AI, Change Fatigue, and the Missing “What” in Transformation

The uncomfortable numbers

Depending on the study you look at, somewhere between 65% and 80% of large transformations fail to fully deliver on their stated objectives. Digital and AI programs are no exception: BCG’s research across hundreds of companies shows that about 70% of digital transformations fall short of their targets.

Recent analyses even suggest the problem is getting worse, with some 2024 executive surveys indicating that close to 9 out of 10 large-scale business transformations fail to achieve their original ambitions. The report published by MIT in August 2025 showed 95% of AI transformation pilots do not show any ROI.

For all the frameworks, playbooks, and methodologies, change fatigue is rising, while sustained impact remains rare.

How we misdiagnose “failed change management”

When these programs stall, the diagnosis is almost always the same:

• “People are resistant.”
• “Change fatigue is too high.”
• “Our culture is not ready.”

So organizations double down on the psychology of change. They move change management under HR, hire coaches, psychologists, or neuroscientists, and launch new training, communications, and engagement campaigns.

All of this matters. But in many boardroom conversations, the entire topic of change management gets compressed into the how of change: communication, resistance, mindset, incentives. The deeper problem is more basic and more structural: the what is not defined clearly enough for any of those tools to work.

The missing “what”: clarity before psychology

In every transformation that actually sticks, there is ruthless clarity on what is supposed to change. Not at the slogan level, but at the level of operating reality.

That means answering questions such as:

• What specific business problems are we solving?
• What outcomes will be different for customers, employees, and shareholders?
• What processes, decisions, and workflows will be redesigned?
• What roles will change, disappear, or be created?

Research on failed change efforts consistently highlights vague or unrealistic goals as a primary reason initiatives stumble. Without a sharp definition of what is changing, organizations end up with beautiful slide decks and town halls—but very little real movement on the ground.

Only when the “what” is concrete can leaders communicate in a coherent and cohesive way why change matters and what it actually means for different parts of the business

Why AI transformations struggle even more

AI amplifies this problem.

On paper, almost every large company now has an AI strategy, but the value realization gap is striking. BCG’s 2024 research shows that about three-quarters of companies struggle to achieve and scale value from AI, and that roughly 70% of the obstacles are people- and process-related rather than algorithmic. Other workforce surveys show that only a small minority of employees consider themselves confident AI users, with most having had less than a few hours of training.

Underneath these numbers, a common pattern emerges:

• AI is framed as “we need to use AI” instead of “we need to solve this specific business problem where AI can help”.
• Investments concentrate on tools and pilots, but frontline teams cannot answer a simple question: “What exactly is going to be different in how I work and how we create value?”
• Leaders over-index on building AI literacy and underinvest in redefining AI-enabled ways of working and AI-enabled business models.

In other words, organizations are doing a lot of work on the how of AI (training, experimentation, tools) without first deciding the what in enough detail.        

A simple 2×2: “What” vs “How”

This simple picture reinforces a hard truth: even world-class change management cannot rescue a transformation that is vague about what is actually changing.        

So where should CEOs start?

From conversations with CEOs and C-suites wrestling with AI and broader transformation, the most effective leaders seem to do three things differently:

1. They define the “what” in business language first
2. They treat the “how” as a design discipline, not just psychology
3. They are honest about bandwidth and blind spots

A question for leaders driving AI

If your AI or transformation program feels heavy, slow, or exhausting for the organization, the root cause may not be “resistance” at all. It may be that people are being asked to change without a clear sense of what is truly changing and why it matters for the business and for them.

When you look at your own AI agenda today, which box does it really sit in on the “What vs How” 2×2 and what would it take to move it into sustainable transformation?

About the author Reshma Ramachandran is an industry leading expert of designing and leading successful business transformations, specifically the ones that leverage technology. She has worked in 13 countries across 3 continents and has worked with/advised several Fortune 500 companies on transformation.