Interviewing
Tell Me About Yourself, With No One on the Other End
Answer it as 'why am I plausible for this job', not as a biography. Nobody is there to follow up on the interesting part, so the opener has to carry its own relevance. Say what you do now, name one problem from the posting you have already worked on, and give one example with a real result. Close with why this role. Everything after the first sentence is evidence for it.
The takeMost advice for this question was written for a room, and the room is where all of it works. Read the interviewer, adjust, stop when they lean in. Take the room away and 'keep it conversational' is a mood rather than an instruction. The honest replacement is unglamorous: decide in advance which of the posting's problems you are claiming, and say it before you say anything about yourself. That is a writing task, and it happens before the clock starts.
Where Olive fits
Open a role and see what the work shows
An Olive assessment is not an interview and there is no camera in it: you work a real assignment openly with an AI assistant, think aloud or type, and a person writes or edits the findings on what happened, in words. You are granted the same report the employer reads, free.
Assess your shortlistWhat Should a Recorded Opener Actually Do?
It has to carry its own relevance, because nobody is going to draw it out of you. In a live interview the opener is a handshake and the real content arrives when someone asks a follow-up. A recording has no follow-ups. So the opening is where a claim about your fit has to go, and the rest of the take is evidence for it.
There is evidence that the front of an interview does real work. In structured mock interviews with 189 accounting students, the impression an interviewer formed during the small talk, before any structured question, correlated .42 with that same interviewer's structured score 6. When a different interviewer supplied the structured score, it fell to .25 and .24. The route ran through rated competence.
That was a live interview with students, and a correlation is not a decision time.
Nothing in it shows anyone making up their mind in the first minute, and it should not be read that way. What it supports is narrower. The opening of an interview leaks into the scored part, and the thing that leaks is an impression of capability rather than an impression of warmth. On a recording, where warmth has nobody to land on anyway, that points to what the first sentence should be made of.
Which is the job, stated as something you have already done.
"I run payroll for a two-hundred-person company and I have moved that payroll onto a new provider" is a claim about the posting. "I graduated in 2019 and I have always liked numbers" is a claim about you.
Only one of them gives a reviewer something to check.
What Does an AI Scorer Do to Your Answer?
Believing one will score you shortens the answer, and that has been measured. In an online mock interview, 124 participants recorded spoken answers, half of them told beforehand that an algorithm would evaluate the recording 3. That half gave shorter answers, engaged in less deceptive impression management, and felt they had fewer opportunities to perform. Shorter is the part that costs you something.
Those were students in a low-stakes study with no job on the line, run before 2020.
The instinct behind it is reasonable. There is no point performing for something that cannot be charmed. But on a system that reads what you said, a short answer simply hands the reader less to work with.
A second finding names a sharper version of the same instinct. In a paper reporting six studies, people rated AI as less capable than humans at judging interpersonal skills, with no equivalent gap on analytical skills 4. In one, 201 participants put AI at 3.28 against 5.24 for humans on a seven-point scale. In another, 107 people who had been through both kinds of process later reported putting less weight on interpersonal skills when the assessor was an AI 4.
That last part is recalled behaviour rather than observed behaviour, from samples in a market where AI video interviewing is unusually established.
The adaptation is backwards on any system that reads a transcript. HireVue's 2025 explainability statement says its AI relies only on what the candidate says 1. It uses no video analysis and no other audio characteristics, and specifically does not assess facial expressions, body language, background and surroundings, or tone of voice. Take the collaboration out of your opener and you have deleted the only evidence of it a scorer could have used.
That is one vendor describing its own product, and a description is not an audit. It says nothing about what any other platform reads.
And none of it means nobody watches, because the recording still exists for a recruiter to play.
How Long Should the Answer Run?
Use the time you are given, up to about three minutes. In a mock asynchronous video interview with 185 participants, the response limit per question was set at one, three or five minutes 5. Performance, rated by people rather than software, followed a curve: highest under the three-minute limit and lower at one and at five. Speaking longer went with lower anxiety and higher ratings, and speaking time levelled off from three minutes.
What moved was clarity and relevance. Judged competence stayed stable across all three limits, which reads as the extra time buying a clearer account of the same ability rather than a better one 5.
The experiment moved a setting the employer controls, not a choice you make inside it. The authors are also clear that offering more than three minutes did not improve anything.
So a tidy forty-five seconds is a habit from the room, not a target.
Length is only half of the timing question, and the rest is decided before you speak. HireVue's statement lists preparation time per question as an employer setting, from zero to five minutes or unlimited 2. Question retries are a separate Yes or No switch, with unlimited as the recommended default, and the vendor says plainly that how a candidate experiences the assessment is configurable by the company deploying it.
Recommended is not guaranteed, and none of it carries over to another platform. The instructions screen is where you find out whether you get five minutes to think or none, and whether a second take exists at all.
Read that screen before you plan the answer.
Then write for the version you actually have. Most of the work in preparing for a one-way video interview where nobody is on the other end is exactly this kind of deciding, done while the clock is still yours.
Write the Opener Before You Open the Link
Draft it on paper and say it out loud until it uses the time you are given without running past it. A recorded opener is a writing problem in an interview's clothes, because everything a live conversation would let you adjust in the moment has to be decided first. Four beats do the job, in this order.
- Now. The work you do and the scale you do it at, in one sentence.
