Data Analyst Interview Questions

What they’re assessing: The screener is separating hands-on analysts from people who only consumed dashboards someone else built.

How to answer: Lead with your toolset and one quantified outcome: 'I work primarily in SQL and Power BI; my churn dashboard helped the retention team cut monthly churn by 8%.' Name the business domain (fintech, e-commerce, SaaS) because domain match matters heavily in Indian analyst hiring. Close with why this company's data problems interest you.

What they’re assessing: With so many career-switchers entering analytics in India, screeners test whether your interest survives contact with grunt work.

How to answer: Avoid 'data is the new oil' clichés. Tell a specific origin story — an Excel model you built that changed a decision, or a college/work project where the numbers contradicted intuition. Acknowledge the unglamorous 70% (cleaning, reconciling, stakeholder back-and-forth) to show you know the real job, then say why the remaining 30% makes it worth it.

What they’re assessing: Analysts who understand the business ramp up months faster; ignorance here is an easy rejection filter.

How to answer: Do 30 minutes of homework: know the company's core metric (GMV for e-commerce, AUM for fintech, occupancy for hospitality) and one recent development. Then connect: 'Your quick-commerce expansion means dark-store level unit economics analysis — that is exactly the granularity of work I want.' Specificity beats flattery every time.

What they’re assessing: Standard Indian screening triage — budget fit and joining timeline before panel time is invested.

How to answer: State current CTC honestly (it will surface in background verification) and give a researched range for expectations rather than a hike percentage. On notice period, share your official duration plus any buyout or negotiation option. If you are a fresher or from a bootcamp, quote the market band for entry analysts in your city and say you are flexible for the right learning curve.

What they’re assessing: Many Indian analyst roles support US/EU counterparts; HR is screening out mismatched expectations early.

How to answer: Answer honestly with your real constraints, but frame flexibility positively: 'Occasional overlap calls are fine; I have handled a 4-hour overlap with a UK team before.' Ask a clarifying question back — how frequent, which geography — because doing so shows you treat commitments seriously rather than saying yes to everything.

What they’re assessing: This is the core value test — analysts who cannot cite one decision they influenced are seen as report generators.

How to answer: Use STAR with the decision as the punchline: 'My cohort analysis showed 60% of churn came from one onboarding path; the product team redesigned it and 90-day retention rose 12%.' Include the moment of persuasion — how you presented it and to whom. If your work was one input among several, say so; honest partial credit is more convincing than claimed sole credit.

What they’re assessing: Every analyst ships a wrong number eventually; they are testing ownership and your correction process.

How to answer: Pick a real error — a join that duplicated rows, a filter applied inconsistently, stale data. Walk through: how it was caught, how fast you corrected and communicated it, and the process change you made (a QA checklist, reconciling totals against a source of truth before sending). Ending with 'I now sanity-check every output against a known benchmark' shows the lesson stuck.

What they’re assessing: Requirement clarification is half the job; they want to see you interrogate the ask instead of guessing.

How to answer: Describe your clarifying framework: what decision will this inform, what grain (daily/SKU/region), what time window, and what does 'sales' mean here — bookings, invoiced, or collected revenue (a classic trap in Indian B2B). Share how you confirmed scope in writing with a quick mock-up before building. A 15-minute mock saves a week of rework — say exactly that.

What they’re assessing: Analysts sit in a service position; they are testing whether you prioritise by business impact or by who shouts loudest.

How to answer: Show a criteria-based approach: revenue impact and decision urgency first, then effort. Describe communicating the queue transparently to all three requesters with dates — the transparency is what prevents escalations. Bonus point: mention identifying that one request was a repeat need and turning it into a self-serve dashboard so it never lands on your desk again.

What they’re assessing: Insight that cannot be communicated is worthless; storytelling is the most under-tested skill they are probing.

How to answer: Give a before/after: the technical version ('logistic regression showed feature X with 0.7 coefficient') versus what you actually said ('customers who skip step 2 are three times more likely to leave — here are the three fixes'). Mention leading with the recommendation, not the methodology, and keeping method details in an appendix for those who ask.

What they’re assessing: Real analyst work is 60-70% cleaning; they want evidence you have survived genuinely dirty data.

How to answer: Be specific about the mess: inconsistent date formats, city names spelled five ways, duplicate customer records across systems, or Excel sheets maintained by hand across branches — very common in Indian mid-size companies. Walk through your cleaning logic and, critically, how you documented assumptions ('treated blank status as inactive, flagged for business confirmation'). Documented assumptions are the senior signal.

What they’re assessing: Window functions plus GROUP BY logic is the single most common live-SQL test in Indian analyst interviews.

