A data analyst needs SQL, a spreadsheet, a visualisation tool, some Python, statistics, and the ability to explain a result to a manager. This guide shows the level expected for a first job, how recruiters check it, and how AI is changing the work.
In short: data analyst skills fall into two families. Hard skills are SQL, a spreadsheet, Power BI or Tableau, Python, statistics and data modelling. Soft skills are business understanding, rigour, critical thinking and the ability to explain a result. Recruiters check them through tests, projects, a portfolio or a certification.
Data analyst skills at a glance
Two public French references describe the job: the ROME M1419 sheet from France Travail and the Onisep career sheet. Both cite databases, statistics, programming, visualisation and communication. Translated into what a first employer expects:
| Skill | Expected for a first job | How to show it |
|---|---|---|
| SQL | Write queries alone, with joins, aggregations and window functions | Commented queries on GitHub, technical test |
| Spreadsheet (Excel, Google Sheets) | Pivot tables, lookup functions, cleaning with Power Query | A documented analysis file |
| Power BI or Tableau | Clean data model, correct measures, readable dashboard | Published dashboard, vendor certification |
| Python | Prepare and analyse data with pandas, draw charts, automate a task | A reproducible notebook |
| Statistics | Descriptive statistics, confidence intervals, tests, simple regression | An A/B test explained step by step |
| Data modelling | Star schema, granularity, keys and relationships | Model diagram in your portfolio |
| Business understanding | Turn a team’s question into indicators | A one-page project brief |
Hard skills: the technical side
SQL comes first. Most company data sits in relational databases, and many recruiters test SQL live. In the 2025 Stack Overflow Developer Survey, 58.6% of respondents said they use SQL and 57.9% Python. You should be able to:
- filter and sort, and handle NULL values correctly;
- join several tables and spot the duplicates a bad join can create;
- aggregate with GROUP BY and filter aggregates with HAVING;
- structure long queries with subqueries or common table expressions;
- use window functions for rankings, running totals and period comparisons;
- check your result by counting rows at each step.
Spreadsheets stay. They are what business teams use: pivot tables, lookup functions such as XLOOKUP or INDEX and MATCH, and Power Query to automate repeated cleaning.
Power BI and Tableau: model before you draw. The chart is not the hard part. The model and measures make an indicator correct. Microsoft’s PL-300 exam, in its version of 20 April 2026, gives a useful map of the skills:
| PL-300 exam domain | Weight | Examples |
|---|---|---|
| Prepare the data | 25 to 30% | Connect sources, profile and clean, create fact and dimension tables |
| Model the data | 25 to 30% | Relationships, date tables, DAX measures, performance |
| Visualise and analyse | 25 to 30% | Choose visuals, report narrative, accessibility, anomaly detection, Copilot |
| Manage and secure Power BI | 15 to 20% | Workspaces, scheduled refresh, row-level security, sensitivity labels |
Python automates what spreadsheets cannot. The useful base is pandas, NumPy, matplotlib or seaborn, Jupyter notebooks and simple scripts to automate an export or call an API. R remains a solid alternative in statistics-heavy teams.
Statistics separate analysts from chart makers. Know descriptive statistics, outliers, sampling and confidence intervals, hypothesis tests and A/B tests, correlation versus causation, and linear regression.
Data modelling keeps dashboards reliable. Microsoft recommends the star schema: dimension tables describe entities and filter data, while fact tables store events and are summed. Always know the granularity of a table: what does one row represent?
Data quality and privacy matter. Analysts often handle personal data. The GDPR’s minimisation principle limits processing to what is necessary. In practice, prefer aggregated or pseudonymised data, record your sources, and flag anomalies rather than silently fixing them.
Soft skills: what makes the work useful
In its Future of Jobs Report 2025, based on a survey of more than 1,000 employers, the World Economic Forum says analytical thinking is the most sought-after core skill: seven companies in ten consider it essential in 2025.
| Soft skill | What it means day to day | How to train it |
|---|---|---|
| Business understanding | Know what margin, churn or delivery time mean in this company | Read the team’s reports; ask for written definitions |
| Problem framing | Turn “sales are falling” into checkable questions | Write a one-page brief before each analysis |
| Critical thinking | Doubt a result that looks too good | Systematic checks, peer review |
| Communication | One message, the evidence, a recommendation | Summarise in three sentences before making charts |
| Rigour | Reproducible queries, tracked versions | Git, README, data dictionary |
| Prioritising | Balance urgent requests against high-impact work | Rank requests by impact and effort |
Hard skills are often the filter at the door. Soft skills create the value afterwards: a perfect query that answers the wrong question is useless.
How to prove your skills
Build three strong projects, not ten weak ones:
- a SQL analysis on a public database, with precise questions and commented queries;
- a Power BI or Tableau dashboard that answers a business question, with the data model shown;
- a Python notebook that cleans a real dataset and runs a statistical analysis.
Give each one a README: the question, the data and its source, the method, the result and the limits. Kaggle, data.gouv.fr and Insee all offer datasets. If you are changing career, a project from your earlier sector shows both technique and business sense.
