Data Analyst
This Data Analyst interview question bank covers 15 questions across Data storytelling, SQL, ETL, Statistics, Visualization, Python. Each one mirrors a entry to mid level screen, so you can rehearse the exact areas a hiring panel digs into and walk in ready.
What this interview tests
- Data storytelling
- SQL
- ETL
- Statistics
- Visualization
- Python
Data Analyst interview questions
- Question 1Focus area: Data storytelling
A product manager asks you why daily active users dropped 15% last week. Walk me through how you would investigate this and what you'd present back.
What a strong answer covers
A strong answer covers structured investigation (segment by platform, geography, cohort), checking for data pipeline issues first, funnel analysis, cohort comparison, clear communication of findings with confidence levels, actionable recommendations. Be ready to discuss: How would you distinguish between a real drop and a data quality issue? What visualizations would you include in your report? How would you frame your findings for a non-technical audience?
- Question 2Focus area: SQL
Write a SQL query to find the top 10 customers by revenue in the last 90 days, but only those who have made at least 3 purchases. How would you structure this?
What a strong answer covers
A strong answer covers proper GROUP BY with HAVING, date filtering with appropriate functions, understanding of aggregate vs window functions, index awareness for large tables, handling of NULL values and edge cases. Be ready to discuss: How would you modify this if the orders table has billions of rows? How would you handle customers in different time zones?
- Question 3Focus area: ETL
Describe a situation where you had to clean and transform a messy dataset before analysis. What were the main issues, and how did you handle them?
What a strong answer covers
A strong answer covers systematic approach to data profiling (missing values, duplicates, outliers, type mismatches), reproducible cleaning pipeline (not manual edits), documentation of assumptions, pandas/SQL proficiency, awareness of how cleaning choices affect downstream analysis. Be ready to discuss: How did you document your data cleaning decisions? What tools and libraries did you use?
- Question 4Focus area: Statistics
How would you design an A/B test to measure whether a new checkout flow increases conversion rate? What statistical considerations are important?
What a strong answer covers
A strong answer covers hypothesis formulation, sample size calculation (power analysis), randomization strategy, choice of statistical test, significance level and power, understanding of p-values vs practical significance, awareness of multiple comparison problem, duration considerations. Be ready to discuss: How would you determine the sample size needed? What would you do if the results are directionally positive but not statistically significant?
- Question 5Focus area: Visualization
Tell me about a dashboard you built that had a real impact on business decisions. What metrics did you choose and why?
What a strong answer covers
A strong answer covers metric selection tied to business goals (not vanity metrics), information hierarchy in dashboard design, appropriate chart types, interactivity for exploration, user-centric design (knowing the audience), measurable impact on decisions. Be ready to discuss: How did you decide which metrics to highlight versus which to make available on drill-down? How did you handle stakeholders who wanted to add more and more metrics?
- Question 6Focus area: ETL
You're given two datasets that need to be joined: one from the CRM and one from the billing system. They share a customer ID, but you notice discrepancies. How do you handle this?
What a strong answer covers
A strong answer covers data reconciliation approach, LEFT vs INNER join implications, investigating root cause of mismatches (timing, data entry errors, different identifiers), documenting coverage and limitations, fuzzy matching techniques if needed. Be ready to discuss: How would you quantify the mismatch and decide if the join is reliable enough? What would you do with records that exist in one system but not the other?
- Question 7Focus area: Statistics
Explain the difference between correlation and causation with a real-world example from your work. How do you communicate this distinction to stakeholders?
What a strong answer covers
A strong answer covers clear articulation of confounding variables, Simpson's paradox awareness, techniques (quasi-experiments, propensity score matching, instrumental variables), ability to explain without jargon, diplomatic but firm communication about limitations. Be ready to discuss: What techniques can help you move closer to establishing causation from observational data? How do you handle pushback when someone insists a correlation proves their hypothesis?
- Question 8Focus area: Python
How would you use Python (pandas) to analyze user engagement data and identify the features most correlated with user retention?