How to answer: Structure the answer: aggregate revenue by category and product, apply RANK() or DENSE_RANK() OVER (PARTITION BY category ORDER BY revenue DESC), filter rank <= 3 in an outer query or CTE. State the WHERE/HAVING difference precisely — WHERE filters rows before aggregation, HAVING filters groups after. Mention the tie-handling choice (RANK vs DENSE_RANK vs ROW_NUMBER) unprompted; interviewers score that distinction heavily.

What they’re assessing: Fan-out from one-to-many joins is the most common real-world analyst bug; they are testing practical join intuition.

How to answer: Diagnose out loud: the join key is not unique on one side (for example, orders joined to a payments table with multiple rows per order), so rows fanned out and the SUM double-counted. Fixes: aggregate the many-side to the join grain first in a CTE, or dedupe with ROW_NUMBER. State your habit: always check row counts and totals before and after a join.

What they’re assessing: Excel remains the daily driver in most Indian companies; they are testing working fluency, not certification trivia.

How to answer: Give the practical reasons: VLOOKUP breaks when columns are inserted and cannot look left; INDEX-MATCH and XLOOKUP handle both, and XLOOKUP defaults to exact match. For pivots, note they aggregate fast but struggle with row-level lookups and multi-condition logic where SUMIFS or Power Query fits better. Mentioning Power Query for repeatable cleaning steps signals you work beyond formulas.

What they’re assessing: This tests whether you design for decisions or decorate with charts.

How to answer: Start with questions, not charts: what decisions does the sales head make weekly — territory allocation, rep coaching, pipeline risk? Then propose a hierarchy: top row of 4-5 KPIs versus target (revenue, pipeline coverage, win rate), trend view, and drill-down by region/rep. Say what you would leave out and why — every chart must answer a question or it is noise. Name your tool fluency (Power BI or Tableau) with one real example.

What they’re assessing: Stats literacy separates analysts who can validate a claim from those who just chart whatever moves together.

How to answer: Keep definitions crisp: the p-value is the probability of seeing data this extreme if there were truly no effect — it is not the probability the hypothesis is true. For correlation/causation, use a business case: ice-cream sales and AC sales rise together because of summer, not each other; or app opens correlate with purchases because engaged users do both. Close with the fix: A/B tests or controlling for the confounder.

What they’re assessing: Metric-drop cases test structured thinking under ambiguity — the defining analyst skill.

How to answer: Show a decision tree: first rule out data issues (tracking bug, pipeline failure, definition change) — roughly half of real 'drops' are instrumentation. Then segment: platform, app version, geography, acquisition channel, new versus returning users. Then check external causes: a recent release, outage, competitor launch, or seasonality (exam season and festivals move Indian consumer metrics visibly). Narrate hypotheses in order of likelihood and how each is cheap to test.

What they’re assessing: The manager is checking you can own the full loop independently, not just execute assigned queries.

How to answer: Pick your best project and structure it in five beats: the business question, how you scoped and clarified it, data work (sources, cleaning, method), the insight, and the decision plus measured outcome. Spend the most time on scoping and outcome, the least on SQL details — that weighting itself demonstrates seniority. Keep it under four minutes and invite drill-downs.

What they’re assessing: One wrong number in a leadership review destroys an analyst's credibility; managers hire for reliability first.

How to answer: Describe a concrete QA routine: reconcile against a known source of truth (finance totals, GA sessions), check row counts at each transformation step, test edge months, and have definitions documented so 'revenue' means one thing. Mention version-controlling queries or maintaining a metrics dictionary if you have done it. A specific checklist beats 'I am very careful.'

What they’re assessing: This probes composure and whether you treat challenges as audits rather than attacks.

How to answer: Show a de-escalation sequence: first genuinely re-verify — seniors often know a context you lack, like a definition change or an offline sales channel your data misses. Then walk them through methodology transparently and try to reconcile their mental number with yours; the gap usually has a findable cause. If the data holds, present it as 'here is what would need to be true for both to be right.' Never make it a win-lose.

What they’re assessing: Junior analysts add; senior analysts subtract. This question directly tests which one you are.

How to answer: Give your filter: if a chart does not change a decision or answer a recurring question, it goes. Describe a real instance where you cut a stakeholder-requested widget and the pushback conversation — offering the cut content as an on-demand view is the diplomatic move. Cite the cost of clutter: attention is the scarcest resource in a Monday review meeting.

What they’re assessing: The manager is checking fit against the team's growth paths and how long you will realistically stay.

How to answer: Pick a genuine direction and connect it to this role: 'I want to go deeper into analytics engineering — owning pipelines and metrics infrastructure — and this team's dbt adoption is exactly that path.' It is fine to keep two options open, but name concrete next skills (advanced SQL modelling, Python, experimentation) rather than saying 'wherever the company needs me.'