Change ManagementBeginner5 min read

Lewin Three Stages

Kurt Lewin's three-stage model is the original (1947) and most enduring framework for organizational change: (1) Unfreeze — destabilize the existing equilibrium by challenging current beliefs, surfacing dissatisfaction, and creating a sense of urgency. (2) Change — introduce new behaviors, processes, and structures while people are in a malleable state. (3) Refreeze — institutionalize the new state through reinforcement, new norms, and updated systems so it becomes the new equilibrium. The model's central insight: humans default to homeostasis. You cannot just 'add' new behavior to an organization — you must first weaken the old equilibrium, then install the new one, then lock it in. Most change failures happen because leaders skip Unfreeze (assume people are already ready) or skip Refreeze (declare victory too early).

Also known asUnfreeze-Change-RefreezeLewin's Change ModelThree-Step Change Model

Challenge a friend Browse library

The Trap

The biggest trap is the modern dismissal: 'Lewin is too simple, the world moves too fast, organizations should be in constant change.' This is wrong on two counts. First, simple models endure precisely because the underlying dynamics are simple — humans require unfreezing, regardless of how fast the environment moves. Second, 'constant change' organizations almost always have terrible adoption because nothing ever Refreezes — every initiative is half-melted. The other trap: rushing Unfreeze. Leaders create urgency through fear (layoffs, threats) which destabilizes too violently — productivity collapses and engagement craters before the Change phase can begin. Effective Unfreeze is uncomfortable, not catastrophic.

What to Do

For any major change, explicitly plan all three phases with distinct success criteria. Unfreeze (4-12 weeks): goal is shifting beliefs and creating openness. Measure: % of org expressing 'we need to change' in pulse surveys. Tactics: data dump on competitive threat, customer pain stories, executive vulnerability. Change (3-12 months): goal is new behaviors performed visibly. Measure: behavior adoption rate, not training completion. Tactics: pilots, coaches, new tools, daily routines. Refreeze (6-24 months): goal is the new state becoming 'the way we work here.' Measure: behavior persistence after sponsor attention is removed. Tactics: institutionalize in performance reviews, hiring criteria, promotion paths, and onboarding.

Formula

Change Durability = Quality of Refreeze ÷ Time Since Active Sponsorship — without refreezing, behavior decays at ~30% every 6 months

In Practice

Lou Gerstner's turnaround of IBM (1993-2002) is a textbook Lewin three-stage execution. Unfreeze: Gerstner inherited an IBM in existential crisis — losing $8B/year, executives still in denial about how broken the mainframe-only strategy was. He used the financial reality as the Unfreeze lever, famously telling the executive team 'the last thing IBM needs right now is a vision' — meaning, first acknowledge the brutal facts. Change: He restructured the company around services and consulting, killed the planned breakup, and shifted the operating model. Refreeze: He rebuilt the leadership team, redesigned compensation around customer outcomes, and created the now-famous 'IBMer values' that institutionalized the new culture. By 2002, IBM was profitable, services-led, and culturally transformed. Gerstner's book 'Who Says Elephants Can't Dance?' is essentially a case study in Lewin's three stages applied at $80B scale.

Pro Tips

01
Unfreeze is harder than it looks. Sustained, comfortable organizations have powerful immune systems — managers actively rationalize 'we're fine, we tried that before, we're different here.' Effective Unfreeze almost always requires bringing in external data the org can't dismiss: customer interviews on video, lost-deal analyses, side-by-side competitor benchmarks. Internal data gets re-narrated; external data is harder to wave away.
02
Refreeze is where most KnowMBA-tier organizations fail. The new behavior is happening, the leader celebrates the milestone, attention shifts to the next priority — and within 12-18 months the behavior has reverted. Refreeze requires that the new behavior become the only path to promotion, the only way bonuses get paid, and the default in onboarding for new hires. If a new joiner can succeed without doing the new behavior, you didn't Refreeze.
03
The middle phase (Change) is the most expensive but the least leveraged of the three. Most change budgets pour into Change-phase activities (training, tools, communications) while skimping on Unfreeze (insufficient urgency-building) and Refreeze (no system updates). Rebalance: spend roughly 25% on Unfreeze, 50% on Change, 25% on Refreeze. Most orgs spend 5/90/5.

