One-Way Video Interviews: How They Work and When They Beat a Phone Screen

A one-way video interview is a screening round where the employer sets the questions in advance, the candidate records answers alone on their own device with no interviewer present, and the hiring team watches the recordings later. Vendors call it an asynchronous video interview or a pre-recorded video interview. Recruiters just call it a one-way. It sits between the CV screen and the first live conversation, and depending on how it's configured it either buys a team back a week of calendar time or quietly removes a chunk of the pipeline nobody notices leaving.

Both outcomes are common. The difference sits in a handful of settings that take four minutes to change and that almost nobody changes.

How does a one-way video interview work from the recruiter's side?

You build a template: a set of questions, a recording time limit per question, a preparation window before recording starts, a rule on whether candidates may re-record, and a deadline. The ATS or the interview tool sends a link. Candidates open it, usually on a phone, work through the questions, and submit. Reviewers watch on their own schedule and score against a rubric.

The best public picture of how organisations actually configure these comes from Patrick Dunlop, Djurre Holtrop and Serena Wee, who analysed archival data from an Australian video interview vendor and published it in the International Journal of Selection and Assessment in 2022. Their dataset covered 2,550,105 recorded responses from 627,999 candidates, answering 52,623 questions across 12,105 interview templates. Most interviews contained four or five questions. Candidates typically got 30 seconds to prepare and two minutes to answer. Previewing questions or re-recording answers was rarely permitted.

Two findings from that study matter more than the rest. First, employer settings closely tracked the platform's defaults, which means most hiring teams are running whatever the vendor shipped rather than anything they chose. Second, the median interview had only ten candidates in it. This is not just a bulk-hiring tool for 4,000 campus applicants. Plenty of teams use it on a shortlist of a dozen.

How many questions should a one-way video interview have?

Four. Five if the role is genuinely varied.

Ottilie Tilston and colleagues, writing in Human Resource Management in 2024, examined the experiences of 27,809 applicants across 11 countries and found that adding more questions was associated with worse candidate reactions, an effect sharpest among applicants over 31. Younger applicants, under 30, cared more about having enough time to answer. Nobody in that sample reacted well to a long interview.

Structure the four like this. Question one is a warm-up that you barely score, because the first answer anyone records is their worst answer and you do not want to burn a real competency on it. Ask something like "tell us what caught your attention about this role." Questions two and three carry the weight: behavioural or situational prompts tied directly to a line on your scorecard. Question four is role-specific and concrete, the one a bluffer cannot answer.

What should never appear in a one-way is anything a text field answers better. Notice period, current and expected CTC with the fixed-variable split, base location, willingness to relocate, work authorisation. Making someone record 90 seconds of video to say "60 days' notice" wastes their evening and yours. Collect that in the application form and reserve the recording for judgement.

What response window and retake limits actually do to completion

The retake question splits practitioners, and the purist position is wrong. The argument against re-recording is that it lets candidates polish and therefore rewards preparation over substance. In practice, forbidding retakes mostly punishes people whose toddler walked into frame or whose first take was ruined by nerves. Tilston's data points the same way: more preparation time and the option to re-record were both associated with meaningfully better applicant reactions.

Allow retakes. Cap them at two or three per question if you want a limit that sounds defensible. Tell candidates the cap up front, on the invitation, not when they discover it mid-recording.

On the response window, three to five days spanning at least one weekend is the sensible default. Anything under 48 hours penalises the exact candidates you want most, the ones currently employed and busy. Send the invitation on a Tuesday or Wednesday with a Monday deadline, and set an automatic reminder at 48 hours out. A meaningful share of completions in most systems arrive on the final day, which means the reminder is doing more work than the invitation.

How do you review 40 recordings without losing an afternoon?

Forty candidates at four questions and two minutes each is over five hours of footage. Watched end to end, that is a day and a half. It's the single reason one-way rounds get abandoned after two cycles.

The fix is to stop watching candidates and start watching questions. Reviewers instinctively open candidate one, watch all four answers, form an impression, then move to candidate two. That's the slowest possible order and the most bias-prone, because by candidate thirty you're comparing against a memory of candidate four. Instead, pick your highest-signal question, usually question two, and watch that single answer for all forty people in one sitting. Roughly ninety seconds each with playback at 1.5x speed and captions on, and you're through the pool in about an hour with a directly comparable set of impressions.

That first pass produces a yes, no, or maybe. Take the yes and maybe pile, typically twelve to fifteen people, and only then watch their full submissions. Fifteen candidates at six minutes is another ninety minutes. Total: under three hours for a pool that would have taken forty scheduled phone screens.

Never send a hiring manager forty recordings and expect anything back; send eight, with a one-line note on each. And if your tool supports clipping, share the 40-second segment that made you shortlist someone rather than the whole file. Managers will watch a clip between meetings. They will not open a five-minute video.

How do you score consistently across reviewers?

