Short answer: For most frontline roles, a structured voice interview on the candidate's phone is a better first step than the one way video recorded through video interview software. It asks less of candidates, gives recruiters the same job evidence and leaves out appearance, which says little about the work. Offer video only as an option, where seeing the person matters.
Key takeaways
- Both formats let candidates answer in their own time. The difference is what gets recorded: face and voice, or voice only.
- Structure matters more than the medium. The same questions for everyone, tied to the job, is what the research supports. A camera adds nothing to that.
- Video asks more of a frontline candidate (camera, light, privacy, comfort on screen) and puts appearance in front of the reviewer, a known source of bias.
- Recording an interview is not banned as such, though the GDPR applies, and the Dutch data protection authority says employers may not require a video application. AI that infers emotions from a face or voice has been banned at work since 2 February 2025, and the European Commission's guidelines, which are not binding, apply that ban to candidates.
- In any format, an answer is still a claim. Check certificates, identity and right to work before the offer, and let a person make the decision.
On this page: What is the difference between one way video interview software and a voice interview? · What does each format ask of a frontline candidate? · What evidence does each format give the recruiter? · What do EU and Dutch rules say about video and voice interviews? · Which format should you choose for frontline roles? · What should staffing agencies and frontline employers watch for? · What does the evidence not tell us? · How Radius helps · Frequently asked questions
What is the difference between one way video interview software and a voice interview?
Both let candidates answer in their own time, but video also records their face.
A one way video interview, also called an asynchronous video interview, shows the candidate a set of questions on screen. The candidate records an answer to each on camera, usually against a time limit, and a recruiter watches the recordings later.
A voice interview drops the camera. The candidate answers by voice on their phone, again at a moment that suits them. Some voice interviews read out fixed questions in order. Others are conversational: an AI interviewer asks a follow up question when an answer is vague, such as "which certificate, and until when is it valid?" The recruiter then gets the recording, a transcript or a summary, depending on the tool.
The labels hide a lot of variation. Lukacik, Bourdage and Roulin (2022) point out that asynchronous video interviews differ widely in design, and that research on them lags well behind how much employers use them [2]. So compare designs, not labels: how many questions, what time limits, whether a candidate can record an answer again, whether there is any follow up, and what the recruiter actually sees.
What does each format ask of a frontline candidate?
Video asks for more: a camera, light, privacy and comfort on screen.
Assume the interview happens on a phone. In the Netherlands, 96 percent of people aged 12 and over used a smartphone or mobile phone to go online in 2025, against 73 percent who used a laptop (NL, CBS, published 10 February 2026) [1]. An order picker, cleaner or security officer will most likely do your interview on that phone, after a shift, in a shared room or on a break.
Going online by device, NL, 2025
Share of people aged 12 and over
Both formats work on a phone; the difference is what else the candidate needs:
- For video: somewhere to prop the phone, decent light, a background they are happy for an employer to see, enough mobile data or wifi to upload video, and the confidence to talk to their own face on a screen.
- For voice: a quiet enough moment and a connection.
Research on applicant reactions points the same way, although none of it was done with frontline samples. In an experiment with 113 participants, Langer, König and Krause (2017) found that people saw recorded digital interviews as creepier and less personal than videoconference interviews, and reported more privacy concerns. The authors warn that applicants may drop out as a result [3]. A review of 12 studies by Blacksmith, Willford and Behrend (2016) found that applicants reacted less favourably to interviews held through technology than to face to face interviews [4].
Neither study compared video with voice only, so they do not prove voice is better. They do show that recording yourself on camera for a stranger is not a neutral request.
Two cautions before you drop video:
- A voice format is a barrier for candidates who are deaf or hard of hearing, or who have a speech impairment. Give them another route, such as written questions or a call with a recruiter who can adapt.
- Some candidates are more at ease on camera than on the phone. Either way, offer a clear alternative to anyone who cannot or would rather not use the standard format.
What evidence does each format give the recruiter?
Both give job evidence, but only when the questions are structured.
The strongest finding in selection research is about structure, not media. Sackett, Zhang, Berry and Lievens (2022) took a fresh look at decades of studies on how well selection methods predict job performance, and structured interviews came out on top [5]. A structured interview asks every candidate the same questions, tied to what the job needs, and judges the answers against the same criteria. That can happen on video, by voice or in a room.
