"Change can be managed."

Most meaningful change isn't a linear project; it's a complex adaptive system, and those can't be controlled — only influenced. The evidence is brutal. MIT's State of AI in Business 2025 report found that roughly 95% of corporate AI initiatives fail to deliver measurable impact — and, critically, the failures were overwhelmingly organisational, not technical.

 

The models mostly worked; what broke was integration, adoption, and the human system around them. That's the whole lie in one statistic: organisations treated a complex human transformation as a technical rollout, and it collapsed. Meanwhile the people expected to absorb all this change are exhausted.

 

Gartner's research found the average employee faced around ten planned enterprise changes in a single year, up from just two in 2016 — and willingness to support organisational change has fallen from 74% in 2016 to 43%. This is "change fatigue," and it's not resistance to any single change; it's the cumulative weight of constant, poorly-led disruption.

 

And here's the part the linear models miss entirely: change moves through people the way grief does — the "change curve" of shock, resistance, exploration and, eventually, commitment. You cannot Gantt-chart someone through denial. A person mid-resistance isn't behind schedule; they're exactly where the human process puts them. Treating that as a project-management failure, rather than a normal stage to be led through, is how transformations die.

 

The uncomfortable truth is that change can be led, shaped, and navigated — but it cannot be managed in the tidy, controllable sense the word implies. The illusion of control is the thing that keeps sinking the ship.

Trade the plan for a direction — set a clear destination and guiding principles, then adapt the route as reality responds. Rigid multi-year change plans are obsolete the moment they meet a live, reacting organisation. Sense and respond; don't dictate and enforce.

 

Lead people through the curve, don't manage them past it — expect shock, resistance and exploration as normal stages, not obstacles. Meet people where they actually are, name the discomfort honestly, and give the process time. Skipping the curve doesn't speed change; it buries it.

 

Respect the change budget — people have finite capacity for disruption, and you've probably already overspent it. Before launching the next initiative, ask what you can stop, pause, or protect people from. Fewer, better-led changes beat a constant churn nobody believes in.

 

Be honest, especially when you don't have the full picture — change fatigue is fed by spin and false certainty. "Here's where we're going, here's what we don't yet know, here's how we'll figure it out together" builds the trust that carries people through uncertainty. Manufactured confidence destroys it.

 

Treat AI change as human change — the 95% failure rate is the warning. The technology is rarely the hard part; the organisation is. Invest in adoption, learning, and the human system around the tool at least as much as in the tool itself, or you'll join the 95%.