Every change programme seems to arrive with its own map. A curve to climb: denial, resistance, acceptance. A sequence to follow: unfreeze, change, refreeze. Eight steps to execute in order. Five building blocks to install in each person, one at a time. Kübler-Ross’s change curve, Lewin’s three-step model, Kotter’s eight steps, Prosci’s ADKAR: different vocabularies, the same underlying promise. Follow the stages, in the right order, and the change will land.

It is an appealing promise, especially to leaders who are anxious about a transformation they don’t fully control. A staged model turns something messy and human into something that looks manageable: a project plan with milestones, not an uncertain unfolding of people’s beliefs, habits, and relationships. It gives facilitators a slide, gives sponsors a sense of progress, and gives everyone involved a shared language for what is otherwise a disorienting experience.

These models also persist because there is a kernel of truth in each of them. People do resist before they accept. Old habits do need loosening before new ones take hold. Momentum does need building before it can be sustained. Awareness does need to come before desire, and desire before ability. The stages describe things that genuinely happen. What they get wrong is the promise built on top of that description: that because these experiences occur, they occur in order, on a predictable timetable, and that leading change is therefore a matter of moving a group through the correct sequence.

The evidence, and the origins of these models, tell a more complicated story. Each one was built for a narrower purpose than the one it is now asked to serve, and each has been stretched well past the conditions under which it was first observed.

This article explores why the myth of predictable stages endures, what it costs organisations that rely on it, what the research and the models’ own origins actually show, and what leaders can do instead when they are asked to guide a change that will not sit still long enough to be staged.

Why the myth endures

Part of the appeal is simply relief. Leading change without a map is uncomfortable, and a staged model offers something to hold onto when the outcome is genuinely uncertain. If a leader can say “we are in the resistance phase, this is expected, it will pass,” anxiety drops, for the leader and often for the room. The stage becomes a container for feelings that would otherwise seem alarming or personal. This is not a trivial benefit. But it is a benefit that comes from naming an experience, not from predicting one.

Part of the appeal is institutional. Organisations reward the appearance of control. A change plan built around clear stages photographs well in a steering committee deck. It gives sponsors something to report against: “we have completed the awareness phase and moved into desire.” It allows consultancies to sell a repeatable methodology rather than a bespoke, uncertain engagement. None of these incentives have much to do with whether the stages actually predict what will happen in a specific organisation, with specific people, in a specific moment. They have to do with what is legible to the people funding and overseeing the change.

Part of the appeal is historical residue. Kurt Lewin developed his three-step model in the 1940s, studying groups in relatively stable, slower-moving social and organisational settings. Unfreeze the current state, move to the new one, refreeze it into place: this made real sense in a world where organisations could plausibly expect to arrive at a new steady state and stay there for a while. That expectation of eventual stability has not disappeared from how change is talked about, even though very few organisations now experience anything like a steady state to refreeze into. Markets, technology, and workforce expectations keep moving underneath every “final” state a change programme declares finished.

Part of the appeal is borrowed authority. The change curve most facilitators still draw on a whiteboard traces back to Elisabeth Kübler-Ross’s 1969 work with terminally ill patients: denial, anger, bargaining, depression, acceptance (Kübler-Ross, 1969). It carries the weight of a landmark, humane piece of writing about death. But it was not a validated psychological model, let alone one built for organisational change. Kübler-Ross described her book as a series of dialogues intended to help clinicians understand what dying patients needed, not a stage theory to be applied elsewhere. She later cautioned that the stages “were never meant to help tuck messy emotions into neat packages.” Not everyone who is dying moves through all five stages, and those who do rarely move through them in that order. Management writers lifted the shape of the curve out of that context and relabelled the object of grief as a merger, a restructure, or a new system rollout. The stages stayed the same; what people were supposedly grieving simply changed.

