Lewin's Change Model — Unfreeze, Change, Refreeze

A team I worked with tried to adopt async-first comms. They announced the change Monday, mandated it Tuesday, regressed by Friday. They forgot to unfreeze — to make staying the same feel harder than changing. The new behavior was the only thing they tried to teach; the old behavior was still the cheapest option in the room. Cheap behaviors win every time.

Why it matters

Kurt Lewin's model is 75 years old and still the cleanest mental model for behavior change. Everything else (Kotter, ADKAR, Prosci) is downstream of it. It works because it describes the underlying physics: behavior sits in an equilibrium of forces, and you can either weaken the forces holding the old behavior in place or strengthen the forces pulling toward the new one. Most change programs do the second and ignore the first.

The other reason it matters: 'refreeze' is the step everyone forgets in modern, fast-moving orgs. The fashion is to celebrate change as continuous; the reality is that unstable, half-frozen new behaviors regress to the old equilibrium within months because nobody made the new state cheaper than the old one to stay in.

The 3 phases

Force field analysis

Lewin's key insight: weakening restrainers works better than amplifying drivers. Pushing harder against resistance creates more tension; removing the resistance lets the system shift. Most change programs spend 80% of effort on drivers and 20% on restrainers. The high-leverage move is the opposite ratio.

Example

Buffer's 2013 switch to fully transparent salaries unfroze the old norm by publishing the formula and reasoning publicly — making the old (secret) state untenable in their stated culture. They changed: everyone's salary visible, formula on the website. They refroze by writing it into onboarding, hiring offers, and review processes. Ten years later, still the default. The refreeze is what made it permanent; many companies have tried transparency and regressed within 18 months because they never embedded it.

Apply on Monday

• For your current change, list driving and restraining forces.
• Pick the top 3 restrainers — design a release plan for each.
• Identify the 'refreeze' systems (hiring, reviews, tools) — schedule the changes.
• Don't announce the change until you've unfrozen.
• Block calendar time for refreezing — most leaders skip it because the change feels 'done'.

Common mistakes

• Skipping unfreeze — announcing change without disturbing equilibrium.
• Pushing harder when you should release a restrainer.
• Forgetting to refreeze — and regressing in 3-6 months.
• Treating change as a project instead of a state transition.
• Releasing only the visible restrainers and leaving the cultural ones intact.
• Refreezing too early — before the new behavior is actually working.

Reflection prompts

1. Which current change have I not actually unfrozen?
2. What's the strongest restrainer I'm pushing against instead of releasing?
3. Which system would I need to update to refreeze this for real?
4. Where am I confusing 'announced' with 'changed'?

Takeaways

• Unfreeze. Change. Refreeze. Skip any and you regress.
• Release restrainers; don't just add drivers.
• Refreeze in systems (hiring, review, tools), not slogans.
• Behaviors return to the cheapest equilibrium. Make the new one cheaper.