Data analyst assessment
A data analyst skills assessment that shows how candidates work.
Updated 26 September 2026.
The short answer
Trial Day is a sixty-minute skills assessment for data analyst candidates. Each candidate works on a company of their own, in SQL, Python or Excel, and you get a report that scores five things, from whether they checked the data to whether they caught a mistake in an AI's analysis. You can shortlist from it the same day.
What a data analyst assessment should test
Most analyst tests check one thing: can this person write a query that returns the right rows. That matters, but it is a small part of the job. Day to day, an analyst has to:
- Check the data before trusting it. Duplicates, gaps and odd definitions are normal. The analyst who looks first avoids the confident wrong answer.
- Read somebody else's work. A dashboard or report built by a colleague, which may be wrong.
- Give a number under pressure. A manager on the way into a meeting wants one figure, now.
- Check an AI's analysis. More and more of the first draft is written by a model. Someone has to know when it is wrong.
- Follow a problem to its cause. Not stop at the first plausible explanation.
An assessment that only tests the first query misses the rest. Trial Day puts all of them in one hour.
How it works
- Create the role. Pick the seat (revenue operations, marketing, finance, product or retention), the seniority and the tools.
- Send one link. In your company's name, with your logo. Nothing to install. Sixty minutes, six decisions.
- Shortlist from the report. Every result lands in one list, sortable by score, and two candidates can be compared side by side.
What is scored
| Skill | The question it answers |
|---|---|
| Judgement | Did they call it right? |
| Rigour | Did they check the data first? |
| Calibration | Did they know when they were sure? |
| Depth | Did they reach the cause? |
| Supervision | Did they check the AI's analysis? |
Being right is only one of the five. No AI decides the score. The report adds a recommendation, strengths, watch-outs and questions to ask in the interview. See a sample report.
Why candidates cannot prepare or hand it to an AI
- Every candidate gets a different company, drawn from a pool of forty, with six to nine tables of realistic data. There is no answer sheet to find.
- The evidence has to be gathered inside the product, and pasting is blocked.
- One question hands them an AI's analysis and asks whether it holds up.
- They write up their findings in 40 to 150 words, and you read it.
Questions hiring managers ask
How long does the assessment take?
Sixty minutes, in the browser, with nothing to install.
Which tools can candidates use?
You choose SQL, Python or Excel for each role.
Is it the same test for every analyst?
No. There are five seats with their own questions, and each candidate gets a company from a pool of forty, measured to be equally hard.
Does it decide who to hire?
No. It is a first filter that shows how someone works. You make the decision in the interview.
What does it cost?
Three assessments free for fifteen days, no card. Then packs of assessments that last twelve months. No subscription and no per-seat charge.
More for hiring managers: Take-home alternativeSQL test alternativeTrial Day vs TestGorilla