Most change programs do not fail because the plan was wrong. They fail because the people who had to work differently never actually did, and everyone quietly drifted back to the old way within a quarter. As COO, your job is not to announce the change but to make the new way easier to follow than the old habit, and to keep it that way long after the launch email.

The practical test is simple. Six months after a change, walk the floor and watch what people actually do. If they use the new process without being reminded, you led the change. If they use the old one and hide it, you ran a communications campaign.

This guide covers the sequence that gets you from the first to the second: real urgency, a coalition with weight, changing behaviour one person at a time, handling resistance without flattening it, and measuring adoption instead of activity. It leans on two well-established models — John Kotter's eight-step change model and the ADKAR model from Prosci — because they cover the two halves of the problem: the organizational push and the individual transition.

Start with a reason people believe, not a slogan

Kotter's first step is creating a sense of urgency, and it is the one leaders most often fake. A slide that says "we must transform to stay competitive" creates no urgency, because it is abstract and could have been written any year. Real urgency is concrete: a customer you lost last month, a margin that slipped two quarters running, a competitor who now ships in half your lead time.

Strong looks like a COO who can state the reason for change in one sentence a warehouse supervisor repeats correctly to their own team. Weak looks like a townhall where the "why" is a mission statement, and the room nods politely and forgets it by lunch.

Tie the change to something the affected people already care about. Consolidating two order systems? Do not lead with "operational efficiency" — lead with "you are keying every order twice, and it is why you stay late on Fridays." Urgency built on a real, felt problem pulls people toward the change; urgency built on fear of the future just makes them defensive. A broader business transformation program should name the concrete pain it removes before it names the target state.

Build a coalition with weight, not a project committee

Kotter's second step, the guiding coalition, is where most COOs under-invest. A coalition is not the project team and it is not a list of names on a RACI chart. It is a small group with enough combined authority, credibility, and floor-level respect that when they back the change, resistance elsewhere runs out of oxygen.

The mistake is stacking it with titles. A vice president who is disliked adds a logo to your deck and nothing to your momentum. The supervisor who has been there fourteen years and whom everyone trusts adds real weight without one. You want both formal authority — to remove barriers and fund the work — and informal authority, so the change is credible to the people doing it.

Test the group against three questions. Can it actually clear obstacles, or only escalate them? Does it include one person the sceptics respect? Would the change survive if you went on leave for a month? If the honest answer to the last one is no, you have a sponsor, not a coalition. Making that group work is a cross-functional leadership problem, because the coalition usually spans functions that do not normally share goals.

Change happens one person at a time: the ADKAR model

Kotter describes what the organization does. ADKAR describes what each individual goes through, and it is the sharper lens when a change is technically small but behaviourally large. ADKAR stands for Awareness, Desire, Knowledge, Ability, and Reinforcement, and the key insight is that they are sequential. You cannot train someone (Knowledge) into a change they do not want (Desire), and you cannot reinforce a behaviour they were never made able to perform (Ability).

Diagnose a stalled change by finding the earliest letter that is missing. If you assume it is a training gap but the real gap is Desire, more training makes it worse. Here is how a COO acts on each stage.

The most common failure is treating Reinforcement as optional. A change that is not reinforced decays, because the old habit is still available and still easier. The strongest reinforcement is structural — turn off the old system, remove the old form — so there is no easier option to fall back on.

Sequence the rollout so early wins are real

Kotter's step on short-term wins is widely quoted and widely faked. A short-term win is not a status update saying the project is "on track." It is a visible, unambiguous result a sceptic would accept as proof the change works — a queue that got shorter, an error rate that dropped, a Friday people got to leave on time.

Sequence the rollout to generate one of these early and on purpose. Pick a first slice small enough to succeed and visible enough to matter. A mid-sized firm rolling out a new fulfilment process might start with one product line in one location, get the error rate down, and let that win travel by word of mouth. A simultaneous "big bang" across every team maximizes the chance that one visible failure poisons the effort before any win lands.

Be honest about which wins count. A win only the project team can see, or that needs a spreadsheet to explain, does not build momentum. It has to be legible to the people you still need to convince.