Write the rubric before you write the questions. Three or four dimensions maximum, each with behavioural anchors describing what a 1, a 3 and a 5 actually look like for this role. Vague dimensions like "communication" produce vague scores; "explains a technical decision to a non-technical listener without jargon" produces defensible ones.

Then three rules. Reviewers score independently and cannot see each other's scores until they've submitted their own, or you'll get anchoring dressed up as consensus. Calibrate on three recordings together at the start of every new requisition, comparing scores and arguing about the gaps, before anyone reviews alone. And score each candidate on the same question in the same sitting, which is the same batching logic as above applied to fairness rather than speed.

There's reassurance in the research. A meta-analysis by Rosemary Hyejin Moon, Jiyoung Park and Hyung In Park in Current Psychology, pooling 24 studies across 31 independent samples, found interview ratings from asynchronous formats did not differ significantly from synchronous ones. The ratings hold up. How candidates feel about you afterwards does not.

Why do candidates abandon a one-way video interview?

This is the part vendors leave off the slide. The same meta-analysis found asynchronous interviews were associated with lower organisational attractiveness than synchronous formats, more negative behavioural intentions such as willingness to accept an offer or recommend the employer, and negative justice perceptions where live interviews showed positive ones. In plain terms: the format works for assessment and costs you goodwill.

Candidate preference data points the same direction. An American Staffing Association poll of 2,019 US adults, reported by SHRM in 2023, found 70% preferred in-person interviews, 17% video and 9% phone. Meanwhile iCIMS' 2025 State of Frontline Hiring Report found 60% of workers start applications they never finish, with the interview stage accounting for the largest share of drop-off at 32%. Appcast's 2025 benchmark data showed applications taking under five minutes completed at 12.47% while those over fifteen minutes completed at 3.61%. Length kills completion, and a one-way round is length.

There's a counterweight worth knowing. Barbara Brenner, Tuulia Ortner and Doris Fay, in Frontiers in Psychology in 2016, found participants rated asynchronous video interviewing broadly favourably, in the same band as other non-interactive screening steps such as personality tests, with perceived usefulness the strongest predictor of attitude. Candidates don't hate the format. They hate being given it without explanation.

Concrete ways to reduce drop-off:

1. Say why in the invitation, in two sentences, and be honest: it saves the candidate a scheduling round and lets more than one person on the panel hear them.
2. Publish the total time commitment on the invitation. "Four questions, about eight minutes, three days to complete."
3. Show the questions before recording starts, or at least tell candidates they'll see each question before the timer runs.
4. Allow retakes and say so up front.
5. Make it work on a phone over mobile data, and test it on one before launch, not after.
6. Give a named human contact for technical trouble, not a support form.
7. Commit to a response date in the invitation and hold to it. Cronofy's 2024 candidate research found 42% of candidates had withdrawn from a process because of scheduling delays; silence after a recording feels identical.

When is a one-way better than a 20-minute phone screen?

The appeal of a one-way interview for recruiters is arithmetic. Yello's recruiter survey, cited by interviewstream, found 67% of recruiters spend between 30 minutes and two hours scheduling a single interview, and 35% named scheduling the most time-consuming part of their job. Hireflix's analysis of 1,820 reviews of one-way interview software, which is vendor-published and should be read as such, reported reviewers cutting per-candidate screening from 30 to 45 minutes down to roughly five, with 61% citing the removal of scheduling back-and-forth as the main gain.

Two rules of thumb hold up. Under about fifteen candidates for a specialist role, the phone is faster and better. Above thirty for a defined role with a stable brief, the one-way wins on both time and comparability. Treat it as a video screening interview that replaces the first filter, never as a replacement for a conversation before an offer.

What do you owe candidates on recordings and AI scoring?

Recordings are personal data with a face and a voice attached. India's Digital Personal Data Protection Rules were notified in November 2025 with a phased runway, the consent manager framework operational from 13 November 2026 and full compliance due by 13 May 2027, and they require purpose-limited retention and deletion once the purpose is served. Decide your retention period now, write it into the invitation, and delete on schedule.

If anything automated scores or ranks the recordings, disclosure is no longer optional in the EU: the AI Act's Article 50 transparency obligations, including telling people they are interacting with an AI system, applied from 2 August 2026, even though the high-risk obligations covering recruitment systems were deferred to 2 December 2027 under the digital omnibus agreement. Telling candidates plainly what reviews their video is also, separately, one of the cheapest ways to protect completion rates.

Platforms including Xakal, whose Xara AI Interviews product sits alongside a built-in ATS at thexakal.com, now bundle asynchronous rounds with scoring and pipeline tracking in one place, which removes the export-and-email step that makes batch review painful. The tooling matters less than the settings, though. Four questions, retakes on, five days, a rubric written before the questions, and an honest sentence explaining why you're asking will outperform any platform run on defaults.