What Are the 4 Stages of Change in Organizations?
The 4 stages of organizational change are: unfreezing (recognizing the status quo is no longer viable), transition (experimenting with new ways of working), refreezing (anchoring new behaviors), and evaluation (measuring what has actually changed). This model, rooted in Kurt Lewin’s work, remains the most operational framework for leading a real transformation.
Stage 1: Unfreezing — Accepting That Things Can’t Stay the Same
Lewin called it “unfreeze.” The idea is straightforward: before changing anything, you need to create the psychological and organizational conditions that make change possible.
In practice, that means your teams need to understand why the current model is insufficient. Not just hear it in a PowerPoint presentation. Actually understand it.
In the AI integration projects I work on, this is consistently the most underestimated stage. Leaders want to jump straight to the tools. They forget that if their teams don’t see the problem, they’ll never embrace the solution.
A successful unfreezing produces a clear signal: people start asking questions. Not silently resisting.
Stage 2: Transition — The Most Uncomfortable Phase
This is where everything is unclear. Old ways of working no longer function. New ones aren’t yet mastered. This is where most transformation projects fail.
Why? Because managers confuse speed with progress. They push on deadlines at the exact moment teams need time to experiment, make mistakes, and adjust.
In the AI context, this phase corresponds to what I observe with clients: teams test tools, some adopt, others work around them. Unsanctioned AI proliferates. Processes become hybrid, sometimes incoherent.
That’s normal. Even healthy. As long as clear guardrails were established from the start.
I’ve built a 6-dimension diagnostic framework to assess exactly where your organization stands in this transition phase. Download the AI Board Pack 2026.
Stage 3: Refreezing — Anchoring What Works
Lewin called this “refreeze.” The goal: stabilize new behaviors before the organization reverts to old reflexes.
This is the stage that Kotter, in his 8-step model, breaks into several sub-phases: consolidating gains, producing short-term wins, institutionalizing new practices. Kotter is right on the substance. But for a leader managing a real transformation, the central question remains the same: what do we make permanent?
Concrete answers: newly documented processes, redefined roles, performance indicators aligned with new ways of working. Not slides. Decisions.
In AI projects, refreezing often happens through AI governance: who validates what, who is accountable for algorithm-assisted decisions, which use cases are authorized. As I explained in my analysis of the 4 types of AI, the category of AI you deploy directly determines the level of governance you need.
Stage 4: Evaluation — The Only Honest Stage
Most change management models stop at refreezing. That’s a mistake.
Evaluation isn’t a compliance audit. It’s a simple question asked without complacency: have behaviors actually changed, or did we just change the presentation slides?
The indicators to watch aren’t deployment indicators (how many employees were trained, how many tools were installed). They’re real adoption indicators: are decisions being made differently? Are processes producing different results?
This is where skills development takes on its full meaning. Based on market signals I track, 2027 budgets are concentrating on rare skills and adaptation to new tools. Organizations that succeed at this evaluation stage will be those that invested in AI culture, not just software licenses.
On this topic, my article on which jobs will survive AI provides complementary insight on what this means for your teams.
What These 4 Stages Mean for You, Concretely
Lewin’s model isn’t academic theory. It’s a diagnostic tool.
When a transformation fails, it almost always fails at one of these 4 stages. Either unfreezing didn’t happen (teams didn’t understand the why). Or the transition was rushed (not enough time, no guardrails). Or refreezing wasn’t formalized (new behaviors weren’t anchored). Or evaluation was cosmetic.
Identifying where you are is half the work.
If you’re a CHRO or CEO looking to structure your change management approach, particularly in an AI integration context, request a free diagnostic.
FAQ
What is the difference between Lewin’s model and Kotter’s?
Lewin proposes 3 phases (unfreezing, transition, refreezing) with a psychological and organizational logic. Kotter develops 8 more operational steps, centered on leadership and communication. Both are complementary. Lewin provides the framework, Kotter provides the levers.
How long does each stage of change take?
There is no standard duration. Unfreezing can take a few weeks in a reactive SME, several months in a large corporation. The transition is often the longest. What matters is not burning through stages under deadline pressure.
How do you manage resistance to change in teams?
Resistance is almost always a signal of incomplete unfreezing. People resist when they don’t understand the why, or when they don’t feel heard during the transition phase. The answer isn’t to force. It’s to listen to what the resistance is saying.
Do the 4 stages of change apply to AI projects?
Yes, and that’s precisely where they’re most useful. AI projects rarely fail for technical reasons. They fail because unfreezing didn’t happen, because the transition wasn’t supported, or because AI governance wasn’t formalized at the refreezing stage.