Handle resistance as information, not an obstacle

Resistance is usually treated as something to overcome. Most of the time it is something to listen to. The frontline person resisting a new process often knows an edge case your plan missed, and if you flatten their objection with a slide, you lose both the fix and their trust.

Separate the two kinds. The first is substantive: the change genuinely breaks something and the resister is right. Fix the plan and credit them publicly — that single act converts your loudest critic into evidence that objections get heard. The second is transitional: the person is losing status, comfort, or a skill they were proud of. That deserves a direct, honest conversation, not a pretence that no one loses anything.

Strong resistance-handling looks like a COO who can name, out loud, who loses what in a change, with a specific answer for each. Weak looks like relentless positivity insisting the change is good for everyone — which the people losing something read as naive or dishonest. Sustained resistance also wears on the people leading the change, so protecting the engagement of the teams carrying it is part of the job.

Measure adoption, not activity

The metric that kills change programs is activity: emails sent, sessions delivered, people trained. All of these can read 100% while adoption is near zero. Measure the behaviour, not the effort to produce it.

The cleanest adoption metrics are behavioural and hard to fake. What percentage of orders now go through the new system without a workaround? How many days since the last use of the old process? What is the gap between the best-adopting team and the worst, and is it closing? These tell you whether the change is real; a training-completion number does not.

Set a handful before launch and watch them longer than feels necessary. The dangerous window is the two to three months after go-live, when launch energy fades and habits reassert themselves. A change at 80% adoption in week two and 40% in week ten did not succeed; it decayed, and catching that decay is what your metrics are for. For a cultural shift rather than a process change, pair these with the slower signals of a cultural transformation program.

Using Kotter and ADKAR together

You do not choose between the models; you use them for different jobs. Kotter is your plan for the organization; ADKAR is your diagnostic for a specific group that is stuck, telling you which stage they are failing at so your next move is targeted instead of generic "more communication." Keeping every affected group informed is a stakeholder engagement discipline, and the software and frameworks that support the work are in the companion guide to change management tools.

Key takeaways

• Change fails on adoption, not design. Judge it by what people do six months later unprompted, not by a smooth launch.
• Build urgency on a concrete, felt problem people already care about — not a slogan or a fear of falling behind.
• A guiding coalition needs formal authority to clear barriers and informal authority to be credible on the floor. Titles alone add a logo, not weight.
• Use ADKAR to diagnose a stall: find the earliest missing stage. A Desire gap looks like a training gap but worsens with more training.
• Reinforcement is not optional. The strongest form is structural — remove the old option so nothing easier remains.
• Treat resistance as information, and measure behavioural adoption through the two-to-three-month window where habits reassert themselves.

Frequently asked questions

What is the COO's role in change management versus the change manager or HR? The COO owns whether the change sticks operationally — the sequencing, the coalition, and the decision to turn off the old way. A change manager or HR partner runs the mechanics: training, communications, and individual support. Your distinct value is authority: only someone at your level can reallocate resources, override a blocking manager, and make the new process the only one. Delegating the mechanics is fine; delegating the sponsorship is how changes quietly die. Should I roll change out all at once or in phases? Phase it in almost every case. A small first slice designed to succeed gives you a credible early win that travels by word of mouth and builds momentum for the rest. A simultaneous "big bang" maximizes the chance that one early failure poisons the whole effort before any win lands. The main exception is a change that genuinely cannot be partial, such as a compliance cutover with a hard legal date. How do I handle a respected senior person who resists the change? Work out which kind of resistance it is first. If the objection is substantive and the plan really does break something, fix the plan and credit them openly — that turns a loud critic into proof that objections are heard. If they are losing status or a skill they were proud of, have an honest conversation about that loss. Do not win the argument publicly and leave them humiliated, because a respected sceptic you flatten becomes a permanent source of drag. What metrics tell me whether a change actually worked? Behavioural metrics that are hard to fake: the share of work now flowing through the new way without a workaround, days since the last use of the old process, and the gap between your best- and worst-adopting teams. Avoid activity metrics like emails sent or training completed — all of them can read 100% while real adoption sits near zero. Set a few behavioural measures before launch and track them well past the point where the change feels done.