Data analyst interview questions and sample answers

If you’re preparing for a data analyst interview, expect questions that test technical skills, statistical thinking, and how you communicate insights. Interviewers want evidence you can turn data into decisions.

Below are common question types you’ll see—SQL, data quality, statistics, business interpretation, and stakeholder communication—with short sample answers and tips you can adapt to your own experience.

SQL: structure your answer and show intent

SQL questions are almost guaranteed. Interviewers look for correct syntax, but more often they care about approach: how you join tables, handle NULLs, and optimize for clarity. If you get a live coding problem, narrate what you’re doing.

Sample question: "Write a query to find the top 3 products by revenue in the last quarter."

• Sample short answer (conceptual): "Filter orders by date for the last quarter, join to products to get price, compute revenue per order line (quantity * price), group by product, sum revenue, and order descending with LIMIT 3."
• Concrete SQL example: SELECT p.product_id, p.name, SUM(ol.quantity * ol.unit_price) AS revenue FROM order_lines ol JOIN products p ON ol.product_id = p.product_id JOIN orders o ON ol.order_id = o.order_id WHERE o.order_date BETWEEN '2023-01-01' AND '2023-03-31' GROUP BY p.product_id, p.name ORDER BY revenue DESC LIMIT 3;
• Tips: mention assumptions (time zone, inclusive dates), talk about handling returns or discounts, and suggest indexes or covering indexes if performance comes up.

Data quality: show a detective mindset

Questions on data quality test how you detect, quantify, and communicate issues. Interviewers want to see that you won’t blindly trust a dataset and that you can triage problems.

Sample question: "How would you investigate a sudden drop in daily active users (DAU) in our dashboard?"

• First checks: verify data pipeline health (job failures, schema changes), check upstream events (tracking, SDK updates), and compare raw event counts vs. processed metrics.
• Run quick diagnostics: sample raw rows for missing or malformed fields, plot volume over time by region/device, and check if particular cohorts are affected.
• Answer example: "I’d check pipeline logs and recent deployments, compare raw event ingestion to downstream counts, and isolate by client/app version. If pipeline and versions look OK, I’d sample events to see missing user IDs or timestamps. I’d document findings, estimate impact (e.g., 30% fewer events from mobile), and propose fixes—rollback change, patch the tracking SDK, or backfill missing events—along with a communication plan for stakeholders."
• Tip: quantify impact and propose both short-term mitigation and long-term fixes (validation checks, alerting, and a schema contract).

Statistics and inference: keep it practical

You don’t need to lecture on p-values, but you should show you can choose the right test and explain results in business terms. Expect A/B testing or hypothesis questions.

Sample question: "We ran an A/B test and saw a 2% lift in conversion with p=0.04. What do you tell the product manager?"

• Answer example: "A p-value of 0.04 suggests the observed effect is unlikely under the null hypothesis; it’s statistically significant at the 5% level. But I’d also report the confidence interval of the lift, the baseline conversion, sample sizes, and power. If the lift is 2% on a 10% baseline, that’s a relative 20% increase—meaningful. I’d check for optional stopping, metric definitions, and any segmentation effects before recommending a rollout."
• Practical points: always report effect size and uncertainty, check metric quality, consider practical significance, and guard against multiple testing and peeking.

Business interpretation: translate numbers into decisions

Interviewers hire analysts who can move from data to recommended action. They want concise conclusions, trade-offs, and a next-step proposal.

Sample question: "You see revenue per user dropped but average order value increased. How do you explain this?"

• Answer example: "An increase in AOV with falling revenue per user suggests fewer users are transacting while those who do spend more. I’d segment by cohort and channel to see if acquisition dropped or if a specific segment churned. Next steps: identify the lost cohort, check acquisition sources for recent changes, and run retention and LTV analysis to decide whether to adjust acquisition spend or focus on re-engagement."
• Tip: offer one clear recommendation and an experiment or metric to validate it (e.g., run a re-engagement campaign and track cohort retention over 30 days).

Stakeholder communication: be concise and action-oriented

Interviewers want to know you can tailor your message: technical depth for engineers, business implications for PMs, and clear next steps for executives.

Sample question: "How do you present a complex analysis to non-technical stakeholders?"

• Structure: start with the headline (one sentence), then the key evidence (one or two charts), and finish with recommended action and risks.
• Example script: "Headline: recent pricing change reduced trial-to-paid conversion by 15%. Evidence: conversion curve and cohort table showing drop in younger cohorts. Recommendation: rollback the pricing test for three weeks and run a targeted experiment. Risk: revenue per user may fall; offset with targeted upsell messaging."
• Tip: always bring a clear ask—what decision you want from the stakeholder—and at least one metric to measure after implementation.

Practical answer examples for common interview prompts

Below are ready-to-adapt responses for common prompts. Use your own numbers and context when you speak—interviewers can tell when an answer is generic.

Practice a short version (30–60 seconds) and a one-minute expanded version that includes an impact statement.

• Tell me about a time you found a bug in production: "I noticed a daily revenue metric dropped 40%. I traced it to an order_line JOIN that excluded refunded transactions. I created a hotfix to include refunds, backfilled a week of data, and added a unit test and monitoring alert. Impact: restored accurate reporting and prevented wrong billing decisions."
• Walk me through an analysis: "Goal: identify churn drivers. I defined churn as 30 days of inactivity, pulled cohort tables, ran a Cox regression for time-to-churn with usage features, and found declining engagement with a new onboarding flow correlated with churn. Recommendation: revert onboarding changes for a subset and measure retention over 45 days."
• Handle 'I don't know' gracefully: "I don’t have the exact number on hand, but here’s how I’d get it: query the events table for the metric, join to users for demographics, and segment by cohort. If you want, I can sketch the query now."

Focus practice around these question types and personalize the sample answers with your own metrics and stories. Before interviews, pick two examples (one technical, one business) and rehearse concise versions that include impact.

During the interview, be explicit about assumptions, quantify uncertainty, and always end with a concrete next step or recommendation.