Data literacy

What is data literacy?

Data literacy is the ability to read, question, interpret, and communicate with data. A data-literate person does not just look at a chart and accept it. They ask where the number came from, what was counted, what was left out, and whether the conclusion follows from the evidence.

You do not need to be a data analyst to be data literate. The point is not to build every report yourself. The point is to understand enough to use a report responsibly, spot misleading patterns, and ask useful questions before making a decision.

Microsoft's Fabric and Power BI adoption guidance describes data literacy as the ability to correctly interpret, create, and communicate data. For most teams, interpretation is the first bottleneck: opening a dashboard is easy, drawing the right conclusion is the skill.

What does it look like at work?

Data literacy shows up in small habits.

• Ask how a number is defined
  Does revenue include VAT? Are credit notes subtracted? Is the date based on order date, invoice date, or payment date? Many business arguments are definition arguments in disguise.

• Read charts critically
  A truncated axis can make a small change look dramatic. A percentage on a tiny base can sound bigger than it is. Complaints rising by 50 percent means something different when the count moved from two to three.

• Separate correlation and cause
  Customers who read the newsletter may buy more, but that does not prove the newsletter caused the purchase. Loyal customers may simply be more likely to read it.

• Question averages
  One large order can pull the average away from what most customers actually do. The median often tells a better story when values are skewed.

• Check freshness and source
  A report based on last month's extract can be perfectly calculated and still be the wrong input for today's decision.

Average or median?

Imagine ten webshop orders. Nine customers spend 100 euros. One business customer spends 3,000 euros. The average order value is 390 euros, which looks healthy in a management slide. The median is 100 euros, which better represents the typical order.

If you set stock, marketing, or free-shipping thresholds around the average customer in this example, you are designing for a customer who hardly exists.

This is not an exotic statistics lesson. It appears in salaries, delivery times, support tickets, customer lifetime value, and process lead times. A data-literate colleague asks what the distribution looks like before trusting the average.

The hidden bottleneck in BI projects

A dashboard has value only when people can read it and trust it. You can invest in a data warehouse, semantic models, and beautiful Power BI reports; if the team does not understand the numbers, decisions drift back to gut feeling and private spreadsheets.

Data quality solves only part of the problem. A correct number can still be misread. A clean trend can still be mistaken for causation. A precise KPI can still be meaningless if nobody knows what it includes.

Self-service analytics depends on data literacy even more. If people build their own reports without a shared understanding of definitions, grain, filters, and quality, self-service becomes several versions of the truth with nicer visuals.

Data literacy and AI literacy

Data literacy and AI literacy are close relatives. Data literacy is about interpreting numbers, charts, and datasets. AI literacy is about understanding what AI systems can and cannot do, where outputs can fail, and when human review is needed.

The shared habit is healthy scepticism. Someone who asks where a number came from is more likely to ask where an AI answer came from too.

There is one difference. Data usually arrives as raw material that invites interpretation. AI output often arrives as a polished answer in confident language. That makes review feel less necessary precisely when it still matters.

In the EU, AI literacy is now a legal concern for organisations that deploy or use AI systems under the AI Act. Data literacy has no matching general legal obligation, but it is often the practical foundation for using AI and analytics well.

How can an SME build data literacy?

You do not need a six-month training programme. A few steady habits are usually more useful.

1. Write down definitions. Revenue, margin, active customer, churn, open order, and overdue invoice should have shared meanings.

2. Add context to reports. Show the source, refresh time, KPI definition, and known caveats. Three clear sentences can prevent ten confused conversations.

3. Make questions welcome. A colleague who asks whether a percentage is meaningful is protecting the decision, not slowing the meeting down.

4. Use real reports as training material. Discuss one number from the monthly dashboard and unpack where it came from.

5. Start with people who already decide from numbers. Finance, sales, planning, operations, and management habits spread through the company.

The Flemish government's data strategy, for example, treats data literacy as one of its core pillars. An SME does not need the same scale, but the lesson is the same: tools alone do not create a data-driven organisation.

What to watch out for

One-off training fades
People remember data concepts when they use them. Tie learning to the reports and decisions they already handle.

Different roles need different depth
A finance lead needs margin definitions. A salesperson needs pipeline definitions. A manager needs trend, variance, and KPI interpretation. One generic session rarely fits everyone.

Do not weaponise dashboards
If numbers are used mainly to blame people, people learn to defend numbers instead of understand them.

Tool skill is not the same thing
Someone can be excellent in Excel or Power BI and still draw the wrong conclusion. Someone with no formula skills can still ask the right question.