What is candidate screening? Methods, mistakes and what the research says

Candidate screening is the step between receiving applications and interviewing: checking each applicant against the job’s requirements to decide who is worth a closer look. It usually means some combination of screening questions, reading resumes, short phone calls and, for some jobs, a short test or work sample. Background and reference checks are screening too, but many employers run them near the end, often after a conditional offer.

Screening decides who you never meet, so its mistakes are invisible. A bad interview is noticed. A good applicant screened out on day one isn’t.

Screening, selection, recruiting and sourcing

The terms overlap, and people use them loosely. A workable set of definitions:

Term	What it covers
Sourcing	Finding people who might apply, or approaching people who haven’t
Recruiting	The whole job of filling an open role, from posting to offer
Screening	Narrowing the applicants to those worth assessing in depth
Selection	Assessing the remaining candidates in depth and choosing one

Screening and selection use some of the same tools. The difference is depth: screening asks “could this person do the job?” across many applicants; selection asks “who would do it best?” across a few.

The main screening methods

Method	What it tells you	Cost to you	Watch out for
Knockout questions	Yes-or-no facts: a licence, right to work, a shift	Almost none	Using them for judgements (“5+ years?”) instead of facts
Resume and application review	What the person says they’ve done	Minutes per applicant	Reading for polish, keywords or proxies rather than evidence
Phone or video screen	Interest, logistics, a first check on a key requirement	15–20 minutes each	Turning it into an unstructured interview
Work sample or skills test	Whether they can do a piece of the actual work	Setup time, applicant time	Long unpaid tasks; tests unrelated to the job
Background and reference checks	Whether stated facts are true	Fees and time; consent rules	Running them on everyone, early; asking referees for opinions instead of facts

Our guides cover several of these in depth: screening a large pile of resumes, structured interviews and the hiring process from start to finish.

What predicts performance, and what doesn’t

Methods differ a lot in how well they predict how someone will do in the job. A 2022 re-analysis of selection research by Sackett and colleagues estimated these correlations with later job performance, where 0 means no relationship:

Method	Estimated validity
Structured interview	.42
Job knowledge test	.40
Work sample test	.33
Unstructured interview	.19
Years of experience	.07

Two lessons follow for screening. First, the easiest thing to screen on, years of experience, is among the weakest predictors. Second, the stronger methods are ones you can’t run on every applicant, so the job of the first screen is to avoid ruling out good people cheaply, not to find the best one.

Common screening mistakes

• Screening on proxies. Degrees, year counts and well-known employers are quick to check, which is why they get used. They stand in for something you could ask about directly.
• Treating “not mentioned” as “doesn’t have”. An application that’s silent on a requirement is unclear, not a fail. In a survey of 2,275 employers, Harvard Business School and Accenture (2021) found that 88% of those using software to filter or rank applicants said it screened out qualified high-skills candidates at least some of the time.
• Matching keywords. Someone who “closed the books every month” has done the month-end close, whether or not they used those words. See what AI resume screening should actually do.
• Changing the bar as you go. The fiftieth application gets read more harshly than the fifth. Write the requirements down first and read every application against the same list.
• Screening on things the law protects. See below.

The legal line

Every screening method has to be about the job. In the United States, the Equal Employment Opportunity Commission’s guidance on employment tests and selection procedures explains that Title VII prohibits “neutral tests or selection procedures that have the effect of disproportionately excluding persons” in protected groups where they are “not ‘job-related and consistent with business necessity.’” Determining whether a procedure has that effect “ordinarily requires a statistical analysis”, and the guidance notes that when an employer uses a test developed by someone else, “the employer is still responsible for ensuring that its tests are valid”.

Other jurisdictions have their own rules. Since 1 January 2026, for example, most Ontario employers with 25 or more employees must leave Canadian-experience requirements out of publicly advertised postings, and must say in the posting whether AI is used to screen, assess or select applicants. Background checks have their own consent rules; in the U.S. the Fair Credit Reporting Act requires a stand-alone written disclosure and the applicant’s written permission. Our articles on job posting disclosure rules and offer letters link to the official sources. This is not legal advice.

Screening with software

Most applicant tracking systems and many screening tools can filter or rank applicants automatically. The questions to ask of any of them are the same as for a person doing the screening: what is it checking against, can you see why it made each call, what happens to applicants it isn’t sure about, and who makes the final decision? Can AI help you identify the best applicants? looks at what the evidence supports.

Where SortTheStack fits

SortTheStack is a screening tool, not an applicant tracking system. It reads every application against your Must have and Nice to have requirements and marks each one met, not met or unclear, with the evidence quoted. Unclear is its own outcome, so applicants aren’t dropped for leaving something out. Knockout questions are yes-or-no rules you write, with no AI involved, and a person decides who to interview. See how the screening works.

Sources

• Fuller, J. B., Raman, M., Sage-Gavin, E., & Hines, K. (2021). Hidden workers: Untapped talent. Harvard Business School Project on Managing the Future of Work and Accenture.
• Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology, 107(11), 2040–2068. doi.org/10.1037/apl0000994
• U.S. Equal Employment Opportunity Commission, Employment tests and selection procedures (2007), accessed 19 October 2026.
• Ontario Ministry of Labour, Immigration, Training and Skills Development, Requirements related to publicly advertised job postings, accessed 12 October 2026.
• Federal Trade Commission, Background checks: What employers need to know, accessed 13 October 2026.