Data Analyst Interview Questions And Answers

You've cleared the resume screen, and now a hiring manager at Swiggy or Zerodha wants to see if you can actually wrangle messy datasets and explain what they mean. The data analyst interview in India has evolved into two distinct tracks: one that tests SQL and Excel competence for traditional analytics roles, and another that assumes you can write Python, work with cloud pipelines, and talk intelligently about how LLMs might automate parts of your own job.

What interviewers actually test

Expect three layers. First, the technical screen: SQL queries, usually live or on a shared screen. You'll be asked to pull cohort retention numbers, calculate month-over-month growth, or identify duplicate records in a mock orders table. If the company is Razorpay, Nykaa, or another product-led firm, add a take-home case study where you analyze a CSV, build a dashboard in Tableau or Power BI, and present insights in a deck.

Second, the business sense round. You'll get a vague prompt like "our checkout drop-off rate increased 8% last week, what would you investigate?" or "how would you measure the success of a new feature?" They want structured thinking, not just tool knowledge. Name your hypotheses, the data you'd pull, and the stakeholders you'd loop in. This round separates analysts who can execute queries from those who understand why the query matters.

Third, the culture and scenario fit. Smaller startups and growth-stage companies ask behavioral questions wrapped in analytics context: "Tell me about a time your analysis was ignored" or "How do you prioritize when three PMs want dashboards by Friday?" At companies like Meesho or PhonePe, expect questions about working with regional datasets, handling Hindi or vernacular metadata, or explaining churn patterns across tier-2 and tier-3 cities.

SQL and technical questions you'll see

Write a query to find the second-highest salary in an employee table. Classic warm-up. Use DENSE_RANK() or a subquery with LIMIT and OFFSET. Interviewers watch how you handle edge cases: what if two employees share the top salary? What if there's only one row?

SELECT DISTINCT salary
FROM employees
ORDER BY salary DESC
LIMIT 1 OFFSET 1;

Or with window functions:

SELECT salary
FROM (SELECT salary, DENSE_RANK() OVER (ORDER BY salary DESC) AS rnk FROM employees)
WHERE rnk = 2;

Calculate 30-day rolling retention for a user activity table. You'll have a table with user_id, event_date, and event_type. The interviewer wants to see if you understand cohort logic and can self-join or use window functions without Googling mid-interview.

Given an orders table and a returns table, find the return rate by product category. Straightforward join and aggregation, but they're checking whether you remember to handle NULL categories, whether you use LEFT JOIN correctly, and if you can explain the business implication of a 15% return rate in electronics versus 3% in groceries.

Python-specific: Load a CSV, clean missing values, create a new feature (like order_hour from a timestamp), and plot a histogram of orders by hour. If the role mentions "analytics engineering" or lists dbt in the job description, expect questions about data modeling, slowly changing dimensions, or how you'd structure a star schema for an e-commerce warehouse.

Case study and take-home assignments

Most product companies send a dataset and a prompt. Recent examples from Indian startups:

• E-commerce: You get three months of transaction data with user_id, order_value, category, city, and payment_method. The prompt: "Identify user segments and recommend one retention initiative for each."
• Fintech: A loan application funnel with drop-off at each stage. "Where is the biggest leak, and what data would you need to diagnose why?"
• Edtech: Course enrollment and completion rates across geographies. "Should we expand to Tamil Nadu or Karnataka next?"

Interviewers expect a slide deck (usually five to eight slides), a clean notebook or SQL script, and a two-minute Loom video or live presentation. They're scoring clarity of insight, not complexity of method. A well-labeled bar chart that shows tier-2 cities have 40% higher repeat purchase rates beats a random forest model with no interpretation.

Common mistakes: spending four hours on feature engineering and two minutes on the executive summary, forgetting to state assumptions (like how you handled missing city values), or building a dashboard with 12 panels when the brief asked for three actionable insights.

Business-sense and product questions

"Daily active users dropped 10% yesterday. Walk me through your investigation." Start with data quality: was there a logging outage, a bot purge, or a definition change? Then segment by platform (Android, iOS, web), geography, and user cohort. Mention you'd check for app crashes, server errors, or a new release. Finally, compare to the same day last week and last month to rule out seasonality. The structure matters more than the perfect answer.

"How would you measure the success of a referral program?" Define success first: is it new user acquisition, revenue, or long-term retention? Then name metrics: referral conversion rate, cost per referred user, 90-day retention of referred versus organic users, and contribution margin. If you've read the company's blog or earnings updates, tie your answer to their known priorities (for example, Cred cares about high-intent users, not volume).

"Explain a technical concept to a non-technical stakeholder." Pick something you actually know: p-values, A/B test sample size, or why correlation isn't causation. Use an analogy. For instance, "A p-value is like a confidence score. If it's below 0.05, we're saying there's less than a 5% chance this result happened by luck, so we trust the pattern." Avoid jargon, watch for nods, and pause to let them ask questions.

Behavioral and situational prompts

"Tell me about a time your analysis led to a decision." Use the STAR format: situation, task, action, result. Be specific. "At my last internship, I noticed that 60% of cart abandonment happened on the payment page. I ran a funnel analysis, found that COD users completed checkout 25% faster, and recommended adding UPI as a one-tap option. The product team tested it, and checkout completion rose 12% in the first month."

"How do you prioritize when you have five requests and two days?" Mention impact and effort. Ask stakeholders for context: is this for a board deck, a bug fix, or exploratory research? Flag dependencies. If you can't finish everything, communicate early and propose a phased delivery. Interviewers want to see you're not a ticket-taker.

"Describe a time you made a mistake in your analysis." Own it. Maybe you forgot to filter test users, or you used the wrong join and inflated revenue by 30%. Explain how you caught it, how you communicated the correction, and what you changed in your workflow (like adding a sanity check or peer review step). Humility and learning matter more than perfection.

Salary and role expectations

Entry-level data analyst roles at startups and mid-sized product companies typically range from ₹4 LPA to ₹8 LPA, with Bengaluru and Gurgaon on the higher end. Analysts with two to four years of experience and strong SQL, Python, and dashboard skills report offers between ₹8 LPA and ₹15 LPA. Senior analysts or those moving into analytics engineering roles at firms like Razorpay, Flipkart, or Groww can see ₹15 LPA to ₹25 LPA, especially if they bring dbt, Airflow, or cloud warehouse experience.

Equity is less common than in engineering roles, but growth-stage startups sometimes offer ESOPs worth 0.05% to 0.2% for mid-level hires. Always ask about the vesting schedule and the last valuation.

If you're preparing across multiple roles, compare expectations in adjacent fields by reviewing business analyst interview questions and answers or exploring data science interview questions and answers to understand where the boundaries blur.

Key takeaways

• Interviews split into three parts: technical (SQL, Python, dashboards), business sense (structured problem-solving), and behavioral (communication and prioritization).
• SQL questions focus on joins, window functions, and cohort logic; practice on real datasets, not just LeetCode-style puzzles.
• Take-home case studies reward clear insights and clean presentation over complex models; always state your assumptions and limitations.
• Business-sense prompts test whether you can translate data into decisions; structure your answer with hypotheses, data sources, and stakeholder impact.
• Salary for early-career analysts typically ranges from ₹4 LPA to ₹8 LPA, scaling to ₹15 LPA to ₹25 LPA for senior roles with cloud and pipeline experience.

Ready to put your prep into practice? Browse the latest openings on UnoJobs data analyst roles and apply with confidence.