AI Candidate Screening: Five Methods Compared for High Volume Frontline Roles

What is AI candidate screening?

AI candidate screening is software that sorts, questions or assesses applicants first.

It covers very different tools, and all of them can reply to a thousand applicants at 21:00 on a Sunday.

The better question is what the recruiter knows about each applicant afterwards. A warehouse, logistics, cleaning, security or hospitality job usually comes down to a handful of requirements: a valid certificate, the shifts the person can work, a start date, enough of the language on the floor to follow safety instructions, and the right to work. A method is useful in proportion to how much of that list it covers, and how much of what it reports a recruiter can check.

Why compare screening methods by evidence rather than speed?

Because all five methods are quick, but they show different things.

Sackett, Zhang, Berry and Lievens write that their findings are not a mandate to prefer one method regardless of circumstances: validity, cost, time, volume, differences between groups and applicant reactions all need weighing together (Sackett et al., 2023). Their broad finding still helps: methods built around the specific job, such as structured interviews, job knowledge tests, work samples and empirically keyed biographical questionnaires, tended to predict overall job performance better than general measures of ability or personality.

We add one practical test for frontline roles: for each requirement, does the method give the recruiter something to look at, or only a yes or no from a machine?

How do the five screening methods compare?

Each method answers a different question about the applicant.

See how a structured voice interview reads each answer against your requirements on the AI Interviewer page.

What does a CV filter actually check?

It checks what applicants wrote about themselves, not what they can do.

Whether it uses keywords or a language model, a CV filter only works with the words on the page. For frontline roles that page is often short, out of date, in a second language or missing, because many people apply from a phone. A filter that favours well written CVs partly filters for writing, and AI writing tools now make a polished CV easier to produce.

The methods that top the selection research are built around the job, such as structured interviews, job knowledge tests and work samples; reading a CV is not one of them (Sackett et al., 2023). We look at what years of experience can and cannot tell you in Three Years on Reach Trucks Tells You Less Than You Think.

A CV filter helps as intake, pulling out stated certificates and contact details. It hurts as the step that decides who is never read, more so when it sorts everyone into one ordered list, as we discuss in why we count evidence per requirement.

Where do knockout questions help, and where do they hurt?

They help on a few hard facts and hurt everywhere else.

Answer "no" to "Can you work nights?" and the application stops. If a role is nights only, that is reasonable. The trouble starts when knockout questions multiply or get vague. Someone with two years of order picking may answer "no" to "Do you have warehouse experience?" because they think it means forklift work. At volume nobody sees these cases, because nobody reviews a rejection the system made on its own.

Legally, the GDPR does not ban automated screening. It gives candidates the right not to be subject to a decision based solely on automated processing that significantly affects them, such as an automatic rejection, unless an exception applies: necessity for a contract, authorisation by law, or explicit consent (GDPR, Article 22). Where an employer relies on the contract or consent exception, candidates must at least be able to get human intervention, give their view and contest the decision. The Dutch Data Protection Authority (AP) says that in hiring assessments, a fully automated score that decides how an application proceeds affects the applicant significantly in almost all cases (Autoriteit Persoonsgegevens).

The practical rule: keep knockout questions to the few facts that truly end an application, word them plainly, and send a "no" or a missing answer to a person who can overrule it.

What can chat screening add?

Chat adds speed and reach, but answers stay short and self reported.

A chat can open a conversation minutes after an application, at any hour, and works well for availability and simple logistics. But applicants less comfortable typing in the chat language are at a disadvantage even when they would do the job well. Everything a chat collects is the applicant's own statement: enough for "I can start on Monday", not enough for "I hold a valid reach truck certificate".

Since 2 August 2026 the EU AI Act has required providers to design AI systems that talk directly with people so that those people are told they are dealing with AI, at the latest at the first interaction, unless that is obvious from the context (AI Act, Article 50(1) and (5)).

Why do structured interviews hold up best in the research?

Structured interviews had the highest average validity in recent research.

In their 2022 reanalysis of earlier summaries of selection research, Sackett and colleagues found that structured interviews had the highest average validity (how closely a method's results relate to overall job performance) of the methods they compared: 0.42 (Sackett et al., 2022). They also stress the spread. The 80% credibility interval runs from 0.18 to 0.66, so they suggest reading it as "0.42, plus or minus 0.24" (Sackett et al., 2023). The authors read that spread partly as a call to invest in careful design, so a badly built structured interview may predict far less. The research detail is in Structured vs Unstructured Interviews, and the method in How to Run Structured Interviews at High Volume.

Structure means the same questions for every applicant, taken from the role's requirements and rated against rules written in advance. Run by voice, it suits frontline applicants: speaking is easier than typing on a phone, follow up questions turn "yes, I have a certificate" into "reach truck, valid until 2028", and a recording lets a recruiter check what was said.

