Pre-employment assessments: a validity-first guide to testing the right things

There is a quiet irony in how most companies hire: they will spend weeks debating a candidate's résumé and background — the inputs that predict performance least — and then skip the one step that predicts it most, which is watching the person actually do a version of the work. Pre-employment assessments are how you fix that inversion. Done well, they replace speculation about whether someone could do the job with evidence of whether they can.

But “assessment” is a broad and abused word. It covers everything from a thoughtfully designed work sample to a gimmicky personality quiz with no bearing on the role, and the difference between those is not cosmetic — it is the difference between a tool that improves your hiring and one that just adds friction and legal risk. The organizing principle that separates them is validity: does this assessment actually measure something the job requires, and does that measurement predict who will succeed? Everything in this guide hangs off that question.

Start with validity, not with a tool

It is tempting to start by choosing a platform or a test type, but that is backwards. Start with the role. Decompose it into the three to five things a person genuinely has to be good at to succeed, expressed as tasks rather than traits — “can debug an unfamiliar codebase,” “can de-escalate an upset customer,” “can structure a financial model from messy inputs.” Those tasks are your specification. A valid assessment is one that puts the candidate in front of a faithful miniature of those tasks; an invalid one measures something adjacent and hopes it correlates.

This is why the research on selection methods is so consistent. Across decades of meta-analysis, the assessments that best predict on-the-job performance are the ones that most resemble the job: work-sample tests, structured interviews built around real scenarios, and validated tests of the general ability and job knowledge the role actually draws on. Unstructured interviews, years of experience and résumé screening sit much lower, because they measure proxies rather than the work. When you choose by validity, you are simply choosing to measure the thing instead of a shadow of it.

The assessment types worth your time

A handful of formats do most of the useful work. Work-sample tasks are the gold standard: a scoped, realistic piece of the actual job — a short take-home, a live exercise, a portfolio review — judged against a rubric. They predict well precisely because they are the job in miniature. Structured interviews are close behind and cheaper to run at volume; treated as an assessment with defined questions and scoring rather than a chat, they are one of the most reliable tools available (we cover the mechanics in the structured interview guide).

Job-knowledge and skills tests have their place for roles with a clear body of required knowledge, as do validated general-ability assessments when they are genuinely job-related and applied to everyone equally. The formats to be wary of are the ones that drifted in from elsewhere: unvalidated personality typologies used as pass/fail gates, brain-teasers that test composure-under-pressure more than competence, and anything where you cannot articulate the line from the test to the work. If you cannot explain why a strong score predicts strong performance, the test is decoration.

Fairness and candidate experience are part of validity

A common mistake is to treat fairness and candidate experience as soft considerations layered on top of a “real” technical decision. They are not separate — they are part of whether the assessment works. An assessment that is inconsistently administered, inaccessible to some candidates, or so long and unpaid that strong people drop out is not measuring ability cleanly; it is measuring ability tangled up with availability, accommodation and patience. That noise degrades the very prediction you are trying to make.

So the fairness checklist is also a quality checklist. Keep assessments tightly job-related, so you are never screening on something irrelevant. Administer and score them the same way for everyone, ideally against a rubric written in advance, so the result reflects the candidate and not the grader's mood. Respect people's time — a two-hour unpaid take-home filters for free time as much as for skill. And monitor outcomes across groups, so that if a test is producing large unexplained differences, you catch it and check whether it is genuinely measuring the job. These are not concessions to niceness; they are how you keep the assessment valid.

Running assessments at scale without the cost spiral

The practical objection to good assessment is cost. Work samples take time to design and grade; structured interviews take interviewer hours; doing either consistently across hundreds of candidates is exactly where most processes quietly revert to résumé-skimming. This is the gap that AI interviews are built to close — running the same structured, skills-focused conversation with every candidate, transcribing and scoring it consistently, so the assessment that used to be reserved for your top few can be offered to everyone without the cost scaling linearly with the queue.

The point is not to replace human judgment but to spend it well: let a consistent first-round assessment surface the people worth a deep human look, instead of burning your best interviewers on triage. That is skills-based hiring made operational.

How Spoon turns assessment into a default, not a project

Spoon bundles the validity-first approach into the product. Every candidate sits the same structured AI interview — a work-relevant conversation scored consistently — and recruiters receive an anonymized, skills-ranked shortlist rather than a stack of résumés. There is no test-design project to staff and no per-grader inconsistency to police, because the assessment and the scoring are the platform.

For higher-stakes roles, recruiters can layer in their own job-specific tests on top. See how it works for companies, or read the skills-based hiring pillar for how assessment fits the wider process.

Frequently asked

Keep reading