Renata Castellanos leads a 40-person operations department at a mid-size insurance firm. When the company decided to bring AI into claims processing, leadership handed her the tools, a license budget, and a deadline, then assumed her team would simply start using them. Four months in, adoption sat at 22%. The tools were deployed; almost nobody used them. Worse, a quiet cynicism had set in: "We do these big initiatives and nothing ever actually changes." Renata realized her mistake was not the technology choice. It was that she had managed a software rollout when she needed to lead a human change. She restarted the effort using real change-management frameworks, and over the next year adoption climbed past 80%. This lesson is the method she used, scaled to the level of a manager leading her own team or department, not the whole company.

What This Lesson Covers

Technology does not drive transformation. People do. You can deploy the best AI tools available and still see nothing change, because the systems get installed but used poorly or abandoned, teams resist even good initiatives, cynicism builds, and the return on investment never materializes. The opposite is equally true: with skilled change leadership, teams embrace new capabilities, momentum builds, the change sticks, and the intended outcomes actually arrive. As a manager, you are the change leader for your team. Your ability to manage the human side determines whether your AI initiatives succeed.

This lesson covers the human side specifically. You will learn to apply Kotter's eight-step model to an AI rollout in your department, to use the ADKAR model for individual adoption, to understand the change curve people move through emotionally, to engage resistance instead of fighting it, and to segment your team by readiness so you lead each group differently. Throughout, we follow Renata as she turns a stalled 22% rollout into a thriving one.

Why the Human Side Is Most of the Work

A useful rule of thumb: the technology is maybe 30% of a transformation, and the change management is the other 70%. Renata learned this the hard way. The tools were fine. What was missing was urgency (her team did not feel why this mattered now), a coalition (she was pushing alone), a vision (no one could picture what good looked like), and any visible early win to prove the effort was real. Fix the human side and the same tools that sat idle suddenly get used.

Resistance is not a sign the change is wrong. It is a sign that people are processing what it means for them. Your job is not to overpower it but to understand it.

Kotter's Eight Steps, Applied to Your Team

John Kotter's eight-step model is a widely used framework for leading organizational change. Here is how it maps onto an AI rollout inside a single department.

1. Create urgency. Help people feel why this matters now, not someday. "Our claims-processing time has to come down to stay competitive, and AI is how we get there." Without urgency, change is optional and gets deprioritized.
2. Build a coalition. Recruit a small group of respected people who will actively champion the change: an influential senior peer, a couple of well-liked team members (ideally former skeptics who got convinced), and a voice from each affected function. They do not just nod in public; they advocate and help solve problems.
3. Form a clear vision. Describe what success looks and feels like. "In 18 months, AI handles routine claims triage and our team focuses on the complex, judgment-heavy cases. We are faster, customers wait less, and the work is more interesting." Make it inspirational but believable.
4. Communicate the vision, relentlessly. Through many channels, with consistency. People need to hear a message many times, often 10 or more, before it sinks in. Weekly is the floor, not the ceiling.
5. Empower broad-based action. Remove the barriers: a fast approval path to try a tool, training and support, protected time to experiment, and explicit permission to fail safely while learning.
6. Generate quick wins. Engineer visible successes in the first three to six months and celebrate them publicly and by name. "Mariam's pod cut handling time 30% using the assistant." Wins convert skeptics faster than any argument.
7. Consolidate gains. Do not stop at the first win. Use the momentum to expand to more of the team and into more complex use cases, deepening capability rather than coasting.
8. Anchor it in the culture. Make it "how we work here": include AI fluency in how you hire and evaluate, and tell the transformation story to new joiners so it becomes the norm, not a project.

ADKAR: Change Happens One Person at a Time

Kotter operates at the level of the team. ADKAR zooms in on the individual, and it explains why a person stalls. Each person needs to move through five stages in order: Awareness (they understand why the change is needed), Desire (they actually want to take part, the "what's in it for me" is answered and fear is reduced), Knowledge (they know how to do the new thing), Ability (they have practiced and can actually do it), and Reinforcement (the change is reinforced so it sticks).

The practical power of ADKAR is diagnostic. When someone is not adopting, find the stage they are stuck at. If they resist, it is often a Desire gap (unaddressed fear), not stubbornness, and more training will not help. If they try and fumble, it is an Ability gap and they need practice, not another lecture. Renata found most of her stalled team members were stuck at Desire: they understood the business case fine but quietly feared the AI was there to replace them. No amount of Knowledge-stage training was ever going to move them until that fear was addressed directly.

