A data analyst mock interview with AI works best as three short sessions rather than one long one: SQL and spreadsheets, a business case with a chart, and stakeholder and behavioral questions. That mirrors what analyst interviews check, which is less about algorithms and more about getting the numbers right and explaining them to someone who doesn't live in the data. A chatbot can set the questions, play a confused stakeholder and critique your explanations; you still write the SQL and build the pivot table yourself, in real tools.
This guide covers what analyst loops test, 12 questions to rehearse, the three sessions with copy-paste prompts, a worked business case, a 60-second explain-it drill, a rubric and a two-week plan.
What data analyst interviews check
| Skill | How it's usually tested | What to mock |
|---|---|---|
| SQL | Live queries or a timed online test: joins, aggregation, window functions | Two queries per session, in your own editor |
| Spreadsheets | Lookups, pivot tables, cleaning a messy column; sometimes a live Excel task | A real workbook with a few hundred messy rows |
| Visualization and BI | Choosing a chart, critiquing a dashboard, explaining a Tableau or Power BI build | Describe the chart you'd build and why, then build it |
| Business case | “Revenue fell; why?” or “Which metrics would you track for this team?” | One case per session, structured out loud |
| Communication | Explaining a finding to a non-technical stakeholder | The 60-second drill below |
| Behavioral | Deadlines, messy data, numbers that don't match someone else's | Two STAR stories per session |
Analyst loops commonly run two to five interviews after a recruiter call, often with a SQL test or screen first; our data analyst onsite interview questions page walks through a typical final round. One employer publishes its first step: Amazon's business intelligence engineer process, as described on its hiring pages (checked by our team November 17, 2026), starts with an online assessment that includes a SQL challenge. Ask your recruiter whether any round includes a live Excel or SQL task, and in which tool.
12 questions to rehearse
SQL (in your own editor)
- Monthly active users for each of the last six months.
- The top three products by revenue in each region.
- Customers whose first order was in January and who ordered again within 30 days.
Spreadsheets
- You need to match two sheets on customer ID, but some IDs have trailing spaces. What do you do? (Clean with TRIM, then look up, then count the rows that still don't match.)
- Build a pivot of revenue by month and channel, then tell me one thing it shows.
- VLOOKUP, XLOOKUP or INDEX/MATCH: when would you use each? (VLOOKUP only looks to the right and defaults to approximate match; XLOOKUP looks either way and defaults to exact match but isn't in older Excel versions; INDEX/MATCH works everywhere.)
Visualization
- Which chart would you use to show conversion through a five-step signup funnel, and why not a pie chart?
- A manager says your dashboard is “confusing.” What do you check first?
Business case
- Weekly orders fell 8%. What do you do? (Worked below.)
- Which three metrics would you put on a dashboard for a customer-support team, and what would make you worried?
Stakeholders and behavior
- Tell me about a time your numbers didn't match someone else's.
- Explain a finding to a manager who has five minutes. (The drill below.)
Written answers for most of these are on our data analyst interview questions page; for the tool questions, see Excel interview questions, Tableau interview questions and Power BI interview questions.
Run it as three 20- to 25-minute sessions
| Session | Time | Content | Open on your desk |
|---|---|---|---|
| A: numbers | 25 min | Two SQL questions and one spreadsheet task | SQL editor, a spreadsheet, a timer |
| B: case and chart | 25 min | One business case, one chart choice, one dashboard critique | Paper for your structure |
| C: people | 20 min | The stakeholder drill, two behavioral questions, your questions for them | Story notes, closed while you answer |
Short sessions keep each skill sharp and fit around a job. Run all three back to back only in the final week, when you want to feel the full loop.
Prompt for session A
You're interviewing me for a data analyst role at [type of company]. The data available: [paste table names and columns]. Give me two SQL questions and one spreadsheet task of the kind asked in analyst interviews, one at a time. I'll do them in my own tools and paste my query or describe my formula. Then ask one follow-up about an edge case, such as duplicates, blanks or a date boundary. Don't solve anything for me.
Why it works: you get interview-style tasks without the answers, and the edge-case follow-up mimics the second question a real interviewer asks.
Prompt for session B
Play a hiring manager giving me a business case for a data analyst interview. State the situation in two sentences, with no data. Answer my clarifying questions with realistic, made-up numbers, and reveal a breakdown only when I ask for it by name. If I jump to a cause without checking the data first, ask how I know. End by asking for my recommendation in one sentence.
Why it works: revealing data only on request rewards a structured search instead of guessing.
Prompt for session C
Play a busy sales director with no analytics background. I'll explain a finding in under 60 seconds. Interrupt once with a skeptical question such as 'So what do I do on Monday?' Afterwards, tell me which sentence lost you, if any, and which word you'd want explained. Don't rewrite my explanation.
Why it works: it tests the skill most analyst interviews weigh heavily, translation, and leaves the rewriting to you so the words stay yours.
Worked example: “Weekly orders fell 8%. What do you do?”
Weak answer: “I'd look at the data to find out why. Maybe there was a marketing change or a bug. I'd make a dashboard of orders over time, talk to the team, then find the root cause and fix it.”
