More AI doesn't mean less control

Why 95% of organizations fail to get measurable ROI from AI

Almost every organization is experimenting with AI these days. Yet, very few can actually prove what those experiments are delivering. The conversation often revolves around models, features, and time savings. For executives and managers, a different question is more important: how do you translate AI into measurable business value?

That question is more relevant than ever. According to Rabobank , productivity growth in our region (Regio Zwolle) is lagging 14% behind the rest of the Netherlands. At the same time, the MIT study The GenAI Divide (2025) concludes that 95% of organizations are not getting measurable returns from generative AI.

That makes AI a striking paradox. The need for productivity improvement is high. The technology is available. Yet, most organizations still struggle to realize demonstrable value.

During our AI inspiration lunch in Zwolle, the focus wasn't on the technology itself, but on why the step from experiment to result proves so difficult.

Searching and summarizing isn't the same as making decisions

Generative AI is great at processing large amounts of information. Summarizing documents, retrieving information, and identifying patterns are now standard capabilities. However, business value usually doesn't come from just finding information. In practice, employees have to make choices every day. For example, regarding a customer query, a proposal, an exception to a process, or an internal agreement.

That creates a different need for information:

• Which agreement applies in this situation?
• Which source is the definitive one?
• What is a proposal based on?
• Which parts require human judgment?

That’s the difference between an AI tool and an AI colleague. A tool provides information. An AI colleague supports an employee in preparing for a decision.

Why usage alone doesn't tell the whole story

Many organizations start their AI journey with pilots, experiments, and initial use cases. That makes sense. New technology needs to be explored before it becomes part of daily processes.

At the same time, MIT research shows that using generative AI on its own doesn't automatically lead to measurable ROI. An organization can use AI without actually improving its processes. Employees might save time, but that time doesn't necessarily contribute to customers, quality, or innovation in any visible way.

That’s why the question isn't just whether AI is being used, but rather where the freed-up capacity is going and which results are actually improving as a result.

The dilemma between speed and accountability

Discussions about AI often hit the same snag. Organizations want to speed up their work, but they also want to retain accountability for the outcomes. These don't have to be mutually exclusive. The question isn't how much autonomy an AI solution gets; it's what role AI plays within a process. A reliable AI colleague doesn't take over responsibility—it prepares the ground for it.

An AI colleague that knows where to draw the line

Imagine a director asks how their organization can become AI-proof. To put together a solid proposal, a consultant needs to analyze the conversation, review follow-up emails, consult previous projects, and apply internal policies. The necessary knowledge usually already exists; the problem is that it’s scattered across different sources. An AI colleague can pull that information together, identify the real need behind the request, and prepare a first draft.

But there remains a crucial difference between supporting and deciding. Choices regarding pricing, exceptions, risks, or commitments remain the domain of human judgment. Not because AI falls short, but because organizations must define where accountability lies. When that boundary is clear from the start, AI can actually handle more of the preparatory work.

The button isn't the solution

With AI, a lot of attention goes to the visible application: the chat interface, the assistant, or the button a user clicks to ask a question. But successful implementations usually require more than just a good user interface. They rely on three foundations:

1. AI information management

Which sources are authoritative? Who owns the information? How is the information kept up to date?

2. Making knowledge AI-ready

AI can only work with the information that’s available. Furthermore, the status of that information must be clear. After all, a draft document means something very different from an approved work instruction.

3. Supporting daily operations

Only when information is well-organized can AI effectively support employees in their work. This shifts the focus from technology alone to the combination of technology, information, and ownership.

Tech + Touch + AI

At TeamValue, we look at AI through three interconnected lenses. Tech provides the technical environment where AI can access relevant information. Touch stands for human ownership, collaboration, and decision-making. AI supports employees by pulling information together and prepping proposals.

This combination aligns with how TeamValue uses technology: not as an end in itself, but as a tool to deliver real, measurable value.

What happens when knowledge becomes explicit

Many organizations are sitting on a goldmine of knowledge. Some of it is tucked away in documents, systems, and processes, while the rest lives in the heads of your team. When you start capturing that knowledge more explicitly, the benefits are huge. Decisions become easier to explain, you’re less reliant on specific individuals, and past experiences stay available for future challenges. Plus, you can use feedback to keep refining your processes. These are great wins even without AI, but they also happen to be the foundation for reliable AI support.

Don't start with the tool

So, the first question shouldn't be: Which AI tool should we buy? A better place to start is with a recurring process or decision. Then ask:

1. What knowledge is needed to prepare this decision properly?
2. Which information sources are relevant for that?
3. Where is human judgment still essential?

Only then does it become clear what role AI can actually play.

No choosing between speed and control

AI doesn't have to force organizations to choose between speed and accountability. The real challenge is organizing your knowledge, sources, and ownership. Once it’s clear what information is being used, what proposals are based on, and where the human decision-making line is drawn, you create a workflow where AI supports the team while employees stay in the driver's seat. Or, to put it simply: AI suggests. You decide.

The true value of AI isn't about how many prompts you run or how many people use it—it’s about making better decisions, streamlining processes, and getting better results.

Getting started with AI

AI delivers the most value when it’s built on a solid foundation of knowledge, processes, and ownership. Organizations that invest in these areas create the right conditions to use AI successfully and responsibly. Curious what that looks like for your organization? Get in touch with Christian Sinke via the contact form or email [email protected].