Recruitment tests are useful when they answer a question you actually have about a candidate. Chosen well, a short test before interview tells you which of eighty applicants are worth an hour of your time. Chosen badly, it screens out good people on something the job never needed.
This guide lists the twelve types of test employers use most, but organises them by the question each answers, because that is how the choice should be made. Every test is one input; none of them replaces the interview, references or a work sample, and no test should be the only reason someone is rejected.
Start from the question, not the test
| The question you have | Test types that answer it | Time for a candidate |
|---|---|---|
| Can they do the specific task? | Software skills, coding, role-based skills | 10 to 20 min per test |
| Can they work things out under pressure? | Numerical, verbal, logical reasoning, cognitive ability | 10 min per test |
| How would they handle real situations here? | Situational judgement, remote-work assessment | 10 min per test |
| How do they tend to behave? | Personality, behavioural | 10 to 20 min, untimed |
| Can they work in the language the job uses? | Language proficiency | 10 min |
| Is what they told us true? | Background and reference checks | Days, not minutes |
Pick one or two rows, not all six. An assessment of more than about 40 minutes loses candidates, and a battery of six tests rarely tells you more than three well chosen ones.
Tests that answer "can they do the task?"
1. Software skills tests check practical knowledge of a named tool: Excel, QuickBooks, Salesforce, Photoshop. They are the most direct screen for roles where the tool is the job. They tell you nothing about judgement, and they go out of date, so check the test covers the version and features you use.
2. Coding tests check a language or framework through multiple-choice or short problems. Good for screening volume before a technical interview; a poor substitute for reviewing real code. Pair with a short take-home or a pairing session for the final few.
3. Role-based skills tests combine two to four short tests that match a job, such as numeracy, spreadsheets and accuracy for a finance assistant. They are the quickest way to a defensible shortlist for common roles, provided you can see and edit which tests are inside.
Tests that answer "can they work things out?"
4. Numerical reasoning presents tables and charts in a business context. Use it for any role that handles figures. Do not use it as a general intelligence proxy for roles that do not.
5. Verbal reasoning asks whether a statement follows from a passage. Useful where reading accurately matters: contracts, policies, briefs. It is a comprehension test, not a writing test.
6. Logical reasoning uses shapes and sequences to test rule-finding without language or numbers. Strong for analysts, engineers and graduates; unfair as a filter for roles that are mostly known procedure.
7. Cognitive ability tests mix the three above into one score. Convenient, but the single number hides which skill is weak, so prefer the separate tests when the role leans on one of them.
Tests that answer "how would they act here?"
8. Situational judgement tests describe a realistic workplace situation and ask for the most and least effective response. They are the best short screen for judgement, customer handling, teamwork and problem solving, and they are hard to game because there is no lookup. Scenarios must fit your kind of work; a retail scenario tells you little about a lab technician.
9. Remote-work assessments are situational judgement tests written around working apart: communication, self-management, asking for help. Use them only if the role is remote.
Tests that answer "how do they tend to behave?"
10. Personality questionnaires describe preferences, not ability, and have no right answers. Use them to shape interview questions and onboarding, never as a cut-off. Be wary of any report that presents personality as a predictor of performance without saying how that was established.
11. Behavioural assessments are similar in intent but framed around work drives and motivations. The same rule applies: a conversation starter, not a filter.
Tests that answer "can they work in this language?"
12. Language proficiency tests check reading and grammar in the working language. Necessary where the role demands it; discriminatory where it does not, so tie the requirement to the job.
Checks that are not tests
Background, right-to-work and reference checks establish facts, not ability. They belong after a conditional offer, not in the screening stage, and in most countries the rules on what you may check are strict.
Choosing well: five checks before you send anything
- Can you say what a low score would stop this person doing? If not, drop the test.
- Is the time reasonable? Under 40 minutes total for a first screen.
- Will every candidate see the same thing? Same questions, same time limit, same instructions.
- Do you know how it is scored? You should be able to see every answer, not only a number, and any comparison should say who it is against.
- What happens next? Decide before you send it which scores lead to an interview and what you will ask.
If you want to see what a result looks like before choosing, open a sample report. Then start from the role: the finance, logical reasoning and maintenance assessments on this site are ready-made batteries you can edit before sending. Your first five candidates' results are free, with no card needed; after that you pay one price per candidate whose results you unlock, billed monthly.
