Open app

Interview questions · Data analyst

Data analyst interview questions and how to answer them

Updated 3 November 2026

A data analyst interview usually mixes three kinds of question: technical ones (SQL, spreadsheets, basic statistics), a small case where you work through a business problem out loud, and behavioural ones about explaining your findings. Many companies also send a separate SQL or take-home test. These are the questions that come up most, with what the interviewer is checking and an example answer.

Reading answers helps, but saying them out loud is what makes them stick. When you are ready, run a free mock interview with these questions.

Try 3 questions now. No sign-up.

Answer out loud, like in the real interview, and get instant feedback on each answer.

The 12 questions

  1. Walk me through a data analysis project from start to finish.
  2. How do you handle missing or messy data?
  3. What is the difference between an INNER JOIN and a LEFT JOIN?
  4. How would you find duplicate rows in a table?
  5. What is the difference between WHERE and HAVING?
  6. Sales dropped 10% last month. How would you investigate?
  7. How would you explain a technical finding to someone non-technical?
  8. Tell me about a time your analysis changed a decision.
  9. Which tools do you use, and how do you choose a chart?
  10. What is the difference between correlation and causation?
  11. Tell me about a time you found a mistake in your own work.
  12. How do you prioritise requests from several teams?

1. Walk me through a data analysis project from start to finish.

What they are checking: Whether you understand the whole process, not only the tools.

How to answer: Go in order: the business question, where the data came from, how you cleaned it, what you analysed, what you found, and what changed because of it. Keep tools in the background; the decision is the point.

Example answer: “Our support team wanted to know why tickets were rising. I pulled six months of tickets with SQL, cleaned duplicate entries, and grouped them by category and product version. Most of the rise came from one release, so the product team fixed that flow and tickets dropped back the next month.”

2. How do you handle missing or messy data?

What they are checking: Whether you check data before trusting it.

How to answer: Explain that you first find out why it is missing, then choose: remove rows, fill values, or flag them, depending on how much is missing and what the analysis is for. Say you note what you did so others know.

Example answer: “First I check how much is missing and whether it is random or follows a pattern, like one store never sending data. If it's a few random rows I usually exclude them. If it's a pattern, I talk to whoever owns the data, because filling it in would hide a real problem. I always write down what I changed.”

3. What is the difference between an INNER JOIN and a LEFT JOIN?

What they are checking: Basic SQL, and whether you know when each one matters.

How to answer: An INNER JOIN keeps only rows that match in both tables. A LEFT JOIN keeps every row from the left table and fills the right side with NULL when there is no match. Give a use for each.

Example answer: “An inner join gives me only customers who have placed orders. A left join gives me all customers, with NULLs for the ones who never ordered, which is what I'd use to find customers who signed up but never bought.”

4. How would you find duplicate rows in a table?

What they are checking: Practical SQL you will use every week.

How to answer: Group by the columns that should be unique and keep groups with more than one row: GROUP BY email HAVING COUNT(*) > 1. Mention that you would check why duplicates exist before deleting anything.

Example answer: “I'd group by the columns that should be unique, say email, and use HAVING COUNT(*) greater than one. Then I'd look at a few duplicates to understand where they came from, because a sign-up bug and a data import mistake need different fixes.”

5. What is the difference between WHERE and HAVING?

What they are checking: Whether you understand when filtering happens in a query.

How to answer: WHERE filters rows before grouping. HAVING filters groups after aggregation, so it can use results like COUNT or SUM.

Example answer: “WHERE filters individual rows, for example orders from 2026. HAVING filters the grouped results, for example only customers whose total spend is over 500.”

6. Sales dropped 10% last month. How would you investigate?

What they are checking: How you structure a problem you have never seen.

How to answer: Think out loud. Check the data is right first, then break the drop down: by region, product, channel, new versus returning customers. Look for a change that lines up in time, like a price change or a site outage. Say what you would check first and why.

Example answer: “First I'd confirm the drop is real and not a tracking problem. Then I'd split sales by region, product and channel to see where it is concentrated. If it's all one channel, I'd check what changed there last month, like a campaign ending or a checkout change, before looking wider.”

7. How would you explain a technical finding to someone non-technical?

What they are checking: Whether your work will actually be used.

How to answer: Lead with what it means for them, then the one number that supports it, then what they should do. Use one simple chart, not five. Save the method for questions.

Example answer: “I'd start with the decision: we should move budget from channel A to channel B. Then one chart showing B brings customers at half the cost. I keep the method in the appendix and only go there if someone asks.”

8. Tell me about a time your analysis changed a decision.

What they are checking: Impact, not only activity.

How to answer: Use the STAR order: situation, task, action, result. Make the result a decision or a change, and say what you personally did.

Example answer: “Our team planned to discount a product that was selling slowly. I found it sold well in two regions and almost nothing elsewhere because it was not stocked there. We fixed distribution instead of discounting, and sales rose without cutting the price.”

9. Which tools do you use, and how do you choose a chart?

What they are checking: Whether your skills match their stack, and whether you think about the audience.

How to answer: Be honest about your level with each tool (spreadsheets, SQL, Python or R, Tableau or Power BI). For charts, match the chart to the question: trends over time as lines, comparisons as bars, parts of a whole only when there are few parts.

Example answer: “Day to day I use SQL and Excel, and Power BI for dashboards. I'm comfortable with pandas in Python for cleaning. For charts I pick based on the question: lines for change over time, bars for comparing groups, and I avoid pie charts with more than three or four slices.”

10. What is the difference between correlation and causation?

What they are checking: Whether you will over-claim what the data shows.

How to answer: Correlation means two things move together. Causation means one causes the other. Give an example where a third factor explains both, and say how you would test cause, such as an A/B test.

Example answer: “Ice cream sales and sunburn rise together, but neither causes the other; hot weather drives both. If I needed to show that a change caused an effect, I'd want an A/B test or at least a fair comparison group.”

11. Tell me about a time you found a mistake in your own work.

What they are checking: Honesty, and whether you have habits that catch errors.

How to answer: Pick a real mistake, say how you found it, how you fixed and communicated it, and what you check now so it does not happen again.

Example answer: “I once shared a report where revenue was double-counted because of a join. I noticed when the total didn't match finance's number, told my manager the same day, and sent a corrected version. Now I always reconcile totals against a known source before sharing.”

12. How do you prioritise requests from several teams?

What they are checking: Whether you can manage your own workload.

How to answer: Explain how you weigh impact and urgency, and that you are open with people about timing instead of quietly falling behind. Mention agreeing priorities with your manager when two requests clash.

Example answer: “I ask what decision each request feeds and when that decision happens. Urgent, high-impact work goes first, and I tell the others when to expect theirs. If two teams both say urgent, I bring it to my manager rather than guessing.”

Practise these out loud. Enter the company and the role, paste your CV, and Nail My Interview plays the interviewer: it asks questions like these, listens to your answers and scores them afterwards. Your first full mock interview is free.

Start a free mock interview

More interview questions