A phone screen exists to answer one question: can this person explain something clearly enough that the next conversation is worth booking. It is a good question. It is a poor use of a synchronous half hour, repeated across everyone who applied, in timezones that do not overlap.
Asynchronous video answers the same question without the calendar. The candidate records when they can, you watch when you can, and everyone gets asked the same thing.
What video is actually good at
Video is worth using for exactly one class of thing: how a person explains something to another person. Whether an argument holds together out loud. Whether a technical idea survives being told to someone who does not already know it. Whether they can be concise when nobody is interrupting them.
It is a poor tool for everything else. Do not use video to test knowledge that a written answer would show better, and do not use it to form impressions about someone's manner, background or room. That is not evaluation, and in several jurisdictions it is not lawful either.
Writing questions worth answering
The default video question is "tell us about yourself", which produces a rehearsed answer that separates nobody. Better questions have a subject.
- Explain something from your last role to someone outside your field. You learn whether they can drop jargon, which is most of what senior communication is.
- Describe a decision you got wrong and what you changed. Rehearsed answers stop being rehearsed here.
- Walk through how you would approach a problem the role actually contains. The reasoning matters more than the answer.
- Give the one piece of context a new colleague would need to work with you. Short, specific, and hard to fake.
Let them re-record
A single-take rule tests composure under an artificial constraint, not communication. Allow a retake and you measure the thing you meant to measure, from people who are not panicking.
Scoring without drifting
Video invites impressionistic judgement more than text does, because a face is on the screen. That makes a written scale more necessary here, not less. Score each response on the same named dimensions, one at a time, and score every candidate on a dimension before moving to the next.
Standardising how answers are evaluated is one of the structuring components with the clearest support in the interview literature, independently of standardising the questions themselves.
Watch several before you decide on any. The first candidate you watch sets your calibration whether you intend it or not, and things that seemed disqualifying in isolation often turn out to be what everyone does.
Reviewing faster without losing the thread
Watching at 1.5x speed cuts the time you spend reviewing by a third.
1 minute of video takes 1 / 1.5 = 0.67 minutes to watch, so the saving is 33%.
The comparable figure for reading rather than watching is worth knowing when deciding which format to ask for: adults read English non-fiction at a mean of 238 words per minute silently, which is considerably faster than anyone speaks. If the content is information rather than delivery, ask for it in writing.
That is the real trade. Video costs more of your time per unit of content than text does. Spend it only where the delivery is the thing you are evaluating.
Who a video step quietly excludes
A video requirement is not neutral. It asks for a quiet room, a working camera, enough bandwidth to upload, and comfort being recorded. Those are not evenly distributed, and none of them is the skill you are hiring for.
The fix is not to drop the step, it is to say what the step is for and offer an equivalent. A written answer to the same question, judged on the same dimensions, costs you a few minutes of extra reading and keeps candidates in the process who would otherwise have quietly stopped.
- Publish the question list and the time budget before the candidate starts, so nobody is surprised into a bad first take.
- Offer a written alternative on request, without asking why. A person should not have to disclose a disability to answer a question about their work.
- Caption or transcribe the responses. This helps reviewers with hearing loss and, incidentally, makes every response searchable.
- Never score presentation, lighting, accent or background. If it is not the dimension you wrote down, it is not a dimension.
Recordings are personal data
A video response is one of the more sensitive things a candidate will ever hand a company. It carries their face, their voice and their home, and unlike a CV it cannot be anonymised after the fact.
Treat it accordingly. Say before recording what you will do with the file, who inside the company can open it, and how long you keep it. Set a retention period and let it actually expire, rather than accumulating a library of faces belonging to people you did not hire. If you use automated analysis on the recording, say that too, in plain words, before they press record.
The test for a retention policy
If you cannot say today how long you keep a rejected candidate's video, the answer is forever, and that is a decision you made by not making it.
Where video screening goes wrong
- Asking for video when a written answer would do. The candidate spends longer and you learn less.
- Too many questions. Four short ones are answered honestly; ten become a chore that only the desperate finish.
- Never giving feedback. People spent unpaid time on your process. A short reply is the minimum, and it is what they tell others about.
- Treating it as a full replacement for interviews. It replaces the screen, not the conversation.