20 scenario-based interview questions and answers for a

data analyst role

focusing on practical situations you might face in a professional setting:

1. Scenario: Handling Missing Data

Q: You have a dataset with many missing values in key columns. How would you handle this
situation? A: I would first analyze the extent and pattern of missing values. If the missing data is
significant, I would explore imputation methods such as using the mean, median, or mode. For
categorical data, I might use the most frequent value. If it's not essential or minimal, I would consider
removing those rows or columns. In case the missing data is systematic, I might investigate the cause
and address it accordingly.

2. Scenario: Analyzing a Drop in Sales

Q: Your company has experienced a sudden drop in sales over the last quarter. How would you
analyze the cause? A: I would start by collecting relevant data from various departments, such as
sales, marketing, customer feedback, and competitor analysis. I would perform trend analysis on the
sales data, segment it by region, product, and customer demographics to identify any patterns.
Additionally, I'd examine external factors such as market conditions or seasonal trends and use
statistical tests to see if the change is statistically significant.

3. Scenario: Choosing Key Metrics

Q: Your manager asks you to track the performance of a marketing campaign. Which metrics would
you choose? A: I would choose metrics based on the campaign's objectives. For example, if the goal
is awareness, I’d track reach, impressions, and engagement rates. If the goal is lead generation, I’d
focus on conversion rate, click-through rate (CTR), and cost per acquisition (CPA). Sales-related
campaigns would focus on ROI, customer lifetime value (CLTV), and the number of conversions.

4. Scenario: Data Cleaning

Q: You find that the dataset contains duplicate records. How would you handle them? A: I would first
identify the duplicates using functions like drop_duplicates() in Python (Pandas) or SQL queries. Once
identified, I would determine if these duplicates represent valid repeated transactions or are errors.
Based on the context, I’d either remove the duplicates or aggregate them by summing or averaging
depending on the situation.

5. Scenario: Analyzing Customer Churn

Q: You need to identify factors contributing to customer churn. How would you approach this? A: I
would begin by defining churn and selecting relevant data, such as customer demographics,
purchase history, service usage, and support interactions. Then, I’d conduct exploratory data analysis
(EDA) to find patterns in churned vs. non-churned customers. I would use logistic regression or
decision trees to model the likelihood of churn and identify significant factors like service quality,
price, or support issues.

6. Scenario: Data Visualization

Q: Your manager asks for a visualization of quarterly sales performance for various regions. How
would you present this? A: I would use a combination of bar charts and line graphs to show trends
over time. A clustered bar chart could display sales by region per quarter, while a line graph could
show the overall sales trend. I would use color coding to differentiate between regions and possibly
add interactive features in Power BI or Tableau for a deeper dive into the data.

7. Scenario: Presenting Insights to Non-Technical Stakeholders

Q: How would you present complex analysis results to non-technical stakeholders? A: I would
simplify the analysis by focusing on key takeaways and actionable insights. Instead of using technical
jargon, I’d use clear visualizations like charts and graphs that convey trends and patterns. I’d provide
context and relate the data to business outcomes, explaining how the insights can help make
decisions.

8. Scenario: Outlier Detection

Q: How would you deal with outliers in your dataset? A: First, I would determine whether the
outliers are due to data entry errors or represent genuine rare events. If they are errors, I’d correct or
remove them. If they are legitimate, I’d explore whether to keep them or transform them (e.g., log
transformation) based on their impact on the analysis. I might also consider running models with and
without outliers to understand their effect.

9. Scenario: Feature Engineering

Q: How would you create new features from raw data to improve model accuracy? A: I would analyze
the dataset for potential relationships between variables and derive new features. For instance, if I
had a date column, I could create features like day of the week, month, or time since the last
purchase. I’d also use domain knowledge to combine features or create interaction terms that might
improve model performance.

