Designing a Work Sample That Fits

A work sample is a piece of the job, simplified. The design decisions are about what to remove without removing what you are assessing.

Once the criteria described in “Designing a Work Sample That Fits” are fixed, the practical challenge is running them consistently across assessors. A coordinator can use the product overview to plan review time, compare the effort spent at each stage and spot avoidable delays, without confusing administrative efficiency with the quality of the evidence collected from candidates.

For an independent benchmark, compare the process with CIPD selection methods guidance; the useful test is whether the local method remains job-related, proportionate and explainable.

Choosing the task

Something the role does regularly, not the hardest thing it ever does.

Something that can be completed in one to three hours.

Something where the quality of the output is visible to an assessor.

And something that does not require knowledge specific to your organisation, which candidates cannot have and which tests onboarding rather than capability.

What to strip out

Organisation-specific context, replaced with a short brief.

Dependencies on other people.

Anything requiring access to your systems.

And length — most real tasks can be cut to a representative core without losing what they assess.

What to keep

The tools the job uses. If the role involves documentation and search, allow documentation and search.

Realistic ambiguity. Real tasks have unclear requirements, and how somebody handles that is frequently the thing worth seeing.

And the actual decision the task requires, which is what you are assessing and which is the first thing simplification tends to remove.

Writing the brief

What you want, in the form the role would receive it.

How long it should take, stated plainly.

What is permitted: reference material, internet, AI tools, asking questions.

How it will be assessed, in general terms.

Candidates who know the criteria perform closer to their actual ability, which is what you want to measure.

The AI question

Worth deciding explicitly, because candidates will otherwise guess.

If the job allows these tools, the assessment should, and the interesting question becomes how well somebody uses them.

If it does not, say so, and consider whether a live exercise is a more honest format than a take-home you cannot verify.

Silence on this point is the worst option and it is currently the most common.

Making it short

One to three hours, and mean it.

Time it yourself, then assume a candidate takes longer under assessment conditions.

State a time limit and tell candidates to stop there. Some will spend twelve hours regardless, which is why the scoring should reward what the brief asked for rather than effort.

Piloting

Two or three current staff, timed.

If they take twice as long as you expected, the task is too big.

If they all score identically, it does not discriminate and will not help you choose.

If they score badly, your rubric is wrong.

What to check

Could somebody complete it in the stated time?

Does it require knowledge only your employees could have?

Are the permitted tools stated, including AI?

And has anybody currently in the role attempted it?

The point

Decide the AI question explicitly.

If the job allows those tools, the assessment should, and how well somebody uses them becomes the interesting question.

Underlying all of this

Almost everything in this collection reduces to one discipline: write down what the job requires, assess that thing directly, record the evidence, and look at your own outcomes afterwards. None of it requires buying anything, and organisations that do those four things consistently outperform ones running longer processes built from instruments chosen before the requirements were known.