SiteThe interview went well. The CV says PLCs, drives, hydraulics, the lot. They talked through a gearbox change like they'd done fifty. You shook hands and they started on the Monday.
Three weeks later the case packer stops on nights. Servo fault. Your new engineer opens the panel, looks at the drive for a long time, and rings the shift manager to ask who usually does these.
Nobody lied. The CV listed kit they'd stood next to. The chat tested how well they talk about work. It never tested their hands.
So you find out what someone can really do at 2am, on your line, with the trailers waiting. That's the most expensive place to learn it.
What a chat does tell you
We'd never drop the interview. A good chat shows you how someone thinks. Do they ask what the fault history says before they guess? Do they talk about isolation without being prompted? Do they say "I don't know, but here's how I'd find out"? Those are good signs.
What a chat can't show you is whether they can read your drawings, find a fault on a live-looking panel without touching anything they shouldn't, or set up a coupling so it doesn't eat bearings. For that, you need them in front of the kit.
There are good questions that get closer. Our interview questions that show real skill post has a list. But a question still gets you an answer. A test gets you a result.
The law expects you to check
You don't have to trade test anyone. But you do have to know what they can do before you hand them the job.
The Management Regulations say an employer must "take into account their capabilities as regards health and safety" when giving people tasks (regulation 13). PUWER's code of practice says maintenance work "should only be done by people who are competent to do the work" (L22, paragraph 79).
HSE sums up competence as "the combination of training, skills, experience and knowledge" plus the ability to use them safely (HSE, what is competence). Training is the bit a certificate shows. The rest you have to see.
For electrical work the bar is set out more closely. Under the Electricity at Work Regulations, nobody should do work that needs technical knowledge or experience to prevent danger unless they have it, or are properly supervised (regulation 16). HSE's guidance lists what that covers. It includes "adequate understanding of the system to be worked on and practical experience of that class of system" (HSR25, paragraph 239).
Read that again. "That class of system." A sparky who's spent ten years on domestic rewires is competent at domestic rewires. That says little about your 400V motor control centre.
Build the test from your own kit
A good trade test takes about an hour. Keep it close to the jobs they'll really get. Here's how we'd set one up.
Electrical. Take an old starter panel or a training rig. Put one fault in it: a blown control fuse, a broken wire on an overload contact, a coil gone open. Hand them the drawing and a meter. Then watch. Do they isolate and prove dead before they put a hand in? Do they prove the tester on a known source first? Do they work the drawing, or just poke about? Our safe isolation post covers what good looks like.
Mechanical. A pump and motor on a baseplate, slightly out of line. Ask them to check it and say what they'd do. Or a gearbox with a worn bearing and a box of parts. Ask how they'd get it out and what they'd check before it goes back.
Fault-finding on paper. Give them a real alarm history from one of your machines and the fault log for the month. Ask what they'd look at first. A good engineer narrows it down step by step. You'll see it in how they talk. Our post on half-split fault-finding shows the method you're hoping to see.
Controls, if the job needs it. Laptop, the right software, a copy of the program. Ask them to find why an output isn't firing. Online access only, no changes. If they go straight for forcing bits, that tells you something too.
Use the same test for everyone going for the same job, so you're comparing like with like.
Watch how they work
Getting the right answer matters less than you'd think. Plenty of good engineers miss a planted fault in an hour on strange kit. What you're watching for is the order they work in.
Good signs:
- they isolate before they open anything, without being asked
- they ask for the drawing, or ask what changed last
- they say what they're checking and why
- they stop and ask when they're not sure
Bad signs:
- they skip proving dead "because you told me it's off"
- they start swapping parts before they've found anything
- they can't explain what they just did
That last list is the one that costs you. The engineer who swaps parts at random will get some machines running. They'll also run up your spares bill and leave you with the fault that keeps coming back.
Agency and contract engineers too
The same duty applies to someone you bring in for a shift. You still need to know they're competent for the job you hand them. A full trade test for a one-night callout isn't realistic, but a check is. Ask for the exact kit they've worked on, as well as the trade. Ask what they'd do first on your fault. Then watch the first hour. Our post on what a good cover engineer looks like on the first shift has the signs.
Where someone is close but not quite there, the law gives you another route. HSE's guidance on regulation 16 allows for "some degree of supervision" for people without enough knowledge or experience (HSR25, paragraph 241). That works on days with a senior engineer about. It doesn't work on nights when they're on their own.
Our view
Run a trade test for every permanent hire. One hour on your kit beats any amount of talking. Count the cost the other way. A wrong hire costs a few months' wages, and some breakdowns that ran hours longer than they should have. Put your own downtime cost per hour against that and the hour spent testing looks cheap.
What to do on Monday
- Pick one electrical and one mechanical fault you can set up on spare kit, and write them down as your standard test.
- Write a short score sheet for how they work: isolation, method, explaining, asking.
- For every engineer already on nights alone, check you've seen them work on the kit they cover.
* Paragraph numbers refer to L22, fourth edition (2014, amended 2018), and HSR25, third edition (2015). Correct as of 9 September 2026. Check HSE's competence pages for the latest.