Kotter’s eight-step model carries its own kind of borrowed authority: a widely repeated statistic. Kotter’s original research found that roughly 70% of major change efforts fail to achieve their goals (Kotter, 1995), a figure a 2013 McKinsey survey later found similar support for. That statistic has been repeated so often, in so many keynote decks, that it now functions as settled fact, and the model built to fix it inherits some of that certainty by association. Few of the people quoting the 70% figure have read the systematic review that later tested the model itself against the wider literature.

Together, these threads, relief, institutional reward, historical residue, and borrowed authority, keep the myth in circulation. It is not sustained by evidence that the stages reliably predict outcomes. It is sustained by how useful the idea is to the people telling the story.

The cost of believing this works

When leaders treat staged models as forecasts rather than descriptions, several costs follow.

The first is misdiagnosis. If a leader believes a team should have moved from “resistance” to “acceptance” by now, and it hasn’t, the model invites a judgment about the people rather than a question about the system. Individuals get labelled as stuck, difficult, or not on board, when what is actually happening may be a legitimate response to conditions that have shifted since the plan was drawn up. The stage becomes a diagnosis of character instead of a signal to investigate.

The second is false closure. Lewin’s refreeze step assumes a system can be paused, altered, and then resolidified into a durable new state. Leaders who believe this declare victory, dismantle the temporary structures built to support the transition, and move attention elsewhere. In a genuinely dynamic environment, this handoff often happens just as new pressures are emerging that the “refrozen” state cannot absorb. What looked like the end of the change project turns out to be the start of a new, unplanned one, except now nobody is watching for it.

The third is wasted specificity. ADKAR narrows the focus to the individual: Awareness, Desire, Knowledge, Ability, Reinforcement. That focus is genuinely useful for helping a single person move through a discrete transition, and it gives managers something concrete to check for. But applied to organisational or cultural change, it can create the illusion that if enough individuals have been walked through the five letters, the organisation has changed. Critics of the model point out that it can neglect the group-level and organisational dynamics, culture, power, incentives, informal networks, that determine whether an individual’s new behaviour survives contact with the wider system. A change project can tick every ADKAR box for hundreds of people and still fail, because the thing that needed to shift was never at the individual level to begin with.

The fourth cost is the one Kotter’s model itself has quietly absorbed. Appelbaum and colleagues conducted a systematic review of the literature relating to each of Kotter’s eight steps, checking how much support existed for the model fifteen years after publication (Appelbaum et al., 2012). They found reasonable evidence for several individual steps, but no study that validated the model as a whole, as a sequence that reliably produces successful change when followed in order. In other words, the most famous failure statistic in change management launched a model whose own sequential logic has never been shown to be the thing that prevents that failure. Organisations that follow the eight steps in strict order, believing the sequence itself is the active ingredient, may be getting the benefit of some genuinely useful individual practices, urgency, coalition-building, communication, while gaining nothing from the ordering they have been told is essential.

The fifth cost is cultural. When a model implies there is a correct emotional and behavioural sequence, deviation from it starts to look like failure, resistance, or lack of readiness, rather than information. People whose response to a change doesn’t match the expected stage learn to perform the expected stage rather than report their actual experience. The organisation loses exactly the signal it most needs: an honest account of what is really happening, as opposed to what the model says should be happening by now.

What the research says about how change really works

The research on organisational change has moved a long way from the assumption that change is a controlled sequence executed against a group of people. It increasingly describes change as something that emerges from ongoing interaction, not something installed in stages from outside.

Cummings, Bridgman, and Brown revisited Lewin’s own writing and found that the crisp three-step “unfreeze-change-refreeze” model widely taught in management courses is largely a later simplification, built up by textbook writers over decades, of a more nuanced and continuous view of group dynamics that Lewin himself held (Cummings, Bridgman and Brown, 2016). The model most leaders learned is not quite the model Lewin proposed; it is a flattened, more linear version that better suited the tidy diagrams of later management texts than the messier original theory.

