How do you manage resistance to change in the workplace — and how do you deal with employees who resist it?

Managing resistance to change in the workplace means reading the resistance before trying to remove it, choosing a method that fits what the reading shows, and giving the people who know the work a role in the change rather than a briefing about it. Resistance is not one thing. Some of it is missing information, some is a real loss that nobody has priced, some is a disagreement about whether the change will work, and some is a defense of standing or control. Each responds to a different move, and the common failure is to apply the move for the first kind, more communication, to all four. The research, from Paul Lawrence in 1954 to John Kotter and Leonard Schlesinger in 1979 and the operating record since, points one way: the manager who treats resistance as a signal about the change manages it; the manager who treats it as a fault in the people does not.

What are the reasons for resistance to change in the workplace?

The reasons are covered in an earlier post on why people resist change even when it is positive, and the short form is that a change takes specific things from specific people now and offers general things to the organization later. For the manager who has to act, the more useful distinction comes from Paul Lawrence's 1954 Harvard Business Review article, "How to Deal with Resistance to Change," which the magazine reprinted in 1969 with a retrospective by the author. Lawrence's finding, from watching industrial engineers introduce new methods on factory floors, was that people rarely resist the technical change. They resist the social change that rides along with it: the alteration in who gives instructions to whom, whose judgment counts, who is consulted and who is told. An engineer who could not understand why operators fought a better jig had usually not noticed that the jig arrived with a message about their competence, delivered by a stranger who had asked them nothing.

That distinction does the first piece of managerial work. If the resistance is to the social change, redesigning the technical change will not touch it, and more explanation of the technical merits makes it worse, because every explanation confirms that the explainer thinks the problem is the listener's understanding. Kotter and Schlesinger, in "Choosing Strategies for Change" in the same journal twenty-five years later, sorted the sources into four: parochial self-interest, misunderstanding and lack of trust, different assessments of the situation, and a low tolerance for change itself. The sorting matters because the four do not share a cure. Misunderstanding yields to information. Self-interest does not; it yields to a trade. A different assessment, where the people doing the work believe the change will fail, is sometimes right, and the manager who overrides it has thrown away the one free piece of due diligence on offer.

What are the strategies for managing resistance to change?

Kotter and Schlesinger listed six, and the list has not been improved on in the nearly fifty years since. Education and communication, where the resistance comes from inaccurate information or none. Participation and involvement, where the initiators lack information they need and the resisters have the power to block. Facilitation and support, training and time and a hearing, where the resistance is adjustment anxiety. Negotiation and agreement, where a group stands to lose something real and has the power to make the loss expensive. Manipulation and co-optation, giving a resister a desirable role in designing the change in exchange for endorsement. And explicit or implicit coercion, where speed is essential and the initiators hold the power to impose.

Two things about the list are usually left out when it is quoted. The first is that the authors did not rank the six from good to bad. They gave each a situation in which it fits and a cost it carries: participation and education are slow; negotiation can become expensive once others learn to negotiate; co-optation backfires if people feel tricked; coercion is fast and leaves people angry at the initiators. The second thing left out is the authors' strategic continuum. At one end sits a change planned clearly in advance, with little involvement from others, moved fast, running over whatever resistance appears. At the other sits a change not fully planned at the start, involving many people, moved slowly, with resistance minimized along the way. Where to stand on the continuum depends on four things: how much resistance is anticipated and of what kind, how much power the initiator holds relative to the resisters, where the information needed to design the change actually sits, and what is at stake if the change is slow, or if it is wrong.

That last pair is the operator's real decision, and it is why the strategies cannot be chosen from a list of best practices. A payroll system due before a regulatory deadline, imposed on a staff who cannot block it, sits at the fast end; participation there costs the deadline. A new way of handling complaints, where the people on the phones know things the designers do not and can quietly refuse what they are given, sits at the slow end; coercion there produces a system technically live and operationally dead. Most changes sit between, and the error the authors described was choosing the point by temperament rather than situation: the fast mover who imposed where the knowledge was on the floor, or the consensus builder who consulted while the deadline passed.

How do you deal with employees who resist change?

