Skills-based hiring: what it is and which pre-employment assessments predict performance

Skills-based hiring means choosing candidates on evidence of the skills a job needs, such as a work sample or a structured interview, wherever they learned those skills. A degree or a number of years in a similar role becomes one clue among several. SHRM's 2026 research describes candidates being evaluated "primarily on their competencies and skills, regardless of how the skills were acquired." The tools that gather that evidence are pre-employment assessments, and the research on them finds that job-specific ones predict job performance best, with structured interviews at the top.

How many employers hire on skills

About a third do it routinely in the United States. In a SHRM Foundation press release dated 12 February 2026, "34% of organizations reported often or almost always using skills-first strategies in their hiring processes." Of the organisations that rarely or never hire this way, 55% said they were interested in starting. About two-thirds of HR professionals who use these strategies said they get more qualified applicants.

SHRM surveyed 1,184 US HR professionals between 13 and 27 October 2025, alongside supervisors and workers. Its report lists how often the HR respondents' organisations have used three practices:

In the same report, about 85% of HR professionals said their organisation would be more likely to treat a candidate without a college degree as meeting the job's requirements if a reliable assessment showed the skills. Organisations that know how to assess candidates' skills were more likely to meet or exceed their financial objectives, 63% against 51%. That is a survey correlation, and it does not show that assessing skills caused the result.

The OECD sees the same gap across its member countries. Its report A Skills-First Labour Market, published on 18 June 2026, says: "Recruitment and career progression continue to rely heavily on qualifications and professional experience as proxies for skills." It adds that employers "face practical difficulties in assessing skills reliably and at scale".

Skills-based hiring compared with degree and experience screening

Years of experience predict little on their own. Dropping a degree requirement is not enough either. The OECD report says that "removing qualification requirements alone is insufficient to broaden opportunities", and that inclusive skills-first hiring needs "alternative entry routes that allow candidates of all backgrounds to demonstrate competence." An employer that removes the requirement still needs a way to check the skills.

Pre-employment assessments, and how well each predicts performance

A validity figure is the correlation between the score on an assessment and later job performance. Zero means the score tells you nothing, and 1 would mean it predicts performance perfectly. The figures below come from meta-analyses, which pool many studies and correct for measurement problems.

The classic summary is Schmidt and Hunter's 1998 review of 85 years of research. In 2022, Paul Sackett and three colleagues re-ran the numbers. They found that earlier meta-analyses had applied a statistical correction (for range restriction) to studies that did not need it, which inflated many estimates. Their 2023 open-access article prints both sets side by side:

Source: Sackett, Zhang, Berry and Lievens (2023), Table 1. That article later raises assessment centres to .33, using a reliability figure for managerial jobs.

The work sample figure had been revised before. Roth, Bobko and McFarland's 2005 meta-analysis found a mean observed correlation of .26 between work sample tests and job performance, or .33 after correction, against the .54 that Hunter and Hunter reported in 1984.

Four points from the research matter when you choose an assessment:

1. The job-specific methods lead. Sackett and colleagues write that the top predictors "are those specific to individual jobs, such as structured interviews, job knowledge tests, work sample tests, and empirically keyed biodata."
2. Averages hide a wide spread. For structured interviews, the authors give an 80% credibility interval of .18 to .66, and say the validity "should really be viewed as '.42, plus or minus .24'". A carelessly built interview can sit at the bottom of that range.
3. Some methods need experienced applicants. The same article says work samples and job knowledge tests are "not generally applicable for entry-level hiring" where skills are learned after hire, and that structured interviews are not readily suited to high-volume screening. It points to research on automated scoring of asynchronous video interviews as one way to adapt interviews for volume.
4. Combining methods still works. Across every combination of one to five widely used predictors, mean validity was .47 with the 2022 figures, against .51 with the older ones.

Where assessments sit in candidate screening

Candidate screening is the step between application and the hiring team's interviews. In a skills-based process, an assessment does most of the filtering that a CV read used to do. One workable order:

1. Define the skills the role needs and how each will be scored. SHRM found 76% of organisations had defined skills and competencies for each role.
2. Write the job ad around those skills, and keep only the requirements the job really has, such as a licence or the right to work.
3. Give every candidate who meets those requirements the same assessment, early. A structured interview suits most roles; add a work sample or a knowledge test when candidates should already be able to do the work.
4. Score every answer against the same rubric, and keep the evidence behind each score.
5. Bring the strongest candidates to interviews with the hiring team, and have a person make the decision.

Speed is a real constraint. In SHRM's survey, 23% of HR professionals named pressure to hire quickly as a barrier to skills-first hiring, and 29% named the lack of tools or processes for assessing skills.

Common pitfalls

Testing the wrong skill

An assessment predicts only if it measures what the job uses. Sackett and colleagues recommend "a careful job analysis" and measures that sample job behaviour, either directly through a work sample or through "tell me about a time when" interview questions.

Adverse impact

In the US data Sackett and colleagues use, cognitive ability tests show the largest gap between groups, and "a system relying solely on cognitive ability will generally produce substantial adverse impact." With the 2022 figures, they found combinations of assessments with and without a cognitive test reached almost the same validity, so an employer can keep validity while reducing that gap.

An unvalidated tool

A vendor's average is not your result. The authors advise employers to look for validity studies "in situations similar to that under current consideration", including in test publishers' archives. Ask a vendor what evidence it has for jobs like yours.

Changing the label and nothing else

If the job ad drops the degree but the screener still reads CVs for one, nothing changes. The OECD's warning about removing requirements "alone" applies here.

Inconsistent use

In SHRM's survey, 25% of HR professionals named inconsistent hiring practices across teams as a barrier.

Hong Kong and Singapore notes

The guidance in both places is about how candidates are treated during selection and how their personal data is used. Work rights are a separate check: in Singapore, the Employment Pass salary floor applies to an Employment Pass application whatever the assessment shows.

In Singapore, TAFEP's recruitment pages explain "how to recruit fairly, from job advertisements to job application forms and job interviews." If an assessment uses AI on candidates' personal data, the PDPC's Advisory Guidelines on use of Personal Data in AI Recommendation and Decision Systems (issued 1 March 2024) apply. Under the PDPA, consent is generally needed and people must be told the purpose. The OECD report also notes that Singapore, like France, offers skills passports that combine records of employment and training with skills information, which individuals can share with employers.

In Hong Kong, the Privacy Commissioner for Personal Data's AI framework of June 2024 lists the assessment of job applicants among AI uses that may be higher risk, and expects a person to stay in control of high-risk decisions. The rules for AI in hiring in both places are set out in what is an AI interview.

This is general information, not legal advice; check the official source for your case.

How ROUND.1 fits

ROUND.1 is a hiring marketplace for Hong Kong and Singapore, built by h.work. Each candidate takes one AI first-round video interview, in a web browser with a camera and a microphone. It lasts about 20 minutes by default (roles can be shorter), 30 at most, and each role is interviewed in one language, English or Mandarin Chinese.

The interview is scored competency by competency, and each score is backed by quotes from the transcript. Accent, pronunciation, grammar, fluency, personality traits and protected characteristics (such as age, gender, race, religion, disability or marital status) are not scored. The same interview is reused for the other roles the candidate chooses, so a candidate is not interviewed again for each company.

A company first sees the scored evaluation without the candidate's name or contact details, which open when it shortlists the candidate. A person at the company decides whom to hire. The validity figures above come from published research on assessment methods in general. None of them is a measurement of ROUND.1's interview.

The full list of facts and limits is on ROUND.1, explained. How the format works for candidates is in one-way video interviews and the STAR method, and agency pricing is compared in how much recruiters charge.