Appelbaum and colleagues’ review of Kotter’s model, discussed above, reached a similarly deflating conclusion for a different reason: individual steps have support, but the sequence as a totality does not (Appelbaum et al., 2012). Change efforts that succeed tend to involve many of the ingredients Kotter names, urgency, a guiding coalition, a vision, short-term wins, but not necessarily in the fixed order the eight-step diagram implies, and not as a guarantee that doing them in order produces the result.

The ADKAR model’s own creator, Jeff Hiatt, built it specifically as a model of individual change, drawing on research into how people personally adopt new behaviours (Hiatt, 2006). Later critics, applying it at organisational scale, have noted that its sequential structure lends itself to a known endpoint and a linear process, when real organisational change usually involves stages that overlap, occur simultaneously, or need to be revisited as circumstances shift. The model was never designed to carry the weight of whole-system transformation, and much of its apparent failure in that role is really a mismatch of scale rather than a flaw in the original idea.

Ralph Stacey’s work offers the clearest alternative account. Stacey describes organisations as complex responsive processes: patterns of ongoing interaction between people, not machines that can be repositioned from one state to another by an external operator (Stacey, 2011). On this view, change is not something a leader does to a system from outside; it is something that emerges from the continuing conversations, relationships, and negotiations that make up the system itself. Staged models implicitly treat the organisation as an object being moved through a sequence. Stacey’s account treats it as a living process that a leader participates in, and can influence, but cannot fully script in advance.

The Cynefin framework, developed by Dave Snowden, offers a practical way of seeing why staged models sometimes work and sometimes don’t (Snowden and Boone, 2007). In complicated situations, where cause and effect are knowable in advance through analysis or expertise, staged plans and established good practice genuinely help; a checklist can carry you through a known sequence reliably. In complex situations, the ones most organisational, cultural, and strategic change actually live in, cause and effect are only clear in hindsight. There is no reliable staged sequence to follow, because the system itself is still forming its response as the change unfolds. The appropriate move in a complex situation is to probe, sense what is emerging, and respond, repeatedly, rather than to execute a pre-agreed script from step one to step eight.

Taken together, this research doesn’t dismiss the value of naming stages of experience. It relocates that value. Stages describe things that happen to some people, some of the time, in some circumstances. They stop being reliable the moment they are treated as a forecast for a specific, complex, still-unfolding situation.

What leaders can do

If the stages cannot be trusted to predict how a specific change will unfold, what can leaders actually do differently? The answer is not to abandon these models, each names something real, but to change how they are held and used.

Six practices support this shift.

1. Use the models as vocabulary, not itinerary

Kübler-Ross’s stages, Lewin’s unfreeze, Kotter’s steps, and ADKAR’s building blocks are useful language for naming what someone might be experiencing. They are far less useful as a fixed order you expect a whole organisation to move through on schedule. Introduce them to a team as possible experiences, not as a itinerary: “some of you may recognise some of these reactions; you may not experience all of them, or in this order, and that’s normal,” rather than “we are now entering the acceptance phase.”

2. Watch for multiple stages happening at once

Expect different parts of the organisation, and different individuals within the same team, to be in different places simultaneously, and expect some to be in several states at once. Build your check-ins to surface this variation rather than to confirm the group has “moved on” together. A single team meeting can contain people who are still resisting, others who are already experimenting with the new way of working, and others who have quietly reverted. Staged thinking hides this; paying attention to it reveals what is actually needed next.

3. Replace the refreeze with an ongoing rhythm of sensing

Rather than declaring a change complete and moving attention elsewhere, build a standing rhythm of short reviews that keep asking what is emerging, what has drifted, and what new pressures have appeared since the plan was written. Treat any apparent stability as temporary and worth checking on, not as a finish line. This is closer to what Cynefin recommends for complex situations: probe, sense, respond, on a loop, rather than plan, execute, freeze.

