Decisions about people, with data and with care
Misread surveys, turnover figures without context and AI recruitment tools used without understanding them. What each HR profile should know about data, the most common mistakes and a 30-day training plan.
The 5 most common mistakes
In HR an interpretation mistake affects specific people. That is why rigour and privacy matter more here than elsewhere.
Drawing conclusions from small surveys
Six people in a team answer and the result is generalised, or percentages from groups of very different sizes are compared.
How to avoid itAlways check how many people answered and do not publish results for very small groups.
Turnover without context
Turnover rises, but voluntary leavers, contract ends and a reorganisation are lumped together.
How to avoid itSplit by leaving reason, area and length of service before concluding.
Mistaking correlation for cause
Teams with more training perform better, and the conclusion is that training explains everything.
How to avoid itLook for other explanations and test them before deciding.
Using AI in recruitment without understanding it
Tools that filter CVs or score candidates without knowing what data they learned from. These uses are high-risk under the AI Act.
How to avoid itKnow what data it uses, how its decisions are reviewed and who is accountable.
Sharing more personal data than needed
Reports with names when aggregate data would do, or files emailed without protection.
How to avoid itUse aggregate or pseudonymised data wherever possible.
Sound familiar?
Measure your team’s level with the free self-assessment: 20 statements scored by dimension.
Download the self-assessment →What each profile should know
The levels come from the data literacy framework: basic, use, analysis and management. Not everyone needs to reach the same one.
| Profile | What they should be able to do | Level |
|---|---|---|
| HR administration | Record data properly, protect personal information and spot errors | Basic / Use |
| Recruitment | Use recruitment tools with judgement and understand their limits | Use |
| People analytics / analyst | Analyse surveys, turnover and headcount rigorously and with respect for privacy | Analysis |
| HR director | Decide about people with data and explain the indicators to the board | Management |
Your data training, in three steps
Short training using the team’s real data. Each step links to the guide and the template.
Measure the level
Give each profile the free self-assessment and pick the two or three gaps that most affect the area’s work.
Train with real cases
Short sessions: reading a survey, analysing turnover by leaving reason and using aggregate instead of personal data.
Make it stick
Rules for publishing results (a minimum group size), a common definition of each indicator and a review of the AI tools in use.
Frequently asked questions
Why is data use so sensitive in HR?
Because it is personal data and decisions affect people’s careers and pay. An interpretation mistake, or exposed data, has bigger consequences than in other areas.
What is the minimum group size for publishing survey results?
There is no single figure. Many companies set a minimum number of responses per group so nobody can be identified. Decide yours and always apply it.
Does HR coordinate data training for everyone else?
Often, together with leadership and the data team: it identifies needs, organises sessions and measures progress.
What does the AI Act say about AI in recruitment?
AI systems used to recruit or make decisions about workers are among the high-risk ones, with additional obligations. Whoever uses them must understand their limits and oversee their outputs.
What level of data literacy does your team have?
Free five-minute assessment: your score in the four dimensions and what to work on first.