Where the formats really differ:
- Appearance. Video puts a face, clothes, a home and a background in front of the reviewer. A review of experimental studies found that attractive people did better on a range of job outcomes, and that professionals were as susceptible to this bias as students (Hosoda, Stone-Romero and Coats, 2003) [6]. For an order picker or a cleaner, none of that is evidence. Voice does not remove every cue: accent, tone and fluency are still heard, so the review needs the same discipline.
- Follow up. A one way video cannot ask "and is that certificate still valid?" A conversational voice interview can, if it is built to. That turns a loose "yes, I have experience" into something a recruiter can check.
- Consistency. Blacksmith and colleagues found that interviewers rated candidates lower when the interview was held through technology, and warn against using different modes for different applicants [4]. Langer and colleagues also found that ratings differed between recorded and live video interviews [3]. Pick one standard format per stage for everyone applying to that role, with the same alternative route open to all.
| One way video interview | Structured voice interview | |
|---|---|---|
| What the candidate needs | A phone with a camera, light, a background they are willing to show, data to upload video | A phone, a quiet moment, a connection |
| Follow up on a vague answer | No, the questions are fixed | Possible, if the interview is conversational |
| What the reviewer sees or hears | Face, surroundings, voice and the answers | Voice and the answers, often with a transcript |
| Cues unrelated to the job | Appearance, home and clothing, plus voice and accent | Voice and accent |
| Recruiter review | Watch each answer | Read a transcript or summary, listen where needed |
| Main barrier for | Candidates uneasy on camera, low data, no private space | Deaf or hard of hearing candidates, speech impairments, noisy homes |
| Best fit | Roles where seeing the person matters for the work, offered as an option | Most frontline first interviews |
Want to see what a structured voice interview looks like for a frontline role? See how the AI Interviewer works.
What do EU and Dutch rules say about video and voice interviews?
Recording is not banned, but emotion reading AI is, and video must stay optional.
Five rules matter when you choose a format; the EU AI Act guide for recruitment covers the wider duties.
- Emotion recognition. Since 2 February 2025 the EU 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 (Article 5(1)(f)) [7]. The European Commission's guidelines on prohibited practices, which are not binding, say this ban also covers candidates during recruitment, so emotion reading video interview tools should not be used [8].
- Telling people they are talking to AI. Since 2 August 2026 the 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, unless that is obvious from the context (Article 50(1)) [7]. For an AI voice interviewer, the safe practice is to say so clearly at the start.
- High risk hiring systems. AI systems intended to analyse and filter job applications or evaluate candidates are listed as high risk in Annex III of the AI Act [7]. The obligations for those systems apply from 2 December 2027, a date set by the Digital Omnibus on AI (Regulation (EU) 2026/1744), which moved it from 2 August 2026 [9].
- Collecting only what you need. The GDPR requires personal data to be "adequate, relevant and limited to what is necessary" for its purpose (Article 5(1)(c)) [10]. If your decision does not depend on seeing the candidate, recording their face by default is worth discussing with your privacy officer before you choose video.
- Video must stay optional in the Netherlands. The Dutch Data Protection Authority (Autoriteit Persoonsgegevens) says employers may not require applicants to send a video application, because a video can reveal special categories of personal data. A candidate's consent to a video is valid only if they can also apply another way, for example by letter, without negative consequences [13]. This is regulator guidance, not a statute, so treat one way video as one of two routes, never the only one.
When the AI Act's rules for interview tools apply
- 2 February 2025Article 5(1)(f)AI that infers emotions from a face or voice is banned at work; the Commission's guidelines, not binding, apply this to candidates
- 2 August 2026Article 50(1)AI that talks with people must be built to tell them it is AI, unless that is obvious
- 22 October 2026Position as of this article
- 2 December 2027Annex IIIObligations for high risk hiring systems apply; moved from 2 August 2026 by the Digital Omnibus on AI
Which format should you choose for frontline roles?
Choose by what the job needs you to know, not by what the tool can record.
