Question 1
Which metric in this posting would you trust, and which would you refuse to put on a dashboard?
Approach. Define the trustworthy metric in one sentence and say how it can be gamed.
Data analyst interview questions from a job description
Turn a data analyst job description into likely SQL, metric, and stakeholder questions, grouped by the line that triggered them.
Matched in your browser from the requirements in the posting and Luna’s interview-question hub. These are practice prompts, not a claim that this wording will be asked.
Requirements in the posting
Data Analyst
Turn product questions into numbers a team can use.
You will:
- Write SQL against a warehouse and explain the grain of the result
- Build dashboards in Looker or Tableau, and retire a KPI that makes the team worse
- Use Excel when a stakeholder needs a check they can follow
- Support experiments with a clear hypothesis and the caveat
- Collaborate with product managers who will challenge the number
We will ask you to walk through an analysis that changed a decision.
Turn product questions into numbers a team can use.
You will:
- Write SQL against a warehouse and explain the grain of the result
- Build dashboards in Looker or Tableau, and retire a KPI that makes the team worse
- Use Excel when a stakeholder needs a check they can follow
- Support experiments with a clear hypothesis and the caveat
- Collaborate with product managers who will challenge the number
We will ask you to walk through an analysis that changed a decision.
Likely questions
Role
Analytics
Data Analyst
Question 2
A stakeholder disagrees with your number. How do you walk them through it?
Approach. Start from the grain of the data, then the filter, then the decision the number can support.
Question 3
Tell me about an analysis that changed a roadmap, not just a slide.
Approach. Name the decision that moved and the check you left so the team could see if it worked.
Role
Experiments and statistics
Support experiments with a clear hypothesis and the caveat
Question 4
What would you experiment on in the first month of this job, and what metric would decide it?
Approach. One change, one metric, and the confound you would not ignore.
Question 5
Tell me about a result that was statistically neat and still the wrong product call.
Approach. Say who the metric left out and the decision you made anyway.
Question 6
How do you explain uncertainty to a partner who wants a yes or no?
Approach. Give the range, the decision it still supports, and what more data would not change.
Coding
SQL and data modeling
Write SQL against a warehouse and explain the grain of the result
Question 7
What tables would you need for the core object in this posting, and what is the grain of each?
Approach. Name the entity, the key, and one query you must keep fast.
Question 8
Write the shape of a query this team would run every day. Where would it get slow?
Approach. Say the join, the filter, and the index or rewrite you would try first.
Behavioral
Collaboration
Collaborate with product managers who will challenge the number
Question 9
Tell me about a disagreement with someone who did not report to you. What did you do with their objection?
Approach. Name the objection, what you changed, and what you did not. Do not flatten it into “we aligned.”
Question 10
How do you bring a partner along when this posting’s work depends on their team?
Approach. Say the ask, the artifact you left them, and what you refused to take over.
What to prepare
- Bring one chart you would delete because it made a team worse, and what replaced it.
- Write one experiment: the metric, the unit of randomization, and the result that would stop you.
- Practice one join out loud: the grain of the result, then the filters.
- Have one story where someone disagreed and you kept the working relationship.
- Mark the three requirements you cannot yet support with a story.
- Say each answer out loud once, in under two minutes.
Analyst postings are judged on whether you can define a metric and defend its grain. SQL is the screen. Dashboards are the portfolio. Stakeholder bullets are the “someone disagrees with your number” question. Experiment language means they will ask what the test cannot prove.
A long tool list (Looker, Tableau, Excel) is one theme: can you get a trustworthy number in front of a decision. You do not need a story for every logo. You need one chart you would delete.
Practice a query out loud in layers: grain, join, filter, check. The sample is a product-analytics role.
Sample job description
The posting below is already in the tool. The questions are the ones its requirements point to. Replace it with a posting you are actually interviewing for.
Related: Airbnb role-specific questions · SQL interview pacing · Microsoft coding questions