How to review video interview answers fairly as a team
Watching recorded answers is quick. Rating them consistently, across three reviewers and twenty candidates, is the hard part. A simple process: agree the bar first, rate independently, write notes that are evidence, then decide together.
Recorded interviews fix one problem straight away: everyone on the team can watch the same answer. They do not, on their own, fix the next one. If three people watch the same two-minute answer and one gives it five stars, one gives it two and the third did not really form a view, you have not learned much about the candidate. You have learned about your reviewers.
This is a practical guide to how to evaluate video interviews consistently as a team: what to agree before you press play, how to rate, what to write down, and how to turn individual ratings into a decision you can all stand behind.
Agree what a good answer looks like before you watch
This is the step most teams skip, and it is the one that matters most. For each question, write one or two lines describing what a strong answer includes. Not a model answer, just the ingredients. For example:
"Tell us about a time something went wrong at work." A strong answer names a real situation, is honest about the candidate's own part in it, describes a specific action, and says what they would do differently.
"Why this role?" A strong answer mentions something specific about the job or the company, and connects it to the candidate's own experience or goals.
Then decide what the stars mean. One to five is plenty, but spell it out: five is "clearly strong, I would want to talk to them", three is "fine, nothing stands out either way", one is "did not answer the question". Ten minutes of this before the first recording saves an hour of argument after the last one.
Write these notes where every reviewer can see them. In AsyncView, the simplest place is the note you add to each question when you build the interview, or a shared document the team keeps alongside the question set.
Rate independently first, then compare
If the hiring manager rates first and everyone else sees it, you have one opinion with three signatures. Have each reviewer watch and rate on their own, without seeing the others' scores, and only then look at everything together.
Two habits make this work:
Rate each answer, not just the whole interview. An overall impression hides the detail. Someone who is excellent on experience and weak on motivation is a different hire from someone who is average on both, even if the overall stars match.
Rate before you read anyone's notes. Notes are persuasive. Form your own view, then read.
AsyncView is built around this. Each reviewer gives one to five stars per answer and overall, plus a recommendation: Strong yes, Yes, Maybe or No. Everyone's ratings, recommendations and notes for the same candidate then appear side by side on one screen, so the comparison happens after the independent view, not instead of it.
Write notes that are evidence, not impressions
"Seemed confident" is an impression. "Gave a specific example of handling a refund dispute, explained the policy clearly, and said what she would change next time" is evidence. The second kind of note is useful in two ways: it tells your colleagues why you rated the way you did, and it tells you, three weeks later, why you shortlisted this person.
A simple rule: every note should point at something the candidate actually said or did. If you cannot, ask yourself what you are reacting to. Often it is accent, background, lighting or a nervous first ten seconds, none of which predicts how someone will do the job.
Notes in AsyncView are private to your team and attached to the specific answer, so the evidence stays next to the thing it describes. Candidates never see them.
Watch out for the usual traps
The first ten seconds. Nerves show early. Judge the answer, not the warm-up.
Production quality. A good webcam and a tidy room tell you about someone's equipment, not their ability. The practice question and device check give everyone a fair start, but phones and small flats are still a fact of life.
Similar-to-me. We warm to people who remind us of ourselves. Your criteria exist to counter exactly this.
Order effects. The candidate you watch after a weak one looks better than they are. Take a break every five or six interviews, and consider watching in a different order from your colleagues.
Fluency over substance. Some people speak smoothly and say little. Go back to your ingredients list: did they actually include them?
Tab-switch notes. If a candidate left the interview tab during an answer, your review screen says so and for how long. Treat it as context, not a verdict. A glance at a calendar is not the same as reading a script, and the answer itself will usually tell you which it was.
Speed up without losing fairness
Fair does not have to mean slow. A few things that make reviewing twenty candidates an afternoon's work rather than a week's:
Playback speed. Most answers are perfectly clear at 1.5×. Drop back to normal speed for the ones that are close. AsyncView remembers your speed and volume between answers so you are not resetting it every time.
Keyboard navigation. Move between questions with the arrow keys or J and K without reaching for the mouse.
Review in question order, not candidate order, when you are calibrating. Watching eight answers to the same question back to back makes the differences obvious and keeps your bar steady. Switch to candidate order once you are confident.
Do not rewatch to decide. If your notes are evidence, you rarely need to. Rewatch only when two reviewers disagree and the stakes are high.
Make the shortlist decision together
Now compare. Put every reviewer's ratings side by side and look for three things:
Agreement. Where everyone gave four or five, you have a shortlist candidate. Where everyone gave one or two, you have a decision. Do not spend meeting time on these.
Disagreement. Where ratings split, this is where the conversation belongs. Read each other's notes first; usually one reviewer noticed something the other missed. If not, watch the answer together and settle it on the evidence.
Patterns. If one reviewer is consistently a star harsher than everyone else, that is fine as long as they are consistently harsher. It is a problem only if it moves around.
Then decide. In AsyncView you shortlist or reject from the review screen. Rejecting is final, so you are asked to confirm, and it is worth a short, kind email to anyone you turn down; they gave you twenty minutes of their time.
The process on one page
Before watching: one or two lines per question on what a strong answer includes, and what each star rating means.
Each reviewer watches and rates independently, per answer and overall, before reading anyone else's notes.
Notes point at what the candidate said or did, never at how they seemed.
Use 1.5× playback and keyboard navigation, take breaks, and calibrate by watching one question across candidates.
Compare side by side; spend the meeting on the disagreements only.
Shortlist, reject, and tell people.
If you want to see the review screen with your own questions, create an interview and record a sample answer yourself. The full list of review features is on the features page, and if you are still deciding which rounds to run async, start with async vs live interviews: which rounds suit each.
