For a long time, I thought progressing in Data was mainly about accumulating technical skills.
Python.
SQL.
Cloud.
Machine Learning.
MLOps.
Then I discovered an important limitation:
**technical skills allow you to build a solution.
They do not necessarily help you build the right solution.**
Translate before executing
Every business function has its own language.
Its constraints.
Its metrics.
Its priorities.
A technically excellent model solving the wrong problem is still the wrong solution.
The first skill is therefore the ability to understand before building.
Define “good”
A Data Scientist may talk about precision, recall or F1-score.
A business team may talk about lead time, customer satisfaction, cost or risk.
Both perspectives are legitimate.
The Data leader’s role is to connect them.
Before measuring model performance, define what “good” actually means for the business.
Orchestrate without controlling everything
Working across different functions does not mean becoming an expert in all of them.
It means knowing:
- who to involve;
- which questions to ask;
- when to bring in an expert;
- how to make trade-offs;
- how to create shared language.
Data becomes powerful when it creates bridges.
Between technology and business.
Between exploration and production.
Between data and decisions.
The real advantage
In a world where AI can generate code in seconds, raw execution becomes less differentiating.
Judgement becomes more important.
Asking the right questions.
Identifying the real problem.
Understanding constraints.
Knowing what not to build.
The rarest skill may no longer be knowing how to produce. It may be knowing what to produce, why and for whom.