In a CoPilot Interview mock interview, every answer you give out loud gets a whole-number score from 1 to 10. The AI model you picked as the interviewer gives the score, together with what worked, what to improve, the points you didn't cover, a strong answer and the follow-up a real interviewer would likely ask. At the end of the round, the report shows how many questions you answered and your average score to one decimal place.
CoPilot Interview is an AI interview assistant for Windows and macOS, and Mock Interview is one of its four modes, included on the free plan. This article explains where the numbers come from and how to read them. To set up a round, see the step-by-step Mock Interview guide.
What each score is based on
When you click ✅ Done answering, press Ctrl+Shift+A (⌘+Shift+A on a Mac), or stop talking for 8 seconds on a non-coding question with auto-finish on, the app sends four things to the model chosen under 🤖 Interviewer AI: the question, a short list of points a strong answer should cover, the transcript of what you said, and your 🎯 Level. For a coding question it also sends any code you typed. The model is told to ignore transcription slips, filler words and grammar, and to grade substance, structure and communication the way a real interviewer would.
| Question type | What the grader looks for |
|---|---|
| 🗣 Behavioral | A concrete situation, your own actions, a measurable result and what you learned |
| 🔧 Technical | Substance, structure and communication; technical questions are written to need explanation and trade-offs, not a yes or no |
| 🏗 System Design | Clarifying the requirements, a coherent high-level design, the key trade-offs, and how the design scales and fails |
| 💻 Coding | Correctness, complexity, edge cases and readability of the code you typed; if you only talked through an approach, the approach plus what a working solution still needs |
Because your level is part of every grade, a 7 at Senior is a stricter bar than a 7 at Junior.
What the feedback card shows
The grade appears in the answer panel as a card titled like “📊 Feedback · Q2 · 7/10”, and the header above the question changes to “Question 2 of 5 · 7/10”. The card lists, in this order:
- Score: 7/10, followed by a one-sentence verdict
- 🟢 What worked: the parts to keep next time
- 🔴 Improve: specific, actionable fixes
- ⚠️ Not covered: points from the question's list that your answer missed
- 💡 A strong answer: a concise model answer; see how to use the model answer
- ❓ Likely follow-up: the probing question a real interviewer would ask next
- 🗣 You said: the start of your transcript (up to 600 characters), so you can check what the app heard
How the round report and average work
- The report heading reads like “Mock interview report — 4 of 5 answered · average 6.8/10”.
- The average covers graded answers only. A question you skip counts as not answered. If grading fails for an answer, that answer has no score and is left out of the average.
- ↺ Retry asks the same question again, and your new attempt replaces the earlier score in the report.
- You can click 📊 Summary at any point after your first answer to get the report early. Questions you haven't reached are marked as not reached.
- Under the heading, the AI writes a debrief in five parts: Overall (how ready you are for this role today), Per question (each score with the single most important fix), Top 3 things to work on, What already lands and Next round (which question types to repeat, and at what level).
- The round ends with “Round complete — average 6.8/10. Start a new round for fresh questions on the same role, or End the session.” 🎧 New round tells the AI not to repeat questions from earlier rounds in the same session.
Limits of the score
| Item | Detail |
|---|---|
| Scale | Whole numbers from 1 to 10 for each answer; the round average shows one decimal place |
| Who grades | The model in 🤖 Interviewer AI. The free plan uses free models; Pro and Pro+ unlock stronger models |
| Rubric | No fixed rubric across rounds: the AI writes each question's list of points when it plans the round |
| Language | Feedback is written in the 🌐 Lang language you chose (10 languages) |
| Meaning | A coaching signal for your next attempt, not a prediction of a hiring decision. AI feedback can be wrong, so check it against your own judgment |
For how answer scoring works across AI practice tools in general, read what a score out of 10 does and does not tell you.
Other scores in CoPilot Interview
- Browser demo: the browser demo grades a typed answer to one question from its bank of 145, with a score from 1 to 10, strengths, weaknesses and a likely follow-up. It doesn't show a strong answer or a round report.
- Candidate mode 📊 Summary: a report card with an overall score and five areas (Content & accuracy, Structure (STAR/clarity), Communication, Problem-solving, Role fit). It scores only what you said while your microphone was on; with the microphone off, it is a prep report without scores.
- Interviewer Mode: scores a candidate's answers and adds a Hire, No hire or Maybe recommendation; see Interviewer Mode reports.
If a score looks wrong
- Check 🗣 You said first. The grade covers what the app heard. If words are missing, fix the microphone before your next answer; see microphone and voice settings.
- “Grading failed: … Retry the question or move on.” The model didn't return a grade. Click ↺ Retry, or pick another model under 🤖 Interviewer AI before your next round.
- Feedback feels generic. Add your résumé and the full job description. Without them, the app runs a general interview at your level, on your Focus topic or, if you left Focus empty, on software engineering.
- Your score dropped after a retry. The report keeps your latest attempt, so retry until the fix holds.
Frequently asked questions
Is 7 out of 10 a good score in a CoPilot Interview mock interview?
The app doesn't publish score bands. Compare a score with your own earlier answers at the same level and question type, and use the 🔴 Improve list to decide what to change. A rising average across rounds tells you more than any single number.
Why did a similar answer get a different score?
Each grade is a fresh call to an AI model, and each round's questions come with their own list of points to cover. Scores can vary between attempts and between models, so judge progress across several answers rather than one.
Does the browser demo score the same way?
It uses the same 1-to-10 scale, but it grades a typed answer to one question from a fixed bank. It returns strengths, weaknesses and a follow-up, with no strong answer and no round report.
Are my mock interview scores saved?
Yes, on your computer. Every session is saved automatically as a Markdown file in Documents › CoPilot Interview › Sessions, including the feedback cards and the round report.
Practice this interview before the real one
The AI mock interview builds questions from your resume and the job description, reads them aloud and scores each spoken answer out of 10. Included on the free plan.
Start a mock interview