Data Analyst Onsite Interview Questions: 30 Examples by Round

Data analyst onsite interview questions usually cover five areas: a SQL exercise you solve in front of the interviewer, an Excel or BI case, a stakeholder or business case, a presentation of your take-home analysis, and behavioral questions. Below are 30 example questions grouped by round, what a strong answer shows, and sample answer outlines we wrote for the questions that trip people up most.

These are examples we wrote to match common patterns, not a list from any particular company. Analyst loops vary more than most: a marketing analytics role may focus on dashboards and attribution, while a product analyst role looks closer to a junior data scientist loop. Ask your recruiter which rounds you will have and which tools you will use. For a general question bank, see our data analyst interview questions guide.

What does a data analyst onsite interview include?

A data analyst onsite is commonly three to five interviews of 30 to 60 minutes, in person or on video. Each interviewer typically scores one area. The table shows the usual rounds and how to prepare for each.

For the logistics side of the day, see our hub on how to prepare for an onsite interview.

SQL exercise questions

Interviewers watch how you work as much as whether the query runs. Say what each table contains before you write, build the query in steps and describe how you would verify the result.

1. "Find the total revenue by month for the last 12 months." Shows date truncation and filtering. Mention time zones if timestamps are stored in UTC.
2. "List customers who ordered in January but not in February." Tests LEFT JOIN … IS NULL or NOT EXISTS.
3. "Find the second-highest order value for each customer." Tests window functions and how you handle ties.
4. "Calculate the share of orders that used a discount code, by region." Watch for integer division, which can quietly return zero in some databases.
5. "Show month-over-month growth in active users." Use LAG(). Define "active" before writing.
6. "Why does this query return more rows than expected?" Usually a join on a non-unique key. Compare row counts before and after the join.
7. "Find users whose first order was over $100." Tests ranking within a group and filtering after ranking.
8. "How would you handle NULLs in a revenue column?" Explain the difference between a missing value and a zero, and ask how the data is produced.

Sample answer outline we wrote for question 2: "I'll confirm that orders has one row per order with a customer_id and an order_date. I'll build two sets: customers with any January order and customers with any February order. Then I'll select from the January set where no matching customer exists in February, using NOT EXISTS because it handles NULLs more predictably than NOT IN. To check it, I'd count January customers, February customers and the overlap, and confirm the result equals January minus the overlap."

More drills are in our SQL interview questions guide.

Excel and BI case questions

Some analyst onsites hand you a spreadsheet or dataset and a question, then watch you work. Others ask you to describe how you would build a dashboard. Either way, structure matters: inputs, calculations and outputs should be easy to follow.

9. "Here is a sales file. Which product line grew fastest last quarter?" Clean the data first, then use a pivot table, and state how you define growth.
10. "Build a chart that shows the trend in support tickets by category." Choose a chart type that fits; a line chart for trend, not a pie chart.
11. "Match these two lists of customers that use different ID formats." Tests lookups and text cleaning. Say how you would handle records that do not match.
12. "Design a dashboard for the head of sales." Start with the three decisions they make each week, then choose metrics and filters for those decisions.
13. "Two dashboards show different revenue numbers. What do you do?" Compare definitions, filters, refresh times and data sources before assuming either is wrong.
14. "How would you automate a weekly report that takes you four hours?" Describe the source, the transformation step and a scheduled refresh in the BI tool you know.

Business and stakeholder case questions

The business case tests whether you can turn a vague request into a useful analysis. Start by asking what decision the stakeholder needs to make. Then define the metric, break the problem down and finish with a recommendation, not just a finding.

15. "Our conversion rate dropped last week. Walk me through how you'd investigate." Check tracking first, then segment by channel, device, region and new versus returning visitors.
16. "The marketing team wants to know which channel works best. How do you answer?" Define "best" (cost per acquisition, customer value), and mention the limits of last-click attribution.
17. "A manager asks for 'all the data on customer churn.' What do you do?" Ask what decision the data will support and agree on a narrower question.
18. "How would you measure the impact of a price increase?" Compare revenue, volume and churn before and after, and look for a comparison group.
19. "Store A has higher sales than Store B. Is Store A better managed?" Raise size, location and foot traffic; normalize before comparing.
20. "What metrics would you track for a new loyalty program?" Enrollment, repeat purchase rate among members versus similar non-members, and program cost.

