Skills-based hiring: stop interviewing the CV

Skills-based hiring means making decisions from evidence that someone can do the work and develop into the role. Removing a degree requirement is one possible step. It is not the entire method. If the interview still rewards familiar companies, fluent generalities and a manager's unexplained feeling, the old process has simply changed its heading.

LinkedIn's Future of Talent Report 2026 discusses a move beyond conventional hiring towards verifying and deploying capability. CIPD's September 2026 selection guidance provides a practical foundation for structured assessment. The useful question for an employer is narrower than a market slogan: what evidence would make this appointment a responsible decision?

Define the first outcomes before writing requirements

Describe what the person should make possible in the first few months. For a product role, that might mean clarifying a customer problem, creating an agreed priority decision and bringing a small improvement into use. For an operations role, it might mean making a recurring handoff reliable. Choose outcomes that reflect the job you can actually offer.

Then identify the capabilities needed: analysis, communication, delivery judgment, relevant domain knowledge and learning. Distinguish must-have entry skills from things the company can teach. A job description that requires mastery of everything may conceal the fact that nobody has decided what matters most.

Keep essential professional qualifications where the role or law requires them. Skills-based hiring does not mean ignoring credentials that genuinely establish eligibility. It means making each requirement earn its place rather than inheriting it from a job description last updated during a different business model.

Choose a small sample of real work

Give candidates an anonymised or synthetic scenario with enough context to act. Ask for a decision and its reasoning, not a complete unpaid consultancy engagement. A realistic exercise can reveal how someone frames uncertainty, asks questions and handles a trade-off.

For a hypothetical product leadership vacancy, provide a short set of customer complaints, a capacity constraint and two conflicting stakeholder requests. Ask the candidate what they would investigate first and what they would postpone. Avoid an answer key that silently requires knowledge of your internal politics.

Agree a time limit and tell candidates what a sufficient answer looks like. If you need substantial original work, discuss payment and appropriate arrangements. An elaborate take-home task can exclude good people with existing work, caring responsibilities or limited spare time. The exercise should improve the signal without demanding a second job.

State the AI rules before the exercise

If the role uses AI, consider allowing it during a suitable part of the task. Ask candidates to explain what they used, what they checked and what they changed. A polished answer is weak evidence if nobody can discuss how it was produced.

Use a live follow-up to introduce a new constraint. Ask how the recommendation changes and why. The goal is to see understanding and adaptation. It is not to catch someone in a theatrical trap or penalise them for declining to expose private accounts and unrelated files.

Greenhouse's 2026 US hiring research illustrates the trust questions around AI on both sides of recruitment. Tool use, authorship and dishonesty are different issues. Define your standard clearly enough that a candidate does not have to guess which one you are testing.

Score evidence before discussing impressions

Use a short rubric with observable indicators. For example, problem framing might require identifying the user and intended outcome. Prioritisation might require explaining an opportunity cost. Communication might require giving a recommendation the audience can act on.

Have interviewers record evidence independently before the group discussion. Otherwise, the most senior person's opinion can become everyone's memory of the interview. Keep a place for uncertainty: “we did not test this capability” is more useful than filling the gap with a confidence score.

Do not reduce the entire person to an arithmetic average. Some criteria are essential, some can be developed, and some are relevant only to a particular assignment. The rubric structures judgment; it does not eliminate the need to make and explain a decision.

Design fairness into the process

Tell candidates what will happen, which technology is used and how to request adjustments. Use comparable core questions and assess job-related evidence. Avoid treating camera presence, accent, speed of speech or social similarity as automatic proof of ability.

Review where candidates drop out. A lengthy process may be filtering for availability rather than competence. Check whether the exercise assumes equipment, software access or domain familiarity that the role does not actually require. Fairness improves when constraints are examined before they quietly become selection criteria.

CIPD's 2026 resourcing research offers wider context on recruitment practices. It cannot tell you that your particular exercise predicts performance. That requires following what happens after appointment and checking whether the evidence you valued was actually useful.

Connect the decision with onboarding

Turn the assessment into a development starting point. Record demonstrated strengths, unresolved questions and the support agreed for early work. Do not promise to teach a gap during recruitment and then treat the same gap as a surprising failure in week two.

Review the process after several hires. Ask whether the rubric predicted useful contribution, whether capable candidates withdrew and whether managers interpreted criteria consistently. Small samples call for caution; they can still reveal obvious process defects worth fixing.

For a specific leadership role, see how to assess an AI product leader. For the transition into work, read why onboarding fails. My experience across product, engineering and teaching is relevant when a company needs the role and its evidence made clearer. Bring the job you are struggling to define.

Questions people ask

What is skills-based hiring in practice?

Define the outcomes of the role, identify the capabilities needed and assess relevant evidence through structured questions and proportionate work samples. Removing degree requirements may widen access, but it does not by itself change how decisions are made. Retain qualifications genuinely required for the work or legal eligibility.

Should candidates be allowed to use AI in a work sample?

State the rule in advance and match it to the job. If AI is allowed, ask how the candidate used it, checked the result and adapted the answer. A follow-up with a new constraint can reveal understanding. Do not rely on uncertain AI-text detectors as proof of dishonesty.

How long should a hiring work sample take?

Keep it proportionate to the evidence needed and explain the expected time. Use synthetic or anonymised material and avoid extracting free production work. If a substantial assignment is necessary, discuss payment and appropriate arrangements. Review whether the task excludes capable applicants for reasons unrelated to the role.