Data analyst interview questions

Data analyst interviews mix three things: technical checks (usually SQL, sometimes Python or Excel), a case where you turn a vague business question into an analysis, and questions about how you explain results to people who do not work with data. The strongest answers always end with the decision the analysis changed.

8 questions with sample answers

  1. Tell me about an analysis that changed a business decision.

    Sample answer

    Our subscription team wanted to cut prices because churn was rising. I segmented churn by acquisition channel and found it was concentrated in users from one discount campaign, who left at three times the normal rate. Instead of a price cut, we stopped that campaign and redesigned onboarding for discount users. Churn fell from 7.5% to 5.8% a month within a quarter.

    What the interviewer is checking: Whether your work leads to action, and whether you can tell the story in business terms rather than as a list of queries.

  2. What is the difference between INNER JOIN and LEFT JOIN, and when would you use each?

    Sample answer

    An INNER JOIN keeps only rows that match in both tables. A LEFT JOIN keeps every row from the left table and fills in NULL where there is no match. If I'm listing customers with their orders and want to include customers who have never ordered, I use a LEFT JOIN and then filter or count the NULLs, which is also how I find inactive customers.

    What the interviewer is checking: SQL fundamentals and whether you can connect syntax to a real question.

  3. How do you handle missing or messy data?

    Sample answer

    First I find out why it's missing, because that decides the fix. A tracking bug on one app version is different from customers skipping an optional field. Depending on the cause, I exclude the rows, fill values with a documented rule or flag the gap in the report. I always note how many rows were affected so nobody reads the result as more precise than it is.

    What the interviewer is checking: Judgement and honesty about data quality, not just technical cleaning steps.

  4. How would you investigate a sudden 20% drop in daily active users?

    Sample answer

    I'd check whether the drop is real before anything else: a tracking change, a broken dashboard filter or a public holiday. If it's real, I'd split it by platform, country, app version and acquisition channel to see where it's concentrated. A drop limited to one Android version after a release points to a bug; a drop across everything points to something external like an outage or seasonality.

    What the interviewer is checking: A structured approach that starts with data validity and narrows down by segment.

  5. How do you explain a complex finding to a non-technical audience?

    Sample answer

    I lead with the answer and the recommendation, then show one chart that proves it, and keep the method in an appendix. For a recent pricing review I opened with "customers on annual plans are worth 2.4 times more, so we should push annual at checkout", followed by a single cohort chart. The detail was there for anyone who wanted it.

    What the interviewer is checking: Communication. Analysts who bury the conclusion under methodology lose their audience.

  6. How would you set up and read an A/B test?

    Sample answer

    I agree the primary metric and the minimum effect worth detecting up front, then calculate the sample size so the test runs long enough, usually full weeks to cover weekday patterns. I don't stop early when the result looks good. When it ends, I check the confidence interval, not only the p-value, and look at guardrail metrics like refunds to make sure we didn't win one number by losing another.

    What the interviewer is checking: Statistical literacy and awareness of the common traps: peeking, underpowered tests and ignoring side effects.

  7. Which tools do you use, and why?

    Sample answer

    SQL for almost all extraction, Python with pandas when the logic gets too complex for SQL or I need to automate a recurring report, and Power BI for dashboards that business teams use daily. I pick the simplest tool that the people maintaining the work after me will understand.

    What the interviewer is checking: Fit with their stack, and whether you choose tools for the problem rather than for your CV.

  8. Tell me about a time you found an error in your own work.

    Sample answer

    Situation: I had shared a revenue dashboard with leadership. Task: two days later I noticed refunds were being counted as sales. Action: I told my manager that morning, fixed the query, added a reconciliation check against the finance ledger and sent a short correction note. Result: the numbers were corrected before any decision used them, and the reconciliation check has caught two more issues since.

    What the interviewer is checking: Integrity and ownership. They want to hear that you raise mistakes fast and prevent repeats.

How to prepare

  • Practise SQL by hand: joins, GROUP BY with HAVING, window functions and date handling.
  • Prepare two stories where your analysis changed a decision, each with a number.
  • Look up the company's product and guess its key metrics, so you can discuss them in a case question.
  • Have one dashboard or notebook you can describe from question to recommendation.

Questions to ask the interviewer

  • Who are the main stakeholders for this role, and how do they request analysis?
  • How reliable is the data today, and who owns data quality?
  • Which decision would you most like this role to inform in its first six months?
Practise this interview with AppliQuestions built from your job posting and a spoken mock interview, with feedback on every answer in the Apply4Me app.