The 7 Key Steps of Change Management
The 7 key steps of change management are: diagnosing readiness, building a leadership coalition, defining a concrete vision, communicating continuously, removing structural obstacles, investing in skills development, and measuring value capture. Applied to AI integration, these steps determine whether an initiative survives beyond the pilot phase.
Many companies across Morocco, Belgium, and France are deploying AI tools. Few succeed in making adoption stick. This is not a technology problem. It is a human one.
1. Diagnose Before You Decide
Before talking about AI, ask a simple question: is your organization ready to change? Not ready to buy a tool. Ready to change its habits, processes, and reflexes.
This diagnosis covers several dimensions: team maturity, data quality, managers’ capacity to carry a change message, and latent resistance. What I observe with my clients is that this step is systematically rushed. Organizations jump straight to solution selection.
The result: deployments that stop after three months because no one anticipated that frontline teams would not trust the algorithm’s recommendations.
2. Build the Coalition Before You Launch
Here is the counter-intuitive step. Most executives believe an AI initiative is driven from the top. It is not. It is driven by a coalition.
This coalition must include a visible executive sponsor, convinced line managers, and profiles with credibility among operational teams. Without this triangle, the initiative remains the concern of the IT department (DSI, Direction des Systèmes d’Information) or consulting firms. It never becomes the organization’s own.
According to Le360, consulting and audit firms in Morocco are redefining their own working methods under the influence of AI. Those succeeding in this transition share one trait: partners who carry change internally, not just tools deployed by management.
3. Formulate a Vision Everyone Understands
A change management vision is not written in consultant jargon. It must answer the question every employee is asking: what does this change for me, concretely, in my work tomorrow?
If your answer fits in three slides and two acronyms, start over. The vision must be formulated in operational language. Not “optimizing decision-making processes through AI”. Rather: “you will spend less time searching for information and more time deciding”.
This is the vision framework I structure in my 2 to 3-week AI Governance Sprint. Learn more about my services.
4. Communicate Continuously, Not Once
Change communication is not a launch email. It is a flow. Before, during, and after deployment.
Teams need to hear the same message multiple times, through multiple channels, from people they respect. When a CHRO communicates only once in a plenary meeting with no follow-up, a large part of the organization is left uncertain and disengaged.
In the AI deployments I accompany, the strongest resistance rarely comes from the least qualified teams. It comes from middle managers who were not involved in the decision and who perceive the tool as a threat to their role as information filters.
5. Remove Structural Obstacles
A change management initiative that does not touch existing processes changes nothing. It is a coat of paint.
If you deploy a conversational agent for customer service but your teams are still evaluated on manually handled call volumes, the issue is not the tool itself: it is the misalignment between your performance indicators and your new operating model. Validation processes, reporting habits: all of this must be re-examined alongside the technology deployment.
As I explained in my article on integrating AI in recruitment, process redesign always precedes durable tool adoption.
6. Invest in Skills Development
AI culture is not decreed. It is built, progressively, through training and experimentation.
This does not mean training everyone in Python. It means every employee understands what AI can do within their scope, what it cannot do, and how to interact with it critically. This is what AI literacy means, and it is currently one of the most underestimated gaps in organizations.
Unmanaged AI is a real risk. EcoActu.ma documents this explicitly: companies deploying tools without formal accompaniment expose themselves to risks that no one anticipated. That warning applies in Casablanca as much as in Brussels.
7. Measure Value Capture, Not Adoption
Final point, and the one most leadership teams forget. Organizations measure tool adoption rates. They do not measure what the tool actually generated in terms of impact on decisions and organizational outcomes.
These are two different things. A tool used by 80% of teams can produce zero impact on decisions made. Conversely, a tool used by 20% of teams on the right use cases can transform an entire department.
Define your outcome indicators before deployment. Not after. And connect them to concrete business objectives, not usage metrics. For further context, see my analysis on the types of artificial intelligence and what they mean concretely for your processes.
If you are a CHRO or CEO and want to structure your change management approach around AI, request a free diagnostic.
FAQ
What is the first mistake in AI-related change management?
Skipping the diagnosis. Most organizations jump straight to tool selection without assessing their teams’ actual maturity, data quality, and existing resistance. This shortcut is costly.
How long does change management for an AI deployment take?
There is no standard duration. A deployment on a limited scope can be structured in a few weeks. A transformation across an organization of several hundred people requires several months, with distinct phases of piloting, adjustment, and scaling.
Are middle managers really a barrier?
Not by nature. But they are often the first to perceive AI as a challenge to their role. When they are not involved in deployment design, they become passive or resistant. When engaged early, they become the strongest change ambassadors.