People First · AI Adoption
Organizational Change Management for AI and System Rollouts
A new system can go live on time, on budget and fully tested, and still fail. It fails because the people who have to use it every day were never brought into the change.
That is the gap organizational change management closes. I came to it through conversion work: years of building the websites, content and funnels that turn a visitor into a customer. The forces that decide whether a buyer says yes (motivation, value, incentive, friction and anxiety) are the same forces that decide whether a team adopts a new way of working. Raising the value of the new way and removing the friction and anxiety around it is the same job whether the person is a customer or a warehouse supervisor.
People first. AI with a human touch.
Three principles shape every engagement, and they matter most when the change involves AI.
Secure the people. Protect the people who carry the know-how. Long-tenured staff often hold the knowledge that keeps the operation running, and a rollout that treats them as an obstacle loses that knowledge at exactly the wrong moment.
Solve with the people. The people closest to the work define the problem. They know where the process breaks, which workarounds exist and why, and what will actually get in the way on day one.
Scale through the people. Adoption is the only multiplier. A capability nobody uses returns nothing, however well it was built.
Start with what people actually think
The most useful work in any change programme happens early: short, structured interviews with the people affected, usually fifteen to twenty minutes and six to ten questions each. Four themes carry most of the value.
- Expectations. What do they think is actually going to happen?
- History. What has and has not gone well the last time this organization changed?
- What keeps them up at night. Their real concern, in their own words.
- Incentives and barriers. If people have no reason to change and real obstacles in the way, they will not change. Incentives have to be created and barriers removed.
Those answers drive the rest of the strategy. Without them, communications and training are built on assumptions.
Change impact, stakeholders and the change network
From the interviews comes a change impact assessment: which roles are affected, what changes for each of them, and how much. Stakeholder mapping sets out who has influence, who is ready, and where resistance is likely to come from and why.
A change network of champions inside each team then carries the change in the language of that team. People believe a respected colleague more readily than a project announcement, and champions give the programme an early warning when something is not landing.
Training on the real work
Training fails when it teaches the system instead of the job. People need to practise the tasks they will actually do on day one, in the order they do them, with support that is withdrawn gradually as they prove they can do the work. That approach draws on Training Within Industry, the method developed to train very large numbers of people quickly and to a consistent standard.
Where Lean Six Sigma fits
Change and process improvement belong together. Automating a broken process encodes the problem in software, so the process is mapped and fixed with the people who run it before anything is automated. The process improvement work and the change work share the same interviews, the same maps and the same measures.
AI adoption specifically
AI changes the equation in two ways. It raises anxiety, because people reasonably ask whether it replaces them, and it changes how work is checked, because an AI draft still needs a human decision. The approach that works is to let AI draft and let people decide, to involve the people closest to the work in choosing where AI is used, and to say plainly what stays with human judgement. The same discipline sits behind the AI consulting work.
Measuring adoption, not just go-live
Adoption measures are agreed at the start: who is using the new process, how often, how well, and whether the old workaround has stopped. Readiness checks before launch and the same measures after it show whether the change took hold. Where the data allows, those measures go into a dashboard leaders will actually open, built on the same data engineering work.
What the engagement includes
- Stakeholder interviews and mapping. Structured conversations with the people affected, and a clear picture of influence, readiness and concern.
- A change impact assessment. What changes for each role, and how much, so effort goes where the impact is.
- A change network. Champions recruited and supported inside each team affected.
- Communications. A plan that explains why the change matters to each audience, in their terms, at the right moments.
- Training on the real work. Role-based, task-first training with support that steps back as people prove the work.
- Resistance management and adoption tracking. Measures agreed up front, reviewed through and after go-live.
How the work runs
- Listen. Interview the people affected and map the stakeholders.
- Assess. Build the change impact assessment and agree the adoption measures.
- Plan. Communications, training and the change network, shaped by what people said.
- Prepare. Train on the real work and check readiness before go-live.
- Sustain. Track adoption after launch and act where it stalls.
Experience
Change work runs through the Analytics AIML and Aptask partnership, alongside Frank Shines, and client work is under NDA. It spans pharmacy benefits, medical devices, diagnostic laboratories, education services and infrastructure construction, as well as ERP and AI rollouts.
Related
Talk about your situation
The first conversation is short and mostly questions. Get in touch and tell me what you are trying to change. Or see how this is priced.
Frequently Asked Questions (FAQs)
What is organizational change management?
The work of bringing people through a change so that a new system, process or tool actually gets used. It covers understanding who is affected and how, communicating the change, training people on the work they will really do, managing resistance, and measuring adoption. Projects deliver the technology; change management delivers the people side of the result.
Why do system rollouts fail when the technology works?
Because go-live and adoption are different events. A system can be configured correctly, tested and launched on schedule, and still fail if the people who use it do not see the value, find it harder than the old way, or are worried about what it means for their role. Those are people problems, and they need to be worked on before launch, not after.
How is AI adoption different from other change?
AI raises the anxiety part of the equation sharply, because people reasonably ask whether it replaces them. It also changes how work is checked, since an AI draft still needs a human decision. Adoption goes better when the people closest to the work help define where AI is used, when the know-how they carry is protected, and when the role of human judgement is stated plainly.
How does a marketing background help with change management?
Conversion work studies exactly why people say yes or no: motivation, perceived value, incentive, friction and anxiety. The same forces decide whether an employee adopts a new way of working. Years of measuring what moves a buyer translate directly into designing communications, training and incentives that move a team, and into measuring whether they did.
When should change management start on a project?
At the start, alongside the technical plan. Stakeholder interviews and the change impact assessment shape the training plan, the communications and often the design itself. Brought in a few weeks before go-live, change management can only announce decisions people had no part in, which is the situation that creates resistance.
How do you measure whether a change actually worked?
With adoption measures agreed up front: who is using the new process, how often, how well, and whether the old workaround has stopped. Usage data, short pulse surveys and readiness checks before go-live, followed by the same measures after it, show whether the change took hold or only went live.