35+ Data Analyst Interview Questions & Answers (2026 Guide)

Data analyst interviews at top tech companies test SQL fluency, structured analytical thinking, statistics literacy, and the ability to turn data into a clear recommendation.

These are the most common data analyst interview questions, sourced from real interviews at Meta, Google, Amazon, Uber, and more.

Top data analyst interview questions

The most common data analyst interview questions fall into five categories. View more questions.

1. Write a query to find employees who earn more than their manager. See 76 answers.
2. There's been a 10% drop in Facebook's daily post views. How would you investigate? See 6 answers.
3. Tell me about a time you made a mistake. See 107 answers.
4. Find the number of users who called three or more people last week. See 13 answers.
5. What is the market size for driverless cars in 2025? See 18 answers.
6. Tell me about a time the business problem wasn't clearly defined. See 5 answers.
7. Find the top 3 posts by number of likes where likes are 100 or more. See 47 answers.
8. Tell me about yourself. See 126 answers.
9. Find the top salaries in each department. See 35 answers.
10. How would you convey insights and methods to a non-technical audience? See answers.

These questions span the following categories:

• SQL and coding questions
• Data analysis and business case questions
• Behavioral questions
• Statistics and experimentation questions
• Excel and visualization questions

SQL and coding questions

SQL is the single most-tested skill in data analyst interviews. Hiring managers at Meta, Amazon, Google, and Uber all say the same thing: if you can't pass the SQL round, you won't get the job. Questions range from basic JOINs and GROUP BY to window functions, CTEs, and query debugging.

1. Write a query to find employees who earn more than their manager. See 76 community answers. Asked at multiple companies. IDE question, medium difficulty.
2. Find the number of users who called three or more people in the last week. See 13 community answers. Asked at Meta.
3. Find the top 3 users by number of posts with 100+ likes. See 47 community answers. IDE question, medium difficulty.
4. Find the top salaries in each department. See 35 community answers. Asked at Tesla. IDE question, hard difficulty.
5. Calculate the monthly post success rate by user type. See 36 community answers. Asked at LinkedIn. IDE question, easy difficulty.
6. Find the top 3 employees by salary. See 68 community answers. IDE question, easy difficulty.
7. Explain SQL stored procedures: what they are, when to use them, and how they differ from functions. See 4 community answers.
8. What's the difference between WHERE and HAVING?
9. How would you calculate a rolling 7-day average of daily active users?
10. How would you deduplicate a table with no unique key?

Sample answer: "Employees who earn more than their manager"

From a community answer on Exponent:

SELECT e.firstname AS firstname, m.salary AS manager_salary FROM employees e JOIN employees m ON e.manager_id = m.id WHERE e.salary > m.salary;

This is a straightforward self-join problem, but interviewers are watching for how you think through the table relationships. The key is recognizing that the employees table references itself through the manager_id foreign key. Candidates who talk through the join logic before writing code consistently score higher.

Sample answer: "Rolling 7-day average of DAU"

SELECT date, dau, AVG(dau) OVER (ORDER BY date ROWS BETWEEN 6 PRECEDING AND CURRENT ROW) AS rolling_7d_avg FROM daily_metrics;

The detail that separates good from great here: ROWS BETWEEN 6 PRECEDING AND CURRENT ROW gives you 7 days total including today. If there are gaps in dates, the rolling average will silently use fewer data points. Strong candidates flag this edge case and suggest either filling gaps first or switching to a RANGE-based window.

SQL interview tips

Talk before you type. SQL interviews are interactive, and staying silent while you think is a surprisingly common mistake. Say "I'm going to break this down" before you start coding.

Ask clarifying questions early: what time range? How is revenue defined? Are there duplicates to handle? Then check for edge cases. NULLs, duplicates, and date gaps are where candidates separate themselves.

Finally, connect your output to business meaning. Interviewers care that you can explain what the result means, not just that you can write the query.

Data analysis and business case questions

Business case questions test whether you can take an ambiguous problem, structure your approach, and land on a clear recommendation. Interviewers want to see your thought process, not a "right" answer.

11. There's been a sudden 10% drop in Facebook's daily post views. How would you investigate? See 6 community answers. Asked at Meta.
12. What is the market size for driverless cars in 2025? See 18 community answers. Asked at Google.
13. How would you convey insights and methods to a non-technical audience? See answers. Asked at IBM.
14. Sales dropped 25% last month. How would you investigate?
15. What metrics would you use to measure whether a new feature launch was successful?
16. A stakeholder says the dashboard numbers don't match finance's report. What do you do?
17. How would you use cohort analysis to identify retention issues?