Myth vs Reality

Myth

“Lewin's model is outdated for fast-moving modern organizations”

Reality

The model describes human psychology, not organizational structure. Humans still require dissatisfaction with the status quo before they will adopt new behavior, and they still revert without reinforcement. Modern frameworks like ADKAR and Kotter's 8 steps are essentially elaborations of Lewin, not replacements.

Myth

“Refreezing makes the organization rigid and unable to change again”

Reality

Refreeze means stabilizing a new equilibrium so it becomes the normal operating state — it does not mean making the organization unable to ever change. A company can be in active Refreeze for one initiative while in Unfreeze for the next. The alternative — never refreezing anything — is what produces change-fatigued organizations where nothing ever sticks.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Knowledge Check

An IT director rolls out a new ticketing system. Training is complete, the new system is live, and adoption hits 85% in week one due to leadership pressure. Six months later, adoption has dropped to 47% and shadow workarounds are common. According to Lewin, which stage was likely skipped or under-invested?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Change Initiative Investment Mix (% of Total Budget)

Enterprise transformation programs

Recommended Balanced

Unfreeze 25 / Change 50 / Refreeze 25

Common Real-World Pattern

Unfreeze 5 / Change 90 / Refreeze 5

Refreeze-Skipped (typical reversion)

Refreeze < 10%

Source: KnowMBA analysis of Prosci & McKinsey change-management studies

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

💼

IBM (Lou Gerstner Era)

1993-2002

success

When Lou Gerstner became CEO, IBM was losing $8B/year and the consensus was to break the company up. Gerstner executed a textbook three-stage transformation. Unfreeze (year 1): he forced the executive team to confront brutal financial and customer data, famously saying IBM didn't need a vision yet — it needed honesty. Change (years 2-4): he killed the breakup plan, repositioned IBM around services and integrated solutions, restructured the operating model, and brought in new senior leaders. Refreeze (years 5-9): he rebuilt the comp system around customer outcomes, redesigned promotion criteria, and articulated the IBM Values that became the cultural anchor for the next decade. By the time he retired in 2002, IBM was profitable, services-led, and culturally transformed.

Annual loss at start

$8B (1993)

Market cap (1993 → 2002)

~$30B → $170B

Services revenue share

27% → 45%

Cultural transformation duration

~9 years end-to-end

Multi-year transformations live or die by the Refreeze phase. Gerstner could have declared victory in 1996 when the financials stabilized — instead, he spent the next 6 years institutionalizing the change. That patience is what separated IBM's turnaround from temporary turnarounds at peer companies that reverted within 5 years.

Source ↗

🚆

Hypothetical: NorthernRail Logistics

2021 Safety Culture Initiative

mixed

After two serious incidents, the CEO of a 6,000-person rail logistics company launched a 'Safety First' transformation. The launch was excellent: company-wide stand-downs, new safety training, visible signage, and personal stories from executives. Incident rates dropped 40% in the first 9 months. The CEO declared victory at the year-one all-hands. Twelve months later, incident rates had crept back to 80% of the original baseline. A consultant diagnosed: Unfreeze had been strong (the incidents themselves), Change had been well-executed, but Refreeze had been almost entirely absent — promotion criteria didn't reference safety behaviors, hiring screens didn't include safety judgment, and no manager had ever been denied a bonus for a safety lapse. The fix involved restructuring 1,200 manager comp plans and rebuilding the supervisor onboarding curriculum. Incident rates dropped sustainably below baseline within the next 18 months.

Initial incident reduction

40% (year 1)

Incident rebound (year 2)

Back to 80% of baseline

After Refreeze rebuild

Sustainable < baseline

Comp plans restructured

1,200

A loud launch and a quiet death is the most common pattern in change. The CEO mistook initial improvement for permanent improvement and shifted attention. Without Refreeze, every change reverts — fast.

Related concepts