The short answer
Give a realistic, self-contained task on real equipment, tell the candidate in advance what it will cover, observe method rather than only outcome, score against written criteria, and pay for substantial assessment time. Tests designed to catch people out produce less information than tests designed to let them work.
Choosing the task
The task should be representative of the actual job, completable in the time allowed, and safe to observe. Avoid puzzles — you are testing workshop competence, not lateral thinking.
Task options by profile
| Profile | Suitable task | Time |
|---|---|---|
| Automotive mechanic | Diagnose a known fault; brake inspection and report | 60–90 min |
| Mechatronics technician | Diagnose an electrical fault requiring measurement | 90 min |
| Truck mechanic | Air brake system inspection and leak test | 60–90 min |
| Diesel mechanic | Investigate a running fault; interpret measurements | 90–120 min |
| Heavy equipment mechanic | Hydraulic performance complaint diagnosis | 90 min |
| Industrial mechanic | Bearing replacement and shaft alignment | 90–120 min |
What to observe
Outcome is the least informative part. Two candidates may both find the fault, and be very different hires.
- Safety behaviour before starting — securing equipment, PPE, isolation
- Whether they gather information first or start dismantling
- Tool selection and handling
- Whether they measure or assume
- How they respond when a hypothesis is disproved
- Whether they work cleanly and keep track of components
- Whether they verify their own work before declaring completion
- How clearly they explain what they found afterwards
Scoring consistently
Write the criteria before the first candidate arrives, use the same sheet for everyone, and involve more than one assessor where possible. Without this, technical hiring drifts toward whoever presented most confidently.
| Dimension | Weight | Evidence to record |
|---|---|---|
| Safety | Critical — pass/fail | Unprompted safe behaviour; no unsafe shortcuts |
| Diagnostic method | High | Information gathered, measurements taken, reasoning shown |
| Technical execution | High | Correct procedure, tool use, quality of work |
| Recovery from error | High | Response when first hypothesis fails |
| Self-verification | Medium | Checks work before declaring done |
| Communication | Medium | Clarity explaining findings |
| Speed | Low | Context only — never the primary criterion |
Treat safety as pass/fail rather than as a weighted score. A candidate who works unsafely is not a partial match.
Common design mistakes
- Testing an unrepresentative task. A rare fault tests luck, not competence.
- Not telling candidates what to expect. Surprise measures anxiety, not skill.
- Scoring only the outcome. Discards the most useful information.
- Unfamiliar tools with no orientation. Tests tool familiarity, not ability.
- Assessors intervening inconsistently. Makes candidates incomparable.
- Unpaid long assessments. Good candidates decline, and rightly so.
- Using assessments as free labour. Damages reputation quickly.
Assessing international candidates
Where relocation makes pre-hire testing impractical, there are workable alternatives.
- Structured technical interview with scenario questions requiring described method.
- Video walkthrough where the candidate explains a repair they completed.
- Assessment during the initial period, planned and communicated in advance.
- Trial period with a named mentor and defined review points.
Giving feedback
Give every assessed candidate brief, specific feedback. It costs ten minutes and materially affects whether they speak well of your company and whether they reapply when they are stronger. In a small technical labour market, reputation compounds.
Fairness and legal considerations
A practical assessment is a selection instrument, which means it carries the same fairness obligations as any other part of recruitment. The specifics vary by jurisdiction and you should confirm them for your own — but the principles are consistent.
- Give every candidate the same task, tools, time and level of assistance
- Tell candidates in advance what the assessment will cover
- Use written criteria decided before the first candidate is assessed
- Record observations rather than relying on recollection
- Make reasonable adjustments where a candidate needs them
- Do not use assessment output as productive work — that is unpaid labour
- Pay for assessment time beyond roughly an hour
The point about not using the output is worth emphasising. If the “test” is a repair you needed doing anyway and the candidate is not hired, you have had free labour — and word of that travels quickly in a small technical labour market.
A scoring sheet you can adapt
Keep it on one page and complete it during the assessment rather than afterwards.
| Observation | Scale | Note |
|---|---|---|
| Isolated and secured equipment before starting | Pass / fail | Fail ends the assessment |
| Gathered information before dismantling | 1–5 | Asked questions, checked history |
| Tool selection and handling | 1–5 | Correct tools, used correctly |
| Measured rather than assumed | 1–5 | Which measurements, in what order |
| Response when first hypothesis failed | 1–5 | Most informative single item |
| Work quality and cleanliness | 1–5 | Torque, fit, tidiness |
| Self-verification before declaring done | 1–5 | Checked own work |
| Explanation of findings afterwards | 1–5 | Clarity, not fluency |
Key takeaways
- Use a real, known fault. You know the answer, the candidate does not, and the situation is honest.
- Tell candidates what to expect. Surprise measures anxiety, not skill.
- Observe method over outcome. Two candidates who both find the fault can be very different hires.
- Score against criteria written before the first candidate. Otherwise the standard drifts and comparison becomes impossible.
- Pay for substantial assessment time and never use the output as production work. Word of unpaid labour travels quickly in a small labour market.
Testing candidates who are already employed
Most capable technical candidates are currently working, which constrains when and how you can assess them — and employers who ignore that lose people at exactly this stage.
Practical assessments during working hours are often impossible without the candidate taking leave. Offering an early evening or Saturday slot, and paying for the time, signals that you understand the position they are in. Compressing the assessment to a single focused task rather than a half-day is usually enough.
Avoid asking for anything that would require them to disclose the application to their current employer. That is a needless risk for them and a bad start to the relationship.
