Data Analyst interview questions (India, 2026): what to ask and what to listen for

12 interview questions for a Data Analyst role, grouped by what each round should establish, each with what a strong answer contains, plus a five-point scorecard so every interviewer rates the same things.

Last updated 3 September 2026. Competitor pricing and plan limits are re-verified against each vendor’s own pages.

Interviewing a Data Analyst well means asking about work the candidate has actually done, listening for specifics, and scoring every candidate on the same five criteria. The questions below are grouped by what they establish; each one comes with what a good answer contains, which is the part most question lists leave out.

A Data Analyst turns an Indian company's operational data — orders, customers, marketing, finance — into reports, dashboards and answers that change decisions, working in SQL, spreadsheets and a BI tool, and is usually the first person whose whole job is asking what the numbers actually say.

Use the screening questions on the job description page to filter before anyone reaches an interview, and the scorecard at the end so a panel can compare candidates rather than impressions.

Analysis that changed something

Tell me about an analysis that changed a decision. What was the question, what did you find, what did the company do, and what were the caveats?: A strong answer: A real decision, a finding stated plainly, and honest caveats. Analysts who only produce reports cannot answer this.
Describe a time the data did not support what a stakeholder wanted to hear. What did you do?: A strong answer: Showed the evidence clearly, offered what the data could support, held the line politely. This is the integrity test for the role.
Tell me about a metric that was defined inconsistently across teams and how you fixed it.: A strong answer: Recognises definition drift, documents a single definition, and gets teams to adopt it — the unglamorous core of the job.
What was the messiest data source you worked with, and how did you make it usable repeatably?: A strong answer: A cleaning pipeline — scripts, validations, a schedule — not a heroic one-off in Excel.

Technical depth

Here is a small schema [orders, customers, payments]. Write the SQL for monthly revenue by acquisition channel, excluding refunded orders, and tell me what could go wrong with it.: A strong answer: Correct joins and filters, awareness of duplicates, timezones, partial refunds and late-arriving data. Judge reasoning about pitfalls as much as the query.
Retention dropped 5% last month. How would you investigate?: A strong answer: Segments by cohort, channel, product, geography; checks for data issues first; forms hypotheses and tests them; distinguishes a real change from noise.
Explain to a founder why the average order value went up while revenue went down.: A strong answer: Plain-language explanation of mix effects and volume, without jargon. Communication is half the role.
How do you decide whether a difference between two groups is real or noise?: A strong answer: Practical statistical sense — sample sizes, variance, a confidence interval or a test — and the humility to say 'not enough data'.

Working with stakeholders

Show me a dashboard you built. Who used it, how often, and what did you remove from it?: A strong answer: Built for a specific reader with a purpose; evidence of use; and the discipline to remove what nobody looked at.
How do you handle a queue of ad-hoc requests from five teams?: A strong answer: Prioritises by decision impact, turns repeat requests into self-serve dashboards, and says no or not yet with reasons.
What would you want to instrument or collect in our product that we probably are not?: A strong answer: Thinks ahead to next quarter's questions — events, attribution, cohorts — and knows collection has a cost.
What do you want to get deeper in — engineering, statistics, a domain, product — and does this role offer it?: A strong answer: A direction a first-analyst role can plausibly serve — building the data setup, deep knowledge of one business domain, moving towards product or analytics engineering — with awareness that there is no data science team to grow into here. Alignment reduces the one-year churn common in analyst roles.

Scorecard for Data Analyst interviews

Score each candidate on these five criteria, one to four, straight after the interview and before talking to the other interviewers. Written scores taken independently are what make a panel decision defensible; a discussion first produces one opinion with several signatures.

Decision impact: analysis that changed something, with honest caveats.
Integrity: holds the line when data contradicts a stakeholder.
SQL and data craft: correct queries, awareness of pitfalls, repeatable cleaning.
Investigation and statistical sense: segments, hypotheses, noise versus signal.
Communication: plain-language explanations; dashboards built for readers.

Screen before you interview

Most Data Analyst interviews that go badly were avoidable at the application stage. The job description template for this role carries five screening questions that settle the deal-breakers — notice period, location, the one or two hard requirements — before anyone books a slot.

Do you write SQL — joins, window functions, CTEs — as a daily part of your work, and have you for at least 2 years? — knockout
Have you built a dashboard or report that leadership used to make a decision, and can you describe the decision? — for context
Are you able to work [from our City office X days a week / remote with overlap hours]? — knockout
What is your current notice period, in days? — for context

Questions, answered straight

What is the best Data Analyst interview question?

Ask about an analysis that changed a decision — question, finding, what the company did, caveats. Then ask about a time the data contradicted what a stakeholder wanted. Together they test impact and integrity, the two things that separate analysts from report-builders.

Should a Data Analyst interview include a SQL test?

Yes, a short live one on a small schema — monthly revenue by channel excluding refunds, say — with the follow-up 'what could go wrong with this query'. The pitfalls they name matter more than the syntax.

How do I assess communication in a Data Analyst?

Ask them to explain to a founder why average order value rose while revenue fell, in plain words. Analysts who can explain mix and volume without jargon will be useful to your leadership team; those who cannot will produce dashboards nobody reads.