Hire a Skilled
Data Analyst
in Days, Not Months

How do you approach cleaning and preparing a messy dataset before analysis?

Strong candidates describe a systematic process: profiling the data first (nulls, duplicates, outliers, type mismatches), understanding the source system before making assumptions, documenting every transformation, and validating output row counts and distributions. Look for analysts who treat data cleaning as a discipline rather than a quick step before "the real work." Candidates who jump to analysis without describing validation steps will deliver insights built on bad foundations.

Walk me through a recent project where your analysis directly influenced a business decision. What did you find and how did you present it?

This separates analysts who produce reports from those who drive decisions. Look for a clear arc: question framing, methodology, finding, and the specific action taken based on the insight. Strong candidates describe tailoring the presentation to the audience — an executive summary for leadership, granular breakdowns for operators. Analysts who struggle to name a specific business outcome from their work signal they've been producing outputs, not driving impact.

How do you validate the accuracy and integrity of your analysis before sharing results with stakeholders?

Look for a layered validation approach: cross-checking totals against known values, comparing outputs against prior period results for reasonability, peer review of SQL logic, and sanity checks at each aggregation level. Strong candidates describe specific cases where they caught an error in their own work before it went to leadership. Analysts who describe publishing results without a validation step are one bad filter condition away from a significant business mistake.

Which visualization tools have you used, and how do you decide what type of chart or layout to use for a given audience?

Strong candidates describe tool selection based on context — Tableau or Power BI for self-service dashboards, Python for exploratory analysis, simple Excel charts for one-off executive updates — and chart type selection based on what the data is trying to communicate: trends (line), composition (bar/pie), distribution (histogram), correlation (scatter). Look for candidates who describe simplifying visuals for executive audiences and adding detail for operational ones. Analysts who use the same chart type for everything regardless of message signal limited visualization judgment.

Write a SQL query to find the top 5 customers by revenue in the last 30 days, excluding any who had a refund in that period.

This is a practical screen for SQL fundamentals — aggregation, filtering, date logic, and exclusion patterns. Look for candidates who write clean, readable SQL with appropriate aliasing, use a WHERE clause or LEFT JOIN / IS NULL pattern to exclude refunds, and correctly apply date filtering. Bonus points for window functions if they're aware of alternatives. Analysts who can't construct this query without significant coaching will struggle with ad hoc reporting requests in a real environment.

How do you handle a situation where two data sources give you conflicting numbers for the same metric?

This is a real-world problem in every data environment. Strong candidates describe a structured reconciliation process: identify the definition differences (are both sources measuring the same thing?), trace each number back to its source table and transformation logic, document the discrepancy and its root cause, and engage the data owners to establish a single source of truth. Analysts who pick whichever number is "closer to what leadership expects" without investigation are creating trust problems downstream.