Data analyst roadmap: what to study, in what order and how to gain experience

What a data analyst does day to day

Before choosing what to study, it helps to know what for. A data analyst takes business questions, such as why sales dropped in a region or which customers are most likely to leave, and turns them into queries, calculations and charts someone can use to decide. A large share of the time goes into getting and cleaning the data, not the analysis itself. Another important part is explaining results to people who are not technical.

In the data projects we run with clients, the analyst is the link between the business side and the people building the systems. They know what can be measured, spot when a figure was entered wrongly and turn a vague need into a concrete indicator. That is why this roadmap mixes technical tools with practice on real questions: one without the other falls short.

What to study and in what order

The order we suggest starts with what companies use most and moves towards the more specialised. Excel and SQL first, because they are in almost every team that works with data. Then visualisation, because analysis that is not communicated is useless. Statistics comes along from the start in small doses, and Python arrives once you know what you want to automate. The times in the table assume eight to ten hours a week and are rough estimates.

If you already work with Excel every day, you can shorten the first stage; if you come from a role with no contact with data, give it more time. The important thing is not to study every tool at once. People who try to learn SQL, Power BI and Python in parallel usually end up with three half-learned skills and no finished project.

• Excel: conditional functions, XLOOKUP, pivot tables and Power Query.
• SQL: queries, aggregations, JOIN, subqueries and window functions.
• Visualisation: Power BI or a similar tool, with a data model and measures.
• Descriptive statistics: means, medians, spread, distributions and correlation.
• Python: pandas to clean and transform data, plus a charting library.
• Communication: presenting findings with context and a concrete recommendation.

Statistics: how much you really need

To start as an analyst you do not need a university statistics course. You need a solid grasp of descriptive statistics: when to use the median instead of the mean, what the standard deviation tells you, how to read a distribution and why a correlation does not prove that one thing causes the other. That alone avoids most of the wrong conclusions we see in business reports.

Later on, some inferential statistics will help, such as confidence intervals and hypothesis tests, especially if you will analyse experiments or surveys. But that part pays off more once you can pull and prepare data. Studying it at the start, with no data of your own to apply it to, tends to leave you with formulas you quickly forget.

Python: when it comes in and what for

Python becomes useful for an analyst when tasks show up that Excel, SQL and Power BI handle poorly: cleaning files with irregular structures, combining data from an API, automating a report that is currently built by hand or running finer statistical analysis. The pandas library is the entry point and covers most of those cases. Jupyter or a similar environment lets you work interactively and document the analysis as you go.

What we do not recommend is starting the roadmap with Python. Without SQL or a sense of how data is organised in a company, Python turns into a general programming course that takes a long time to pay off. On the other hand, people who arrive knowing SQL pick up pandas much faster, because the operations are the same with different syntax: filtering, grouping and joining tables.

Projects to build a portfolio

An analyst portfolio does not need to be long. Three or four well-explained projects say more than ten notebooks with no context. Each one should start from a question, show where the data came from and how it was cleaned, and end with a conclusion someone could act on. Publish them on GitHub or on a profile where they can be reviewed without downloading anything.

• A sales analysis in SQL on a sample database, with findings by product, region and season.
• A Power BI dashboard with indicators for a process, explaining which decision it supports.
• Cleaning and analysis in Python of a public dataset from your country, with a written summary of the findings.
• A process currently done by hand in Excel, rebuilt with Power Query or Python, showing the before and after.

Common mistakes along the way

Most people who drop this path do not do so for lack of ability, but because of how they go about it. Collecting certificates without projects, or spending months watching videos without writing a query of your own, feels like progress without being progress. These are the stumbles we see most in people who are starting out:

• Studying several tools at once and finishing none.
• Skipping Excel and SQL to go straight to machine learning.
• Building projects on datasets that are too clean to teach you how to prepare data.
• Showing charts without saying what they mean or which decision they support.
• Waiting to feel ready before applying or offering to help with data in your current job.

How to choose courses for this path

Instead of a single course that promises to cover everything, one focused course per stage usually works better: one for Excel, one for SQL, one for Power BI and one for Python for analysis. All-in-one courses tend to stay on the surface of each tool. For each one, check that the syllabus reaches intermediate topics, includes exercises and projects, and has been updated recently.

Language matters differently depending on the tool. For Excel, pick a course in the same language as your installation, because function names change. For SQL and Python, since the code is in English anyway, a Spanish-language course works just as well to get started. And be wary of courses that promise to make you an analyst in a few weeks: the tool is learned quickly, judgement is not.

When the analyst role is not enough

Over time, some people find that what they enjoy most is not answering questions but building the data infrastructure. That path leads to data engineering or analytics engineering, where modelling, pipelines and the cloud carry more weight. Others prefer the predictive side and move towards data science, which demands more statistics and programming. The foundation you build on this roadmap serves both directions.

There is also a limit on the business side. When a company depends on an analyst rebuilding the same report every month from loose spreadsheets, the problem is no longer analysis but systems. That is usually where we come in at TintoDev, automating data capture and integration so the analyst can spend their time interpreting, not copying and pasting.

Where the courses are

Frequently asked questions

Do I need a university degree to become a data analyst?

Not always. Many people reach the role from finance, operations, engineering or administration, backed by a portfolio and experience with data in their own job. Some companies do require one, so check job ads in your city or in the kind of company you are interested in.

How long does it take to become a data analyst?

At eight to ten hours a week, roughly six to eight months to work through Excel, SQL, Power BI, basic statistics and Python and have a portfolio. If you already work with data or put in more hours, it can be less. Landing the first job is a separate process that depends on your market.

Which should I learn first, SQL or Power BI?

SQL, if you have to choose. It is how you get data out of systems and the basis of any serious report; Power BI is quicker to learn once you understand how tables relate. If your current job needs dashboards right now, you can start with Power BI and study SQL alongside it.

Python or R for data analysis?

Python, for most people starting out. It works both for analysis and for automating processes and connecting systems, and it has a very large community. R is still very good for statistics and in academic settings, so if your field uses it, it makes sense to learn it.

How do I get experience if I have never worked as an analyst?

Start in your current job: almost every department has a manual report that can be improved with what you are learning. Document that project, with the data anonymised, and add it to your portfolio. Projects with public data from your country also work, as long as they start from a concrete question.