Interviewing
How to Use Notes in a Recorded Interview Without Reading Them
Your assessment invitation carries the rule that actually governs, so read its instructions screen before you record. Where that screen says nothing about notes, paper on the desk is not the problem. HireVue and Spark Hire let employers set preparation time before each question, and in mock interviews the gain from that time came from planning the answer. What costs you is reading: a script can flatten the specifics a rubric rewards, and reported Amazon guidance lists appearing to read among its interviewers' tells for AI use.
The takeThe advice that tells you to memorize instead is worse than it sounds. Memorized prose fails the same way a script does, and it fails under pressure, when the third question arrives in a shape you never rehearsed. Notes at eye level are the version that survives a question you did not expect, because a keyword can be reordered mid-sentence and a paragraph cannot.
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, on your own clock, thinking aloud in speech or in writing. A person writes or edits six findings on what happened, and the candidate is granted the same report the employer reads, free.
Assess your shortlistIs Having Notes Against the Rules?
Only if the instructions for your round say so, and that screen is easy to click past. The instructions on your assessment invitation are the employer's actual rule for that round. Some employers write a rule down, and plenty never mention notes at all. Read that screen before you record. If it says nothing about notes, a page on the desk breaks no rule you were given.
One employer's written rule, reported in detail, is about AI tools rather than paper. IT Pro reported in March 2025, relaying internal documents seen by Business Insider, that Amazon's recruiter guidelines tell candidates not to use generative AI during an interview unless explicitly permitted 1.
The same guidelines warn that failing to follow them may result in disqualification from the recruitment process 1. That wording reaches you third-hand, through one outlet reporting another's read of internal documents.
Paper on the desk is not what that rule targets.
Hiring-side guidance works the same question from the other chair, and its answer is less harsh than you might expect. The hiring-side piece on a candidate whose eyes keep moving off-camera tells interviewers that gaze is not evidence of cheating, and that notes written before the call are one of the ordinary explanations. It also warns them that acting on eye movement risks screening on a disability.
Knowing what the other side has been told beats guessing at it from here.
What Is the Preparation Timer For?
The preparation timer is for planning the answer, and the measured benefit points at planning specifically. Basch and colleagues tested the preparation time given before each question in asynchronous video interviews. A pilot found it helped performance, and the main study replicated the effect and found it ran through active response preparation, with no mediation through strain or through impression management 2.
Those were mock interviews rated by people, from one German research group, and the sample sizes and effect sizes sit in the full text rather than the abstract. Treat it as a direction, not a dose. The direction is the useful part: the timer rewards deciding what you will say.
Paper is how you hold a decision you have already made.
How much of that time you get is an employer setting rather than a property of the product. HireVue's 2025 statement lists preparation time per question among the configurations the deploying company chooses, from zero to five minutes or unlimited, next to a retries switch whose recommended default is unlimited 3. Spark Hire's help documentation for employers puts think time between 30 seconds and 5 minutes or leaves it unlimited, and says that limiting think time also limits the candidate to one take 4.
Where that clock is limited on Spark Hire, the take you are about to record is the only one 4.
So the first thirty seconds are not spare time. Use them to pick the example and its ending, which is the case what to do with a thirty-second prep timer makes at length. Your notes exist to make that choice fast: a glance that finds the right line beats a minute spent recalling which project had the number in it.
Write Three Words per Example, Not a Paragraph
Where notes are allowed, keep each example to three or four words on one line, and put the page where a glance finds it fast. A keyword prompts you to say the thing. A sentence invites you to read it aloud, and the voice tends to follow the eye. Tape the card just under the camera or beside the lens, so the glance is short.
This is not about hiding the notes. Nothing here is a way to pass a read answer off as spontaneous. The placement matters for a plainer reason: a prompt you find in half a second does not pull you out of the answer.
Four things earn a line on the card:
- the situation in two or three words, like the migration weekend
- the number the result turned on, because that is the first thing a timer takes from you
- the decision you made, rather than the story around it
- the question you want to ask at the end, if there is a slot for one
What does not earn a line is a full sentence you plan to say word for word. The moment a whole sentence sits on the page, the eye starts tracking it, and the answer turns into a reading. That mechanism is the honest core of every article telling you never to use notes, and it is a fair warning aimed at the wrong object.
The problem was never the paper.
Who Notices a Read Answer, and What It Costs
A scoring model and a person watch the same recording, and they notice different things. HireVue's 2025 statement says its AI reads only the transcribed words of an answer, running no video analysis and assessing no facial expression, body language or tone of voice 5. That is the vendor describing its own product, not an audit, and it covers that one scorer. The recording still exists, and people still watch it.