4. Pair individual-level tools with organisational-level attention

If you use ADKAR or a similar model to support individuals through a transition, pair it deliberately with attention to the group and system around them: incentives, informal norms, power structures, and the behaviour of other leaders. An individual’s awareness, desire, knowledge, ability, and reinforcement can all be present and still fail to produce lasting change if the surrounding system rewards the old behaviour. Ask not only “has this person moved through the stages?” but “does the system around them make the new behaviour easy to sustain?”

5. Treat the 70% failure statistic as a prompt to investigate, not a reason to follow a formula more strictly

Kotter’s original finding, and the figures that have echoed it since, describe how often change efforts fail; they don’t validate any particular sequence as the fix. Rather than tightening adherence to eight steps in strict order, use the individual practices with good evidence behind them, building urgency, forming a genuine coalition, communicating relentlessly, and stay alert to whether the specific change you are leading needs them in a different order, or repeated, or run in parallel.

6. Build the capacity to sense, not just the capacity to plan

Complex change asks less for a better plan and more for a better sensing apparatus: genuine channels for hearing what people are actually experiencing, forums where patterns can be named honestly, and leaders willing to treat surprising feedback as useful information rather than evidence that someone hasn’t caught up with the stage they’re supposed to be in. This is slower and less tidy than a diagram with eight boxes and an arrow through them. It is also closer to how change actually happens.

At the heart of these practices is a different idea of what a leader is doing when they lead change. It is not moving people through a known sequence toward a known end state. It is staying in close enough contact with a living system to notice what it is telling you, and responding as it unfolds.

Closing reflection

The staged models of change; Kübler-Ross’s curve, Lewin’s freeze and refreeze, Kotter’s eight steps, ADKAR’s five letters, each came from somewhere specific: a clinician’s dialogues with dying patients, a mid-century study of relatively stable groups, a ten-year survey of failed transformations, a model built for individual adoption. Each has since been generalised far beyond that origin, and dressed up as a universal sequence that any organisation can expect to move through in order.

The research doesn’t support that generalisation. Kübler-Ross’s own stages were never validated as a predictable sequence, even in the context they were built for. Lewin’s tidy three steps turn out to be a later flattening of a more continuous original idea. Kotter’s individual steps have evidence behind them; the sequence as a whole does not. ADKAR does what it was built to do for one person at a time, and strains when asked to carry a whole organisation.

None of this means the language of stages is worthless. It means the language should be held loosely, as a way of naming what might be happening, not as a forecast of what will happen next, in order, on schedule. The organisations that navigate change well are not the ones that follow the diagram most faithfully. They are the ones that stay closely enough in touch with what is actually unfolding to respond to it, again and again, long after the slide with the eight boxes has been put away.

Reflection questions

  • Which of these models do you find yourself reaching for most often, and where has it let you down?
  • Have you found a better way of naming what people are going through during change, without pretending you can predict it?
  • Do you have resources or practices that have helped you sense what’s actually happening, rather than what a model says should be happening?

Thanks for reading!

Sources

Appelbaum, S.H., Habashy, S., Malo, J.L. and Shafiq, H. (2012) ‘Back to the future: revisiting Kotter’s 1996 change model’, Journal of Management Development, 31(8), pp. 764–782.

Cummings, S., Bridgman, T. and Brown, K.G. (2016) ‘Unfreezing change as three steps: Rethinking Kurt Lewin’s legacy for change management’, Human Relations, 69(1), pp. 33–60.

Hiatt, J. (2006) ADKAR: A Model for Change in Business, Government and Our Community. Loveland, CO: Prosci Learning Center.

Kotter, J.P. (1995) ‘Leading Change: Why Transformation Efforts Fail’, Harvard Business Review, 73(2), pp. 59–67.

Kübler-Ross, E. (1969) On Death and Dying. New York: Macmillan.

Snowden, D.J. and Boone, M.E. (2007) ‘A Leader’s Framework for Decision Making’, Harvard Business Review, 85(11), pp. 68–76.

Stacey, R.D. (2011) Strategic Management and Organisational Dynamics: the challenge of complexity to ways of thinking about organisations. 6th edn. Harlow: Pearson.