Data analyst interview questions: SQL, Excel, stats and sample answers

What do data analyst interviews assess?

A data analyst turns raw data into answers someone can act on. Interviews are built around that chain: finding the right data, cleaning it, analyzing it correctly and communicating the result. Each round tends to test one link, and a strong candidate shows they can carry a question all the way to a recommendation.

The U.S. Bureau of Labor Statistics profile of data scientists, a closely related occupation in its Occupational Outlook Handbook, describes the same chain: determining which data are available and useful, cleaning raw data so software can read it, presenting findings with visualization software and making business recommendations to stakeholders. It lists analytical, computer, communication and problem-solving skills as important qualities, including conveying results to technical and nontechnical audiences. Analyst interviews probe those qualities directly.

Many employers also use structured interviews, where every candidate gets the same questions and is scored against benchmarks. The U.S. Office of Personnel Management even uses “Describe a situation where you analyzed and interpreted information” as its example of a behavioral question, which is close to the core of this job. Read the posting closely as part of your interview preparation: the tools it names (SQL dialect, Excel, Tableau, Power BI, Python or R) and the teams it mentions tell you which rounds to expect.

A typical process includes a recruiter screen, a technical screen with live SQL or spreadsheet questions, a case study or take-home exercise, and a final round with the hiring manager and stakeholders. Smaller teams may combine these into one or two conversations.

SQL and database questions (and what a good answer shows)

SQL questions are usually asked live, in a shared editor or on a whiteboard, against a small schema such as customers, orders and products. Interviewers care about correctness and reasoning more than memorized syntax. Before writing, restate the question, name the tables and keys you will use and say how you will handle nulls, duplicates and ties.

Window functions come up often because they separate analysts who can only aggregate from analysts who can compare rows. The PostgreSQL documentation explains the key idea: a window function calculates across a set of related rows, like an aggregate, but does not collapse them into a single output row, so each row keeps its identity. It also notes that window functions cannot be used in a WHERE clause, so filtering on a rank requires a subquery, a detail interviewers like to probe.

• “What is the difference between an inner join and a left join?” Shows you know a left join keeps every row from the left table and fills missing matches with nulls, and when each is the right choice.
• “Why did your row count go up after a join?” Shows you recognize a one-to-many relationship duplicating rows and know to check the grain of each table first.
• “What is the difference between WHERE and HAVING?” Shows you understand that WHERE filters rows before grouping and HAVING filters groups after aggregation.
• “Find the top three products by revenue in each category.” Shows you can rank within groups with a window function, filter in an outer query and explain how you handle ties.
• “Calculate each customer’s running total of spend over time.” Shows you can partition by customer and order by date inside a window rather than writing a self-join.
• “How would you find duplicate records in a table?” Shows you group by the columns that define a duplicate, count them and ask which record should survive.
• “How does COUNT(column) differ from COUNT of all rows?” Shows you know counting a column skips nulls, which quietly changes averages and rates.
• “A query that used to run in seconds now takes minutes. What do you check?” Shows practical habits: filtering early, selecting only needed columns, checking join keys and asking about indexes or partitions.

Excel, statistics and data cleaning questions

Spreadsheet questions are common for business-facing analyst roles, and some employers hand you a workbook to work through on screen. Statistics questions are usually conceptual rather than mathematical: interviewers want to know you will not mislead anyone. Data cleaning questions test the judgment behind every analysis you will ever deliver.

• “When would you use XLOOKUP or INDEX-MATCH instead of VLOOKUP?” Shows you know lookup limitations, such as VLOOKUP only searching to the right and breaking when columns move.
• “How would you summarize 50,000 rows of sales by region and month?” Shows you reach for a pivot table, and that you check the source range and blanks before trusting the totals.
• “Mean or median for reporting typical order value?” Shows you check the distribution and prefer the median when a few very large orders skew the mean.
• “Sales and ad spend move together. Did the ads work?” Shows you separate correlation from causation and suggest seasonality, other campaigns or a controlled test as explanations to rule out.
• “Explain a p-value to a marketing manager.” Shows you can say plainly how surprising the result would be if there were no real effect, without overstating certainty.
• “How would you set up and read an A/B test?” Shows you decide the metric and sample size before launch, randomize properly and avoid stopping early when a result looks good.
• “You receive a file with missing values, inconsistent date formats and duplicate IDs. Walk me through it.” Shows a repeatable process: profile the data, decide rules with the data owner, document every change and keep the raw file untouched.
• “How do you handle outliers?” Shows you investigate before deleting, since an outlier may be an entry error or the most important record in the file.