Sample answer: "10% drop in Facebook's daily post views"

From a community answer by Steve Y. on Exponent:

Define: How is daily post view calculated? Isolate Issues: Data issue, Time period, Geo, iOS vs Android vs Web. Correlated Metrics in the funnel: DAU, Time spent/scrolls, Engagement (likes, comments). External factors: Competitor actions, Big events. Internal factors: Product launch, Feature change.

This answer follows the structure interviewers want to see: start by clarifying the metric definition, then systematically decompose the problem. The candidate segments by platform and geography to isolate the driver, checks correlated metrics in the funnel to distinguish between fewer users vs. fewer views per user, and scans for external vs. internal causes. Rule out data issues first, because a broken tracking pixel could explain the entire drop.

Sample answer: "Market size for driverless cars"

From a community answer by Mark on Exponent:

Questions: Should we consider US or Global? Is there any existing market insights/research around driverless cars adoption? Can we assume that driverless car technology is production/public ready by 2025? Is this focus on commercial or consumer use?

The interviewer isn't looking for a precise number. They're evaluating your estimation framework. This candidate starts by scoping the problem (US only, consumer focus), then builds up from population to driving-age adults to car ownership rate to replacement rate to willingness-to-pay premium for autonomous. Each assumption is stated explicitly so the interviewer can push back.

Business case interview tips

Use the PACE framework: Plan (ask clarifying questions, scope the problem), Analyze (funnel analysis, cohort breakdowns, segmentation), Construct (synthesize into 1-2 clear insights), Execute (recommend specific next steps).

Segment before you diagnose. Break the problem down by region, product, channel, and cohort before looking for a root cause. And lead with "so what": state the insight first, then explain how you got there.

Behavioral questions

Many candidates who ace the technical rounds fail the behavioral round. Interviewers are evaluating communication clarity, cross-functional collaboration, ownership, and how you learn from mistakes.

18. Tell me about a time you made a mistake. See 107 community answers. Asked at Airbnb, Amazon, American Express, and 12+ more companies.
19. Tell me about yourself. See 126 community answers. Asked at Accenture, Amazon, Airbnb, and 33+ more companies.
20. Tell me about a time you solved a complex problem. See 28 community answers. Asked at Amazon, GitHub, and 7+ more.
21. Tell me about a time the business problem wasn't clearly defined. How did you handle it? See 5 community answers. Asked at Anthropic and Uber.
22. Tell me about a time you had to make a decision with a lot of uncertainty. See 2 community answers. Asked at Tinder.
23. How do you prioritize tasks? See 70 community answers. Asked at Accenture, Adobe, Amazon, and 13+ more.
24. How would you approach learning about a task you're completely unfamiliar with? See 2 community answers. Asked at Apple.
25. Tell me about your past projects. See 4 community answers. Asked at Accenture, Amazon, and 17+ more.

Sample answer: "Tell me about a time you made a mistake"

Adapted from community answers on Exponent:

Early in my career, I built a churn model that predicted high-risk users based on login frequency. The model looked great on paper, but when we acted on it, the intervention didn't move retention at all. I dug back in and realized I had correlation without causation — users weren't churning because they stopped logging in. They stopped logging in because they'd already decided to leave. The real leading indicators were things like failed transactions and support tickets. I learned to always validate assumptions with a small test before scaling, and to think harder about the causal story behind the data.

This answer works because it shows vulnerability, a specific technical mistake, and a concrete lesson. The interviewer isn't looking for perfection. They want to see that you reflect honestly and change your approach. Answers that dodge the question or describe trivial mistakes signal a lack of self-awareness.

Behavioral interview tips

Use the STAR framework: Situation (set context briefly), Task (your specific responsibility), Action (what you did, and where most of your answer should live), Result (quantify impact).

Prepare 5-7 experiences you can adapt to different questions. Know your resume inside and out, because interviewers will dig into details. Own your mistakes. Interviewers want reflection and growth, not perfection, and saying "I learned X and now I do Y" is more compelling than a flawless narrative.

Statistics and experimentation questions

You don't need to be a statistician, but you need a working understanding of hypothesis testing, confidence intervals, and A/B testing mechanics, especially for roles involving product analytics.

