Topic HubUpdated December 29, 2025

Top SQL Data Manipulation Interview Questions

SQL data manipulation is one of the most commonly tested skills in technical interviews for data roles. Companies like Meta, Google, Amazon, and other top tech firms use SQL questions to evaluate your ability to query, transform, and analyze data effectively. In a typical data manipulation interview, you'll be asked to write queries that JOIN multiple tables, aggregate data using GROUP BY and window functions, handle NULL values, and transform data using CASE statements and CTEs. Interviewers assess not just whether your query works, but also its efficiency, readability, and how well you communicate your approach. Whether you're preparing for a Data Scientist, Data Analyst, Data Engineer, or Analytics Engineer role, mastering SQL data manipulation is essential. The questions in this collection cover real interview problems from top companies, ranging from basic JOINs to complex multi-step transformations.

Browse All SQL Questions

Questions by SQL Concept

JOIN

INNER, LEFT, RIGHT, FULL OUTER JOINs - combine data from multiple tables efficiently.

Window Functions

ROW_NUMBER, RANK, DENSE_RANK, LAG, LEAD for advanced analytics.

CTE & Subqueries

Common Table Expressions and subqueries for readable, modular SQL.

Aggregation

COUNT, SUM, AVG, MIN, MAX with GROUP BY for data summarization.

CASE WHEN

Conditional logic for data transformation and categorization.

Date/Time

Date arithmetic, timestamps, and time zone conversions.

String Functions

CONCAT, SUBSTRING, UPPER, LOWER, TRIM, and pattern matching.

NULL Handling

NULL semantics, COALESCE, NULLIF, and proper NULL comparisons.

UNION & Set Operations

Combine result sets with UNION, UNION ALL, INTERSECT, and EXCEPT.

HAVING Clause

Filter aggregated results with GROUP BY and HAVING for post-aggregation conditions.

Common Patterns in Interviews

Top-N per Group

Find the top N records within each group using ROW_NUMBER() or RANK() with PARTITION BY.

Deduplication

Remove duplicate records using ROW_NUMBER(), DISTINCT ON, or self-joins with aggregation.

Rolling MAU/WAU

Calculate rolling active users using window functions with date range conditions.

Retention Cohorts

Track user retention by joining signup dates with subsequent activity dates.

Funnel Conversion

Measure drop-off at each stage using conditional aggregation or CTEs.

Sessionization

Group events into sessions using LAG/LEAD window functions with running sum.

Common Pitfalls to Avoid

Join Explosion

Multiplying rows when joining without proper cardinality. Always verify join conditions.

WHERE vs ON

For OUTER JOINs, WHERE filters after the join, ON filters during. Know the difference.

NULL Semantics

NULL != NULL. Use IS NULL or IS NOT NULL for comparisons.

Time Zones

DATE truncation can give unexpected results without timezone awareness.

Integer Division

5/2 = 2, not 2.5. Cast to DECIMAL or FLOAT for accurate division.

Frequently Asked Questions

What SQL topics are most commonly tested in interviews?

The most commonly tested SQL topics include JOINs (especially LEFT JOIN), window functions (ROW_NUMBER, RANK), aggregate functions with GROUP BY and HAVING, subqueries and CTEs, and CASE WHEN statements. Companies also frequently test your ability to handle NULL values and perform date-based analysis.

How do I prepare for SQL data manipulation interviews?

Start by mastering the fundamentals: JOINs, aggregations, and window functions. Practice on real interview questions from companies you're targeting. Time yourself to build speed. Learn to write clean, readable queries using CTEs. Finally, practice explaining your thought process out loud.

When should I use a CTE vs a subquery?

Use CTEs when you need to reference the same derived table multiple times, when breaking complex logic into readable steps, or when building recursive queries. Use subqueries for simple, one-time derivations. CTEs generally improve readability and are preferred in interviews.

How are window functions used in interviews?

Window functions are used to rank items, calculate running totals, find previous or next values (LAG/LEAD), and compute moving averages. Common interview patterns include finding top N per group, calculating cumulative sums, and comparing each row to an aggregate.

What difficulty level should I practice?

Start with Easy problems to build confidence, then focus on Medium difficulty which represents most interview questions. Practice Hard problems for senior roles or companies known for difficult interviews. Aim to solve Medium problems in 15-20 minutes.

Can I run SQL code during an interview? What is the typical interview format?

In most SQL interviews, especially at FAANG-level companies, you generally cannot run your SQL code. The interview is usually conducted in a shared editor such as CodePad or CodeSignal. The interviewer will show you the data schema and explain the problem statement, and you are expected to write the SQL query directly in the editor. Your solution is evaluated based on logical correctness rather than execution results, so you need to reason through joins, aggregations, filters, and edge cases without being able to test the query. Because the code is not executable, interviewers focus on whether your SQL correctly answers the question, handles corner cases properly, and whether you can clearly explain your thought process and assumptions as you go.

How does practicing SQL on Prachub compare to a real interview?

The questions on Prachub are based on real interview questions and are designed to closely mimic the structure and difficulty of actual interviews. The main difference is that on Prachub, the SQL is executable in our console. This allows you to validate your logic, spot mistakes, and iterate more effectively while practicing. Over time, this helps you build strong intuition so that when you are in a real interview environment where execution is not available, you can confidently reason through your solution without relying on running the code.

Can I look up SQL functions online during an interview?

In most live interview settings—especially technical screens for data science, analytics, or database roles—you are generally not allowed to look up functions online during the actual interview. Interviewers typically expect you to write SQL without internet access and assess your ability to reason and recall syntax under pressure. Many interview processes do not provide a browser or internet access in the coding environment, and relying on outside resources can be viewed as a lack of preparedness for the fundamentals of the language. Candidates are usually evaluated on their ability to recall and apply core SQL concepts logically rather than search for syntax during the interview. The exception is Robinhood and DoorDash, which do allow online search. However, searching online can take up a significant amount of your coding time and may leave questions unfinished, so it is generally better not to rely on it.

Why do data scientists, data analyst, data engineer, and business intelligence engineer get tested on SQL?

SQL is tested because working with data is a core part of a data scientist’s job. Data scientists regularly use SQL to retrieve, join, aggregate, and transform data for analysis, modeling, and decision-making. SQL interview questions evaluate not only syntax knowledge but also whether a candidate can reason about data relationships, edge cases, and business logic in a realistic, production-like setting.