Data analyst assessment practice

What the real test usually looks like

There is no single data analyst assessment. Each employer chooses its own test, often from an assessment vendor, and can usually change the length, the topics and the difficulty. Your invitation or recruiter is the best guide to yours. Three public examples show the range:

• iMocha. iMocha lists its data analyst test at 10 questions in 20 minutes, covering data querying such as SQL, visualization tools, statistical methods and tools such as Excel. It says the test is fully customizable by job role, difficulty and preferred tools, and that employers can add their own coding, multiple choice and whiteboard questions.
• Adaface. Adaface describes a data analysis test of 15 scenario-based multiple choice questions in 30 minutes, split across data analysis, SQL and data interpretation with charts and graphs. It says the test is customized to the job description, and that the length of the test an employer sends will differ from its sample.
• Alooba. Alooba's free data analytics test is a wider example: 50 multiple choice questions in 50 minutes, each with five answer options, covering basic statistics, SQL, Python, R, data wrangling, reporting and visualization.

Content areas commonly reported across vendors:

• SQL: reading a query and saying what it returns, choosing the query that answers a business question, joins, grouping and filtering.
• Spreadsheets: conditional totals and counts, lookups, IF logic and summarizing a list.
• Statistics basics: mean, median and spread, percentages and percent change, and correlation versus cause.
• Charts and tables: reading values, comparing groups, spotting trends and calculating from a table.
• Sometimes Python, R or a dashboard tool, or a hands-on task where you write a query that runs against a real database.

What our mocks cover. Our full mocks have 25 questions in 40 minutes, in four sections: SQL (6), spreadsheets (6), statistics basics (6) and charts and tables (7). You may use a calculator, and you can go back and flag questions. Every question is multiple choice: our mocks do not run SQL you write, and they do not cover Python, R, Tableau or Power BI. If your invitation includes those, add hands-on practice in the tool itself.

iMocha, Adaface and Alooba are trademarks of their owners. Intellect Council is not affiliated with, endorsed by or sponsored by any of them.

Question types with worked examples

One or two worked examples for each section, with the trap most people fall into. None of these repeat the free sample.

1. SQL: WHERE or HAVING. You need regions where at least one single order was over $500. Do you filter with WHERE or HAVING? WHERE, because the condition is about individual rows: WHERE amount > 500. HAVING filters groups after GROUP BY, such as regions whose total is over $500. Trap: both sound right. Ask whether the condition is about one row or about a group.

2. SQL: the missing customers. You join customers to orders with an inner JOIN and count customers. The count is lower than the number of customers in the table. Why? An inner JOIN keeps only customers who have at least one order. A LEFT JOIN from customers keeps everyone, with NULL for those without orders. Trap: reading the shorter result as "some customers were deleted".

3. Spreadsheets: counting with a comparison. Column B holds the region and column C the order value. To count East orders worth 100 or more, use =COUNTIFS(B:B,"East",C:C,">=100"). The comparison goes inside quotes. If the limit sits in a cell, say E1, join it on: ">="&E1. Trap: typing ">=E1", which looks for the text "E1" and counts nothing.

4. Statistics: the average of averages. Store A had 100 orders averaging $20. Store B had 10 orders averaging $40. What is the average order across both? Not $30. Total sales are 2,000 + 400 = $2,400 over 110 orders, so about $21.82. Trap: averaging two averages that cover very different numbers of items. Go back to totals and counts.

5. Statistics: percent or percentage points. A conversion rate goes from 4% to 5%. That is a rise of 1 percentage point, which is also a 25% increase (1 / 4). Trap: answering "1%" when the question asks for the percent change. Check which one the question wants.

6. Charts: the shortened axis. A bar chart shows satisfaction of 92 for Team A and 96 for Team B, and B's bar looks twice as tall. Check the axis: if it starts at 90, the bars show 2 and 6 above that line, not the full values. The real gap is 4 points, about 4%. Trap: reading bar height instead of the numbers.

7. Tables: read the units first. A table header says "Revenue ($ thousands)" and a question asks for Q2 revenue, shown as 95. The answer is $95,000. Trap: picking the option that says $95. Read the title, the headers and any notes before the first question on a table.

How to prepare

1. Read your invitation first. Find out which vendor or format your employer uses, how long it is, and whether it asks you to write SQL that runs, use Excel or Google Sheets, or code in Python or R. Formats vary too much to assume.
2. Learn your pace. Vendor tests are often short. At iMocha's 10 questions in 20 minutes, you have about 2 minutes a question (iMocha). Some questions need a quick sum, so practice with a timer.
3. Cover the SQL core. SELECT, WHERE, GROUP BY with SUM, COUNT and AVG, HAVING, ORDER BY, INNER and LEFT JOIN, DISTINCT, and how NULL behaves in counts and comparisons. Trace small queries by hand on a table of five or six rows.
4. Cover the spreadsheet core. SUMIFS, COUNTIFS, AVERAGEIFS, a lookup such as XLOOKUP, IF and nested IF, and a pivot summary. For more formula practice, try our Excel assessment test.
5. Refresh the statistics basics. Mean, median and mode, when an outlier makes the median better, range, percent change, percentage points, weighted averages, and why correlation does not prove cause.
6. Read every chart slowly. Title, units, axis start, legend, then the question. Most chart mistakes come from skipping one of those.
7. Keep a calculator and scratch paper ready if your test allows them, and know how to work a percentage on both.
8. Take full timed mocks. Practice at the real pace so a 20 or 30 minute test feels normal on the day.

How scoring usually works

Scoring depends on the employer and the vendor. There is no national scale and no universal pass mark. Vendors give employers a report, and the employer decides what result the job needs. One public example:

• Adaface says its report comes with an overall pass or fail recommendation and industry benchmarks, and it suggests employers decide their own threshold score (Adaface).

iMocha and Alooba do not describe their scoring on the pages we checked. Multiple choice questions are usually marked right or wrong. Where a test asks you to write a query, it may be checked against expected results, so an almost-right query can still score nothing.

Our practice score is the number of our own questions you answered correctly, with a breakdown by section: SQL, spreadsheets, statistics basics, and charts and tables. It shows where to focus. It is a practice score: it is not a score on any vendor's or employer's test and does not predict whether you will pass.

Common mistakes on data analyst tests

• Forgetting NULL. COUNT(column) skips NULLs, AVG ignores them, and WHERE email = NULL matches nothing. Use IS NULL.
• Grouping by the wrong thing. Grouping by name merges two different people who share a name. Group by the id, and show the name beside it.
• Joining on the wrong key. Joining orders.id to customers.id looks valid and returns rows, but they are the wrong rows. Match the foreign key, such as customer_id.
• Dividing by the wrong number in percent change. Always divide by the starting value.
• Treating "greater than" as "at least". In SQL and in formulas, > and >= give different answers at the boundary. Check the edge case.
• Mixing the mean and the median. One extreme value moves the mean a lot and the median hardly at all.
• Saying "causes" from a correlation. Two things moving together is a pattern to explain, not proof.
• Rushing the table header. Units, time periods and footnotes change the answer.

How this practice compares with the real test

What we match. The four areas vendors describe most, multiple choice questions in a similar style, a calculator, and real time pressure. Every question has an explanation.

What we cannot match. We do not run a live database, so we cannot check SQL you type. We do not cover Python, R or dashboard tools, and vendors use their own question banks and difficulty levels, which we never see.

How to cover the gap. Use our questions to find weak spots and learn the rules, then write a few queries in a free database tool against a small table of your own, and rebuild the spreadsheet formulas in a real sheet. If your invitation mentions a coding or dashboard task, practice that tool directly.