Deal with the resistance as information first and as a behavior second. Jeffrey Ford, Laurie Ford and Angelo D'Amelio made the strongest version of this argument in a 2008 paper in the Academy of Management Review titled "Resistance to Change: The Rest of the Story." Their claim was that the usual account, rational change agents meeting irrational resisters, leaves out what the agents did: broke earlier promises, communicated badly, misread the situation, then labeled the response as resistance. They also argued that resistance is a resource: it keeps the conversation alive when the alternative is silent non-compliance, and it carries information about the change's weak points. An employee who says the new process will break on month-end volume is doing unpaid testing.

The operating move is to separate the three things an objection can be. Information about the change: a flaw, a missed case, a dependency nobody mapped. Information about the loss: what this person gives up, in competence or standing or control, that the business case did not count. Or a position: a decision to oppose, held regardless of what is said. The first two are answered by listening and then changing the design or the terms. The third is rarer than managers believe and is answered by the later methods, negotiation or, where the stakes warrant it, a plain statement that the change is happening and what blocking it will cost.

Albert Hirschman's 1970 book Exit, Voice, and Loyalty explains why the order matters. People unhappy with an organization can leave or speak. Voice is the response that carries information, and it is the one suppression kills first. A manager who treats early objections as disloyalty teaches the people with the most knowledge to stop objecting; what follows is exit by those who can and quiet non-use by those who cannot, the condition in which a system is reported as adopted and is not.

David Luria's Flatten the AI J-Curve gives the most concrete version of the repositioning move for the hardest case, the experienced employee whose expertise the change appears to devalue. Its example is a geologist of twenty-two years who resisted an AI system that threatened to make his judgment look dispensable. Rather than train him to use the tool, the organization made him its evaluator, grading the system's output against what he knew. His expertise became the reason the system was reliable, and the threat became a role. The book's five-step protocol, identify the quiet adopters, reposition the resisters as reviewers, demonstrate peer wins, formalize the standing through certification, then assign leadership of critical work to managers with results, is a sequence, not a menu.

Resistance to change in the workplace examples

Three examples carry most of what the research knows.

The oldest is the Harwood pajama factory, where Lester Coch and John French in 1948 found that operators told about a method change in the usual way lost output, while operators who helped design it recovered quickly and exceeded their old rate. It is cited here for what is usually left out: the groups did the same work, under the same economics, for the same company. The resistance was in the method of introduction, and it moved when the method did.

The second is the pattern Michael Beer, Russell Eisenstat and Bert Spector reported in 1990, in "Why Change Programs Don't Produce Change," also in the Harvard Business Review. Studying change efforts at six large companies, they found that programs launched from the center, training courses, mission statements, culture initiatives, produced little, while change took hold where a unit manager realigned the actual tasks of the unit around a concrete business problem. They named the error the fallacy of programmatic change: the belief that attitudes change first and behavior follows, when the evidence ran the other way. People given a new job, with new relationships and responsibilities, change their attitudes to fit it; people given a course on the new attitudes and sent back to the old job do not. Most of what a program meets as resistance is the gap between what the course said and what the job still requires.

The third is current. Flatten the AI J-Curve names the layer where enterprise AI adoption stalls the Frozen Middle: directors and senior managers whose standing rests on knowing how things work, and for whom a system that surfaces those answers in seconds is an identity threat before it is a productivity tool. The book cites a 2025 survey by the software firm Writer in which 42 percent of executives said the process of adopting AI was tearing their company apart, and an InvestOps survey of financial firms in which 26 percent named employee resistance as a top AI challenge. Its sharper number is behavioral: when the move to next-day settlement arrived, 65 percent of the firms surveyed expanded their operations teams instead of automating the work, a choice made not at the top but in the middle, by the people closest to the operation choosing the familiar cost over the uncertain one. That is what badly managed resistance looks like from outside. Nothing is refused. Headcount is added, the system is installed, and the old way of working continues inside it.

How do you overcome resistance to change in the workplace — or should you?

The word overcome assumes the resistance is an obstacle between the manager and a change that is right, and the assumption is the problem. Eric Dent and Susan Goldberg, in a 1999 paper in the Journal of Applied Behavioral Science, traced how the phrase resistance to change entered management vocabulary and argued that it had come to do harm: it put the fault in the people and licensed the manager to stop listening. Sandy Piderit, in the Academy of Management Review in 2000, added that most responses to change are not opposition but ambivalence, a person who believes the change is needed and also expects to lose by it, and that treating ambivalence as resistance converts a potential ally into an opponent.