Certifications prove a tool, not a job. Microsoft’s Power BI Data Analyst Associate (PL-300) is a proctored 100-minute exam with a passing score of 700, renewed free of charge every 12 months. The Google Data Analytics Professional Certificate has 9 beginner courses, about 6 months at 10 hours a week. More options are in our guide to AI and data certifications.
Self-assessment checklist
Tick only what you can do alone, without a tutorial, in under an hour.
| Area | I can… |
|---|---|
| Data and SQL | ☐ join three tables without creating duplicates ☐ calculate a running total or ranking with a window function ☐ explain the granularity of a fact table ☐ find and handle missing values ☐ document a dataset’s source and date |
| Tools | ☐ build a pivot table and automate cleaning in Power Query ☐ create a star schema in Power BI or Tableau ☐ write a filtered measure ☐ load, clean and aggregate a file with pandas ☐ version my work with Git |
| Statistics | ☐ choose between mean and median, and justify it ☐ calculate and interpret a confidence interval ☐ analyse an A/B test and say if it is conclusive ☐ explain why correlation does not prove causation ☐ read linear regression coefficients |
| Business and communication | ☐ turn a vague request into three precise questions ☐ define an indicator and how it is calculated ☐ summarise an analysis in three sentences for a director ☐ present the limits of my results ☐ apply data minimisation |
Reading your score. With 16 to 20 ticks, you can apply and show your projects. With 10 to 15, strengthen your weakest area with a portfolio project. Below 10, start with SQL and statistics. This scale is a study aid, not a recruitment standard.
How AI is changing the job
The World Economic Forum places AI and big data at the top of the fastest-growing skills to 2030, and employers expect 39% of workers’ current skills to be transformed or outdated between 2025 and 2030. According to Microsoft’s documentation, Copilot in Power BI can create report pages, summarise reports and write DAX queries. Microsoft also warns that without preparation of the semantic model, Copilot can misread data and give generic or incorrect results.
AI cuts repetitive code and moves the value towards four skills: framing the question, checking what the tool produced, preparing data with clear names and definitions, and respecting the rules on which data can go to which tool. Never paste personal or confidential data into an assistant your company has not approved.
Building these skills at AI2
AI2 (Applied Institute of Artificial Intelligence) is a French school of applied AI and data science at Écouen, 22 minutes from Paris Gare du Nord. Its degree programmes are taught 100% in English and prepare RNCP titles at Levels 6 and 7, recognised by the French State. The MSc Applied AI and Data Science covers data engineering, machine learning and deployment, and the BSc Data Analytics and Applied AI teaches students to turn production data into dashboards, alerts and AI predictions.
| Programme | Who it is for | Tuition | Start |
|---|---|---|---|
| Bachelor in Applied AI, Years 1 and 2 | After secondary school | €9,000 a year | February, May or September 2027 |
| MSc, Year 1 | After a bachelor’s degree | €12,500, from €10,500 with scholarship | February, May or September 2027 |
| MSc, 18-month direct entry | Direct entry, subject to your academic record | €13,000 single fee, from €11,300 with scholarship | See tuition fees |
| Fast-Track Data Science, 11 months | Engineering or scientific profiles | €12,500, or €11,000 before 15 November 2026 | February 2027 |
| Paris AI Summer School, 10 days | Anyone curious about applied AI | From €1,990 | 2027 sessions from 14 March |
Frequently asked questions
What are the key skills of a data analyst?
On the technical side: SQL, a spreadsheet, a visualisation tool such as Power BI or Tableau, Python, statistics and data modelling. On the human side: business understanding, rigour, critical thinking and clear communication.
What are the hard and soft skills of a data analyst?
Hard skills include SQL with joins, aggregations and window functions, data models and dashboards, data preparation with Python and pandas, and basic statistics. Soft skills include business understanding, framing problems, critical thinking, communicating results, rigour and prioritising.
Does a data analyst need to code?
At minimum, you need SQL, which is a query language. Python is not required in every job, but it becomes a real advantage as data grows or tasks repeat. You do not need to be a software developer, but your queries and scripts must be reliable and reproducible.
Which certification should a data analyst choose?
Microsoft Power BI Data Analyst Associate (exam PL-300, 100 minutes, renewed free of charge every year) is tied to a widely used tool. The Google Data Analytics Professional Certificate is aimed at beginners. For proof of the whole job, a recognised qualification assesses all the skill blocks.
How can I prove my skills with no experience?
Build a portfolio of three projects: a SQL analysis, a dashboard and a Python notebook. Document each with the question, the data, the method, the result and the limits. A project drawn from your earlier job also shows business understanding.
Will AI replace data analyst skills?
It is changing part of them. Tools such as Copilot in Power BI can create report pages and write DAX queries, but Microsoft warns they can give incorrect results if the data model is not prepared. Framing questions, checking results and respecting data rules remain human skills.
Sources checked in October and November 2026: France Travail: ROME M1419 · Onisep: data analyst · World Economic Forum: Future of Jobs 2025 · Microsoft: Power BI Data Analyst Associate · Microsoft: Copilot for Power BI · Microsoft: star schema · Coursera: Google Data Analytics · Stack Overflow Developer Survey 2025
Send your file to AI2: admissions reviews it, holds an online interview and confirms your programme, tuition and scholarship in writing.