What a strong answer covers
A strong answer covers feature engineering for retention (cohort-based), correlation analysis (Pearson, Spearman), handling multicollinearity, feature importance via simple models (logistic regression, random forest), normalization/standardization, dimensionality reduction if needed. Be ready to discuss: How would you handle features with very different scales? What would you do if the dataset has 500 features?
- Question 9Focus area: Data storytelling
A stakeholder wants a report showing 'average order value by region.' What questions would you ask before building it, and what pitfalls would you watch for?
What a strong answer covers
A strong answer covers clarifying questions (time period, order type, currency conversion, definition of 'region'), outlier handling, small sample size warnings, median vs mean consideration, segmentation suggestions, context metrics (order count, trend over time). Be ready to discuss: How would you handle regions with very few orders skewing the averages? What additional context would make this metric more actionable?
- Question 10Focus area: ETL
Describe how you would build an automated data pipeline that pulls data from an API, transforms it, and loads it into a data warehouse on a daily schedule.
What a strong answer covers
A strong answer covers orchestration tool (Airflow, dbt, Prefect), incremental vs full load strategy, idempotency, error handling and retries, data quality checks (row counts, schema validation, freshness), alerting on failures, backfill capability. Be ready to discuss: How would you handle API failures or rate limits? How would you monitor the pipeline and alert on data quality issues?
- Question 11Focus area: SQL
Write a SQL query using window functions to calculate a 7-day rolling average of daily revenue. Explain each part of the query.
What a strong answer covers
A strong answer covers correct ROWS BETWEEN syntax, date gap handling (calendar table or generate_series), PARTITION BY if needed, understanding of frame specification, performance implications, CTE organization for readability. Be ready to discuss: How would you handle days with no revenue (gaps in the data)? How does the performance of window functions compare to self-joins for this use case?
- Question 12Focus area: Visualization
How do you choose the right type of chart or visualization for a given dataset and audience? Walk me through your decision process.
What a strong answer covers
A strong answer covers matching chart type to data type (categorical, time series, distribution, comparison), audience-appropriate complexity, avoiding misleading charts (truncated axes, dual axes), accessibility (color palettes, alt text), tool proficiency (Tableau, Matplotlib, Plotly). Be ready to discuss: When would you avoid using a pie chart? How do you make visualizations accessible to colorblind users?
- Question 13Focus area: Data storytelling
You discover that a key metric the team has been reporting is calculated incorrectly due to a logic error in the ETL pipeline. How do you handle this?
What a strong answer covers
A strong answer covers quantifying the magnitude of the error, transparent communication, corrected historical data, root cause analysis, process improvements to prevent recurrence (data validation, testing), empathy for stakeholders who acted on bad data. Be ready to discuss: How do you assess the blast radius of the error? How do you communicate this to stakeholders who have been making decisions based on the wrong numbers?
- Question 14Focus area: Python
How would you segment a customer base to identify distinct user personas using data? What approach would you take?
What a strong answer covers
A strong answer covers feature selection for segmentation, clustering techniques (K-means, hierarchical), elbow method or silhouette scores for choosing K, standardization of features, business interpretability of clusters, actionable segment profiles. Be ready to discuss: How would you validate that your segments are meaningful and not just statistical noise? How would you present the segments to the marketing team?
- Question 15Focus area: Statistics
Explain what a p-value means in the context of an A/B test. A stakeholder says 'the p-value is 0.06, so the test failed.' How do you respond?
What a strong answer covers
A strong answer covers correct definition of p-value (not the probability the hypothesis is true), nuance around 0.05 threshold (arbitrary convention), consideration of effect size and confidence intervals, Bayesian alternatives, practical significance vs statistical significance. Be ready to discuss: How would you explain statistical power to this stakeholder? What alternatives to frequentist testing would you consider?
Questions asked in almost every interview
Ready to practice for real?
Rehearse these Data Analyst questions out loud with an AI interviewer that speaks Arabic and English, then get instant feedback on your answers.
Practice with AI, start free