Skills-Based Hiring in Practice: A Step-by-Step Guide for Employers
Skills-based hiring means choosing people by what they can do, not by degrees or titles. Here is how to run it: define the skills, choose evidence, score a work sample and verify what you see.
The basics
What skills-based hiring is
Skills-based hiring means choosing candidates by what they can demonstrably do, rather than by degrees, job titles or where they have worked. The role is described as a set of skills, and candidates are assessed against them with evidence.
It is easy to announce and hard to practice. A Harvard Business School and Burning Glass Institute study found that many employers who dropped degree requirements from job postings changed little about who they actually hired. Real change needs a different process, not different wording.
- A claim on a resume
- A certificate you cannot check
- A structured interview answer
- A verified credential
- A reviewed work sample
Why employers are doing it
The reasons are practical:
- A wider pool: capable people who learned outside the usual routes become visible.
- Better signal: a degree or a past title is a rough stand-in for ability.
- Fewer surprises: you see the work before you commit to the hire.
- Defensible decisions: shared criteria are easier to explain than gut feel.
- AI changes the picture: polished resumes and cover letters are cheap to produce, so claims on paper carry less weight.
How to do it
Step 1: define the skills the role needs
Start from the work, not the job title. List the three to six skills that decide whether someone succeeds in the first six months, and say what good looks like for each in one sentence. Long lists dilute the signal. Ask people already doing the job what they actually use.
Step 2: choose the evidence for each skill
Match each skill to the strongest evidence you can afford to collect:
- A work sample: the candidate does a small piece of the real job.
- A structured interview: every candidate gets the same questions, scored on the same scale.
- A verified credential: a record you can check with the issuer, such as an Ewance certificate.
- A reference or a portfolio piece you can review.
Research on selection methods has been revisited recently. Sackett and colleagues found that earlier studies had overstated how well many methods predict performance, and that structured interviews came out as the top-ranked single method, with other methods still useful. The practical lesson is to combine a structured interview with a work sample rather than rely on either alone.
Step 3: design a work sample
A good work sample is short, realistic and the same for everyone. Aim for 60 to 120 minutes of candidate time. If it needs more, pay for it. Give a brief, the data or materials needed and a clear deliverable. Do not use real confidential material, and do not use the task to get free work.
Step 4: score it with a rubric
Decide the criteria before you see any submissions, score each one separately, and have two people score independently when you can. A simple five-point scorecard is enough:
| Criterion | 1: Weak | 3: Solid | 5: Strong |
|---|---|---|---|
| Framing the problem | Misses what was asked | States the question and constraints | Reframes it usefully and says what is unknown |
| Quality of the work | Errors or gaps that block use | Correct and usable | Correct, clear and ready to act on |
| Checking and judgment | No evidence of checking | Checks the main claims | Tests assumptions and says what could be wrong |
| Communication | Hard to follow | Clear and organized | Concise, with a clear recommendation |
| Following the brief | Ignores key requirements | Meets the requirements | Meets them and flags trade-offs |
Adjust the criteria to the role. The point is to score what you said mattered, in the same way for every candidate.
Step 5: talk about the work
Use the interview to probe the work sample, not to repeat the resume. Ask how they approached it, what they would change and how they checked the result. If AI tools were allowed, ask how they were used. Someone who did the work can explain it in detail; someone who only pasted a result usually cannot.
Step 6: verify what you can
Check credentials with the issuer instead of accepting a PDF. A credential that carries its own proof can be confirmed in seconds, which is the point of verifiable credentials and certificate verification. Treat unverifiable claims as hypotheses to test in the work sample.
Step 7: pilot it and measure
Start with one role. Track who applies, who passes each stage, and how hires perform and stay. Compare with your previous process. Adjust the task and rubric after the first round, and only then extend to other roles.
Watch-outs
Skills-based hiring in the AI era
Candidates will use AI on take-home tasks. Rather than trying to ban it, which is hard to enforce, allow it and assess what AI cannot do for them: framing the problem, checking the output and explaining their choices. Ask for a short note on how they worked. That makes the work sample a better test of AI fluency as well.
Fairness, law and candidate experience
Skills-based methods can widen access, but only if they are job-related and applied consistently. Use the same task, time and criteria for everyone, offer reasonable adjustments, and check your local employment rules on selection and data. Be respectful of candidates' time: keep tasks short, pay for long ones and give feedback.
Wrap-up
An example
Illustrative case: a logistics firm hiring an operations coordinator replaces a degree filter with a 90-minute task. Candidates reconcile a messy set of shipment records and write a short note for a manager. Two staff score each submission on a five-criterion scorecard. The shortlist includes two candidates the resume screen would have dropped, and in the interview they can explain exactly how they found the errors.
Common mistakes
- Dropping degree requirements from the job post and changing nothing else.
- Listing too many required skills.
- Using a work sample that is too long, unpaid or based on real confidential work.
- Scoring without a rubric, or changing the criteria after seeing submissions.
- Treating self-reported skills or unverified certificates as proof.
- Never checking whether the new process improves hires.
Reference
Frequently asked questions
- What is skills-based hiring?
- Choosing candidates by what they can demonstrably do, using evidence such as work samples, structured interviews and verified credentials, instead of relying mainly on degrees and job titles.
- Does skills-based hiring mean you do not look at degrees?
- No. A degree can still be evidence of some skills. It just should not be the only way in, and it should not replace direct evidence of the skills the role needs.
- How long should a work sample test be?
- Aim for about 60 to 120 minutes of candidate time. If it needs more, pay for the time and keep it realistic.
- How do you stop candidates using AI on a work sample?
- It is hard to prevent and often unnecessary. Allow it, ask for a short note on how they worked, and score the framing, checking and explanation, which AI cannot do for them.
- How do you verify a candidate's skills?
- Review the work itself, run a structured interview about it, and check any credential with its issuer, ideally through a public verification page.
- Is skills-based hiring fair?
- It can be fairer than screening on proxies, if the task is job-related, the same for everyone and scored against set criteria. Check your local rules on selection.
- How do you get started?
- Pick one role, define three to six skills, design a short work sample and rubric, pilot it with a small group and compare results with your usual process.