Data analyst interview questions
Data analyst interviews test three skills: getting the right data (SQL), reasoning about it correctly (statistics and metrics), and turning it into a decision someone can act on. Most rounds include live SQL and at least one open business question.
The questions below are grouped by level, with what interviewers expect a strong answer to include.
What interviewers assess
- SQL: joins, aggregation, window functions and query correctness
- Statistics and metrics: averages, distributions, A/B tests and common traps
- Tools: Excel or Sheets, a BI tool such as Power BI or Tableau, and often Python or R
- Business sense: defining metrics and explaining findings to non-technical people
Basic questions
Explain the difference between INNER JOIN, LEFT JOIN and FULL OUTER JOIN.
What a good answer covers: Which rows each keeps, and what NULLs appear. A strong answer adds how a LEFT JOIN quietly becomes an inner join when you filter on the right table in WHERE.
What is the difference between WHERE and HAVING?
What a good answer covers: WHERE filters rows before grouping; HAVING filters groups after aggregation. Give a quick example such as customers with more than five orders.
When would you use the median instead of the mean?
What a good answer covers: When data is skewed or has outliers, such as income or order value. Mention looking at the distribution before choosing a summary.
How do you handle missing or duplicate data?
What a good answer covers: First find out why it's missing; then decide to drop, fill or flag based on the impact. For duplicates, define what makes a row unique. Say that you document what you changed.
What makes a good dashboard?
What a good answer covers: Built for a specific audience and decision, a few key metrics at the top, consistent definitions, and drill-down for detail. Too many charts is a common failure.
Intermediate questions
Write a query to find the second-highest salary in each department.
What a good answer covers: A window function such as DENSE_RANK() partitioned by department, filtered to rank 2. Discuss ties and why DENSE_RANK vs ROW_NUMBER matters.
Calculate a 7-day rolling average of daily sales in SQL.
What a good answer covers: AVG() OVER (ORDER BY date ROWS BETWEEN 6 PRECEDING AND CURRENT ROW), plus how you handle days with no sales (a date spine).
Our sign-ups dropped 20% last week. How would you investigate?
What a good answer covers: Check the data first (tracking broke?), then segment by channel, device, country and funnel step; compare with seasonality; form and test hypotheses. Interviewers want structure, not a single guess.
How would you define and measure customer churn?
What a good answer covers: Depends on the business: no purchase in N days, cancelled subscription, or inactivity. Discuss choosing the window, cohorts, and leading indicators.
Explain an A/B test and how you'd decide whether the result is real.
What a good answer covers: Random assignment, a primary metric chosen in advance, sample size, statistical significance and confidence intervals. Mention peeking and running too many metrics as traps.
Advanced questions
What is Simpson's paradox? Give an example.
What a good answer covers: A trend that appears in every subgroup reverses when groups are combined, because of different group sizes. Shows why segmenting matters before drawing conclusions.
How would you build a metric framework for a new product?
What a good answer covers: A north-star metric linked to user value, input metrics that drive it, guardrail metrics, and clear definitions everyone agrees on.
A query on a large table is slow. What do you check?
What a good answer covers: The execution plan, filtering early, indexes or partitions, avoiding SELECT *, joins that multiply rows, and pre-aggregated tables for dashboards.
How do you present a finding that contradicts what a stakeholder believes?
What a good answer covers: Check the analysis twice, lead with the business question, show the evidence simply, acknowledge limits, and suggest a next step. This tests communication as much as analysis.
When is correlation in your data not enough to recommend an action?
What a good answer covers: Confounders, reverse causation and selection bias; recommend an experiment or a quasi-experimental method where possible.
Common mistakes
- Writing SQL that runs but silently double-counts because of a join.
- Jumping to a conclusion in a case question without checking whether the data is right.
- Presenting numbers without saying what decision they support.
- Listing Power BI or Tableau without being able to describe a dashboard you built and who used it.
How to prepare
- Practise joins, GROUP BY and window functions until you can write them without notes.
- Prepare one analysis story: the question, the data, what you found and what changed because of it.
- Revise mean versus median, distributions, significance and A/B test basics.
- Practise structuring open questions out loud: clarify, segment, hypothesise, check.
Reading questions only goes so far: the real test is answering them aloud, under follow-up questions. Career Scan builds a mock interview from your own resume and the job, then shows exactly which answers need work.
Frequently asked questions
Is SQL the most important skill for data analyst interviews?
For most roles, yes. Expect at least one live SQL round. Excel and a BI tool are close behind, and Python is increasingly common.
Do data analyst interviews include case studies?
Often. You may get a metric drop to investigate, a dashboard to design, or a take-home dataset to analyse and present.
How technical is the statistics part?
Usually practical rather than mathematical: choosing the right average, reading an A/B test, and spotting misleading comparisons.