During the Data Management Marathon 9.0, many inspiring ideas and experiences in the trenches were shared. In this blog series, I use these presentations as a starting point and connect them with my own experience from Data & AI and governance projects. Where relevant, I complement the speakers’ insights with practical examples and lessons I have learned over the years.
The result is not a representative summary of the event, but rather a personal reflection on the challenges organisations are facing today.
Nowadays, many organisations are investing heavily in AI. New models are being developed, copilot agents are being deployed and data platforms are being modernised. Yet many organisations still struggle to secure sufficient budget, get the needed priority and executive support for these new initiatives.
Why So Many AI Projects Never Make It Beyond the Boardroom
According to Samir Sharma, CEO of datazuum, the problem is not a lack of technical expertise. Data and AI teams build excellent solutions, but they often speak a different language from the people who ultimately decide whether to invest.
During the Data Management Marathon 9.0, he captured that difference particularly well. Data teams enthusiastically explains that:
The board, however, asks very different questions:
The challenge, therefore, is not so much clarity as relevance.
AI Becomes a Business Investment
Samir Sharma’s perspective closely aligns with the BARC Data, BI & Analytics Trend Monitor 2026.
While AI was often regarded as a technology experiment only a few years ago, organisations increasingly expect it to make a demonstrable contribution to growth, efficiency and risk management. As a result, executives are becoming less interested in the technology itself and much more interested in its business impact. In other words: “A board does not invest in AI. A board invests in business value.”
The Problem Is Not the Technology
The difference becomes clear when exactly the same investment is presented in two different ways.
The first presentation might sound like this: “We want to invest in a new data platform, a semantic layer and AI models to modernise our reporting.” The same investment, however, can be presented very differently.
During a recent strategic exercise for an international retail organisation, that distinction became particularly clear. The challenge was not to implement AI. The challenge was to reduce food waste and improve product availability across hundreds of stores.
Starting from that business challenge, a very different business case emerged:
“Because of inaccurate demand forecasts, our stores produce too much or too little every day. This results in food waste, lost sales and higher operational costs. With AI, we can forecast demand for each store based on historical sales, weather forecasts and local events. We expect to reduce food waste by 15%, increase product availability and achieve a return on the investment within two years.”
Suddenly, the conversation is no longer about AI or a data platform. It is about less waste, higher revenue and a demonstrable return on investment. And that is exactly the language the boardroom speaks. Don’t Start with Technology! Samir Sharma translates this principle into a simple strategic model. A successful AI strategy does not start with technology, but with three consecutive steps.
Strategic Framing
Everything starts with the strategic direction of the organisation. One important lesson from many AI and data projects is that an AI strategy should never exist in isolation. It must directly derive from the organisation’s mission and vision.
Where does the organisation want to be in three to five years? Which strategic ambitions does it want to achieve?
Only when that future vision is clear can the right questions be asked:
- Which decisions do we want to improve?
- What value do we want to create?
- How can AI contribute concretely to our long-term strategy?
Strategic Enablement
Only then should we ask which data, people, processes and technologies are required to achieve those strategic objectives. Technology is an enabler, not an objective in itself.
Strategic Readiness
Finally, the organisation needs to prepare for sustainable implementation. This is where governance, the operating model, AI ethics, compliance, risk management and change management become critical. The model makes one thing particularly clear: technology only enters the discussion in the second phase.
First, the organisation needs to understand which strategic choices it wants to support and what business value it intends to create. The AI strategy should therefore not be determined by the capabilities of a technology platform, but by the organisation’s long-term vision. That may be one of the most important lessons from Samir Sharma’s presentation — and from many successful AI initiatives: “Business strategy determines AI strategy, not the other way around!”
What Does a Board Actually Fund?
Ultimately, a board does not invest in technology. It invests in outcomes. A board will rarely allocate budget simply because an organisation wants to build a new data lake, semantic layer or modern data platform. Not because these investments have no value — quite the opposite. They often provide the essential foundation. But that foundation only becomes a compelling investment when it is clear which business outcomes it enables.
Technology is not the objective. It is the means to achieve strategic objectives faster, more efficiently and with less risk.
From Insight to Approach
Samir Sharma’s presentation reinforces an approach that I increasingly consider essential when developing a Data & AI strategy: do not start with technology, but with the business objectives and the decisions the organisation wants to improve.
In my own approach, this means answering a number of fundamental questions before discussing technology, architecture or platform choices. The first question is surprisingly simple: “What business problem are we actually solving?”
If the answer starts with “we are building a new data platform” or “we are implementing AI”, there is a good chance the business case is not yet sharp enough. The starting point should be a concrete organisational challenge.
The next step is to translate that challenge into financial impact. What impact do we expect? Will the investment generate more revenue, reduce costs, decrease waste, improve process efficiency or increase profitability?
Next comes strategic relevance. Does the initiative directly contribute to the organisation’s ambitions for the next three to five years? Does it support growth, a better customer experience, greater operational efficiency or a stronger competitive advantage?
Success also needs to be measurable from the outset. A business case without clear KPIs ultimately remains an ambition. Indicators such as reduced food waste, improved forecast accuracy, faster reporting or higher margins make the expected value tangible and measurable. Only then, in my view, should we ask which data, people, processes, architecture and technology are required to achieve those objectives. And ultimately, the approach should lead to one very concrete question: “What decision are we asking the board to make?” Not the approval of a technology, but the approval of an investment for which the expected business value, risks, timing and payback period are clearly understood and substantiated.
But Has the Belgian Market Made That Shift Yet?
When I compare this perspective with what I currently see in the Belgian market, an interesting contradiction emerges. Data & AI teams rightly devote significant attention to architecture, governance and technology. These foundations remain essential. But in many of the Data & AI profiles I see organisations recruiting for in Belgium today, technology expertise is still positioned as one of the most important requirements. Job descriptions often place strong emphasis on specific platforms, tools and technologies, while experience in translating Data & AI into business value, strategic alignment and organisational change tends to receive comparatively less attention.
This suggests that the shift in thinking has not yet fully taken place everywhere.
We increasingly talk about AI as a business transformation, yet we still often select the people expected to lead that transformation primarily based on their technology expertise. That does not mean technology knowledge is becoming less important. Quite the opposite. Without a sufficient understanding of technology, it is extremely difficult to assess its possibilities, limitations and risks. But we need a better balance.
As Data & AI become increasingly strategic, the ability to connect technology with business priorities, organisational change and measurable value becomes equally important. Perhaps we therefore need to reconsider not only how we approach AI initiatives, but also the profiles we look for to lead them.
From Technology to Business Value
This is where Samir Sharma’s message becomes particularly relevant to me. If we truly want to move from Data & AI initiatives to demonstrable business value, that change cannot take place only in the boardroom. It also needs to become visible in the way we define strategy, assess investments, build teams and select leaders. The real question is no longer which AI solution, data platform or architecture we want to implement, but: “Which strategic business objective do we want to achieve — and what combination of people, data, processes and technology do we need to achieve it?”
And that ultimately leads to perhaps the most relevant question for the Belgian market: If we expect Data & AI leaders to create business value, should we continue to select them primarily based on the technologies they know — or on their ability to turn those technologies into tangible business outcomes?
