How answers are scored
Last updated
VirtualInterview.ai scores every practice answer on relevance, depth, clarity, from 0 to 100, against the rubric of the round you chose, and explains each score. This page sets out what is measured, who measures it and where it stops. For a worked example, see the sample report.
What is scored
Every answer is scored on relevance, depth, clarity, each from 0 to 100, against the rubric of the round you chose. Relevance asks whether the answer addressed the question; depth, whether it gave a specific example rather than a slogan; clarity, whether an interviewer could follow it. The report says what each score is based on, so you can check its reasoning.
The overall score
The overall score is the average of the per-answer scores across the three dimensions. It summarises the session; the per-answer notes are where the useful detail is.
One rubric per round
You pick the round at setup (general, technical, behavioral, recruiter screen, system design, managerial, case study, technical assessment, leadership, panel, salary negotiation, final / executive), and answers are judged against that round's rubric: a system design answer and a recruiter-screen answer are not held to the same standard.
Delivery feedback
Content is what is scored. Delivery feedback is speech-based: speaking pace, filler words (in English answers) and pauses, measured from your spoken answer. There is no facial-expression, eye-contact or posture analysis — the camera is only your self-view.
Which AI models do the scoring
Scoring and feedback are written by third-party AI models. Pro subscribers can choose among 4 providers (OpenAI, Anthropic, Google, xAI) and regenerate feedback with a different one; model versions change, so they are not listed here.
The retry plan
The report ends with a retry plan for the weakest answers: which question to re-answer first and what to add, so the next attempt can be compared with the last.
Limitations
The scoring has limits worth knowing before you rely on it. These are the cases where a score can mislead, and the things the product does not measure at all.
- It is not a hiring prediction. A score says how this answer reads against the rubric; it is not the likelihood of an offer, a visa or an admission.
- It is AI-generated practice feedback, written by a model rather than a person.
- It is most useful on specific, complete answers. A one-line answer gets a one-line critique.
- We do not publish accuracy statistics we cannot show you.
- It is a practice tool: it does not feed you answers during a real interview.
Questions
Answers to what people ask about the methodology: how answers are scored, whether a score predicts a hiring decision, and whether the camera is used to judge your face or body language.
- How does VirtualInterview.ai score interview answers?
- Each answer is scored on relevance, depth, clarity, from 0 to 100, against the rubric of the round you chose, with the reasoning for each score shown in the report.
- Is the score a prediction of whether I will be hired?
- No. It measures how an answer reads against the round rubric. It is not a prediction of any hiring, visa or admissions outcome.
- Does it analyse my face or body language on video?
- No. There is no facial-expression, eye-contact or posture analysis — the camera is only your self-view. Delivery feedback is speech-based: speaking pace, filler words (in English answers) and pauses, measured from your spoken answer.
Where each statement comes from
Each section of this page is tied to the part of the product it describes, so you can check a claim against the setting or report it refers to. The list below maps them.
- What is scored. /sample-report, /faq
- The overall score. /sample-report
- One rubric per round. /about, /faq
- Delivery feedback. /about, /video-interview-ai
- Which AI models do the scoring. /about, /pricing
- The retry plan. /about, /sample-report