What a Data Analyst does

Data analysts answer business questions with data: how much did we sell, why did it drop, which customers are leaving, did the campaign work. Every team hires them, from quick commerce startups in Bengaluru to banks, consultancies and global product companies.

This roadmap is built from 50 job descriptions posted in the last 90 days. It starts with SQL and one BI tool, adds statistics and Python, and ends with the judgement that separates senior analysts: framing problems, testing data and turning findings into decisions.

At about 10 to 12 hours a week, most people reach Job-ready in 6 to 9 months.

We read 50 Data Analyst job descriptions posted between July and September 2026: 32 in India and 18 global remote roles.

• SQL is the entry ticket. 88% of postings ask for it, and first-round tests are usually timed SQL problems.

• One BI tool, well. Power BI appears in 80% of postings and Tableau in 64%. Employers care less about which tool than about a clean data model and dashboards people use.

• Communication and stakeholders matter as much as tools. 76% mention communication and 78% stakeholder management: gathering requirements, presenting findings, working across teams.

• Statistics and Python separate candidates. 58% ask for statistics and 60% for Python, more often for mid-level roles and product companies.

• Excel still counts, especially early. 44% of all postings and 57% of junior ones name it.

• Data quality is part of the job. 44% mention cleaning, validation or data integrity.

Differentiators that appeared less often but signal strength: data modelling (26%), ETL (22%), experimentation (10%), cloud warehouses such as BigQuery and Snowflake (8% each), and dbt.

Method, full frequency tables and every job link are in the research file linked from this page.

Popular free curricula teach tools one at a time. The job descriptions ask for more than that. This roadmap adds four things they often skip:

• Data quality as a habit, with checks and automated tests, because wrong numbers are the fastest way to lose trust.

• Metric definitions and problem framing, which interviews at product companies test directly.

• Experiment design and honest statistics: confidence intervals, sample sizes and the limits of observational data.

• Writing for decisions: memos, briefs and status updates, practised in every project.

Every example runs on real, licensed data (UCI, NYC TLC, World Bank, Olist, Criteo, Stack Overflow), never toy tables.