Your AI Roadmap Is Missing a Step: ADKAR

As a consultancy company we often do assessments where we deliver a (amongst other things) a roadmap as an outcome. Nowadays, we get a lot of questions on AI adoption roadmaps. What we then also see a lot, is that the customer also brings the tools and solutions they envision.

That is the pattern I keep running into. With AI, more than with anything before it, companies arrive with the solution already named. Give me Copilot, give me an agent framework, give me the platform. And because the tooling is so easy to buy, we tend to dive straight into rollout and training, and expect people to pick it up.

Sometimes it flows. More often we hit resistance, or something worse. People who nod along in the workshop and then quietly go back to doing things the old way.

We start in the wrong place

When I look at how most of these roadmaps get built, they start at Knowledge. Here is the tool, here is the training, here is the documentation. We skip straight to teaching people how, without ever establishing why or whether they actually want to.

ADKAR, the change model from Jeff Hiatt and Prosci, lays the stages out in order for a reason: Awareness, Desire, Knowledge, Ability, Reinforcement. It is sequential. You cannot buy your way past an earlier stage by investing more in a later one. A brilliant training program does not fix a team that was never convinced the change was necessary in the first place.

Awareness before anything

Awareness is not a kickoff deck. It is explaining, in terms people actually recognise, what problem you are solving and why now. I like going back to the five whys here, because the stated reason for a change (“we are adopting AI”) is almost never the real one. Keep asking why until you hit something a person on the ground actually cares about: fewer weekends lost to repetitive work, less time spent on things nobody enjoys, a product that ships faster because the boring parts are automated.

Skip this step and Knowledge lands on resistant soil. People will sit through the training and still not use the tool, because nobody ever told them why their current way was insufficient.

Desire is where most roadmaps die

This is the stage almost every transformation initiative underestimates. Desire is personal. It is not enough for leadership to want the change, each individual has to arrive at wanting it too. That means actually talking to people: what do they need, what are they afraid of losing, what would make this worth the disruption. Explain the change in terms that make sense to them, not in terms that make sense to the steering committee.

Teams understand what is changing and how to operate the new system all the time. They just never genuinely committed, because the why never landed for them personally. That gap between understanding and wanting is where most AI rollouts quietly stall.

Knowledge, the part we are good at

Once Awareness and Desire exist, Knowledge is the easy part, and it is the part most organisations are already decent at. Documentation, training, office hours. This is also where I lean on a simple progression: show, do together, do yourself. Demonstrate it, then help people try it in a safe environment where failing costs nothing, then let them run with it on their own.

Ability needs practice, not a certificate

Knowing how is not the same as being able to. I can tell people How to drive a car, they can watch YouTube, but that does not make them able to drive. Ability comes from repetition in real conditions, with room to fail without consequence. A certificate of completion is not ability. A few weeks of using the thing on real work, with someone around to unblock you, is.

Reinforcement, or everyone slides back

This is the stage leadership skips most often, because it looks like the project is already done. Make champions out of the people who got there first, reinforce the message repeatedly, and build in feedback loops that catch people drifting back to the old way before it becomes habit again. Without this, adoption curves look great at launch and quietly erode three months later.

The roadmap question

When I am asked to help sequence a transformation, ADKAR is the model I reach for. For every milestone on the roadmap, I ask which ADKAR stage it actually serves, and whether the stages before it are done. A roadmap that is all Knowledge and Ability work, with no Awareness or Desire built in, is not a roadmap. It is a training schedule that is about to disappoint everyone.

If you want to learn more about this and other models I use regularly, visit the Models section of this site, including the ADKAR Change Model page itself. And if you want the fuller story, the book ADKAR: A Model for Change is the one I keep coming back to.

People only start moving when they want to. Getting the sequence right is how you make sure they want to.