Strong answer: “First I'd check the drop is real: did tracking, the pipeline or the definition of an order change, and is the week complete? Then I'd compare with the same week last year to rule out seasonality. If it's real, I'd split the 8% by channel, platform, region and new versus returning customers, and walk the funnel: visits, product views, add-to-cart, checkout, purchase. If the drop sits in one place, say the app's checkout step after Tuesday's release, I'd take that to the product team with the size of the gap. My update to the manager the same day would be what we know, what we're checking next and when they'll hear from me.”
What changed: the strong answer checks the data before blaming a cause, splits the drop in a fixed order, and ends with how the finding gets communicated. In your mock, ask the AI for the follow-up interviewers love: “Every segment dropped about 8%. Now what?” An even drop points at something that affects everyone, such as tracking, seasonality, a price change, a site-wide problem or a market-wide event, rather than one broken step.
The 60-second stakeholder drill
- Pick a real finding from your past work or a portfolio project. Remove employer names, client details and confidential numbers before you paste anything into a chatbot.
- Say it in three parts: the answer, one number as evidence, the action you recommend.
- Run the session C prompt and time yourself. Over 60 seconds means you started with the method.
- Rework the sentence the AI says lost it, yourself, and repeat until two runs in a row land.
Before (an example we wrote): “So I ran a cohort analysis on second-quarter signups with a 30-day window, and when you look at the retention curves by acquisition channel and first-week behavior, there's a pattern where…”
After: “Customers who used the mobile app in their first week stayed with us longer: about twice as many were still ordering after 30 days. I'd recommend promoting the app in the welcome email and checking the effect in a month.”
The method still matters; it's what you explain when the stakeholder asks “how do you know?”. It just doesn't go first.
Score each answer
| Criterion | Strong | Weak |
|---|---|---|
| Accuracy | Numbers reconcile; totals checked; outliers noticed | Right shape, wrong numbers |
| Data skepticism | Asks how the data was collected before trusting a change | Trusts the first chart |
| Structure | Validate, segment, hypothesize, recommend | A list of random causes |
| Tool fluency | Writes the query or formula without searching for syntax | Stalls on basics |
| Visualization judgment | Chooses the chart for the question; labels axes and units | Uses whatever chart is quickest |
| Plain language | Leads with the answer; no jargon for a non-analyst | Leads with methodology |
| Ownership | Says what happens next and when | Ends at “it depends” |
A two-week plan, and what AI can't check
- Week 1: session A every day; alternate B and C. Build one small project on a public dataset that you can talk about for five minutes.
- Week 2: all three sessions back to back twice; one session with a person for the spreadsheet task; the SQL mock interview script for your weakest query type.
What the AI half can't do:
- See your spreadsheet or dashboard. Describing a pivot isn't building one. If a live Excel task is likely, practice it with a timer and a person watching.
- Check real outputs. Its sample numbers are invented and its predictions about your query's output are guesses.
- Push like a real stakeholder. It accepts plausible answers more readily than a skeptical manager would.
- Help during tests. Online SQL tests and take-homes with rules against outside help are your own work.
Where CoPilot Interview fits
Sessions B and C are where CoPilot Interview, an AI interview assistant, helps most: its AI mock interview builds questions from your résumé and the job description, reads them aloud and scores each spoken answer out of 10. For session A, keep your own SQL editor and spreadsheet open. The free plan gives 15 minutes a day of live help in sessions of up to 15 minutes, with AI mock interviews included. The free practice demo has 145 questions in 10 tracks and scores typed answers out of 10 with strengths, areas to improve and a follow-up question, free with no signup. CoPilot Interview comes as a desktop app for Windows 10 and 11 and macOS 12 or later, a free iPhone and iPad companion on the App Store (iOS 16 or later) and an Android companion in beta. Paid plans are Standard at $24.99 a month and Pro at $39.99 a month.
In live interviews, follow your employer's rules. Online assessments, proctored tests, AI-led rounds and take-homes with stated rules forbid outside help, so prepare beforehand.
Disclosure: CoPilot Interview is our product, so weigh what we say about it accordingly.
Frequently asked questions
How do I do a mock interview for a data analyst job?
Split it into three short sessions: SQL and spreadsheet tasks in your own tools, a business case with a chart, and stakeholder and behavioral questions. Have a chatbot or a friend play the interviewer, answer out loud and score each answer against a rubric.
What questions are asked in a data analyst mock interview?
SQL on joins, aggregation and window functions; spreadsheet lookups and pivot tables; choosing a chart; a metric-drop case; explaining a finding to a non-technical manager; and stories about messy data and conflicting numbers.
Can AI help me practice the business case round?
Yes. Ask it to play a hiring manager who reveals data only when you ask for a specific breakdown, and to push back when you jump to a cause. Keep the structure and the recommendation in your own words.
How long should I prepare for a data analyst interview?
Two focused weeks is enough for most candidates who already use SQL and spreadsheets: one week of daily skill sessions and one week of full mock loops, with at least one session in front of a person.
Rehearse it out loud before the real one
The AI mock interview builds questions from your resume and the job description, reads them aloud and scores each spoken answer out of 10. Included on the free plan.
Start a mock interview