Dashboard, stakeholder and behavioral questions

These questions decide many final rounds, because a technically correct analysis that nobody uses has no value to the business. Answer them with real stories, the way you would in any set of behavioral interview questions: one specific project, your own actions and an honest result.

• “Tell me about a dashboard you built. Who used it and what changed?” Shows you started from the decisions users needed to make, not from the available charts.
• “How do you decide which chart to use?” Shows you match the chart to the comparison: lines for trends, bars for categories and a table when exact values matter.
• “Describe a time you presented a finding to a non-technical audience.” Shows you led with the answer and the decision it supports, and kept method details for questions.
• “Tell me about a time your analysis contradicted what a stakeholder expected.” Shows you double-checked your work, shared the evidence respectfully and stayed open to context you lacked.
• “Describe a time you found an error in data or in your own report after sharing it.” Shows ownership, a fast correction to everyone who saw it and a check you added afterward.
• “A manager asks for ‘all the data on churn’ by Friday. What do you do?” Shows you clarify the decision behind the request and agree on a smaller, useful first deliverable.
• “Why data analysis, and why this company?” Shows genuine interest in the questions this business needs answered, linked to evidence from your own projects.

How do you approach a data analyst case study or take-home exercise?

Case questions give you a business problem, such as “Weekly sign-ups dropped this month. How would you investigate?”, and watch how you structure it. Take-home exercises give you a dataset and a deadline, then ask you to present results. Both are graded on reasoning and communication as much as on the final number, so make your thinking visible.

1. Restate the business question and ask what decision the answer will inform, who the audience is and how much time is expected.
2. Check the data before analyzing: row counts, date ranges, missing values, duplicates and whether the metric is defined the way you assume.
3. For a metric drop, rule out data problems first, such as tracking changes or late-arriving records, then break the metric down by segment, channel, region and date.
4. Form two or three hypotheses and test the most likely or most costly first, instead of slicing every column.
5. Lead the write-up with the answer and a recommendation, then show the two or three charts that support it, then list assumptions and limitations.
6. Keep your work reproducible: comment queries, separate raw and cleaned data, and note anything you would do with more time.

Three worked sample answers for data analyst interviews

The answers below are invented for illustration, including every number in them. They show structure and a useful level of detail; build yours from your own projects, coursework or volunteer work, and never claim a result you did not produce. The STAR method answer generator can help you turn rough notes from a real project into the same shape.

Questions to ask in a data analyst interview

Your questions show how you think about data work, and the answers tell you whether the job is analysis or mostly manual report pulling. Pick two or three that matter to you; the questions to ask in an interview guide covers general ones, and the questions to ask the interviewer generator can tailor a list to the posting.

• “What decisions has this team’s analysis influenced recently?”
• “Where does the data come from, and who owns data quality when something looks wrong?”
• “How much of the role is recurring reporting versus new analysis?”
• “Which tools does the team use day to day, and how are queries and dashboards reviewed?”
• “Who are the main stakeholders, and how do requests reach the team?”
• “What would a successful first 90 days look like for this analyst?”

Common data analyst interview mistakes and how to prepare

Most weak answers fail on reasoning or communication rather than technical knowledge. These are the patterns to avoid.

To practice with questions closer to your real interview, paste the actual job posting into the interview question generator to get questions for that role and interview stage, whether a phone screen, a hiring manager conversation, a panel or a final round. Rehearse SQL out loud against a sample database, and after each interview, send a short note referring to something specific you discussed, as covered in thank-you emails after an interview.

• Mistake: writing SQL in silence. Fix: talk through the grain of each table, your join choice and how you handle nulls before typing.
• Mistake: skipping data checks in a case. Fix: say what you would verify first, even when the interviewer hands you a clean table.
• Mistake: listing tools instead of outcomes. Fix: name the question, what you found and what someone did with it.
• Mistake: overstating certainty. Fix: give your confidence, the main limitation and what would change your conclusion.
• Mistake: using company figures from a past employer. Fix: describe the scale or direction of a result without sharing confidential numbers.