Three cautions apply. The interview is only as good as its questions. Speech recognition can make mistakes with accents and background noise, so the recording must stay available. And there is a hard legal line. Since 2 February 2025 the AI Act has banned AI systems that infer people's emotions from biometric data, such as facial expressions or voice, in the workplace, except for medical or safety reasons (AI Act, Article 5(1)(f)). The European Commission's guidelines, which are not binding, say this ban also covers candidates during recruitment (European Commission, 2025). A voice interview should record answers to job questions and nothing about how the person sounds or looks.

When are assessments worth the extra step?

When a skill can be tested and is needed from day one.

Assessments range from situational judgement questions to job knowledge tests, work samples and ability or personality questionnaires. Job knowledge tests and work samples sit near the top of the research for predicting performance. The catch is in the same paper: the authors write that they are not generally applicable to entry level hiring where knowledge and skills are developed after hire, through training or on the job (Sackett et al., 2023). Many order pickers, cleaners and hospitality staff learn the job on site.

A long test before anyone has spoken to an applicant also asks a lot of them. Use one where a skill is truly needed on the first shift and can be shown, such as a short practical check on a reach truck at the site before the offer.

What can no screening method check on its own?

Identity, right to work and certificates need a check of the document.

All five methods collect claims about documents. A CV lists a certificate and an interview records "yes, I have VCA". None of them shows that the certificate exists, is still valid or belongs to the person. That takes a separate check of the document itself, ideally before the offer rather than on the first day. We set out which checks can happen before the offer in Verifying Frontline Candidates.

What does each method show for one warehouse role?

Run one role through all five and the gaps become visible.

Illustrative example. A staffing agency fills reach truck shifts at a distribution centre for a client. The client sets five requirements. This is what a recruiter would see after each method.

No column is complete. Only a document check confirms a certificate or the right to work, so the order of the steps matters more than the choice of one tool.

How should you combine screening methods for high volume hiring?

Use fast methods to gather information, not to reject people.

A flow that holds up at volume has five steps.

Check any screening setup, built or bought:

What do the AI Act and the GDPR mean for AI screening?

Hiring AI is listed as high risk; those rules apply from December 2027.

Under the EU AI Act, AI systems intended to analyse and filter job applications or evaluate candidates are listed as high risk in Annex III. A narrow exception exists for tools that do not materially influence hiring decisions, but it never applies where the system profiles candidates (AI Act, Article 6(3) and Annex III). The rules for high risk AI in employment and recruitment apply from 2 December 2027. That date comes from the Digital Omnibus on AI, Regulation (EU) 2026/1744, published in the Official Journal on 24 July 2026, which moved it from the original 2 August 2026 (Regulation (EU) 2026/1744; European Commission). The bans on prohibited practices have applied since 2 February 2025, and the GDPR applies in full now. Deployer duties are set out in our guide to the EU AI Act and recruitment.

What are the limits of this comparison?

The research is general, and your own roles may differ.

The best test is your own data: how many applicants you lose at each step, and how the people you hire do in their first weeks.

What does this mean for staffing agencies and frontline employers?

Agencies need evidence that travels; employers need it per site.

An agency screens for several clients, and one applicant may fit roles at two or three of them. A knockout rejection for one client's night shift can lose a good candidate for another client's day shift. Evidence per requirement lets a consultant put someone forward with reasons the client can check.

An employer hiring for several sites needs the same detail per site: nights at one, Dutch on the floor at another, and a start date for the planner. For both, a step that rejects quietly can be the most expensive kind, because nobody sees what it cost.

How Radius helps

Radius Hire is an AI hiring platform for frontline roles in the Netherlands. Its AI Interviewer, Match and Radius Passport sit on one candidate record.

The AI Interviewer gives applicants a short, structured voice interview they can take on their phone, whenever it suits them, in the language set for the role. Candidates are told at the start that the interviewer is automated and that a person reviews the result. With Match, for each requirement Radius shows whether it is backed by a document or a check, rests only on what the candidate said, or is still missing, with the moment in the interview that answered it.

The AI Interviewer does not check documents. Where the employer uses Radius Passport, it checks identity and right to work with a trusted verification partner and records certificates with their expiry dates, so a recruiter can review the evidence before an offer. It does not replace the employer's own legal duty to check identity.

Radius never rejects a candidate on its own. A recruiter reviews the evidence, can overrule anything Radius presents and makes every hiring decision. Radius is new, so this explains how the product works rather than claiming results. More for agencies is on the For agencies page.

Conclusion

AI candidate screening is a set of methods that are not interchangeable. A structured interview tied to the job gives a recruiter the most to check and has the strongest research behind it. None of the methods proves a certificate or the right to work, and none should be the one that says no. Gather evidence fast, check documents before the offer and let a person decide.

General information, not legal advice. Position as of 22 October 2026.