- The claim. Which problem in the posting you have already worked on.
- The evidence. One example, with the decision you made and how it came out.
- The turn. Why this role, named specifically enough that it could not be pasted into another application.
The claim is the beat most people skip, and it is the one doing the work. Everything before it is preamble a reviewer has already read on the resume. Everything after it is only interesting because the claim gave it a reason to be there.
The turn is the other skipped beat, and it is where a generic answer gives itself away.
Employers are reading about this same round from the other chair. One story here tells hiring teams that the async video round is worth keeping for reach rather than judgment, because a scripted answer and a prepared one are hard to separate on substance. Read from your side, that is not an argument for scripting. It is an argument for being specific enough that the pile does not absorb you.
Specific is also what keeps the example from sounding like everyone else's. Rehearsed STAR answers start to sound generic because the four-beat shape costs nothing to produce and the detail underneath it does.
The same rule holds here.
Carry one example, and let it carry the number, the option you turned down, or the part that went badly. Then stop when the example and its result have landed, rather than when the material runs out.
Common questions
How long should a recorded 'tell me about yourself' answer be?
Use most of the time the platform gives you, up to about three minutes. In a mock asynchronous video interview with 185 participants, performance rated by people was highest under a three-minute response limit and lower at one minute and at five, an effect driven by clarity and relevance rather than by judged competence 5. That experiment changed the limit the employer sets rather than what candidates chose to do inside it, and the authors note that offering more than three minutes did not improve outcomes. A tidy forty-five seconds is a habit carried over from a live room.
Should I still smile and make eye contact if nobody is watching live?
Deliver it as though a person will play it, because one can. HireVue's 2025 explainability statement says its AI relies only on the words a candidate says, with no video analysis, no other audio characteristics, and specifically no assessment of facial expressions, body language or tone of voice 1. That is one vendor describing its own product rather than an audit, and it does not mean the recording goes unwatched: it still exists and a recruiter can open it. Spend the preparation on what you say, and let the delivery be ordinary rather than performed.
Can I re-record my answer if the first take goes badly?
The employer decides that, and the platform only supplies the switch. HireVue's statement lists question retries as a Yes or No configuration, with unlimited attempts as its recommended setting, and preparation time per question as a separate employer setting running from zero to five minutes or unlimited 2. It states plainly that how a candidate experiences the assessment is configurable by the company deploying it. So read the instructions screen at the start rather than assuming, and none of this carries over to a different platform.
Should I mention teamwork if an AI is scoring the answer?
Yes. In a paper reporting six studies, people rated AI as less capable than humans at judging interpersonal skills, with no equivalent gap on analytical skills, and 107 people who had been through both kinds of process later reported putting less weight on interpersonal skills when the assessor was an AI 4. On a system that scores the transcript, leaving the collaboration out removes the only trace of it. Say who else was involved, what you handled, and what changed because of the way the group worked.
Is it fine to read the answer off a second screen?
Notes help where the instructions allow them; a full script is the weaker choice. A paragraph written in advance tends to drift toward a generic version of the job rather than the one problem you are claiming, and the specific part is what a reviewer can check. Bullet points for the four beats, said in your own words, keep the answer anchored to what you actually did.
References
- 1. 2025 Explainability Statement hirevue.com Supports this article's claim that on one vendor's AI-scored path the score comes from the transcribed words alone, with no video analysis and no assessment of facial expressions, body language, surroundings or tone of voice, which is why an opener stripped of its interpersonal content loses evidence.
- 2. 2025 Explainability Statement hirevue.com Supports the section on timing: preparation time per question is an employer setting of zero to five minutes or unlimited, question retries are a Yes or No switch with unlimited recommended, and the vendor states the candidate's experience is configurable by the deploying company.
- 3. Is anybody listening? The impact of automatically evaluated job interviews on impression management and applicant reactions emerald.com Supports the claim that participants told an algorithm would evaluate their recorded answers gave shorter answers, used less deceptive impression management and felt they had fewer opportunities to perform, which is the trap this article warns a reader about rather than a tactic.
- 4. Lay beliefs about AI assessment of interpersonal skills in personnel selection nature.com Supports the claim that people believe AI is worse than humans at assessing interpersonal skills with no equivalent belief about analytical skills, and that applicants report placing less emphasis on interpersonal skills when the assessor is an AI, which this article argues removes evidence from a content-scored answer.
- 5. Tick Tock, Tick Tock, Time's Up! How Response Time Limits in Asynchronous Video Interviews Influence Interview Anxiety, Impression Management, and Performance doi.org Supports the answer to how long to speak: rated performance was highest under a three-minute response limit and lower at one and five minutes, an effect driven by clarity and relevance while competence ratings stayed stable, with the limit set by the employer rather than the candidate.
- 6. Initial Evaluations in the Interview: Relationships with Subsequent Interviewer Evaluations and Employment Offers homepages.se.edu Supports the claim that the impression formed before the structured questions correlated .42 with the same interviewer's structured score and .25 across interviewers, running through rated competence, which this article uses as a reason to open with substance and explicitly not as evidence about decision timing.
6 sources, numbered by first appearance. How Olive sources claims
General guidance for hiring teams. What works at one company and one volume may not transfer to yours.
Olive assesses how a person works with AI. It does not detect AI-written documents, and it never produces a score, a ranking, or a match percentage for a person. Candidates read the same report the employer reads.