How to Prepare for a One-Way Video Interview in an Hour

Name the competencies hidden in the posting.

Open the job posting and underline every skill, trait, or responsibility it names, then group them into three or four competencies: something like problem-solving, communication, ownership, working with ambiguity. A one-way round sends every candidate the same questions in the same order: comparability rather than conversation. The same structure is what the selection research rates highest, since a 2022 re-analysis ranked structured interviews top of the methods it compared ¹.

That research is about live interviews rated by people, so it says nothing about video, async delivery, or AI scoring. What carries across is the shape: fixed questions, compared side by side. Do not guess at the questions themselves. Guess at the competencies, since those are visible in the posting's own language, and prepare one strong example for each rather than five generic ones that fit none of them precisely. A posting that names "ownership" once is telling you exactly what one question is about to probe.

If the posting is vague or copied from a template, the team's own product page, a recent release, or a public roadmap usually names what the group actually cares about this quarter more precisely than the listing does. Fifteen minutes reading around the posting is fifteen minutes better spent than fifteen minutes searching for someone else's list of "top ten interview questions," since that list was never written for this competency or this employer.

Write one example per competency, with a number and a decision in it.

A rubric-scored answer rewards specifics, not adjectives. For each competency, write two or three sentences describing one real situation: what the actual problem was, the decision you made, and a number that shows the size of it, whether that is a dollar figure, a headcount, a percentage, or a timeframe.

"I'm a strong communicator" gives a rubric nothing to grade. "I rewrote the onboarding doc after two new hires asked the same question in their first week, and support tickets on that topic dropped" gives it something concrete: a trigger, an action, and a result a stranger can check the shape of. If your real history genuinely has no clean number attached, a specific before-and-after still beats an adjective, since the specificity is what does the work.

Write these out in full sentences first, then compress to the two or three you would actually say out loud. A script read verbatim on camera sounds like a script; an outline you know well enough to say naturally does not, and the difference is audible within the first sentence. Reuse the same example across two different competencies only if it genuinely fits both; forcing one story to answer four questions reads as thin the second time it appears.

Test each answer once, then leave it alone.

Record each answer a single time, at your normal pace, and play it back once, listening for whether the example survives being spoken rather than read. Cut anything that only worked on paper. A specific answer that runs a little long beats a smooth one that says nothing, since the content is the part a competency rubric is built to grade.

What you are checking for is content: did you actually name the problem, the decision, and the outcome, in that order, before the clock ran out. If the answer wanders before it gets to the decision, trim the setup rather than the ending, since the outcome is the part a rubric is built to look for last and reward most.

Resist the urge to record a second, third, and fourth take chasing a perfect delivery. Beyond one review, you are polishing performance rather than fixing content, and content is the part you can still change in the hour you have. Where the round is scored by AI, one major platform states that its model uses only the words a candidate says, with no video analysis and no assessment of facial expressions, body language, background, or tone of voice ². Where a person watches the recording instead, delivery is back on the table, so aim for plain and audible rather than polished. What an AI interview scores, and what it stopped scoring, is knowable rather than guessable, and the answer for the round in front of you is worth more than a general rule.

Check the mechanics that actually cost people the round.

Before you start, read the instructions screen for two things: how many retakes you get per question, and how much prep time you have before recording begins. On at least one major platform, retake and prep-time limits are set by the employer deploying the assessment, not fixed by the software, so two people describing the "same" one-way round can have had genuinely different experiences ². Do not assume you get one shot; do not assume you get unlimited ones either.

Test your microphone before the real questions start, on a wired connection if your wifi has ever dropped mid-call, and close every other tab and application that might send a notification sound. A garbled answer is a mechanical failure, not a content problem, and it is worth a message to the recruiter if it happens: say plainly that the audio cut out partway through, name the question number, and ask whether you can redo it. Asking costs nothing, and the answer is the recruiter's to give.

Give yourself real quiet, not just a quiet room. A doorbell, a roommate, or a phone buzzing mid-answer does not usually stop the recording, and a distracted second half of an answer reads exactly like a distracted candidate, whether or not that is what actually happened, so the fifteen minutes spent silencing a phone is not wasted time. It is the last mechanical variable you actually control before the questions start. Whether the async format is even worth keeping once every answer sounds coached is the question employers are actually asking about this round, and reading it tells you what your recruiter has likely already been told about the format you are about to sit for.

See a sample report >