Why review a recorded mock interview at all?
Because your memory of an answer is mostly a memory of what you meant to say. The recording keeps what you actually said, how long it took and where you hesitated. A review turns one practice run into a short list of specific fixes, and the next recording shows whether those fixes held.
Amazon's interview guide recommends "recording yourself or practicing with a friend" to make STAR answers concise yet detailed, typically two to three minutes each. You can't time yourself or hear your own filler sounds while you are talking. A recording does both for you.
Recording yourself alone needs nobody's permission. If a friend plays the interviewer, ask before you record, keep the file private and delete it when you are done. The rules for recording other people, including real interviewers, are in the guide on whether it is legal to record a job interview.
What do coaches score that candidates miss?
Coaches score against a standard written down before the answer, while candidates tend to score against a feeling. Structured interviewers work the same way as coaches. The US Office of Personnel Management says highly structured interviews are "scored according to benchmarks of proficiency," so the bar exists before anyone speaks.
OPM's example rating scale runs from 1 (Low) to 5 (Outstanding), with 3 as average. At level 1 the candidate handles interpersonal situations "involving little or no tension or discomfort," and at level 5 the situations involve "a high degree of tension or discomfort." The scale rewards how hard the situation was, as well as how clearly you describe it.
Google's guide to structured interviewing, updated March 2026, describes rubrics with examples of what a "poor, borderline, solid, and outstanding answer would cover." Borrow that idea for your own review. Before you press play, write one line for each level of the question you practised.
Here is an example rubric for "Tell me about a time you dealt with a difficult stakeholder," made up for this page:
- 1, poor: no specific situation, or a story where someone else did the work.
- 2, borderline: a real situation, with vague actions such as "we aligned on it."
- 3, solid: a clear situation, two or three actions you took yourself, and an outcome.
- 4, strong: everything in 3, plus a number or a named consequence.
- 5, outstanding: everything in 4, in a situation with real tension, told in under three minutes.
How should you watch the recording?
Watch it three times and give each pass one job. The first pass is for content, the second is for delivery and the third is for timing. Doing all three at once spreads your attention thin, and you tend to notice only the thing that embarrasses you most, which is rarely the thing that would cost you points.
| Pass | What to watch or count | What to write down |
|---|---|---|
| 1. Content | A clear situation, your own actions in "I" statements, a result with a number | A score from 1 to 5 against your rubric |
| 2. Delivery | Filler sounds, the silence before you start, trailing off at the end | A filler tally per answer and the start delay in seconds |
| 3. Timing | Total length, and how long the setup ran before your first action | Total time and setup time |
On the content pass, check the three things Amazon's guide asks for. Those are a STAR structure, "I" statements for your own contributions, and "precise metrics and data" where you have them. If you said "we" all through the action section, the interviewer can't tell what you did, so mark the second where your own action starts.
On the delivery pass, count filler sounds and note any long silence before you began. A 2016 study of recorded job interviews by Brosy, Bangerter and Mayor found that longer delays before answering predicted lower hireability ratings. The pages on filler words in interviews and speaking pace cover what to do with those counts.
On the timing pass, write down the length of each answer and the length of the setup. Amazon's two to three minute range is a sensible ceiling for a story answer. If the situation took a full minute before you did anything, cut it first, as the guide to how long interview answers should be explains.
What should you write down after each answer?
Write three lines per answer: the score, the single moment that cost the most, and the words you will say instead next time. The moment needs a timestamp so you can find it again. The replacement needs to be a sentence you could say aloud, because a note like "be clearer" gives you nothing to practise.
Example notes, made up for this page: Question: "Tell me about a time you missed a deadline." Score: 2 of 5. Moment: 1:40, I said "we pushed it a week" and never said what I did. Next time: "I told the client on Tuesday, two days before the date, and offered to ship the half that was ready."
Keep the list short. One fix per answer is plenty, and one fix per session is often better. The free debrief template has the same three columns if you would rather fill in a form.
If the story itself is weak, no delivery fix will rescue it. The STAR method guide covers how to rebuild a story from the result backwards, and the page on shortening an answer that drifted shows how to cut one down.
How do you know if the fix worked?
Record the same question again and score it against the same rubric. If the score rises and the timestamped problem is gone, the fix held. If a new problem appears in its place, that is normal progress, so keep the fix and work on the new problem in the next round.
Change the wording of the question slightly between rounds so you don't memorise a script. A friend can add a follow-up you did not plan, such as "What would you do differently?" The guide to whether your story changed shows how to keep the facts steady when you retell the same example.
After three or four rounds on one question, move on. The free interview questions tool takes a pasted job and the round you picked, then lists likely questions, why they are asked and what the interviewer listens for. Each of those is a ready starting point for your next rubric.
What mistakes do people make when reviewing themselves?
The common mistakes are watching for appearance instead of content, trying to fix everything at once, and scoring generously because you know what you meant. Each one makes the review feel productive while the next answer comes out the same. A written rubric and a single fix per round guard against all three.
- Judging your face and voice before your content. Start with the content pass, every time.
- Fixing five things at once. Pick the one that would move the score most.
- Scoring what you meant. Score only what is on the recording.
- Waiting a week to watch. Review within a day, while you still remember what you were thinking.
- Never recording again. A review without a second recording is only a critique.
How is a mock review different from a debrief of a real interview?
A mock review looks at practice, where you chose the question and can try again tonight. A debrief looks at a real interview after it ended, with an interviewer whose reactions you can't redo. The method is the same: score against a standard, find the moment and write better words. The stakes and the evidence differ.
The page on interview debrief vs mock interview covers when to use each. If you want practice with a clock, the Mock room has you answer out loud, scores each answer out of 10 and checks it against your resume. It is included in Prept's free plan.