The Change Curve: What Resistance Actually Looks Like

People do not move from old to new in a straight line. They travel an emotional change curve, and recognizing where someone sits on it tells you how to help. The typical path runs: shock or denial ("this won't really affect us"), into resistance and frustration ("this is making my job harder, and I'm not even good at it"), down through a low point, then up into exploration ("maybe there's something to this") and finally commitment ("this is just how I work now").

The mistake managers make is treating the frustration dip as failure and either forcing harder or giving up. The dip is normal and expected; it is the bottom of the curve, not the end of the road. Common sources of resistance you will meet there: loss aversion (fear of losing familiar ways, status, or security), uncertainty ("I don't know what this means for me"), competence anxiety ("I won't be good at this"), workload ("I'm already overloaded"), and value mismatch ("this isn't how I think the work should be done"). Engage each one rather than overpower it. The five-move sequence that works: listen for the real concern beneath the objection, acknowledge it as legitimate, involve the person in shaping the solution, provide concrete support, and recognize them when they engage. Forcing change against resistance produces cynicism and shallow compliance; engaging it converts resistance into commitment.

Lead Three Groups, Not One

A single change approach fails because your team is not one audience. Roughly, expect three groups, and lead each differently. The early adopters (around 20%) should be let loose to lead: give them advanced tools to pilot, ask them to help figure out what works, and make their wins visible. The skeptics (10 to 20%) should never be forced: "I'm not asking you to love this yet, just try it and tell me what you actually think." Listen seriously and adjust where they have a point. The middle majority (60 to 70%) needs clear direction, easy on-ramps (training, support, time), and social proof that their peers are succeeding, with a gradual ramp: routine tasks in week one, half their work in week two, full use in week three. One-size-fits-all bores the early adopters, overwhelms the skeptics, and loses the middle.

Worked Example: Renata's 12-Month Phased Rollout

Here is how Renata restarted her stalled rollout, mapping the frameworks onto a real timeline with real adoption numbers. Her department of 40 splits into roughly 8 early adopters, 6 skeptics, and 26 in the middle.

• Month 1, urgency and coalition (Kotter 1 to 2). She opened with the business case tied to the team's own pain ("manual triage is why we work late every month-end"), and recruited a six-person coalition: two early adopters, one respected skeptic she won over in a one-on-one, and three pod leads. Starting adoption: 22%.
• Month 2, vision (Kotter 3). She wrote a one-page "day in the life in 18 months" and walked every pod through it, naming explicitly that the goal was to remove drudgery, not headcount, addressing the Desire-stage fear head-on.
• Months 3 to 6, communicate, empower, quick wins (Kotter 4 to 6). Weekly updates, a fast tool-approval path, a monthly skills workshop (Knowledge and Ability), and 10% protected experimentation time. She turned her early adopters loose first. By month 6, one pod had cut average triage time 30%, which she celebrated by name in an all-hands. Adoption reached 45%, driven mostly by the early adopters and the keenest of the middle.
• Months 7 to 12, consolidate (Kotter 7). She expanded from simple triage to more complex case-summarization use cases, paired each remaining skeptic with an early-adopter mentor, and ran the middle group through the gradual three-week ramp. Social proof did the heavy lifting: "30 of 40 are using it daily now" pulled the laggards along. Adoption crossed 80%.
• Month 12 onward, anchor (Kotter 8). AI fluency entered her hiring conversations and her development plans, and she started telling the transformation story to new hires so it read as normal, not novel.

The numbers tell the story: 22% to 45% to 80% over a year, not because the tools changed (they did not) but because she led the people through the curve. Note what the phasing bought her: the month-6 quick win at 45% is what convinced the skeptical middle that the effort was real, which is exactly what was missing in her failed first attempt. Quick wins are not a nice-to-have; they are the mechanism that moves the majority.

Anti-Patterns to Avoid

Technology focus without change management. "We deployed the tool; people will figure it out." They will not, and benefits never materialize. Lead the change intentionally with a framework.

Treating resistance as an obstacle to crush. Forced change produces cynicism and surface-level compliance. Resistance is data; listen and engage it.

One-size-fits-all change. The same approach for early adopters, middle, and skeptics serves none of them. Segment by readiness.

No quick wins. Eighteen months with nothing visible kills momentum and feeds cynicism. Engineer wins in the first three to six months.

Change without sustaining. A big push that fades the moment your attention moves elsewhere, and people revert. Anchor it in culture, reinforcement, and ongoing attention so it sticks.

