Data analyst interview questions and answers by stage

Data analyst interview questions and answers usually come in five stages: a recruiter screen, a SQL test, a case or metrics round, a statistics round and a behavioral round. Below are 25 typical questions with how to approach each, two worked examples, an entry-level section and a 3-week plan.

What stages do data analyst interview questions come in?

Formats differ by company, so confirm yours with the recruiter. A common loop looks like this.

StageWhat it testsHow to prepare
Recruiter screenFit, motivation, tools you knowA 60-second story and two projects
SQL roundJoins, aggregation, window functions, NULLsWrite queries daily on real tables
Case or metrics roundStructuring a vague business problemA repeatable framework (below)
Statistics roundExperiments, distributions, biasPlain-language definitions with examples
Behavioral roundCommunication and stakeholder workFour stories in STAR form

Some loops add a take-home analysis or a dashboard exercise. The same habits apply.

Screen questions

Bring a CV or resume that lists two or three projects you can explain end to end.

  1. Walk me through a project you are proud of. Question, data, method, finding, decision. Spend most of the time on the finding and what changed.
  2. Why data analysis, and why this company? Name one thing you read about their product or data. Keep it to 30 seconds.
  3. Which tools do you use? Name SQL first, then what you used it for. Say which tools are strong and which are learning. Honesty here saves a bad technical round.
  4. What is your experience with dashboards? Describe one you built: who used it, what decision it fed, and one thing you cut to keep it simple.

For the opener, adapt a structure from tell me about yourself answer examples.

Data analyst interview questions: SQL

This is the round most people fail by rushing. Our SQL interview questions guide has 25 worked queries. The five that appear most for analysts:

  1. Find the second highest salary or the top N per group. Use DENSE_RANK or ROW_NUMBER with PARTITION BY.
  2. Find customers with no orders. LEFT JOIN, then keep rows where the right side is NULL.
  3. Calculate month-over-month change. Aggregate by month in a CTE, then LAG.
  4. What is the difference between WHERE and HAVING? Rows before grouping, groups after.
  5. Why does this join return too many rows? Check the grain. A one-to-many join repeats rows, so sums inflate. Count rows before and after the join.

Interviewers listen for edge cases: NULLs, ties, duplicates, empty groups. Say them before you are asked.

Case and metrics questions

  1. A key metric dropped 10% this week. What do you do? See worked example 1.
  2. How would you measure the success of a new feature? Define the goal, pick one primary metric, add guardrail metrics, say how you would compare users with and without it.
  3. Which metrics would you put on a dashboard for [team]? Ask who uses it and what they decide. Offer three to five metrics, not twenty.
  4. How would you define an active user? There is no single answer. State a definition (for example, completed a core action in 28 days), then say how another definition would change the number.
  5. Two sources show different revenue totals. How do you reconcile them? Compare definitions (gross or net, time zone, refunds), then row counts, then individual records for one day.

Worked example 1: orders fell 10% last week

"First I would confirm the drop is real: is the data complete, did a tracking change ship, and is last week unusual against the last eight weeks? Then I would split the metric: by region, device, channel, new versus returning customers and product. A drop in one slice points to a cause. A drop everywhere points to something global, such as a site outage, pricing or seasonality. I would line the timing up with releases, campaigns and external events. Then I would propose a next step: if it is one channel, I would talk to that channel's owner. I would report what I found, how sure I am, and what I would check next."

Why it works: verify, segment, correlate with events, recommend. You never need the real answer. You need a clear method.

Statistics questions

  1. What is a p-value? The probability of seeing a result at least this extreme if there were no real effect. It is not the probability that the result is true.
  2. Mean versus median: when do you use each? The median holds up against outliers, so it suits skewed data such as incomes or session times.
  3. How would you design an A/B test? Define the metric and the minimum effect that matters, size the sample beforehand, randomize, run for full weekly cycles and do not stop early on a good-looking day.
  4. What is the difference between correlation and causation? Give a quick example, such as ice cream sales and sunburn, both driven by hot weather.
  5. What is selection bias, and how do you spot it? The sample differs from the population you care about. Ask how rows got into the data.
  6. How do you handle missing data? First ask why it is missing. Then drop, fill or flag it, and say how each choice changes the result.

Worked example 2: reading an A/B test

"Say version B converts at 4.4% and version A at 4.0%, with [number] users in each. B looks better, but I would check three things before saying so. One: was the sample size set in advance, and is it large enough that a 0.4 point gap is not noise? Two: did the test run long enough to cover weekday and weekend behavior? Three: did anything else change for one group, such as traffic source? If the result is significant and those checks pass, I would also look at guardrail metrics, like refund rate, before recommending a launch. I would also say how big the gain is in money or users, because a significant result can still be too small to matter."

The numbers are invented for the example. Use any plausible pair and keep the focus on the method.

