- USD$
- GBP£
- EUR€
- BRLR$
EN
Candidates write and run queries in a hands-on environment, and grading compares the output with the expected result — from basic SELECT to joins and window functions.
| region | month | revenue | rk |
|---|---|---|---|
| Southeast | 2026-08 | 1,284,300 | 1 |
| South | 2026-08 | 642,910 | 2 |
| Northeast | 2026-08 | 508,120 | 3 |
| Midwest | 2026-08 | 311,480 | 4 |
An online SQL test for hiring is an assessment where candidates write queries against a real database and the score comes from comparing their result with the expected one. In Coodesh database challenges, the candidate gets a ready-made database, writes queries in a SQL editor, runs them as often as needed and sees the output as a table.
Multiple-choice questions about JOINs tell you little about someone who has to pull the right numbers out of real tables. Because grading looks at the result, any query that reaches the right answer scores, not just the one the reviewer had in mind. Scoring is automatic, and human review is still available in the assessment results.
Upload a .sqlite or .sqlite3 file with at least one table with data, or write a CREATE TABLE and INSERT script on screen. Uploaded databases can be reused in other questions.
Your query becomes the answer key. Run it before saving and check that the output is exactly what the question asks for.
They see the question and available tables on one side and the editor on the other. Running a query shows a results table that highlights rows that differ, are extra or are missing.
Grading uses the query saved when the candidate finishes the question. You see the submitted query, the score and the video if enabled, and your team decides the next step.
Clear comparison rules make the score predictable for the author and the candidate.
Column name casing is ignored and column order does not affect the score.
Row order matters. Use ORDER BY in the reference query and say so in the question whenever order is part of the answer.
When an answer has several statements, grading uses the result of the last one, so candidates can set up steps before the final query.
Off, scoring is all or nothing. On, the score is proportional to correct fields and rows, which helps when more than one approach is valid.
Turn on video in the Presentation tab and candidates record an explanation of their query right after answering.
With integrity on, SQL answers show how the query was written, and every run is logged on the assessment timeline.
For people who filter, aggregate and join tables all day, the test shows whether they reach the right number on their own.
Pair the SQL question with coding challenges in the same assessment, each section with its own weight in the final score.
Build a small database with realistic, anonymized data that looks like what the role works with every day.
Start for free, set up a database and a reference query, and see how candidates handle real data.
The database runs on SQLite, both for uploaded files and for scripts written in the editor. Commands and functions specific to MySQL, PostgreSQL or SQL Server do not work and will fail when executed. When you create a question, write the prompt and the reference query in SQLite-compatible SQL, and run the reference query before saving to confirm the answer key.
The platform compares the candidate's query result with the reference query result, row by row. Column name casing is ignored and column order does not matter, but row order does. If an answer has several statements, the last one counts. The score is calculated from the query saved when the candidate finished the question, so there is no manual grading queue.
Yes. Upload a .sqlite or .sqlite3 file up to 10 MB with at least one table with data; the platform checks extension, size, file integrity and data on upload. You can also write the creation script in the editor, up to 1 MB. Keep databases small, realistic and anonymized. Uploaded databases stay in a list and can be reused across SQL questions.
Candidates see the question and the names of the available tables on one side and the SQL editor on the other. They only discover column contents by querying, so describe in the question what the query should return, which columns and in what order. Running a query shows a results table highlighting rows that differ, are extra or are missing. Tiny tables invite trial and error, so use a realistic amount of data.
Both. With partial credit off, the answer must match the answer key exactly to score. With it on, the score is proportional to correct fields and rows, which helps when more than one approach is valid. Grading is automatic, but a human reviewer can still review any answer in the assessment results, and the hiring decision is always made by someone on your team.
Our ecosystem of deep integrations makes it easy to streamline your hiring processes.
Create your account and send your first assessment today.
