Classical Change Models and When Each Applies

COMPEL Specialization — AITM-CMD: AI Change Management Associate Article 3 of 11

Pick up any mature change-management practice and you will find a favoured model. Prosci shops run ADKAR. Former management consultants reach for Kotter. Human-resources and coaching practices lean on Bridges. Academic curricula still start with Lewin. Each of the four models is a genuine contribution to the field; each has evidence behind it; none is a fraud. The professional error is not using one of them — the professional error is defaulting to one of them for every situation. An AI change practitioner who can only run ADKAR will misdiagnose an organisational-coalition problem as an individual-awareness problem. A practitioner who can only run Kotter will over-scope a narrow team-level transition into a full-organisation change campaign. This article teaches ADKAR, Kotter 8 Steps, Bridges Transition Model, and Lewin’s unfreeze-change-refreeze, names the problem space each solves best, shows where they agree and where they disagree, and gives the practitioner a decision procedure for choosing between them in AI contexts.

Four models in one sentence each

Before the depth, the summary. ADKAR is a five-stage model of the individual’s path through change (Awareness, Desire, Knowledge, Ability, Reinforcement). Kotter 8 Steps is a sequenced organisational campaign with a coalition, vision, communication, empowerment, wins, acceleration, and anchoring. Bridges Transition Model is a three-phase psychological map of transitions (Ending, Neutral Zone, New Beginning) that runs parallel to any external change. Lewin’s unfreeze-change-refreeze is a three-state framing of organisational change that predates all the others by decades. The four models operate at different levels (individual, organisational, psychological, systemic) and on different timescales. They complement each other more than they compete.

[DIAGRAM: TimelineDiagram — four-models-mapped-onto-transformation-arc — horizontal timeline of a representative AI transformation (pre-launch, launch, early months, sustaining months) with four rows showing ADKAR stages, Kotter steps, Bridges phases, and Lewin states mapped onto the arc; primitive teaches the models as layered rather than alternative.]

ADKAR — the individual path

ADKAR, developed by Prosci in the late 1990s, asks a narrow and useful question: what does any single person need to pass through to absorb a change?¹ Five stages answer the question. Awareness of the reason for the change. Desire to participate in the change. Knowledge of how to change. Ability to implement the change. Reinforcement to sustain the change.

ADKAR is strongest when the practitioner needs to diagnose why a specific individual or team has stalled. An employee using an AI tool sporadically but not integrating it into workflow may be stuck at Ability (training gap) or at Reinforcement (no incentive or feedback loop), and the two diagnoses require different interventions. A manager who understands the programme but has not begun to coach their team may be stuck at Desire (they know the change is coming but do not see how it benefits them or their team). ADKAR’s value lies in its diagnostic precision at the individual level.

ADKAR is weakest when the problem is not an individual problem. If the organisation’s strategic rationale is unclear or the coalition required to carry the change is absent, ADKAR will misdiagnose the situation. Running an awareness campaign when the underlying problem is coalition weakness produces a polished communication programme that does not move the change forward.

The AI application is direct. For role-by-role change where the job changes materially — accountants using a generative draft assistant, customer-service agents using an AI agent-assist tool, developers using code-completion — ADKAR organises the change at the individual level. The practitioner maps each stage to observable evidence, runs interventions for each stage that is stuck, and moves the individual through.

Kotter 8 Steps — the organisational campaign

Kotter’s 8 Steps, articulated in Leading Change (1996, updated 2012), runs at the organisational scale rather than the individual scale.² The eight steps — establish urgency, form the guiding coalition, create the vision, communicate the vision, empower broad-based action, generate short-term wins, consolidate gains and produce more change, anchor new approaches in the culture — describe a sequenced campaign rather than an individual’s journey.

Kotter is strongest when the change is genuinely organisational in scope: a new operating model, a new strategic direction, a culture shift. The coalition step is the distinguishing element — Kotter insists that change of this scope requires a cross-functional coalition of leaders with the credibility and authority to carry it. An AI transformation that affects multiple business units, changes the organisation’s operating model, or requires cultural shifts around human-AI collaboration is genuinely Kotter-scale work.

Kotter is weakest when the change is narrow or primarily technical. Running a full eight-step campaign for a ten-person team adopting a new code-generation tool is overwrought; the coalition, vision, and wins machinery is disproportionate to the problem. ADKAR or a simple rollout checklist serves better in that scenario.

The AI application is programme-level. An enterprise-wide AI transformation that touches strategy, operating model, culture, and technology stack is Kotter’s domain. The practitioner builds the coalition explicitly, authors a vision statement that distinguishes the AI transformation from prior digital ones, designs the communication cascade that Article 6 will develop, plans deliberate early wins that Article 9 will develop, and anchors the change in performance-management and cultural artifacts.

Bridges Transition Model — the psychological parallel

William Bridges observed that external change (the event) and internal transition (the psychological process of letting go of the old and accepting the new) run on different tracks and different timescales.³ His three-phase model — Ending (letting go of the prior state), Neutral Zone (the in-between), New Beginning (engaging the new state) — maps the psychological transition that accompanies any external change.

Bridges is strongest when the external change involves loss — of identity, of role, of competence, of status. AI transformation frequently involves exactly this kind of loss. An experienced accountant who wrote variance commentaries for thirty years is not losing the job when the organisation adopts a generative assistant; the accountant is losing a form of craft identity that the old workflow carried. A call-centre agent whose call volumes drop because of successful automation is losing a form of professional rhythm even if the job title remains. The Ending phase is where much of the resistance taught in Article 4 originates, and a practitioner who does not name the Ending explicitly will often misread the resistance as obstruction.