26. An A/B test shows a p-value of 0.04. What does this mean? Should we ship?
27. How do you determine the right sample size for an experiment?
28. What's the difference between Type I and Type II errors? Which is worse?
29. How would you handle an A/B test where the metric improved but you suspect a novelty effect?
30. When would you use a t-test vs. a chi-squared test?

Sample answer: "A/B test with p-value of 0.04"

A p-value of 0.04 means there's a 4% probability we'd see a difference this large if there were no real effect — so it clears the standard 0.05 significance threshold. But I wouldn't ship on that alone. I'd check the effect size and confidence interval: if the lift is 0.1%, it might not be worth the engineering cost even if it's statistically significant. I'd also look at the test duration and whether we hit the planned sample size, check for novelty effects by plotting the metric over time, and confirm the metric moved in the right direction for the segments that matter most. Statistical significance is necessary but not sufficient.

Statistics interview tips

Always connect stats to business decisions. "Statistically significant" is never the end of the answer. Effect size, cost, and practical impact matter more.

Know the difference between correlation and causation. This comes up in nearly every analytics interview. Have a ready example of a spurious correlation.

Excel and visualization questions

Spreadsheets are still the second most-tested skill after SQL. Visualization questions test whether you can choose the right chart, design for your audience, and defend your choices.

31. When would you use INDEX-MATCH over VLOOKUP?
32. Build a pivot table to summarize customer data by region and product.
33. How would you redesign a cluttered executive dashboard?
34. Which chart type would you use to show user retention over 12 months?
35. How would you visualize A/B test results for a non-technical stakeholder?

Sample answer: "INDEX-MATCH vs. VLOOKUP"

I prefer INDEX-MATCH for three reasons. First, it lets me look up values to the left of my key column, which VLOOKUP can't do. Second, if I insert or delete columns, VLOOKUP breaks because it uses a hard-coded column number, but INDEX-MATCH references the column directly. Third, INDEX-MATCH is faster on large datasets. That said, if I'm doing a quick one-off analysis and the data is simple, VLOOKUP is fine — I pick the tool that matches the context.

Data analyst interview frameworks

Two frameworks cover the majority of data analyst interview questions. Knowing when to reach for each one matters more than memorizing the steps.

PACE framework (business case questions)

Use PACE when you're asked to investigate a metric change, evaluate a product decision, or diagnose a business problem.

STAR framework (behavioral questions)

Use STAR for any "tell me about a time" question. Spend the majority of your time on Action.

Practice data analyst interview questions

The fastest way to prepare is to practice with real questions from actual interviews, timed and with feedback.

• Practice data analyst questions — browse 123 questions from real interviews at Meta, Google, Amazon, and more, with community answers.
• Take the data analytics interview course — structured prep covering SQL, business cases, take-home studies, and behavioral rounds.
• Start a mock interview — timed practice with peers or coaches who give real feedback.

Data analyst interview FAQ

How long does a data analyst interview loop take?

Most loops take 2-4 weeks from recruiter screen to final decision. Expect 4-5 rounds: a recruiter call, a technical SQL screen, a business case round, a hiring manager interview, and a behavioral round. Some companies like Uber and Shopify add a take-home case study. For a full breakdown of the interview process and how to prepare for each round, see our data analyst interview prep guide.

What SQL topics are tested most?

JOINs, GROUP BY with aggregations, window functions (RANK, ROW_NUMBER, LAG/LEAD), subqueries, and CTEs cover the vast majority of questions. At the senior level, expect query optimization, NULL handling, and production-readability. Our SQL question bank has the exact problems companies ask.

Do I need Python for a data analyst interview?

Only if the job description lists it. If it does, expect basic pandas: groupby operations, merging dataframes, and simple data cleaning. If Python isn't mentioned, spend that prep time on SQL. SQL is tested everywhere; Python is role-dependent.

What's the difference between a data analyst and data scientist interview?

Data analyst interviews emphasize SQL, business case reasoning, dashboarding, and communication. Data science interviews go deeper on statistics, machine learning, experimental design, and typically require Python or R. The overlap is SQL and A/B testing.

How do I prepare for a take-home case study?

Treat it like a work deliverable, not a school assignment. State your assumptions upfront, link formulas to source data (never paste raw values), lead with insights rather than methodology, and prepare a 5-minute executive summary. For a detailed walkthrough, see our take-home case study module.