The AI Illusion: Why Most Companies Fail at AI Transformation (Part 2)

May 29, 2026

Published in:

Innovation & Technology

A young woman studies her phone with a look of deep thought on her face. — Photo by Entrepreneurs' Organization

In Part One of this two-part series, we looked at the uncomfortable reality behind today’s AI hype cycle: Despite endless promises from influencers and software vendors, the overwhelming majority of AI initiatives fail to produce meaningful business results.

Research from MIT, McKinsey, the RAND Corporation, BCG, and Harvard Business Review consistently points to the same conclusion: AI success has far less to do with the tools themselves and far more to do with how organizations redesign workflows, train people, and manage change.

So, where should leaders actually focus their energy, time, and investment if they want AI to deliver real value in the real world? Let's jump in.

The Rule That Should Be on Every CEO’s Wall

BCG and MIT Sloan studied hundreds of AI transformations and distilled their findings into a simple framework. It’s not a peer-reviewed law of nature; it’s a rule of thumb from real consulting experience. But it captures the truth better than anything else I have found.

Here is the three-part rule that should be on every CEO’s wall regarding the proportional formula for AI resources that will lead your company to success:

10% goes to algorithms and models — the actual AI tools
20% goes to infrastructure — data, security, integration
70% goes to people and processes — change management, training, workflow redesign

Read those numbers again. The influencers are selling you the 10%. Every “top 10 AI tools” listicle, every flashy demo, every “you need this tool right now” reel. They focus on the part that matters least.

The 70% is the unglamorous stuff. It’s sitting in meetings to figure out who owns what. It’s redesigning how decisions flow through your organization. It’s coaching managers who are scared of change. It’s supporting employees when the new system frustrates them. It’s building governance frameworks that are boring but necessary. Nobody posts that on Instagram. But that is where the actual value lives.

Why Tool-Chasing Is a Trap

Here is the dirty secret of the AI influencer economy: They need you to keep chasing tools. Their business model depends on it. Every three months, there is a new model, a new platform, a new “game-changer.” And every three months, you start over.

S&P Global has already documented what this looks like at scale: Companies abandoning AI initiatives at more than double the rate of the previous year. The hype cycle is not a metaphor. It’s a measurable pattern of waste.

MIT discovered something important: Buying an AI tool from a specialized vendor has about a 67% success rate. Building your own AI succeeds about a third as often. The lesson isn’t that one approach is always better; it’s that companies spend months agonizing over which tool to buy and days figuring out how to actually implement it.

They have the ratio completely backwards.

Companies that succeed, that rare 5 to 10%, don’t start by asking “What tool should we use?” They begin by asking, “What problem are we solving, and how does work need to change to solve it?” The tool comes last, not first.

But What About the Companies That Move Fast and Win?

I want to be fair. There are companies that have moved quickly with AI and gotten real results. Some startups have built AI-native operations from day one. Some teams inside larger organizations have found pockets of genuine automation without launching a full-scale transformation.

These stories are real and important. But look closer, and you will see something: Startups didn’t start with broken workflows to fix. They built their processes around AI from the beginning, which itself is a form of fundamental design. And successful teams in large companies? They nearly always had strong leadership support, clear process ownership, skilled staff, and clean data. They might not have called what they did “transformation,” but they did the hard work anyway.

Speed and experimentation are not the enemy. They are essential. But speed without structural change is just fast failure.

The Uncomfortable Truth

I have become increasingly convinced that the AI influencer industry is doing more harm than good. Not because AI isn’t transformative — it absolutely is, and I use it every day. But because the narrative of effortless transformation gives leaders permission to skip the hard parts. And the hard parts are the only parts that matter.

Here are the four things that the research actually tells us to do:

1. Redesign your workflows.

Don’t glue AI onto broken processes. Go back to first principles. Ask how work should flow if you were designing it today, from scratch, with these capabilities available.

2. Train your people — for real.

Not with a video library nobody watches. In the flow of actual work, with managers who lead by example, with incentives that reward adoption, and with the patience to let people learn by doing.

3. Get your data house in order.

Before you deploy. Before you demo. Before you buy. The plumbing has to work, or nothing built on top of it will.

4. Stop chasing tools.

The next shiny model will not save you. BCG’s 10-20-70 rule is real: 70% of the effort is people and processes. Spend accordingly.

The influencers will keep promising the moon. Whether the true failure rate is 80% or 95%, the research from every credible institution says the same thing: The overwhelming majority of AI projects fail.

The question is not whether you should adopt AI. It’s whether you are willing to do the work that makes it actually deliver.

There are no shortcuts. And anyone who tells you otherwise is selling something.

Contributed by Robert van der Zwart, an EO Netherlands member, who is a coach, keynote speaker, founder of AIPO Network, co-founder of The Clever Innovation Box, which assists companies in successfully transitioning to an AI-enabled organization. Robert recently spent five days as the member host of EO’s inaugural executive education program, the EO Stanford Graduate School of Business: AI Integration Lab and wrote about it in 3 Reasons Most Companies Are Getting AI Wrong (And What To Do About It).