The One Skill That Will Separate Winners From Losers in the Age of AI

For most of my career, the scarce resource in any organisation was data. The people who had access to data — who could extract it, process it, and present it — had power. The analytical bottleneck was supply. This is changing rapidly, and the implications for professional development are significant and underappreciated.

We are entering an era where data is not scarce and analysis is not scarce. AI can generate more analysis in an afternoon than an analytics team could produce in a month. What becomes scarce — what becomes the genuine source of competitive advantage — is the ability to know what question to ask.

Why the Question Matters More Than the Answer

A well-formed question contains within it most of the structure of the answer. A question like "how can we use AI to improve our operations" is so loosely specified that almost any answer would technically be correct. A question like "which of the 4,200 accounts flagged by our anomaly detection model in the northern region represent genuine commercial losses as opposed to legitimate consumption changes, and what is the estimated revenue recovery from investigating the top 500?" is precise enough to be answered definitively, actionable enough to drive a specific intervention, and valuable enough to justify the analytical investment.

The difference between these questions is not domain knowledge or analytical skill — it is the discipline of specificity. And specificity requires something that AI cannot easily replicate: genuine understanding of the operational context, the decision constraints, and the value of different types of information to the person who will act on the answer.

The executives I have worked with who are most effective with data are not the ones who understand statistics or who can evaluate model performance. They are the ones who can look at a business problem and say: here is the specific question whose answer would change what I do next. That skill is worth more than any technical training.

What Good Question-Asking Looks Like

Good questions are specific — they can be answered definitively rather than approximately. They are decision-relevant — the answer changes what someone does, not just what they know. They are feasible — the data and analytical capability required to answer them actually exists. And they are honest — they are asked because the asker genuinely does not know the answer and wants to find out, not because they want to generate evidence for a position already taken.

These criteria are more demanding than they appear. Most questions asked in organisational settings fail one or more of them. They are vague because specificity creates accountability. They are not decision-relevant because they are asked for reporting purposes rather than action. They are infeasible because nobody checked whether the required data exists. Or they are dishonest because the answer is predetermined and the analysis is post-hoc rationalisation.

How to Develop This Skill

The most effective development practice I have found is a simple discipline: before commissioning any analysis, write down the specific decision the analysis is meant to inform, and the specific action you will take if the analysis produces each possible answer. If you cannot specify the decision and the contingent actions, the question is not yet well-formed enough to answer usefully. This discipline, applied consistently, transforms the quality and value of analytical work — not because the analysis improves, but because the questions it is answering become genuinely worth answering.

Dr. Sunny Okonkwo

AI Strategist · Head of Data Analytics at one of Africa's largest energy and utility companies. Author of 7 books including the #1 Bestseller The AI Alchemist. Keynote speaker at IIBA, Big Data Summit Canada, Global Summit, and UNICAF.