Insights

A Nordic lesson on AI

Why do accurate reports not lead to timely decisions?

Last week our Data Masters team spent two days at the Data 2030 Summit in Stockholm, the 11th edition of the event for the Nordic region. Together with around 280 data management leaders from companies across the Nordic countries, we talked about one question: how can trustworthy data become the foundation for AI that actually changes how a business operates?

Almost every conversation came back to the same problem: business value is lost in the gap between recognizing a problem and taking action. AI is increasingly used to speed up development and analysis, but that is not all we need to do. That is why it is worth passing on what we heard, because Nordic companies often raise these questions earlier, and we already see the same problem here in our region.

The reports are accurate

At our stand, people from banking, insurance, retail, manufacturing and distribution told us the same story in different words. The reports are usually accurate and teams notice problems in time. Still, the decision is late, because responsibilities are not assigned, approvals take weeks, and afterwards nobody can measure whether the action worked. AI gets us to the number fast, but the human remains the bottleneck.

We heard the same in the talks, where, session after session, attention shifted from building data platforms to the decisions made on top of them: who makes them, who approves them, and how their effect is measured.

Decisions without a history

At our roundtable we asked two questions. Where does the context behind the most important decisions live today, meaning who approved them, why, and what happened afterwards? And when a decision gets stuck, what holds it back: the data, the missing context, or the approval itself?

The honest answer was that most companies are still struggling with the foundation, which is the data. It is already clear to everyone that we cannot expect to read data straight from the source systems, because they are not ready for advanced analytics. The data exists, but the context behind it is missing, because decisions stay in email, in meetings and in people’s heads. By the end of the conversation our conclusion was simple: companies can trace every number back to its source, but they can rarely trace a decision back to the person who made it and the change it caused.

This gap matters for every company thinking about AI. A system that is supposed to recommend business decisions must learn from the history of previous decisions, the reasons behind them and their outcomes. In most companies that history does not exist, and it is the one thing a company has to build itself. Otherwise AI will have no material to learn from.

Start from the business goal and the outcome

The most useful message we brought back from Stockholm was a simple one. Several speakers, each from their own angle, urged teams to start a project from the final business outcome and work backwards: first the value the business needs, then the action that can create it, then the context the decision requires, and only at the end the data.

Here, many AI projects start from the opposite side, with the data and a tool, and the business benefit is expected later. The company ends up with accurate reports, but the decisions are still late.

Many companies can already ask their data questions in plain language and get an answer immediately. But if the answer ends up in yet another report, the time to action does not get shorter. The growth and the benefit lie in what was not possible before, and I will return to that in one of the next articles.

At Data Masters we start with one specific decision that regularly runs late, that everyone complains about and nobody clearly owns. For it, we appoint a responsible person and set clear criteria, build the data foundation it needs, and record every approval.

We measure the result with a control group of customers, randomly selected at the moment of approval and then left unchanged, to see what change the action caused. The first results arrive within the first month, and every following decision is recorded the same way. Over time, the company builds the history that, according to the conversations in Stockholm, many teams are missing. Every action is approved by a human before it reaches a colleague or a customer, and that is the key.

Which decision to fix first

The message from Stockholm was clear: Nordic teams first ask which decision needs fixing, and only then which AI to buy. Every company in our region is worth asking the same question: which decision in your company is constantly late, and if you made it on time tomorrow, how would you prove it delivered results?

Value Discovery

Bring the decision that is late.

A free Value Discovery, sixty to ninety minutes, built around your industry and your Monday morning.