10. Scenario: A/B Testing

Q: You’ve run an A/B test for a new product feature. How would you determine if the change was
successful? A: I’d start by defining the success metric (e.g., conversion rate, click-through rate) and
ensuring the test was properly randomized. I’d then perform statistical analysis using methods like a
t-test or chi-square test to determine if the difference between the control and treatment groups is
statistically significant.

11. Scenario: Data Integration

Q: You need to combine data from two different sources with different formats. How would you
handle this? A: I’d standardize the data by aligning the formats (e.g., consistent date formats,
merging on a common key) and ensure that data types are compatible. I’d also check for any missing
or mismatched entries during the integration and resolve them appropriately before performing the
merge or join.

12. Scenario: Forecasting

Q: How would you forecast next month’s sales based on historical data? A: I’d first analyze the
historical sales data to detect any seasonality or trends. I would then use time-series forecasting
methods such as ARIMA, exponential smoothing, or Prophet, depending on the data pattern. I’d
validate the model using techniques like cross-validation and evaluate its accuracy based on metrics
like RMSE or MAPE.
13. Scenario: Working with Time Series Data

Q: How would you handle seasonality in time series data? A: I’d decompose the time series into
trend, seasonal, and residual components. This helps me understand the seasonal effects on the
data. If seasonality is significant, I’d consider using seasonal adjustment techniques or incorporating
seasonal components in my forecasting models, such as seasonal ARIMA.

14. Scenario: Dealing with Imbalanced Data

Q: Your dataset is highly imbalanced between classes. How would you address this? A: I would use
techniques such as oversampling the minority class, undersampling the majority class, or using
algorithms like SMOTE to balance the dataset. Additionally, I might choose evaluation metrics like
precision-recall or AUC-ROC instead of accuracy, which could be misleading in imbalanced datasets.

15. Scenario: Root Cause Analysis

Q: A business metric suddenly changes (e.g., a spike in website traffic). How would you identify the
root cause? A: I would first validate the data to rule out any collection issues. Then, I’d perform an
analysis of different segments (e.g., geography, marketing campaigns, or time periods) to identify the
source of the change. Correlation analysis or time series comparisons might reveal whether external
factors or internal actions are responsible.

16. Scenario: Dealing with Unstructured Data

Q: How would you handle unstructured text data in your analysis? A: I would convert the
unstructured text into structured data using techniques like tokenization, stemming, and
lemmatization. I’d then use Natural Language Processing (NLP) techniques such as term frequency-
inverse document frequency (TF-IDF) or word embeddings to extract meaningful features for
analysis.

17. Scenario: Improving Data Quality

Q: You’ve found that the data quality is poor in several key columns. What steps would you take to
improve it? A: I would begin by profiling the data to identify issues such as missing values, duplicates,
or inconsistent formatting. I would then clean the data by imputing missing values, standardizing
formats, and removing duplicates. I might also work with the data source to improve future data
collection processes.

18. Scenario: Automating Reports

Q: How would you automate the generation and distribution of a monthly sales report? A: I would
build a script using Python or Power BI that automates the extraction of sales data, performs the
necessary calculations, and creates visualizations. I’d schedule the script to run at regular intervals
(e.g., using a tool like cron or Power BI's scheduled refresh) and distribute the report via email or a
shared dashboard.

19. Scenario: Optimizing SQL Queries

Q: A query you wrote is running too slowly. How would you optimize it? A: I would start by checking
the query execution plan to identify bottlenecks. I’d look for opportunities to use indexes, reduce
joins, or rewrite subqueries as joins. I’d also ensure that the query is only fetching the required data
by minimizing SELECT * and filtering rows early using WHERE clauses.
20. Scenario: Data Privacy Concerns

Q: How would you handle sensitive customer data in your analysis to ensure privacy? A: I would
ensure that the data is anonymized by removing or masking personally identifiable information (PII).
If working with customer data, I’d follow data protection regulations such as GDPR or CCPA, ensuring
that data access is restricted to authorized personnel only.