to do this properly. It is whether they can afford to keep doing it wrong.

There is also a structural problem worth naming. The commercial model through which most consulting is sold creates a tension that is rarely acknowledged: engagements are scoped as projects, measured by deliverables, and ended when the output is produced — not when internal capability is genuinely established. What is typically transferred is the conclusion. The capacity to generate it independently rarely features in the brief.

A subtler variant is harder to see because it looks like care. Well-intentioned support, extended beyond the point where it is needed, becomes its own kind of disempowerment. The person being supported never gets to discover whether they can manage without it. What began as scaffolding becomes a ceiling — the very presence of support prevents the confidence it was meant to build from taking root. The organisations that understand this are scrupulous about withdrawal — stepping back deliberately, and in time, toward the goal of people who no longer need them.

The most fundamental point goes beyond incentives. Genuine transformation in an organisation is an organic process — it must be propelled from within. External actors can accompany it, can catalyse it, can create conditions that help it along. But they cannot manage it. The evidence for this is empirical as much as theoretical: time and again, the growth that external-led programmes produce holds only as long as the external actor is present. The moment the visits stop, the campaigns end, the consultants leave, the momentum dissipates. What was being sustained by external energy, not internal force, reverts. This is not a failure of execution. It is the predictable consequence of treating a process that must arise from within as if it can be driven from outside.

Changing that requires something more than a new methodology. It requires a different relationship to time, to credit, and to what counts as success.

What this actually asks

I have watched this pattern work for 30 years — not in a controlled experiment, but in contexts navigating real complexity with genuine stakes.

The entry point looks different in every context. But it always begins in the same place: an honest reading of what is actually there — the specific capabilities, relationships, and constraints — and an equally honest sense of the direction the organisation is trying to move. Both together generate the learning questions.

There is a third dimension to what this asks — one that goes beyond organisations and consultants. It asks something of individuals too: not just willingness to participate, but a genuine orientation toward their own development and their colleagues'. The model only functions when people understand these as inseparable — that you develop by helping others develop, that learning shared is learning deepened, that humility about what you do not yet know is not weakness but the condition that makes real learning possible. Patrick Lencioni, in The Ideal Team Player, identifies humility as the foundational quality of effective team members — 'the single greatest and most indispensable attribute of being a team player' — the one without which the drive to perform and the awareness of others cannot fully function. The model described here rests on the same recognition. These qualities cannot be installed by a programme. They are, in fact, exactly what the model, at its best, cultivates. And when this orientation takes hold, something else tends to happen: apathy, possessiveness, paternalism, and competitiveness — the behaviours that corrode genuine learning — begin to recede. Not because they are managed out, but because they are not the behaviours that produce the results people are actually seeking.

These principles are not a menu. They are mutually dependent. Accompaniment without the goal of protagonism produces a more humane form of dependency. Systematising learning without reading current reality produces more sophisticated versions of the same wrong answer. The coherence between the elements is the source of the model's power. An organisation that adopts pieces of this selectively will get partial results. One that understands how the pieces ask for each other will find the whole is greater than the sum.

← Back to Reflections