You can predict employee reactions to change by combining three things: what the change takes away from each group (control, status, certainty, workload), how people tend to respond to pressure and uncertainty, measured through behavioral profiles, and early signals from pilots and conversations. AI behavior models now add a fourth option: simulating likely reactions from each person's profile before the announcement. The goal isn't to manipulate people. It's to prepare the support, timing and communication each team needs.
Why prediction matters
Prosci's research has repeatedly found that projects with excellent change management meet their objectives far more often, around 88% of the time, than those with poor change management, around 13%. Much of that gap comes from anticipating how people will react and preparing them. You may have heard that 70% of change initiatives fail; researchers have challenged that figure for lacking solid evidence. The more useful point is that preparation is something leaders control.
What drives reactions to change
| Driver | Question to ask |
|---|---|
| Loss of control | Do people get any say in how the change lands? |
| Uncertainty | What don't they know, and when will they? |
| Status and identity | Does anyone lose expertise, title or influence? |
| Workload | Is the change landing on already stretched teams? |
| Trust | How did the last change go? |
| Individual tendencies | How do these specific people handle ambiguity and pressure? |
Four ways to predict reactions
1. Impact mapping
For each team, list what changes in their daily work and what they lose or gain. Teams losing the most control or status are where resistance usually starts.
2. Behavioral profiles
People differ in how they respond to uncertainty. Some move fast and want details later; others need clarity first; some resist anything that disrupts routine. Team-level profiles show where those tendencies concentrate, so leaders can tailor communication and support. Cairo's team views show this across 80+ indicators.
3. Pilots and listening
Test the change with one team, run short conversations with influential employees and watch for early signals: questions asked, rumors, drops in participation.
4. AI behavior simulation
Behavior models can now predict how specific people are likely to choose among options, given their profile and the situation. Cairo Rosetta does this for organizations: leaders describe the change and get likely reactions by team, so they can test different announcements or sequences beforehand. In our first public test, Rosetta reached 60% accuracy on 30 held-out real decisions, competitive with frontier LLMs; it's an early, honest result, and predictions should inform judgment, not replace it. See Rosetta Arena results.
From prediction to preparation
- Tailor the message to what each team will lose and what they need to hear first.
- Brief managers early and give them the likely questions.
- Give people some control over how the change is implemented.
- Sequence the rollout, starting with teams most likely to adopt.
- Offer development for new skills the change requires.
- Measure and adjust with short pulses after launch.
Ethical guardrails
Prediction should help people through change, never single them out. Report at team level, use data people knowingly provide, and never use predictions to decide who stays or goes. See employee digital twins for the principles.
Frequently asked questions
Can you predict how employees will react to change?
Partly. Impact mapping, behavioral profiles, pilots and AI behavior models can identify likely reactions by team, which helps leaders prepare. No method is certain.
Why do employees resist change?
Usually because of lost control, uncertainty, threats to status, extra workload or low trust from past changes.
Do 70% of change initiatives fail?
The figure is widely quoted but lacks strong evidence. Research from Prosci shows success rates depend heavily on how well change is managed.
How does AI help with change management?
AI behavior models can simulate likely reactions from employee profiles, so leaders can test messages and plans before launching a change.
Sources
- Prosci, 5 Strategic Decisions for Building Organizational Change Capability.
- Hughes, M., Do 70 per cent of all organizational change initiatives really fail?, Journal of Change Management, 2011.
- Cairo Labs, Rosetta Arena, Experiment 1, 2026.