Before you pick a format, run each role through eight questions:
- Does seeing the candidate tell you something the job needs? If not, the camera adds risk without adding evidence.
- Are the questions the same for everyone, and tied to the requirements of the role?
- Can the interview follow up when an answer is vague?
- Can an applicant complete it on a phone, after a shift, on the day they apply?
- Can candidates answer in a language they are comfortable in, with the language of the work floor tested only where the job needs it?
- Is there a clear alternative for candidates who cannot or would rather not use the format, with no penalty for choosing it?
- Does a person review the evidence and make every decision, with nobody rejected automatically?
- Has the provider confirmed in writing that the tool does not infer emotions, and do you decide how long recordings are kept?
If the answers to questions 1 and 3 are "no" and "yes", a structured voice interview is usually the better first step.
Illustrative example: review time for one week of applicants. A distribution centre hires order pickers and gets 60 completed first interviews in a week. Each interview has six questions. The assumptions below are ours, so replace them with your own figures.
| One way video | Voice interview with transcript | |
|---|---|---|
| Assumption per candidate | 6 answers of 1 minute each, watched once at normal speed | 2 minutes reading the summary, 1.5 minutes listening to the answers that need checking |
| Formula | 60 × 6 × 1 minute | 60 × (2 + 1.5) minutes |
| Recruiter time | 360 minutes, or 6 hours | 210 minutes, or 3.5 hours |
Video review time grows with every second a candidate speaks. Transcript review time grows only with the answers that need a closer look.
The certificate still has to be checked. In both formats, "I have a reach truck certificate" is something the candidate said, and neither format proves it. A follow up question (which certificate, valid until when) makes the claim specific enough to check. Check certificates, identity and right to work before the offer, not on the first morning.
What should staffing agencies and frontline employers watch for?
At volume, the risks are inconsistency, client pressure and candidates dropping out.
- Clients who want to see the candidate. Agencies are sometimes asked for a video so the client can "get a feel" for a person. A face is not evidence that someone can work nights on a reach truck. Answers per requirement and checked documents are. Share the evidence, not the face, and never make a video a condition of applying.
- One format per role, across every branch. If one branch uses video and another uses phone interviews for the same role, candidates are judged on different footing [4].
- Completion by format. Track how many candidates start the interview step and how many finish it, per role and per channel. A step that many people start and abandon is too heavy for the people you are trying to reach.
- What candidates are told. Tell them who reviews the result, what is recorded, how long it is kept and how to ask for another route if the format does not work for them.
- Where the decision sits. A person should make every decision. In the United States, 71 percent of adults opposed AI making the final hiring decision, and 66 percent said they would not want to apply for a job where AI helps make hiring decisions (US, Pew Research Center, survey of 11,004 adults, December 2022) [11]. Opinion in the Netherlands may differ, so treat this as a signal rather than proof.
What does the evidence not tell us?
The research is thin on frontline candidates and on voice interviews.
- The studies cited here did not focus on frontline or hourly roles, and several were experiments rather than live hiring.
- We found no independent study comparing one way video with voice only interviews directly. The case for voice in this article rests on what video adds (appearance, effort, privacy concerns), not on a head to head trial.
- Speech recognition is not equally accurate for every speaker. A US study of five commercial systems found almost twice as many transcription errors for Black speakers as for white speakers (Koenecke and colleagues, 2020) [12]. Where a transcript looks wrong, listen to the recording before drawing any conclusion.
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 that 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. Radius states on its AI Interviewer page that it does not read emotions and does not assess accent, appearance or background.
For each requirement, Match 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. Where the employer uses Radius Passport, identity and right to work are checked with a verification partner and certificates are recorded 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 makes every hiring decision. Candidate data is hosted in the EU. The employer sets how long candidate data, including interview recordings, is kept, as described on the trust page. See how this works for warehouse hiring.
Frequently asked questions
What does a one way video interview mean?
It is a recorded interview with nobody from the employer present: the candidate records answers to questions on screen, often against a time limit, and a recruiter watches them later. It is also called an asynchronous video interview.
What questions work best in a frontline screening interview?
Short questions that each test one requirement of the job work best: which certificates you hold and until when they are valid, which shifts you can work, when you can start, and where you have done similar work. Leave out general questions such as "tell us about yourself", which are hard to compare across candidates.