So some resistance should be overcome and most should be used. The kind to overcome is the position held regardless of evidence, and it is overcome by the fast-end methods, applied openly: here is what is changing, why, and what will happen to those who block it. Everything else is used, by the slow-end methods. A manager who cannot tell which is in front of them should assume the second: mistaking information for opposition costs a worse change; mistaking opposition for information costs a few weeks.

One more caution on the word. An earlier post on leading people through a change they did not choose covered the transition that follows loss. Resistance in that period is grief in work clothes, and it is not overcome; it is supported and waited out.

What are some effective techniques for managing resistance to organizational change?

The techniques that hold up act on the social change Lawrence identified, not the technical one. Six follow; none is new.

Diagnose before choosing a method. Write down, before the announcement, who loses what, who can block, where the knowledge sits, and how much time the business case can survive. Those are Kotter and Schlesinger's four factors, and a change announced before they are known has chosen its method by accident.

Match the method to the resistance, and say which you are using. Information for misunderstanding. A trade for a real loss. A role for the people who know the work. A hearing and time for those who need to adjust. Coercion, named as such, for the rare case where speed is everything and the authority exists. People can tell, and a negotiation dressed as consultation is the manipulation the authors warned would poison the next change.

Put the resisters on the quality side of the change. This is the repositioning move. The person who knows the twelve ways the old process breaks is the person who should be grading whether the new one handles them, and the title should say so. It converts the threat to expertise into a demand for it, and it gets the designers the information they lacked.

Change the task before the training. Beer, Eisenstat and Spector's finding reversed the usual order. Give the unit a concrete problem and a changed set of responsibilities, and let the training follow what the new work demands. A course delivered to people whose jobs have not changed is the fallacy of programmatic change in a conference room.

Price the dip and publish the date. Most workplace change includes a period in which people run the old way and the new way at once, checking one against the other, while their normal output is expected. Flatten the AI J-Curve puts the low point of enterprise AI rollouts around the fourth month, with workload rising by about half before it falls. A manager who announces the dip, suspends the normal targets for its stated duration and names the date on which the change will be assessed has turned the worst period into a scheduled phase. A manager who does not has turned it into evidence the resisters were right.

Make progress visible where people feel it. The book's operating cadence is a weekly squad-level scoreboard: who has been certified, what the cycle time has done, one peer win. The mechanism is social: people change what they do when the people beside them are doing it and the fact is visible. A company-wide adoption figure is abstract; a list of fourteen names with three still at the lowest tier is not. Publish the scoreboard, name and promote the converted skeptics, and after enough months to be fair, give the leadership of critical work to those with results. The book is explicit that this is organizational design, not punishment, and that most holdouts read the signal and enroll within a quarter.

What does it cost when resistance to change is managed badly?

The cost arrives in three forms, and the smallest is the one managers count. The first is delay: a change imposed where participation was needed is installed on schedule and adopted late, after the sponsor has stopped watching. The second is the degraded change, the system that is live but routed around, the double record that an earlier post on AI projects found behind most stalled rollouts. The third, and largest, is the one Hirschman named: every objection punished teaches the organization to stop objecting, and the next change arrives in a building where nobody with knowledge will say what is wrong with it. Flatten the AI J-Curve puts it in its own terms: a dip of several months is survivable, and six months of middle-management resistance on top of it turns a trough into an extinction event. Badly managed resistance does not cancel the dip. It extends it.

Managing resistance to change in the workplace: what to do Monday

Read before announcing: list who loses what, who can block, where the knowledge sits and how much time there is, and choose the point on the continuum from that list rather than from habit. When the first objection comes, treat it as information about the change, information about a loss, or a position, and answer it as that thing; most will be the first two. Give the people who know the work the quality role, change the tasks before the course, publish the dip and the date, and make progress visible to the people who sit near each other. Where a position remains after all of that, say plainly that the change is going ahead and what follows from blocking it. That is the method's last step, used last.

Lawrence's 1954 article made the point managers have been rediscovering since: resistance to change is a signal that something is going wrong, a flag worth reading rather than a fault worth correcting, and the manager's job is to find out what it is pointing at. The signal is not in the people. It is in the gap between the change as it was designed and the change as it lands. The gap is usually closable. It is only ever closed by someone who went and looked.

One email when a book ships. Nothing between.

← All posts