guide
Data Analyst Roadmap for India: a six month study plan
A month-by-month plan that assumes you have a job, limited time and no technical degree — covering what to learn, in what order, what to skip, and how to tell you are ready to apply.
By the Samyak faculty team · Published · 11 min read
Most roadmaps you will find are lists of technologies. They tell you to learn Excel, SQL, Power BI, Python, statistics and machine learning, in a graphic, with arrows. They do not tell you how long each takes, what “learned” means, or what to do when you have finished.
This one is a schedule, and it makes an argument about ordering.
Before you start: the two decisions that matter
Decide how many hours a week you can genuinely give. Not aspirationally. Eight to ten hours a week is enough to be job-ready in six months. Four hours a week is enough, but it will take eleven or twelve months, and knowing that up front prevents the quiet demoralisation that ends most self-study.
Decide that you will build things rather than complete things. Finishing a course gives you a certificate. Finishing a project you chose, with data you had to clean, gives you something to talk about for twenty minutes in an interview. Only one of those gets you hired.
Month 1 — Excel, properly
Start here even if it feels unambitious. Excel is how most Indian businesses still hold data, and being visibly good at it is often what gets you the internal reporting work that becomes your first analytics experience.
Cover lookup and reference functions, conditional aggregation, PivotTables, and — the part most people skip — Power Query. Power Query matters because it teaches you the shape of a data pipeline: source, transform, load, refresh. That mental model transfers directly to every tool that follows.
Done when: you can take a messy exported CSV and produce a refreshable summary report without manual copy-paste.
Month 2 to 3 — SQL, in depth
This is the highest-return period in the entire roadmap. SQL is the single most screened-for skill in analyst hiring in India, and unlike most skills it is tested directly and live.
Work through, in order: SELECT and filtering; GROUP BY and aggregation; the join types and what each does to row counts; subqueries and CTEs; then window functions. Give window functions real time — running totals, rankings and period-over-period comparisons are standard interview material and standard daily work.
Practise against a database with enough rows that a bad query is noticeably slow. Toy datasets hide the entire subject of performance.
Done when: you can answer a question like “which customers increased their spend for three consecutive months” without looking anything up.
Month 4 — Power BI and data modelling
Learn Power BI as a modelling tool, not a chart menu. The chart part takes a week; the part that makes you employable is the star schema, relationship cardinality, and DAX measures.
The specific trap here is building dashboards on a single flat table because it works for the demo dataset. It stops working the moment you have two fact tables, and interviewers ask about exactly this.
Done when: you can explain why your model has the tables it does, and write a DAX measure that compares this quarter with the same quarter last year.
Month 5 — Python and statistics, in that order of effort
Python for analysis is a smaller subject than it looks. You need pandas — loading, cleaning, merging, grouping, pivoting — and enough matplotlib to produce a chart. You do not need decorators, metaclasses, or a web framework.
Alongside it, cover applied statistics: distributions, variance, sampling, and what a confidence interval actually claims. This is the module people skip and later regret, because it is what stops you presenting noise as a finding.
Done when: you can take a raw export and produce a cleaned dataset plus four charts in a notebook that runs top to bottom without manual intervention.
Month 6 — Portfolio and applications
Three projects, not ten. Each one should use data you had to clean, answer a question you chose, and end in a written conclusion with stated caveats.
Then write them up. A repository with a README explaining the question, the method, the finding and the limitation is worth more than the code itself, because it demonstrates the thing employers cannot easily test for: whether you think clearly about evidence.
Start applying at the beginning of month six, not the end. Interview processes are slow, and early rejections tell you what to fix while you still have study time left.
What to skip, and why
Machine learning. It feels like the ambitious choice and it is the wrong one for this goal. Analyst job descriptions rarely require it, interviews rarely test it, and the hours are better spent making your SQL genuinely strong.
A second BI tool. Power BI or Tableau, not both. The concepts transfer; the second tool’s menus are a weekend of work if a job ever needs them.
Certifications, mostly. They are a weak positive signal at best. A defensible project is stronger evidence and takes similar effort.
How to tell you are actually ready
You are ready to apply when you can do three things without preparation: write a window function query from a business question, explain the data model behind one of your dashboards, and describe one of your projects — including what it does not prove — in about two minutes.
If those three are true, apply. Waiting until you feel confident is a trap, because that feeling generally arrives several months after employability does.