Bridges is weakest when the change does not involve meaningful psychological loss — when it is a pure capability addition with no identity or role displacement. A developer adding code-completion to an existing workflow does not typically experience a Bridges-scale transition; the developer simply has a new tool. Over-applying Bridges to such cases produces counselling where coaching would do.

The AI application is diagnostic. The practitioner asks, for each affected role, “what is ending here that mattered?” and designs the change experience to honour the Ending before pushing the New Beginning. Rituals of recognition for the prior craft, visible acknowledgment that the role is changing and not simply “being enhanced”, explicit permission for the Neutral Zone confusion — all of these are Bridges-informed practices that materially reduce the resistance the programme will later have to process.

Lewin’s unfreeze-change-refreeze — the systemic frame

Kurt Lewin, writing in 1947, proposed that organisational change moves through three states: unfreezing the current equilibrium, changing to a new state, and refreezing the new state as the new equilibrium.⁴ The language is old and the model predates the others by fifty years, but the underlying insight — that behaviour is held in place by a force field of reinforcing pressures, that change requires reducing those forces before new patterns can establish, and that new patterns require their own reinforcing pressures to persist — remains sound.

Lewin is strongest as a framing tool for the sponsor conversation. Explaining to an executive sponsor that their organisation’s current behaviour is held in place by a force field of reinforcing pressures (performance incentives, hiring patterns, cultural norms, legacy processes) is a more persuasive intervention than explaining that the organisation needs awareness-building. Lewin’s language moves the conversation to the systemic level.

Lewin is weakest as an operational guide. It is too coarse to organise daily practitioner work. Unfreeze-change-refreeze does not tell the practitioner which stakeholders to engage, what interventions to run, or how to measure progress. For operational work, the practitioner drops back into ADKAR, Kotter, or Bridges.

The AI application is at the sponsor and culture layer. When the sponsor asks why the programme is investing in behaviour-reinforcement mechanisms rather than running another training campaign, the Lewin frame carries the explanation: if the new behaviour is not refrozen, the force field will pull the organisation back to the prior equilibrium.

Where the models agree and disagree

The models agree on three substantive points. Change requires motivation before capability — ADKAR’s Awareness and Desire precede Knowledge and Ability, Kotter’s urgency precedes execution, Bridges’ Ending precedes New Beginning. Change requires reinforcement to persist — ADKAR’s Reinforcement, Kotter’s anchoring, Lewin’s refreeze all address the same phenomenon. Change proceeds in phases rather than as a single event — all four models reject the “big-bang” view.

They disagree on the primary unit of analysis. ADKAR is individual-scale. Kotter is organisational-scale. Bridges is psychological-scale. Lewin is systemic-scale. A practitioner who treats them as rival models picks one and loses the others’ insights. A practitioner who treats them as layered lenses uses each where it sees best.

[DIAGRAM: MatrixDiagram — models-on-focus-axes — 2×2 with axes “Individual focus vs. organisational focus” (horizontal) and “Behaviour focus vs. transition focus” (vertical); ADKAR placed in individual/behaviour, Kotter in organisational/behaviour, Bridges in individual/transition, Lewin in organisational/transition; primitive makes the models’ complementarity explicit.]

A decision procedure for AI contexts

When a practitioner opens an engagement, the choice of model is a choice of lens for the specific problem at hand, not a one-time commitment. A simple decision procedure helps.

If the question is “why has this specific role or team not adopted the AI tool”, start with ADKAR. Diagnose which stage the individual or team is stuck at, and design the intervention for that stage.

If the question is “how do we run a transformation that reshapes the operating model, strategy, or culture across the enterprise”, start with Kotter. Build the coalition, craft the vision, plan the cascade.

If the question is “why are experienced employees resisting this change in ways that don’t look like pure obstruction”, start with Bridges. Diagnose what is ending, honour the Ending, plan for the Neutral Zone.

If the question is “why does our culture keep pulling us back to the old behaviour despite our change efforts”, start with Lewin. Examine the force field, identify which pressures the programme has failed to alter.

A mature practitioner uses all four models across a single programme, moving between lenses as the question changes. A practitioner who runs a single model throughout the engagement is practising a discipline narrower than the credential certifies.

Summary

ADKAR, Kotter, Bridges, and Lewin are four genuine models operating at four different levels of analysis. ADKAR organises individual change. Kotter organises organisational campaigns. Bridges maps the psychological transition. Lewin frames the systemic force field. Each is strong in its domain and weak outside it. A practitioner chooses between them for reasons that match the question at hand, and often uses several across a single programme. Article 4 turns to the AI-specific resistance patterns the models collectively help explain, and teaches the practitioner to distinguish legitimate objection from status-quo bias.

Cross-references to the COMPEL Core Stream:

• EATF-Level-1/M1.6-Art05-Change-Management-for-AI-Transformation.md — primary framework article naming the classical models within the COMPEL cycle
• EATF-Level-1/M1.1-Art09-AI-Transformation-and-Organizational-Culture.md — cultural-change foundations that inform the Lewin and Kotter applications discussed here

Q-RUBRIC self-score: 89/100

© FlowRidge.io — COMPEL AI Transformation Methodology. All rights reserved.