Human Judgment Checkpoints

• Stakeholder clarity: Can you name your sponsor, coalition members, key influencers, and skeptics? If not, you have skipped the political work.
• Vision reality: Can you describe success in 18 months in a way that is both inspirational and believable? Vague or over-the-top visions get tuned out.
• Communication frequency: Are you communicating about the change at least weekly? Less than that is almost certainly not enough.
• Resistance reality: Have you actually talked with your skeptics about what they fear? If not, you do not yet know what you are dealing with.
• Quick-win reality: Have you produced and celebrated a visible success in the first six months? If not, your momentum is leaking.

Leading Change With Dignity

Transformation changes people's work and sometimes touches their roles, so lead it with respect. Be transparent about how the work will change, who it affects, and what support you are putting in place; vague reassurance breeds the exact fear that stalls adoption. Be honest, not evasive, about impact. And include the affected people in designing how the change happens rather than simply announcing it to them. People support what they help build, and inclusion is also where you discover the practical problems you would otherwise hit at scale.

Practice and Reflection

Renata's second attempt worked because she did the planning her first attempt skipped. These five exercises are that planning. Do them on paper for a change you are leading now, and keep the answers where you can revise them as the rollout moves.

1. Assess your change readiness. Before you plan anything, take stock. Who on your team are the early adopters, who are the genuine skeptics, and who sits in the middle? What is actually driving urgency for this change, in terms your team would recognise as their own problem rather than a corporate slogan? Who is your executive sponsor, the senior person whose visible support you can point to? And what are the biggest barriers standing between here and adoption? If you cannot answer these, you are not ready to announce anything.

2. Build your coalition on paper first. Name five or six people who could be change champions. For each, write why you chose them, how you will engage them, and what concerns they are likely to raise in a one-on-one. Renata's most valuable coalition member was a respected skeptic she won over privately before asking him to advocate publicly, so do not restrict the list to people who already agree with you.

3. Craft the vision. Write one page describing your function eighteen months to two years from now, with the transformation complete. What does the work look like day to day? What is different from today? What outcomes have you achieved? And, the question people care about most, how do people feel about their work? Read it back and check it against the vision reality test: is it both inspirational and believable? A vision that fails either half gets tuned out.

4. Write a six-month communication plan. Decide the consistent weekly message you will repeat, because people need to hear it many times before it lands. Plan the monthly celebrations and name which early wins you expect to be able to highlight. Choose whose story you will tell to illustrate the change, the way Renata named a specific pod in an all-hands. And decide which data you will share to show progress, so the claim that this is working has evidence behind it.

5. Map your resistance. For each significant skeptic, write who they are, what they are actually concerned about rather than what they objected to in the meeting, how you could address that concern, and what would genuinely help them adopt. Renata found most of her stalled people were stuck at the Desire stage, quietly fearing replacement, and no amount of training would have moved them. Use ADKAR to name the stage each person is stuck at before you decide what to offer them.

Related Lessons

Leading transformation touches three neighbouring lessons directly.

• Building Organizational AI Culture is what makes a transformation permanent. Kotter's eighth step, anchoring the change in culture, is the whole subject of that lesson, and without it a successful rollout quietly reverts once your attention moves elsewhere.
• Workforce Development and Reskilling is the people half of transformation. Asking a team to work differently without building the skills and career pathways to match is what turns a Desire-stage fear into an exit interview.
• Ethical Leadership in AI Adoption underpins the credibility you are spending here. Your team's willingness to follow you through the dip depends on whether they believe you are being straight with them about impact, which is exactly what that lesson develops.

Key Takeaways

• Change management is most of the work. The technology is roughly 30% of a transformation; leading people through the change is the other 70% and the part that determines success.
• Urgency is the prerequisite. Without a felt reason to move now, the change stays optional and never happens. Tie it to your team's real pain.
• Build a coalition; you cannot do it alone. A few respected champions, including a converted skeptic, carry the change further than any solo push.
• Diagnose stalls with ADKAR. Find the stage a person is stuck at. Resistance is usually a Desire gap (fear), not stubbornness, and training will not fix it.
• Expect the change curve and the dip. Frustration is the normal low point, not failure. Engage resistance with listening and involvement instead of force.
• Lead three groups differently. Unleash early adopters, never force skeptics, and pull the middle along with direction, support, and social proof.
• Quick wins move the majority. A visible, celebrated success in months three to six is what convinces the skeptical middle the effort is real, then anchor the change in culture so it sticks.