Data analyst

A Data analyst collects, prepares and analyses data to inform business teams’ decisions. They turn a question about business activity into an understandable analysis, then present their findings through charts, dashboards or recommendations. Their work connects the available data to a practical decision.

The role starts with understanding the request. The analyst clarifies what teams want to know, the constraints to consider and the information needed. They then extract the data, check its quality and choose a suitable method. Producing a dashboard is therefore one part of the work: calculation rules and data preparation choices also matter.

Depending on the role, they work with marketing, customer relations, production or other operational teams. They may report to a data team manager. Responsibilities shared with other specialists need to be clarified, particularly for automating processing and maintaining the data used.

Data analyst, Data engineer and Data scientist: who does what?

• A Data analyst prepares the data needed to answer a business question, carries out the analysis and explains its conclusions and limitations.
• A Data engineer builds and connects the data pipelines that feed systems and analytical uses. The analyst may pass on requirements for transformation or automation.
• A Data scientist uses methods including machine learning and predictive analysis. The boundary is not fixed: some analyst roles also involve statistical models.

A decision based on data depends on how that data has been prepared and interpreted. Duplicates, missing values or inconsistent formats can undermine the findings. The Data analyst needs to make these problems visible and explain the choices made to address them. A clear chart does not remove the need for these checks.

The second issue concerns the question being asked. A request for a dashboard can cover several expectations: tracking a trend, comparing groups or examining a hypothesis. Without a clear brief, the team risks receiving a technically sound deliverable that does little to inform its decision. Defining the indicators, relevant data and intended audience helps limit this mismatch.

Hypothetical example: a team compares activity across several customer segments to decide where to focus its sales efforts. The analyst checks for duplicates and missing values before calculating differences. They explain the rules used and examine whether the findings allow the segments to be distinguished. If the comparison remains ambiguous, they identify the missing data and suggest further analysis before recommending a priority.

Continuity of work is another consideration. Documented transformations and reusable code make it possible to revisit an analysis and examine the choices behind it. When processing is automated, tests and review contribute to its reliability. The aim is to understand how a result was obtained, including when its author is unavailable.

Hiring can support the quality of business monitoring and decision-making; it guarantees neither a quantified improvement in performance nor a solution to every data problem.

• Clarify the business question, constraints and expected outcomes with the teams involved.
• Identify and extract relevant data from accessible sources, including through SQL queries or APIs.
• Clean data by addressing formats, duplicates and missing values according to explicit rules.
• Explore distributions, relationships between variables and anomalies before choosing the relevant analyses.
• Carry out statistical tests or build models suited to the question and the scope of the role.
• Build charts, dashboards and reports that their users can understand.
• Document and automate processing that needs to be reused.
• Present conclusions, their limitations and information relevant to the decision to business teams.

Useful knowledge combines database querying, data preparation, statistics, visualisation and presentation of findings. These skills make it possible to take a project from the available data through to explaining the results. Technical proficiency covers queries and cleaning rules; it goes hand in hand with the ability to choose an appropriate visual representation and explain its conclusions to a business audience.

The French education pathways cited by Apec include the licence in statistics and information processing or in data mining, masters in statistics, econometrics or business intelligence, and engineering schools specialising in statistics or big data. These prepare people for the role, without a single qualification being required for every position.

Programmes such as the OpenClassrooms Data analyst certification also provide for apprenticeships, continuing education or access through experience, subject to their own conditions.

Use the following steps to assess a candidate against the requirements of your role.

1. Define the criteria and expected level of autonomy

List the business questions to address, accessible sources and expected deliverables. Distinguish essential capabilities from tools the person can learn. Specify who will review analyses and who will resolve competing requests.

Choose observable criteria: problem formulation, data preparation, accuracy of queries, choice of method and clarity of presentation. For an autonomous role, add the ability to organise work and explain their decisions and trade-offs.

2. Examine previous work

Ask the candidate to present an analysis they can share without disclosing confidential information. Have them explain the original question, the data used and their personal contribution.

Invite them to show the cleaning rules, transformations and results presented to users. Ask about a difficulty they encountered and the choices they would make differently. Ask how teams used their conclusions, what decision they made and what the candidate knows about the subsequent action taken.

An explanation that connects technical decisions to the business need is a positive sign. A presentation limited to the final result, without an explanation of calculations or checks, calls for further questions.

3. Set an exercise close to the role

Limit the exercise to one question and a small dataset. State the time allowed and expected deliverable; adjust the workload to examine reasoning without asking for a complete study.

Hypothetical example: provide a simplified dataset covering several customer segments, with duplicates and missing values. Ask for a reasoned comparison, accompanied by the qualifications needed to interpret it.

Observe the questions asked before extraction. Ask the candidate to explain their joins, aggregations and checks on the results. If a statistical method is needed, ask them to justify their choice.

Adapt the exercise to your environment and the level you need. For a role with support, you can provide a clearly defined task. For an autonomous role, also ask the candidate to show how they define the boundaries of the problem.

Look for the ability to flag a limitation and suggest a check. Explore conclusions asserted without checks, ignored duplicates or changes of method the candidate cannot explain.

4. Assess presentation and collaboration

Have the candidate present the findings to a business stakeholder. Ask what the stakeholder can take from them, what remains uncertain and what information would allow further analysis.

Observe whether the candidate adapts their vocabulary and answers questions without losing analytical precision. A chart should help explain the result and its limitations. Put a business objection to their conclusion. Invite them to explain what supports their analysis and what would need checking before revising it.

If the role includes management, explore how they review work, discuss an error and share common rules. For an individual contributor role, focus on interaction with teams.

5. Bring the observations together before deciding

Have the technical work reviewed by someone able to examine the queries and statistical reasoning. If that expertise is unavailable internally, arrange specialist support for the assessment.

With the candidate’s agreement, use references to shed light on their past responsibilities, the support they received and their collaboration with others. Ask questions related to the points that need further exploration.

Finally, compare what you have observed with the criteria defined at the outset. Record the capabilities demonstrated, the learning required and the guidance you will be able to provide. Avoid inferring autonomy solely from a previous job title.