The ApprenticeRoute

Online tests, games and coding challenges: the stage most people go out at

The assessment centre gets all the attention. It's the bit people write about, the bit schools run mock versions of, the bit that feels like the real test.

It isn't where you get rejected. Almost nobody gets rejected at an assessment centre in the sense that matters, because by then the field has already been cut to a handful of people per place. You get rejected in January, by a computer, having spent forty minutes doing something that felt faintly ridiculous.

That's the stage this article is about, and it's worth taking seriously for one reason above all others: it is the only part of the whole process that responds reliably to practice.

What you'll actually be sent

Employers mix and match, but nearly everything you'll meet is a version of one of these five.

Situational judgement tests. You're given a workplace scenario and a set of possible responses, and asked to rank them or pick the best and worst. There's no maths, no trick, and the scenarios are deliberately mundane: a colleague misses a deadline, a client asks for something you can't deliver, you spot an error in someone else's work.

Numerical and verbal reasoning. Timed multiple-choice. Numerical gives you a table or chart and asks percentage-change and ratio questions; verbal gives you a passage and asks whether statements are true, false or cannot be determined. The maths is GCSE level. The difficulty is entirely the clock.

Game-based assessment. Increasingly common, and the one that throws people. You inflate balloons for points, sort shapes under time pressure, tap when a signal appears, choose between a certain small reward and a risky larger one. Nothing looks like work. It typically takes 15–30 minutes.

Coding challenges. For technology and engineering schemes specifically. Usually hosted on a platform like HackerRank or Codility: two or three problems, a timer, an editor in the browser, automated tests that run against your solution.

The one-way video interview. Pre-set questions appear on screen, you get a short prep window and then record an answer to a camera with a countdown running. No human is present. Most give you one take; some give you two.

Situational judgement tests: pick the boring option

The single most useful thing I can tell you about SJTs is that they are not testing your intelligence, your creativity, or your ability to take initiative. They are testing whether you would be a low-risk employee.

So the answer is almost always the boring, procedural, tell-someone-and-check one. Not the heroic one where you stay until midnight fixing it yourself. Not the decisive one where you make the call without asking. Escalate, clarify, follow the process, communicate early. It feels like you're being asked to describe a slightly spineless person, and in a sense you are, because in a regulated business the person who quietly improvises is the expensive one.

Two refinements. First, the options usually contain one obvious wrong answer (ignore it, hide it, blame someone), one obvious right answer, and two plausible middles — the ranking is decided in the middle, so read those two properly. Second, some SJTs are tailored to the employer's own values framework, which is published on their careers site. Twenty minutes reading it beforehand genuinely helps.

The games are not a gimmick

Game-based assessments feel unserious, and candidates treat them accordingly, which is a mistake. They come from behavioural psychology rather than from games: the balloon task measures how you calibrate risk after a loss, the shape-sorting measures how fast you infer a rule and how quickly you abandon it when it changes, the reward choices measure how you trade certainty against value.

There's no way to swot for that and no right answer to find. What you can control is the conditions: do it on a laptop rather than a phone, on a proper connection, in one sitting, when you're not tired. Read the instructions twice — a meaningful number of people score badly because they misunderstood the mechanic for the first ninety seconds, not because they lack the trait being measured.

The claim these tools make for themselves is that they're less biased than a CV screen, because they don't see your school. That claim is contested, and the honest position is that nobody outside the vendors has good public evidence either way. What I'd say is this: these tests were not designed to be intuitive, so finding them strange is not a signal about you.

Coding challenges

If you're applying for a software or data scheme, expect an automated coding test, and expect it to be pitched somewhere around "can you write a loop, use a dictionary, and handle an edge case" rather than at competitive-programming level.

Practical points that matter more than talent:

If you're applying for a tech scheme straight from school with no coding background, this stage is the one that will hurt, and it's worth several weekends well before applications open.

One-way video interviews

Everyone hates these, including the people who send them. Some notes from having done one and from watching the results end of the process:

Look at the lens, not at your own face in the corner. Everyone drifts to their own face, and on the recording it reads as looking away. Cover the self-view with a sticky note if you have to.

Don't script it word for word. Read-aloud answers are obvious, and the eye movement of someone reading a second screen is unmistakable. Bullet points on a card behind the camera are fine.

Use the structure anyway. The competency questions are the same four you'll get in person — teamwork, a setback, why us, learning something hard — so the same STAR stories work. The assessment centre article has more on building those.

Do a full test recording first, with the actual software, and watch it back once. Once is enough; twice and you'll talk yourself out of applying.

Finally, check whether there's a practice question. Most platforms offer one and it doesn't count. Use it.

Ask for adjustments — properly, and early

This is underused and it matters. If you're disabled, dyslexic, autistic, have ADHD, or have any condition that affects how you perform in a timed online test, you can ask for reasonable adjustments, and employers have a legal duty under the Equality Act 2010 to consider them for job applicants, not just employees.

In practice that usually means extra time, a screen-reader-compatible format, or the option of a live interview instead of a recorded one. You ask the recruitment contact, not the test vendor, and you ask before you sit the test rather than after you've failed it. You do not have to disclose a diagnosis to the hiring manager to get an adjustment from the recruitment team, and needing one is not a mark against you.

If you had extra time for your A-levels, that's the obvious signal you should be asking. A surprising number of people don't, because the exam board arranged it automatically at school and nobody told them the workplace version exists.

Practice, and what's worth paying for

Practice works here in a way it doesn't at later stages, because the formats are stable and the anxiety is most of the difficulty.

What's worth doing: the free practice tests on the test vendors' own sites (Aon, SHL, Sova, Arctic Shores and the rest all publish samples), a handful of timed numerical papers so the clock stops being a shock, and one full dry run of any video platform.

What isn't: expensive subscription "test prep" packages. There's a small industry selling these to anxious eighteen-year-olds, and the marginal value over free samples is not worth £80 to someone on a part-time wage. Your school or college careers service may also have a subscription already — worth asking before you spend anything.

And do the tests early in the window rather than on the deadline. Popular schemes fill on a rolling basis, which is covered properly in how to find degree apprenticeships and the application timeline.

The AI question, briefly

Don't. Not primarily for moral reasons, though those apply, but because it's the single fastest-growing thing employers are actively looking for.

The Institute of Student Employers' 2025 student recruitment survey covers 155 employers handling over 1.8 million applications between them. Around 85% of them had encountered some form of cheating in the application and assessment process, and roughly 79% reported redesigning or reviewing their recruitment because of AI. That redesign is why one-way video interviews are being replaced with live ones, why more employers re-test at the assessment centre, and why the gap between your online score and your in-person performance is now something people look at directly.

There's more on where the line sits once you're on a programme in AI and apprenticeship coursework.

The one-line version

The online stage rejects more people than every other stage combined, and it's the only one that reliably rewards preparation: pick the boringly responsible answer on situational judgement tests, do the games on a laptop when you're fresh, get a working solution before a clever one on coding challenges, and record a practice video once so the format isn't new.

Ask for adjustments before you sit anything, not after. And use the free practice tests rather than paying anyone.