Most people who want to become data analysts do not fail because the subject is hard. They fail because they learn in the wrong order: three months of Python syntax, a half-finished course on deep learning, and still no idea how to answer "which city had the most late deliveries last month?". This roadmap fixes the order.
The idea is simple: learn the skill that answers business questions fastest, then add tools that make you faster. An analyst's job is to turn a messy question into a clear number, chart or recommendation. Everything below serves that.
Step 1: SQL first, and properly
Almost every analyst job description mentions SQL, because company data lives in databases and SQL is how you ask it questions. It is also the easiest of the core skills to start with: you can write a useful query on day one.
Go through SQL in this order:
SELECT,WHERE,ORDER BYandLIMIT: picking and filtering rows.- Aggregates and
GROUP BY, thenHAVING: totals and averages per group. - Joins: combining customers with orders, orders with products.
CASE WHEN, NULL handling, string and date functions: cleaning on the fly.- Subqueries and CTEs: breaking a hard question into steps.
- Window functions: rankings, running totals, month-on-month change.
Do not just watch videos. Write queries every day. Our SQL practice track is built in exactly this order, and each question is checked the moment you run it.
A good test of "I know SQL": can you write a query for the top 3 products by revenue in each category, without looking anything up? If not, you are not done with window functions yet.
Step 2: Excel or Google Sheets, just enough
You will still be sent spreadsheets at work. Learn sorting, filtering, pivot tables, lookups (XLOOKUP or VLOOKUP), basic charts and simple formulas. A week of focused practice is usually enough for a fresher, because the concepts (filter, group, join) are the same ones you learned in SQL. Our article on moving from Excel to SQL and pandas shows how the two map to each other.
Step 3: Python with pandas
Python becomes important when the data does not fit nicely in one table, needs heavy cleaning, or the analysis repeats every week. For an analyst, Python mostly means pandas for tables and Matplotlib/Seaborn for charts.
Learn just enough core Python first: strings, lists, dictionaries, loops, functions and comprehensions. Then move to pandas: reading a CSV, selecting and filtering, groupby, merge, cleaning missing values and working with dates. Our guide to the pandas operations you will use every day lists the ones that matter most.
Step 4: Statistics for decisions
You do not need a statistics degree. You do need to know when an average is misleading, what a percentile tells you, why correlation is not causation, and how to say "this difference is probably real" with some confidence. Read our statistics for data analysts guide and then practise with NumPy.
Step 5: One dashboard tool
Pick Power BI or Tableau, not both. Power BI is common in Indian companies that already use Microsoft tools. Learn Power Query for cleaning, a simple star-schema model, a handful of DAX measures and how to lay out a one-page dashboard. Our Power BI for beginners article covers the parts that matter.
Step 6: Projects that look like real work
This is where most freshers have the biggest gap. Course certificates show you sat through lessons. A project shows you can take a vague business problem and finish it. Good projects have:
- A business context (a company, a stakeholder, a decision to make).
- Messy data you had to clean, not a perfect Kaggle file.
- Clear questions and clear answers, with numbers.
- A short write-up or dashboard a manager could read in five minutes.
Our live projects are built this way, across banking, insurance, healthcare, retail, logistics and more. Read how to work through a business case project before you start one.
Step 7: Resume, interviews, applications
Once you have two or three solid projects, write them up on a one-page resume (see how to put projects on a data analyst resume), prepare for the usual interview rounds and start applying. Apply steadily rather than in one burst, and keep practising while you apply, because SQL tests are often the first filter.
How long does this take?
It depends on how many hours you put in each day and where you are starting from, so be wary of anyone who promises a fixed number of weeks. What matters more than speed is that each step ends with proof: solved questions, a finished project, a dashboard you can show. If you keep producing proof, you are moving in the right direction.
What to skip for now
- Deep learning, neural networks and computer vision. Useful later, rarely asked of a fresher analyst.
- Learning three dashboard tools at once.
- Collecting certificates without projects.
- Memorising syntax. You will look things up at work too; what you need is to know what to look up.
Practise what you just read
Reading helps; writing the code yourself is what sticks. Start here: