Data Analyst Career Guide: SQL, Metrics, and Interviews

Analyst hiring rewards a habit that course certificates rarely test: checking whether a number is right before explaining what it means. A new BI tool can be learnt in a few weeks; the instinct to question a join, a time zone or a definition before a figure reaches a director takes much longer to build. The sections below cover the work, the Indian variations of the title, and how panels test for that instinct.

The questions an analyst handles in a normal week

Analysts sit between the people who make decisions (product, marketing, operations, finance, sales) and the data that should inform them. Requests arrive as messages, tickets or meeting actions, and most are vaguer than they look. A typical week at a consumer company might include:

• "Why did repeat orders drop in Chennai last week?" You first rule out data problems such as a late-loading table or a tracking change in the latest app release, then segment by new versus returning customers, platform and store, and find that a delivery-slot change removed the late-evening slots many repeat customers used.
• Reconciling two dashboards that disagree on "active merchants", because finance counts any transaction and sales requires three a month. You write one definition, get both teams to agree, and document it.
• Preparing the weekly business review: the numbers, what changed, and three sentences on why.
• Cleaning an export where the same city appears as Bangalore, Bengaluru and BLR, and order IDs lost their leading zeros because someone opened the CSV in Excel.

The BI developer and data scientist roles are the easiest to confuse with this one. A BI developer builds and maintains the governed reporting layer (semantic models, report performance, access rules) that analysts and business users read from. You use data to answer questions and recommend action, and you may prototype a dashboard that a BI developer later hardens. A data scientist takes the questions that need experiments, causal methods or prediction. "Business analyst" is the most confusing title: at IT services firms it usually means writing requirements and user stories, while at product and analytics companies it often means the job described here. Read the responsibilities, not the title.

Skills employers check, roughly in the order they check them

Skill	What the employer needs	How it is tested
SQL	Joins that do not duplicate rows, GROUP BY, CASE, CTEs, window functions for rankings and period-over-period change, careful date handling	Online test, then live queries
Excel or Google Sheets	Pivot tables, XLOOKUP or INDEX-MATCH, SUMIFS, text cleaning, and Power Query for steps you repeat every week	A timed task on a messy file
A BI tool	Power BI, Tableau, Looker Studio or Metabase: a clear dashboard with sensible filters and a few calculated measures	Portfolio walk-through or dashboard critique
Metric definition and business sense	Conversion, retention, average order value and unit economics, including the edge cases inside each definition	Case round
Basic statistics	Mean versus median, outliers, correlation versus causation, reading an A/B test result	Case round follow-ups
Python with pandas	Automating repeated work and handling files too large for a spreadsheet	Differentiator; sometimes a take-home

Excel remains core in India because finance, operations and sales teams run on it, and many analysts exchange files with them daily. Two details impress interviewers: Excel keeps only 15 significant digits, so long numeric IDs must be stored as text, and a "daily" figure computed in UTC splits the Indian day at 5:30 am IST. uHired's job cards list the skills each posting names, so scanning a page of them shows whether SQL or Excel leads in the roles you want.

Titles, levels, and how MIS roles differ

In the first two years, accuracy matters above all: turning a request into a correct query, a clean sheet or a dashboard update, and asking what decision the number is for. Titles include Data Analyst, Junior or Associate Analyst, and MIS Executive. From three to five years you own a domain's metrics, spot problems before anyone asks, read experiments and influence decisions, usually as Senior Analyst, Product Analyst or Lead Analyst. With six or more years you set metric definitions across teams, decide which requests matter, and build or manage the team as Analytics Manager, Principal Analyst, or Head of Analytics at a smaller firm.

Employers differ sharply. Analytics consulting firms and KPOs hire freshers in batches through aptitude tests and guesstimate rounds, then train them on client work. Product companies and startups hire product analysts who work daily with product managers. GCCs run reporting and analytics for global business units, with more process and some overlap with overseas working hours. Banks, NBFCs and insurers combine analysis with regulatory and management reporting.

MIS roles need their own comparison, because many Indian postings use the title for quite different work:

Aspect	MIS analyst	Data analyst
Main output	Recurring reports in fixed formats: daily, weekly, monthly	Answers to open questions, dashboards, metric definitions
Typical tools	Excel with pivots, lookups and macros; ERP or CRM exports	SQL on a database or warehouse, a BI tool, sometimes Python
Judged on	Timeliness and accuracy	Whether the analysis changed a decision
Often found in	Operations, sales, BFSI back offices, logistics, BPO and KPO teams	Product companies, e-commerce, fintech, consulting, GCCs

Treat MIS work as a starting point, and automate it. Rebuild your reports with Power Query or SQL so they refresh themselves, then add a short "what changed and why" note to each one. Automation plus commentary is what turns an MIS CV into an analyst CV.

How analyst hiring rounds are structured

Most processes combine four or five of these:

• Online test: aptitude and logical reasoning for fresher batches; SQL and Excel questions for everyone.
• SQL round: live queries in a shared editor, or a short take-home on a sample schema.
• Excel task: clean, reconcile and summarise a file in 30 to 45 minutes.
• Case round: a metric moved, or a new feature needs a success metric.
• Guesstimate: common at consulting and analytics firms, occasional elsewhere.
• Dashboard critique or portfolio walk-through: you defend choices in your own work or fix someone else's.
• Hiring manager and HR: stakeholder handling, notice period and expectations.

Interview questions worth rehearsing

1. SQL: "List customers who ordered in January but not in February." Use NOT EXISTS or a left join, and mention that NOT IN returns nothing if the subquery contains a NULL.
2. SQL: "Month-over-month revenue growth per category." LAG over months within each category, with a month calendar joined in so that months with no sales are not silently skipped.
3. SQL debugging: "This LEFT JOIN returns the same rows as an INNER JOIN. Why?" A condition on the right-hand table in WHERE drops the unmatched rows; move it into the ON clause.
4. Metric drop: "Daily orders fell 12% yesterday. What do you check?" Data first (pipeline delay, tracking release, time zone), then the calendar (weekday, holiday, a big cricket match, heavy rain), then segments (city, platform, new versus returning, payment method, since a UPI outage hits checkout success), then the funnel step where orders are lost.
5. Guesstimate: "Estimate two-wheeler service visits in Pune in a month." State the structure (households, ownership, service frequency), sanity-check it from a second direction, and give a range. The logic matters more than the number.
6. Dashboard critique: "Here is the dashboard regional heads use. What would you change?" Ask what decision it serves, then cut KPIs with no comparison period, twelve-slice pie charts, unlabelled dual axes, and missing definitions or refresh times.
7. Excel: "Reconcile this CRM export with the billing export and list the mismatches." Expect IDs stored as text in one file and numbers in the other, trailing spaces, and duplicate customers.
8. Definition: "Define an active user for a grocery delivery app." Which action counts, over what window, excluding test and staff accounts, in which time zone.
9. Stakeholder: "A sales head wants 'all the data' on churned customers by tomorrow." Ask what decision it feeds, and offer a focused cut first.

Making an analyst resume read like analysis

Each bullet should carry a question, a method, a finding, and what happened next. Tools belong inside that sentence, not only in a skills list. These examples are illustrative; replace the numbers with your own.

• Example: "Traced a drop in repeat orders in two cities to a delivery-slot change using SQL cohort analysis; operations reversed the change within a week and repeat rates recovered over the following month."
• Example: "Automated a daily MIS pack of 14 Excel reports with Power Query and SQL views, cutting preparation from three hours to 20 minutes and ending copy-paste errors in figures sent to regional heads."
• Example: "Agreed a single definition of 'active merchant' with finance and sales, documented it, and rebuilt three Power BI dashboards on it, ending a monthly dispute over which number was right."

Common mistakes: "MS Office" and "Advanced Excel" with no evidence of what you did; dashboards described by their number of visuals; certificates listed above work; and bullets that stop at "created a report" without saying who used it or what changed.

Where an analyst career can go next

• Deeper analysis: senior and lead analyst roles in one domain, such as lending, marketplaces or SaaS, where domain knowledge becomes your advantage.
• Analytics engineering: if you enjoy building clean tables more than presenting, learn dbt, Git, dimensional modelling and testing. The data engineer guide covers the layer below you.
• Data science: learn inference, regression and experiment design properly, then ask to analyse the next A/B test at work. The data scientist guide explains what panels expect.
• BI development: if you enjoy building reports that whole departments depend on, the BI developer role adds data modelling, DAX and report performance to what you already know.

Projects a panel will want to ask about:

• A messy public dataset cleaned with every decision written down, three metrics defined with their edge cases, a one-page dashboard, and a 200-word memo ending in a recommendation.
• A metric-drop investigation on a public or simulated dataset, showing the order of checks and what each one ruled out.
• A manual Excel report rebuilt as a refreshable Power Query or SQL pipeline, with the time saved stated plainly.

Freshers should also read How Freshers Can Land a First Software Job in India for off-campus strategy.

Openings and a plan for next week

Live listings for analyst roles: Bengaluru, Mumbai, Delhi, India-wide and remote. Then, over the next week:

1. Write ten SQL queries against a sample schema without looking anything up, two of them with window functions.
2. Rehearse one metric-drop case aloud and keep it to ten minutes.
3. Recast your best three resume bullets as question, method, finding and outcome, using the ATS-friendly resume guide for layout.
4. Before tailoring an application, decide whether the posting is an analyst, MIS or business analyst role; How to Read a Software Job Description shows which lines reveal the real job.