For the person watching, the format itself moves ratings. In an experiment with 114 participants comparing face-to-face and videoconference interviews, interviewees were rated lower in the videoconference condition, and differences in perceived social presence, perceived eye contact and impression management contributed to the gap 6. Ratings made live were also higher than ratings made from recordings 6. It was a 2021 lab study of human raters.
Nothing in it measures an algorithm, and a text-only scorer would be blind to it anyway.
Reported Amazon guidance names appearing to read rather than respond naturally among the signs of generative AI use that interviewers are told to watch for, alongside typing during a question and wandering eyes 1. Those are heuristics, and they describe a nervous or neurodivergent candidate as readily as a dishonest one.
Amazon has not published how they are applied.
The other cost is to the content you meant to give. A written answer drifts toward the tidy structure everyone else writes, and the specific number that made your example yours can get smoothed out on the way to the page. Employers see that convergence in aggregate, which is why the hiring-side argument about async video rounds where every answer is read off a screen is about whether the round still separates anyone.
A rubric cannot reward a detail you left on the page.
Common questions
Am I allowed to have notes during a one-way video interview?
Check the instructions screen for the assessment you were sent, because that is the employer's rule for that round. Where it says nothing about notes, there is usually nothing to break. One employer rule that has been reported in detail, Amazon's, is about generative AI tools during the interview rather than paper, so read the invitation rather than a general career-advice page.
Will the AI know I am looking down at notes?
On HireVue the scorer is not reading your eyes: its 2025 Explainability Statement says the AI reads only the transcribed words of an answer and runs no video analysis. That is the vendor's description of its own product rather than an independent audit, and it says nothing about what other platforms read. The recording still reaches the employer, and a person can play it.
How should I lay the notes out?
Three or four words per example, on one line, placed near the lens so the glance is short. Write the situation, the number the result turned on, and the decision you made. Leave full sentences off the page. As soon as one is sitting there, the eye tracks it and the answer becomes a reading.
What is the preparation timer for?
Planning the answer. In a 2021 study of mock asynchronous video interviews rated by people, preparation time improved performance, and the effect ran through active response preparation rather than through reduced strain or impression management. Use the countdown to pick the example and decide how it ended. On Spark Hire, a limited think time also means a single take, so the countdown is planning time rather than warm-up.
What if the instructions say no notes?
Then follow them. A rule you were given is the rule, and nothing here is worth ignoring it for. Prepare the same way regardless: choose four examples, fix the number and the decision in each, and say them out loud a few times before you open the link. Notes are a fast-recall device, and rehearsal without a page does the same job more slowly.
Does looking away from the camera count against me?
Possibly, where a person watches. A 2021 experiment with 114 participants found videoconference interviews rated lower than face-to-face ones, with perceived eye contact, social presence and impression management contributing. That was a lab study of human raters, not a finding about scoring software. Employer-side guidance also tells interviewers that off-camera glances are not evidence of cheating, since notes, magnification software and a second monitor all produce them.
References
- 1. Amazon banned candidates from using AI tools during interviews itpro.com Supports the article's statement that Amazon's recruiter guidelines bar generative AI during an interview unless explicitly permitted, warn of possible disqualification, and give interviewers 'appearing to read rather than respond naturally' as a tell.
- 2. A good thing takes time: The role of preparation time in asynchronous video interviews doi.org Supports the claim that the benefit of preparation time in asynchronous video interviews was mediated by active response preparation, with no mediation through strain or impression management.
- 3. 2025 Explainability Statement hirevue.com Supports the claim that preparation time per question and interview retries are settings the deploying company configures, with unlimited retries as the recommended default.
- 4. Set think time limits for One-Way Interview questions support.sparkhire.com Supports the claim that Spark Hire think time is set per question between 30 seconds and 5 minutes or left unlimited, and that limiting it also limits the candidate to one take.
- 5. 2025 Explainability Statement hirevue.com Supports the claim that HireVue's AI relies only on the transcribed words and assesses no facial expression, body language, background or tone of voice.
- 6. It Takes More Than a Good Camera: Which Factors Contribute to Differences Between Face-to-Face Interviews and Videoconference Interviews Regarding Performance Ratings and Interviewee Perceptions? link.springer.com Supports the claim that in an experiment with 114 participants, videoconference interviewees were rated lower than face-to-face ones, with perceived social presence, perceived eye contact and impression management contributing, and that live ratings exceeded ratings made from recordings.
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.