Behavioral questions

  1. Tell me about a time you explained data to a non-technical person. Show you changed the message for the listener: fewer terms, one chart, one recommendation.
  2. Tell me about a time your analysis was wrong. Own it, say how you found it, say what check you added afterward.
  3. How do you handle a stakeholder who wants a specific answer? Agree on the question first, show the data as it is, and discuss limits openly.
  4. How do you prioritize when several people ask for reports? Ask about the decision each report feeds and its deadline. Say how you tell people what you are putting off.
  5. Describe a time data changed a decision. Best answer: the recommendation, the action taken, the measured effect.

Prepare each as a STAR story. See STAR method interview examples and the behavioral interview questions list.

Entry level data analyst interview questions for freshers

For entry level data analyst interview questions, including those for freshers, the bar is clean fundamentals and clear thinking, not years of experience. Expect the same stages with easier cases and more weight on projects.

  • "Walk me through a project." Use a course, personal or volunteer project. Show you cleaned data, checked it and drew a conclusion.
  • "How do you clean a messy dataset?" Look at types, duplicates, missing values, outliers and inconsistent labels. Say you document each change.
  • "What is the difference between a JOIN and a UNION?" JOIN adds columns by matching keys. UNION stacks rows.
  • "Which chart would you use for X?" Lines for time, bars for categories, and a table when exact values matter.
  • "What would you do in your first 30 days?" Learn the data sources and definitions, meet the people who use the reports, and fix one small thing.

With no job history, a good portfolio project has a real question, a public dataset, a written finding and the code. Be ready to say what you would do differently.

Example answer for a fresher: walk me through a project

"For my course project I asked which [product category] sells best in each [region]. I took a public sales dataset, removed [number] duplicate rows, and filled or dropped missing values after checking why they were missing. I joined it to a regions table in SQL, then built a bar chart and a one-page summary. The finding was that [category] led in three regions but not in [region]. I would now test whether that gap holds across several years."

Adapt the details and keep the order: question, cleaning, method, finding, next step.

Data analyst interview questions for experienced candidates

For experienced candidates, the questions move from "can you do it" to "what changed because you did it". Expect:

  • "Tell me about an analysis that changed a decision." Give the recommendation, the action and the measured effect.
  • "How do you define and protect a metric?" Cover the definition, the source of truth, data quality checks and who owns changes.
  • "How do you say no to a request?" Tie it to the decision the report would feed and offer a smaller version.
  • "How do you work with engineers on bad data?" Report the problem with a reproducible example, not a complaint.
  • "How have you grown other analysts?" Reviews, templates, shared definitions.

Bring numbers from your own work: time saved, revenue affected, error rate before and after. Do not invent them.

What are your top 3 skill sets as a data analyst?

Name three that match the job posting, with proof for each. A safe set is SQL, statistical thinking and clear communication. Say how you used each on a real project, and what a stakeholder did with the result.

"My top three are SQL, statistics and communication. SQL, because I pull and check my own data, for example [project]. Statistics, because I test whether a change is real before I report it. Communication, because the work only matters when someone acts on it, so I end every analysis with one recommendation."

How can I prepare for a data analyst interview?

Prepare in three weeks: SQL first, then metrics cases and statistics, then stories and mock interviews. Read the job posting, list its tools, and practice every answer out loud with a timer. Use this plan.

WeekFocusDaily work
1SQLTwo or three queries a day: joins, GROUP BY, then window functions. Rewrite yesterday's from memory.
2Cases and statisticsOne metrics case out loud a day. Write plain-language answers for the six statistics questions above.
3Stories and mocksWrite four STAR stories. Do two timed mock interviews. Re-read the job posting and match your examples to it.

Practice out loud. Typing an answer and saying it are different skills, and the interview needs the second one.

A live option if you freeze

Practice does more than any tool, and your employer's and interviewer's rules come first. If help is allowed, Hintly is a desktop app for Mac and Windows that listens to a call and shows live answers from your own notes, with the overlay left out of screen shares. It is pre-launch, and the way in is the waitlist. Read more on the AI interview assistant page.

Questions

What are the most common data analyst interview questions?

Expect SQL (joins, GROUP BY, window functions), a metrics or case question such as why a number dropped, basic statistics (A/B tests, p-values, averages versus medians) and behavioral questions about explaining findings to non-technical people. Many companies also set a take-home task or an Excel or BI exercise.

How many rounds does a data analyst interview have?

It varies by company. A common shape is a recruiter screen, one technical round (SQL, sometimes Python or Excel), a case or business round, and a final round with the hiring manager and team. Ask the recruiter for the exact format and tools.

Do I need Python or R for a data analyst interview?

SQL is almost always required. Python or R depends on the job posting. If the posting lists one, expect a question or a short task in it. If it lists only Excel and a BI tool such as Tableau or Power BI, prepare those instead.

What should I bring to a data analyst interview?

Bring two or three projects you can explain end to end: the question, the data, your method, what you found and what changed because of it. A short portfolio link helps, but you must be able to talk about every line of it.

How do I answer if I do not know the business domain?

Ask clarifying questions. What does the metric mean, who uses it, and what decision does it feed? Interviewers are testing how you structure the problem, not whether you already know their industry.

Your next interview, handled.

Hintly is almost ready. Leave your email and we’ll tell you the day you can download it.

Mac and Windows · No spam