Sample answer outline we wrote for question 15: "First I'd confirm the drop is real by checking for tracking or tagging changes and comparing with order data from the database. If it's real, I'd split conversion into steps (visit, product view, add to cart, checkout, purchase) to find where it fell. Then I'd segment the failing step by device, channel and region. If one segment explains most of the drop, such as mobile checkout after an app release, I'd share that with the product team with the estimated revenue impact and a recommendation to roll back or fix."

Take-home presentation questions

If you completed a take-home, expect to present it for 10 to 20 minutes and then defend your choices. Lead with your answer, then show how you got there. Our guide to the take-home assignment interview covers the build; this round is about the presentation.

21. "Walk us through your findings." Open with the recommendation and its impact, then two or three supporting points.
22. "Why did you exclude those records?" Explain the rule you used and what the result looks like if you include them.
23. "What would you do with another week?" Name the most valuable next analysis, not a long wish list.
24. "How confident are you in this conclusion?" Be specific about limits: sample size, missing data or a confounding factor.
25. "What would you tell the CEO in one sentence?" Practice this before the day.

Sample answer outline we wrote for question 24: "I'm confident in the direction but less in the size. Repeat-purchase rates are clearly higher for customers who used the mobile app, across all three regions. But app users were also more engaged before they installed the app, so part of the gap is likely selection. To size the true effect, I'd compare similar customers before and after install, or test an app promotion with a holdout group."

Behavioral questions for data analysts

Behavioral questions check how you work with stakeholders and whether your analysis leads to action. Answer with the STAR method and put a number in the result.

26. "Tell me about an analysis that changed a business decision."
27. "Tell me about a time you found an error in your own work after sharing it." Show how quickly you corrected it and what you changed in your process.
28. "How do you prioritize requests from several teams?" Describe your criteria and how you communicate trade-offs.
29. "Tell me about a time you pushed back on a stakeholder's request."
30. "Describe a time you explained data to a non-technical audience."

Data science loops overlap with these rounds; compare our data scientist onsite interview questions and the full list of onsite interview questions across roles.

How should you practice for an analyst onsite?

Practice the spoken parts as seriously as the SQL. Many analysts write correct queries but explain them poorly under time pressure.

• Rehearse your take-home presentation aloud twice, with a timer, and write down the three hardest questions you could get.
• Practice case and behavioral answers in a browser-based AI mock interview, which scores each spoken answer out of 10 with what to fix and a model answer.
• For questions tailored to the role, the desktop Mock Interview builds the interview from your résumé and the job description.
• It is free for 15 minutes a day, with no account. Follow each employer's rules on AI and use it to prepare beforehand.

FAQ

What is the hardest part of a data analyst onsite?

For many candidates it is the business case, because it is open-ended. Practice asking what decision the analysis supports, defining the metric and finishing with a clear recommendation.

Do I need to know Python for a data analyst onsite?

It depends on the role. Many analyst onsites test SQL and a spreadsheet or BI tool only, while product analyst roles may add Python or R. The job description and recruiter will tell you.

How long should my take-home presentation be?

Follow the time you are given. If none is stated, aim for 10 to 15 minutes and leave the rest for questions. Lead with your recommendation in the first minute.

Will I write SQL on a whiteboard?

Some companies still use a whiteboard or a plain document where you cannot run the query. Practice writing queries without autocomplete and checking them by reading them aloud.

What should I do if my SQL query has an error during the interview?

Say what you think is wrong, check the joins and filters step by step, and fix it calmly. Interviewers often value clear debugging as much as a first-time correct query.

How many SQL questions are in a data analyst onsite?

Commonly two to four questions in one session, getting harder as you go. It varies by company, so ask the recruiter how the SQL round is structured.