Kotter's 8-Step Change Model: A Modern Take

Kotter's eight steps get read as a checklist, which is why they get done out of order. The model's actual claim is about sequence: each step is load-bearing for the ones after it.

Aldermoor Health's scheduling migration had a project plan that would pass any review. Vision statement, comms plan, training schedule, a named delivery lead, a steering group meeting fortnightly.

It started at step four.

The vision was communicated — thoroughly, in three formats. Teams were empowered to act. Nothing moved for five months. Clinicians attended the briefings and went back to the old system, and nobody could say why, because on paper everything that was supposed to happen was happening.

What was missing was steps one and two, and they were missing for a structural reason: they produce no artifact. A vision statement is a document. A training schedule is a calendar. Urgency and a coalition are neither — they're states of the organisation, they can't be attached to a ticket, and they cannot be marked done. So a plan assembled from deliverables will always start at step three or four.

The steps are a dependency graph, not a list

Read as a checklist, the eight look roughly equal and independently completable. Read as the model actually claims, each one is the precondition for the next, and skipping an early step doesn't delay the later ones — it makes them not work.

Step	Produces an artifact?	If you skip it
1. Urgency	No — it's a shared belief	Everything after is optional to everyone but you
2. Coalition	No — it's a set of relationships	The change is "the delivery lead's project"
3. Vision	Yes — a statement	You get activity without direction
4. Communicate it	Yes — comms plan, briefings	People hear it and don't act, because 1–2 are missing
5. Empower action	Partly — removing blockers	Permission with no motivation changes nothing
6. Short-term wins	Yes — a shipped thing	Momentum decays before the payoff arrives
7. Consolidate	Yes — the next tranche	The first win is treated as the finish line
8. Anchor in culture	No — it's the default behaviour	It reverts; see Lewin on refreezing

The three steps with No in the middle column are the three that get skipped, and they are 1, 2 and 8 — the beginning and the end. That is not a coincidence about willpower; it's a consequence of how plans get assembled from things you can schedule.

The test you can run: open your current change plan and mark each item against a step. If nothing maps to 1, 2 or 8, your plan is the middle of the model with both ends missing.

Urgency is a belief other people hold, and it can't be added later

The most common misreading of step one is that urgency means a deadline. A deadline is yours. Urgency is theirs — the belief, held by the people who have to change, that the current situation is worse than the disruption of changing.

Aldermoor's clinicians did not hold that belief, and had good reason not to. The old scheduling system was slow and ugly and they had twelve years of workarounds for it. It worked. The pain was the organisation's — three sites couldn't see each other's availability, which cost the business real money and cost a clinician nothing.

Manufacturing urgency by announcement doesn't work, and the attempts are recognisable: burning-platform emails, countdown clocks, executive videos. What worked was smaller and specific — the project lead sat with two clinicians while they phoned another site to check availability, twice, and then showed them the same task in the new system. Nine minutes versus forty seconds. That's not a slogan; it's a demonstration in their own terms.

The distinction worth carrying: urgency is not "this is important", it's "the current way costs me something I can feel." If the cost lands somewhere else in the organisation, you don't have urgency, you have a mandate — and a mandate gets compliance, not adoption.

The test: ask someone affected to describe what's wrong with the current way, unprompted. If they can't, or the problem they name is somebody else's, step one hasn't happened.

A coalition is people with standing, not people with titles

Step two is usually satisfied by forming a steering group. Steering groups are frequently the opposite of a coalition: senior, sponsoring, and irrelevant to whether a clinician changes what they do on Tuesday.

What makes a coalition work is standing with the affected group — people whose adoption others read as a signal. That's the person everyone asks when the system misbehaves. Sometimes they're senior; often they're just the one who's been there longest and is generous with help.

Two practical properties:

• Small. Four to six. Large coalitions coordinate rather than act.

• Includes at least one credible sceptic. A coalition of enthusiasts is read, correctly, as the project talking to itself. The sceptic who comes round is worth more than three advocates who were never in doubt.

Aldermoor's coalition was three people: the senior nurse everyone asked for help, one of the seven clinicians who had refused the new system on well-measured grounds, and the person who ran the rota. Not one was on the steering group.

The test: name the person your affected team asks when something breaks. If they're not involved in your change, you have a steering group and no coalition.

Steps 6 and 7: the win that ends the change

Short-term wins exist to prove the direction before the full payoff arrives. The failure mode is subtle: the win arrives, it's celebrated, and it becomes the finish line.

Aldermoor's first win was one site fully migrated in six weeks. It was announced, it was real, and it nearly killed the programme — attention moved to the next initiative while two sites were still on paper and the cross-site visibility that justified the whole project was impossible.

Step 7 exists precisely to counter this, and the framing that makes it stick is to describe the win as evidence rather than arrival: this proves the approach works, here is what it lets us do next. That is a different sentence from "we've done it", and the difference decides whether steps 7 and 8 happen.

The test: after your next milestone, ask whether the announcement described a finish or a proof. If it read as a finish, expect the remaining work to lose its people.

Where Kotter is weak

It is a top-down, organisation-wide, long-horizon model, and it shows its age in three places.

• It assumes an organisation-sized change. For a team-level change, eight steps is ceremony. The useful subset is 1, 2 and 8.

• It's slow. The full sequence assumes months. Continuous delivery has changes shipping weekly, and you cannot run Kotter for each one.

• It says nothing about individuals. Kotter operates on the organisation. Whether a specific person adopts is ADKAR's question, and how long they'll feel unsettled is Bridges'.

Used well, Kotter answers one question: in what order should this organisation-wide change happen? Asking it to answer anything else produces a lot of process and not much movement.

The test: if your change affects one team and ships this month, don't run Kotter. Take steps 1, 2 and 8 and leave the rest.

What to do this week

Take your current change and answer two questions honestly.

Who feels the pain of the current way? If the answer is the business rather than the people being asked to change, you don't have urgency, and no amount of communication will substitute. Go and find a task where the old way visibly costs them something, and show them the difference. Nine minutes versus forty seconds.

Who do they ask when something breaks? That person belongs in your coalition. If they're not in it, your change is a project rather than a direction — and projects end when attention does.

Aldermoor's migration completed in eleven weeks once those two things were true. The vision statement never changed. Nothing about steps three through eight was different. They had simply been running the second half of a sequence whose first half hadn't happened.