Every stalled transformation I have rescued had a stakeholder map. It was a slide, made once, in month one, and never opened again. That map did not predict the trouble. It documented it after the fact.
I run change on large ERP and AI programs at Novartis. When a rollout goes sideways, the root cause is almost never the technology. It is a person with influence who quietly stopped believing, and nobody noticed until go-live slipped.
So I stopped drawing static maps. I built one that moves.
Three axes, not two
The classic 2x2 plots influence against interest. Interest is useless. A regional finance director can be intensely interested and still torpedo you. I score every key stakeholder on three axes instead:
- Influence — can this person accelerate or block the program? Formal authority plus informal network reach. Scored 1 to 5.
- Alignment — do they actually want this outcome? Not what they say in steering committee. What they do when I am not in the room. Scored -2 (actively against) to +2 (champion).
- Readiness — is their team capable of absorbing the change on the current timeline? Process maturity, data quality, spare capacity. Scored 1 to 5.
Influence tells you who matters. Alignment tells you which way they lean. Readiness tells you whether their good intentions will survive contact with reality. You need all three. High influence plus high alignment plus low readiness is still a failure mode: a champion whose team drowns.
The number that predicts trouble
For each stakeholder I multiply influence by alignment. That single figure is my early-warning signal.
A person with influence 5 and alignment -1 scores -5. That is a bigger threat than someone at influence 2 and alignment -2. High-influence lukewarm beats low-influence hostile, every time.
I sum the negatives separately from the positives. On my last SAP wave, the negative total was -14 in March against a positive total of +31. Healthy. By May the negatives had drifted to -22 while positives held flat. Nothing had officially gone wrong. But the map had moved, and I acted six weeks before the first missed deadline.
Make it move
A map is only predictive if it updates faster than the situation changes. Mine gets re-scored every two weeks. Three inputs feed it:
- Direct signal. Fifteen-minute one-to-ones with the top influencers. I am not asking for status. I am reading tone, hesitation, what they avoid.
- Second-hand signal. My change agents in each function report shifts I cannot see. A champion who stops replying to their own team is a leading indicator.
- Behavioural signal. Did they show up? Did they send a delegate? Did the promised data extract arrive? Behaviour is the least flattering and most honest input.
Re-scoring takes me forty minutes a fortnight. It is the highest-return forty minutes on the program.
What the movement tells you
The absolute score matters less than the vector. I watch three patterns:
- A champion sliding toward neutral. Usually means they are absorbing costs they did not sign up for. Fix the load before you lose the ally.
- A blocker's influence rising. A reorg, a promotion, a new reporting line. Their alignment did not change but their weight just doubled. Re-engage immediately.
- Readiness dropping across a whole function. This is not a stakeholder problem, it is a capacity problem wearing a stakeholder mask. No amount of persuasion fixes an overloaded team.
Act on the map, or do not bother
The map is worthless as a record. It is only worth building if it changes what you do this week. Each fortnightly re-score ends with three named actions, each owned, each with a date. If nothing changed in the scores, I still ask why the risky people did not move — silence is data too.
The comforting lie of change management is that resistance appears suddenly. It never does. It builds, visibly, in the influence-and-alignment space, for weeks before it surfaces as a slipped milestone or a failed adoption metric.
A one-off slide cannot see that. A living map can. The difference between the two is about forty minutes every two weeks — and roughly one saved program per year, in my experience.
Cédric Bignet is an AI & ERP Change Management expert at Novartis and founder of AInspire. He writes about change management, AI adoption and enterprise transformation.