Candidates react worse to one-way video interviews than to a live conversation, and field data point to which design choices make it better or worse: more preparation time and a re-record option go with better reactions, more questions with worse ones. No independent study gives a reliable "drop-off rate" for the format, so any single percentage you read in a vendor blog should be treated as a claim, not a fact. This guide sorts the evidence by quality, shows what you can control, and explains when an asynchronous voice or chat interview is the lower-friction alternative. Status: November 2026.
"Asynchronous interview" is the umbrella term: the candidate answers on their own time instead of in a scheduled call. The recording can be video (the classic one-way video interview), audio only (voice) or typed answers (chat). That distinction matters, because the time shift is only part of the issue: in a 2022 Aptitude Research report, one in two candidates said they are not comfortable in front of the camera.
What does the research say about candidate reactions?
The honest summary: reactions are mostly less favourable than for live interviews, the effects depend on design, and the best numbers come from a few large datasets. The table shows what each source can and cannot support.
| Source | What it found | How far to trust it |
|---|---|---|
| Tilston, Krings, Roulin, Bourdage and Fetzer (2024), Human Resource Management 63(2) | Field data from 27,809 real applicants. More preparation time and the chance to re-record went with more favourable reactions; more questions went with less favourable ones. Applicants over 31 reacted especially negatively to more questions; women reacted more positively to extra preparation time. | Peer-reviewed, large sample, real applications. Data come from one video-interview vendor; correlational, not an experiment. |
| Sapia.ai Labs (April 2024, SIOP) | About 1.3 million candidates: 60.3 % of chat finishers completed within 24 hours against 33.4 % of video finishers. Completion was about 79 % for both men and women in chat, 78 % of men and 74 % of women in video. | Large, but the vendor sells chat interviews and the full paper was not available to us. The 24-hour figures count only people who finished; the completion rates are the vendor's own numbers. |
| Niemitz et al. (2024), International Journal of Selection and Assessment | In a small field study (N = 76), asking questions as video instead of text did not change applicant reactions, impression management or interview performance. | Peer-reviewed, small sample. Suggests the video question itself is not the problem. |
| Aptitude Research (2022) | About one in four candidates drops off at the interview stage in general. | Sponsored industry survey, not specific to one-way video. |
| Vendor blogs ("55–60 % complete", "-15 % per extra question") | Widely repeated percentages. | No traceable primary study. We did not use them. |
Earlier work points the same way: applicants tend to react poorly to asynchronous video interviews, a finding the authors of the 2024 study cite as their starting point (see their seminar abstract, the Sapia report and the Niemitz study).
What is the drop-off rate for one-way video interviews?
There is no trustworthy universal number. Completion depends on the role, how the invitation is written, how many questions there are and whether the candidate is already invested. The Aptitude Research survey puts general interview-stage drop-off at roughly one in four, but that is not a one-way figure. Measure your own funnel instead (see the pilot sheet below).
Which design choices help, and which hurt?
The first three points come from the 2024 field data, the last two are our own recommendations. Treat all of them as hypotheses to test in your process:
- Give preparation time. Let candidates read the question and think before the clock starts.
- Allow a re-record. The option to re-record was linked to better reactions.
- Ask fewer questions. More questions went with worse reactions, especially among applicants over 31. The study names no ideal number; we suggest starting with three to five.
- Say who watches and what is scored. Transparency is also a legal point (below).
- Time the reminder from your own data. In the Sapia data only 33.4 % of video finishers were done within 24 hours, so we would start at 48 to 72 hours.
Is video needed, or does voice or chat do the job?
If the goal is a pre-screen on availability, location, experience and motivation, you rarely need a face. A voice or chat interview collects the same evidence with less setup for the candidate: no lighting, no camera, no worry about the background. Niemitz only shows that the format of the question made no difference. That the pressure of being recorded is the real burden is our reading, not a tested result.
| One-way video | Async voice | Async chat | |
|---|---|---|---|
| Candidate effort | Camera, light, quiet place, appearance | Quiet place, no camera | Typing on a phone, anywhere |
| Best for | Roles where presence is the job | Frontline and shift roles, speaking roles, low typing skill | Facts, availability, knockout criteria |
| Main risk | Camera pressure, more personal data | Speech recognition for accents and vocabulary; not for deaf or hard-of-hearing candidates | Not for people who find typing hard |
| Cost in Sprad (list price, 29 Oct 2026) | 80 credits (about €8.00) | 30 credits (about €3.00 at €0.10 per credit), 12 languages | 10 credits (about €1.00) |
In Sprad's AI interview you choose the format per role: chat, voice, phone or video. We recommend always offering a human alternative as well. More on choosing between the formats is in chat or voice and one-way video vs. AI phone screen; for roles see which roles fit an AI interview. Prices are on the pricing page.
How do you measure the candidate experience in a pilot?
Run the new format on one role for four weeks and track four numbers, so you end up with your own drop-off rate instead of someone else's:
| Metric | How to count | What it tells you |
|---|---|---|
| Invite to start | Started ÷ invited | Is the invitation clear and trusted? |
| Start to finish | Finished ÷ started | Is the interview itself too long or too stressful? |
| Time to finish | Median hours from invite | When to send the reminder |
| Candidate rating | One question after the interview, 1 to 5 | How it feels, by role and by age or gender group |
Compare the same role against your previous process (phone screen or live call), and split results by gender and age, because the 2024 field data found differences by age and gender. For a step-by-step process see our guide to running a one-way video interview; for tools see one-way video interview software compared.
What does the law require for asynchronous interviews in the EU?
Three points. First, recruiting AI is a high-risk use under Annex III of the EU AI Act; after the Digital Omnibus (Regulation (EU) 2026/1744) the high-risk duties apply from 2 December 2027. Second, the transparency duty for AI systems that interact with people (Art. 50) has applied since 2 August 2026, so candidates must know they are speaking to an AI. Third, a recording is personal data; Art. 22 GDPR limits purely automated decisions, so a person should decide. Emotion recognition in the workplace is prohibited (Art. 5). This is general information, not legal advice.
Frequently asked questions
What is the difference between a one-way and an asynchronous interview?
One-way usually means a recorded video answer to set questions. Asynchronous is broader: any interview the candidate completes on their own time, as video, voice or text.
Do candidates dislike asynchronous video interviews?
On average they react less favourably than to live interviews, but design matters: preparation time and re-recording help, long question lists hurt.
Are one-way interviews fair to all candidates?
In the 2024 field study, reactions differed by age and gender (applicants over 31 reacted worse to more questions), and the Sapia data showed lower completion for women in video. Offer an alternative format and review results by group.
Can an async interview replace the live interview?
No. It replaces the first screening call, not the conversation where you and the candidate decide whether you fit.
How many questions should an async interview have?
No study sets an exact number. The field data link more questions to worse reactions; we suggest three to five.
Jürgen Ulbrich
Jürgen Ulbrich has more than a decade of experience in developing and leading high-performing teams and companies. As an expert in employee referral programs as well as feedback and performance processes, Jürgen has helped over 100 organizations optimize their talent acquisition and development strategies.