Is an AI voice interview the same as a phone screen with a recruiter?
Not quite: it covers similar ground, but more consistently. Every candidate gets the same questions at any hour, and the recruiter reviews the answers afterwards. A recruiter call can explore and reassure, but it rarely happens within hours for every applicant. Many teams use both: a structured interview first, then a call with the people who move on.
Do voice interviews disadvantage candidates with an accent?
They can, if reviewers judge how someone sounds instead of what they say. Keep the review tied to the requirements of the job, listen to the recording when a transcript looks wrong, and test the language of the work floor only when the job needs it.
Should every applicant for a role get the same interview format?
Yes, at a given stage, with a stated alternative that any candidate can choose without penalty. Research on interviews held through technology found that ratings differ by mode, so mixing formats for the same role makes candidates harder to compare fairly.
Conclusion
Video interview software solved a real problem: candidates could answer when it suited them, and recruiters no longer had to chase calls. For frontline hiring, a structured voice interview keeps that benefit and drops what the camera adds, which is effort for the candidate, appearance in front of the reviewer and more personal data to hold. Whatever you choose, use the same questions for everyone, check documents before the offer and let a person decide.
General information, not legal advice. Position as of 22 October 2026.
Sources
- CBS (Statistics Netherlands). "4. ICT-gebruik personen", Digitalisering en kenniseconomie 2025. Published 10 February 2026. cbs.nl. Checked 22 October 2026.
- Lukacik, E.-R., Bourdage, J. S. and Roulin, N. (2022). "Into the void: A conceptual model and research agenda for the design and use of asynchronous video interviews". Human Resource Management Review, 32(1), 100789. doi.org/10.1016/j.hrmr.2020.100789. Checked 22 October 2026.
- Langer, M., König, C. J. and Krause, K. (2017). "Examining digital interviews for personnel selection: Applicant reactions and interviewer ratings". International Journal of Selection and Assessment, 25(4), 371 to 382. doi.org/10.1111/ijsa.12191. Checked 22 October 2026.
- Blacksmith, N., Willford, J. C. and Behrend, T. S. (2016). "Technology in the Employment Interview: A Meta-Analysis and Future Research Agenda". Personnel Assessment and Decisions, 2(1). doi.org/10.25035/pad.2016.002. Checked 22 October 2026.
- Sackett, P. R., Zhang, C., Berry, C. M. and Lievens, F. (2022). "Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range". Journal of Applied Psychology, 107(11), 2040 to 2068. doi.org/10.1037/apl0000994. Checked 22 October 2026.
- Hosoda, M., Stone-Romero, E. F. and Coats, G. (2003). "The effects of physical attractiveness on job-related outcomes: A meta-analysis of experimental studies". Personnel Psychology, 56(2), 431 to 462. doi.org/10.1111/j.1744-6570.2003.tb00157.x. Checked 22 October 2026.
- Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal of the European Union, 12 July 2024: Article 5(1)(f), Article 50(1), Annex III point 4. eur-lex.europa.eu. Checked 22 October 2026.
- European Commission. Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689, C(2025) 5052 final, 29 July 2025. Not binding. digital-strategy.ec.europa.eu. Checked 22 October 2026.
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal of the European Union, 24 July 2026. eur-lex.europa.eu. Checked 22 October 2026.
- Regulation (EU) 2016/679 (General Data Protection Regulation), Article 5(1)(c). eur-lex.europa.eu. Checked 22 October 2026.
- Pew Research Center. "AI in Hiring and Evaluating Workers: What Americans Think". 20 April 2023. pewresearch.org. Checked 22 October 2026.
- Koenecke, A., Nam, A., Lake, E. and others (2020). "Racial disparities in automated speech recognition". Proceedings of the National Academy of Sciences, 117(14), 7684 to 7689. doi.org/10.1073/pnas.1915768117. Checked 23 October 2026.
- Autoriteit Persoonsgegevens (Dutch Data Protection Authority). "Persoonsgegevens van sollicitanten", updated 2 January 2025. Regulator guidance, not legislation. autoriteitpersoonsgegevens.nl. Checked 23 October 2026.