Why Your AI Strategy Is Failing and 5 Things Leaders Who Succeed Do Differently

March 20, 2026

Published in:

Innovation & Technology

Photo by Entrepreneurs' Organization

Nine out of ten companies are experimenting with AI. Nearly 90 percent of CEOs call AI their top investment priority. So why are only 1 percent of organizations—by McKinsey’s count—operating at anything close to AI maturity? And why do 95 percent of enterprise AI pilots, according to a 2025 MIT study, deliver zero measurable return?

Here’s the uncomfortable answer: The technology isn’t the problem. Your organization is. After reviewing the latest research from MIT, McKinsey, Harvard Business School, and others, the pattern is clear. Companies capturing real value from AI aren’t the ones with the most sophisticated models. They’re the ones with the most thoughtful management. That distinction—between a technology bet and a management commitment—is the entire game.

95% of enterprise AI pilots deliver zero measurable return. — MIT NANDA Study, 2025

The Real Bottleneck Is at the Top

Multiple surveys from 2025 point to the same culprit: Leaders not steering fast enough. Not employee resistance. Not model limitations. Leadership.

Cisco’s 2025 AI Readiness Index found that 99 percent of the companies getting consistent, measurable returns from AI share one trait: a well-defined strategy that embraces organizational change and includes formal programs to help employees adapt. That’s a management discipline, not a vendor decision.

McKinsey’s research found that, among 25 organizational factors tested, workflow redesign has the single biggest effect on whether AI delivers EBIT (Earnings Before Interest and Taxes) impact. Yet only 21 percent of companies have fundamentally redesigned even some of their workflows. The gap is not technical—it is organizational.

The Words You Choose Are Costing You

How you talk about AI to your team determines how much value you get from it. Leaders who frame AI as an efficiency tool—code for headcount reduction—trigger a predictable response: Employees do the minimum required. They use the tools superficially. They don’t experiment. They don’t share what works.

A 2025 study in Humanities and Social Sciences Communications found that AI adoption framed as a threat significantly reduces psychological safety at work, increasing employee depression risk. A separate Harvard Business Review analysis by Davenport and Srinivasan found that companies laying off workers based on AI’s potential—rather than proven performance—were souring the very employees they needed to make AI work, often requiring embarrassing rehires.

The research from Brynjolfsson, Li, and Raymond in The Quarterly Journal of Economics tells a different story about what’s possible. Studying 5,172 customer-support agents at a Fortune 500 company, they found a 14 percent average productivity gain from AI—and a 34 percent gain among newer, less experienced workers. The mechanism: AI democratized expert knowledge, compressing the experience curve. That is not an efficiency story. It is more about employee capability. And it only happens when people feel safe enough to engage deeply.

Newer workers achieve a 34% productivity gain when AI democratizes expert knowledge. —The Quarterly Journal of Economics, May 2025

Your Competitive Moat Is Not the Model

Here’s what the strategy textbooks will tell you in five years, but the best operators are figuring out right now: AI models are commoditizing fast. The gap between providers is narrowing. Your competitors can buy the same model you’re using within months.

What they cannot easily copy is your organizational complement—the proprietary workflows you have built, the domain judgment embedded in your processes, the quality of your data, and most importantly, the speed at which your organization learns and adapts. Those are the assets that pass the strategy test for sustained competitive advantage.

Instead of Which model are we using? a more effective board-level question to ask is: Which organizational capabilities are we building that competitors cannot replicate quickly—even if they buy the same tools?

5 Things Leaders Who Get AI Right Do Differently

Here are five actionable steps that leaders who are succeeding with their organizational AI initiatives are doing differently than the rest:

1. They Own the Workflow Redesign

AI value does not emerge from bolting a tool onto existing processes. It comes from reimagining the process around the tool. This work cannot be delegated to IT or a consulting firm—it requires leaders with deep knowledge of the business.

2. They Make Augmentation the Official Story

The framing is a strategic decision. “AI is here to make you more capable” produces different behavior than “AI will make us more efficient.” Choose the former, and mean it.

3. They Build Accountability Into the Design

When AI contributes to a decision, someone still owns the outcome. Defining those accountability structures before deployment—not after a failure—is what separates mature AI programs from chaotic ones.

4. They Treat Governance as an Accelerator

Companies getting the best results from AI manage more risks, not fewer. Lightweight, adaptive governance that evolves with your use cases enables AI to scale safely. Bureaucratic gatekeeping is not governance—it is paralysis.

5. They Measure Differently

Short-term productivity metrics will often look flat or worse during AI implementation—that’s the J-curve that economists have documented for every major general-purpose technology. Leaders who bail during this phase miss the payoff. Broader metrics and longer time horizons are not excuses; they are economic reality.

The window for building organizational advantage around AI is compressing. First movers in technology adoption are being caught quickly. First movers in organizational redesign are not. The companies that will define their industries in the next decade are managing their way there, not buying their way there.

The question is whether your leadership team is treating AI as the management challenge it actually is—or waiting for a technology solution that is not coming.

Contributed by Robert van der Zwart, an EO Netherlands member, who is a coach, keynote speaker, founder of AIPO Network, and a host and organizer of the virtual EO Global AI Summit #4: Transforming to an AI-First Company which took place on 26 February 2026 (EO members can view Summit videos here).