Data Analyst technical round practice: SQL, Excel and metrics

The Data Analyst technical round is rarely about writing perfect syntax from memory. Interviewers are checking whether you can turn a vague business question into a precise query and defend your logic out loud. Expect three overlapping themes.

• SQL fluency: multi-table joins, GROUP BY with HAVING, subqueries, and window functions such as ROW_NUMBER() and RANK().
• Data manipulation: cleaning messy data, handling NULLs, deduplication, and pivoting long-to-wide.
• Metric definition: being handed a term like 'active user' or 'churn' and being asked to define it unambiguously before you query it.

Many rounds are conducted on a shared screen or a browser SQL editor, so interviewers watch how you think, not just the final output.

A handful of patterns cover most technical rounds. Practising these families is more efficient than grinding random problems. Most Data Analyst / Data Science Jobs at Indian product firms recycle the same three archetypes below, so pattern-matching pays off.

Ranking and top-N per group

'Return the top 3 products by revenue in each category.' This tests window functions. A strong answer uses ROW_NUMBER() OVER (PARTITION BY category ORDER BY revenue DESC) and filters in an outer query. A weak answer tries to force it with a correlated subquery and gets tangled on ties.

Second-highest / gaps

'Find the second-highest salary per department without using LIMIT.' Interviewers watch whether you handle duplicate salaries correctly using DENSE_RANK().

Cohort and retention counts

'How many users who signed up in January were still active in March?' This needs a self-join or conditional aggregation on date-bucketed data.

Question: 'From an orders table, show each customer's total spend and their rank among all customers.'

Strong answer: The candidate first clarifies whether cancelled orders count, then writes SELECT customer_id, SUM(amount) AS spend, RANK() OVER (ORDER BY SUM(amount) DESC) AS spend_rank FROM orders WHERE status = 'completed' GROUP BY customer_id. They explain why RANK over ROW_NUMBER matters for ties and mention indexing if the table is large.

Weak answer: The candidate jumps straight in, forgets the status filter, uses ORDER BY without a window function, and cannot explain how ties are broken. The logic may still return rows, but it silently answers a different question. Interviewers care far more about the clarifying step and the tie-handling than about typing speed.

Structure your prep so the round feels familiar rather than novel.

• Week 1: Rebuild core SQL from joins to GROUP BY; write every query out loud, narrating each clause.
• Week 2: Drill window functions and date logic until top-N-per-group is automatic.
• Week 3: Add Excel/Google Sheets modelling, VLOOKUP/XLOOKUP, pivot tables, and simple what-if models.
• Week 4: Do full timed mocks, speaking your reasoning while you type.

Because this round is spoken as much as written, a free AI voice mock interview on InterviewPrep is a natural rehearsal: it builds a mock from your CV and a real job description, then scores your answers, pace and filler words so you hear where you ramble under pressure.