Consider a composite of a situation that plays out surprisingly often in manufacturing (at least, from our experience!).

A company announces an ambitious connected-factory programme. Headquarters wants every critical machine linked to a common platform and dozens of KPIs visible on a single dashboard. Sounds great! 

During discovery, the implementation partner asks questions such as: Who will use each metric? What decision will it change? What happens when a threshold is crossed? There are clear answers for some, but for the rest, the response is a shrug and “Management wants to see it”. The implementation partner recommends starting smaller: perhaps something with one operational constraint, a few useful measures and a defined response. But no, it’s too late - the programme has already been announced internally, budgets signed off and timelines promised to the board. 

So off goes the partner to begin the development and implementation. It passes acceptance testing, the dashboards look impressive and there is praise all round. For the first few weeks, everyone logs in. Success!

A year later, most alerts have been muted. Operators still rely on familiar calls and spreadsheets. The dashboard is opened mainly before management reviews.

Technically, the implementation succeeded. But anyone who has been through this before, knows it has failed in practice.

More data is not the same as more value

This is a recurring digital-transformation malaise. The ambition often begins with a technology desire rather than a business challenge: “We need Industry 4.0.” “We need an IoT platform.” “We need a digital twin.” And yes, increasingly, “We need AI.”

The project is then measured through machines connected, tags deployed, data points captured and dashboards built. These are legitimate delivery metrics. However, they are not business outcomes. We find a more useful starting question is: What will somebody be able to decide or do better because this data exists?

This is close to the established idea of the “value of information”: information has value when it improves a decision or outcome enough to justify the cost of collecting, processing and using it. NIST describes this as a “data-to-decision pathway”—a useful reminder that data is only the beginning of the journey.

A factory can become more digital without becoming meaningfully better

The gap between implementation and value is well documented. BCG found that only 30% of digital transformations met or exceeded their target value and produced sustainable change. In manufacturing, a World Economic Forum study found that just 39% of surveyed executives had scaled data-driven use cases beyond a single product and established a clearly positive business case.

This is not to say digital transformation rarely works. But what it does mean is that technology or dashboards alone do not complete the transformation.

This malaise is hardly confined to digital transformation. Give enterprises a technology (or trend!) and the forest is often mistaken for the trees.

In the 1990s, companies invested heavily in enterprise systems, sometimes expecting the software itself to create integration. Even then, experts warned that the promise of seamless integration could obscure the organizational and process changes required to realise it.

In the 2000s, data warehouses and business-intelligence programmes accumulated reports that were produced more reliably than they were used. In the 2010s, IoT generated “pilot purgatory”: a Cisco survey found that 60% of IoT initiatives stalled at the proof-of-concept stage.

And as many of us know, AI is providing the newest chapter of this story. McKinsey’s 2025 survey found that nearly nine in ten organizations regularly used AI, but only 39% reported any enterprise-level EBIT impact. 

Tellingly, what made a difference to the high performers was not access to a better technology, but the internal redesigning of workflows around it. And it is from that we must take lessons for digital transformation.

Start with the decision, not the device

We would always recommend a value-first approach to digital transformation. This means there is more emphasis on the “so what” rather than the “how”. Such a project should answer six questions before time and money is committed:

1. What decision or action must improve?
2. Who will use the information and own the response?
3. What uncertainty prevents a better decision today?
4. What data would materially reduce that uncertainty?
5. What is the least complex reliable way to obtain it?
6. How will action—and business value—be measured?

This may lead to a large connected system. It may also reveal that the useful answer is already hiding inside existing systems.

For example, in one metal-industry project we took up, the initial requirement was to digitize a manual process around rolling and cutting coil sheets. But digitizing the entry sheet alone would merely have reproduced the existing process on a screen.

By using cut-length data already available through the PLC, together with width and density, the team could calculate a more accurate view of the material’s weight.

Note that the value did not come from generating the greatest possible quantity of new data. It came from recognizing which information actually mattered to the customer, and finding a practical way to derive it - with or without new digitization.

That is also the larger idea behind our “See The Unseen” philosophy. The unseen is not always another variable buried inside a machine. It may be the decision the data must support, the shopfloor constraint overlooked by the original brief, or the simpler solution obscured by enthusiasm for technology.

Good digital transformation does not begin by asking how much of an operation can be connected, or how many sensors one can sell. 

It begins by asking: What will we do differently once we know?

So before your next digital transformation project, ask whether the cost of implementation is balanced by the value of the data being derived.