Mock Interview Rubric You Can Self-Score
Most people judge a practice session by one question: "Did I solve it?" That hides most of what interviewers actually evaluate. You can solve a problem and still get a weak score for silent coding and no testing. You can also miss the optimal solution and still get positive feedback for clear reasoning and a clean recovery.
This mock interview rubric you can self-score breaks a coding round into six dimensions with simple anchors, so every mock tells you exactly what to work on next. It is not any company's official rubric. It is a practical approximation of what interviewers commonly look for in live coding rounds.
The six dimensions
| # | Dimension | What it captures |
|---|---|---|
| 1 | Problem understanding | Restating, clarifying questions, examples |
| 2 | Approach and reasoning | Brute force, better idea, tradeoffs |
| 3 | Code quality | Correct, readable, reasonable structure |
| 4 | Testing and verification | Dry runs, edge cases, fixing bugs |
| 5 | Communication | Narration, check-ins, handling hints |
| 6 | Composure and time | Pacing, recovery from mistakes |
The 1–4 scale
Use four levels so you cannot hide in the middle:
- 1 – Missing: did not happen or actively hurt the round
- 2 – Partial: happened, but late, shallow, or only when prompted
- 3 – Solid: happened clearly and on your own
- 4 – Strong: clear, early, and noticeably well done
Anchors for each dimension
1. Problem understanding
| Score | Looks like |
|---|---|
| 1 | Started coding without restating; misread the problem |
| 2 | Restated, but asked no clarifying questions |
| 3 | Restated, asked about input size and edge cases, did one example |
| 4 | All of 3, plus caught an ambiguity that mattered |
2. Approach and reasoning
| Score | Looks like |
|---|---|
| 1 | Jumped to code with no stated plan |
| 2 | Stated one approach without complexity |
| 3 | Stated brute force and a better approach with complexity |
| 4 | All of 3, plus compared tradeoffs and chose deliberately |
3. Code quality
| Score | Looks like |
|---|---|
| 1 | Did not reach working code, or code was very hard to follow |
| 2 | Mostly working, with messy names or tangled logic |
| 3 | Working, readable, sensible names and helpers |
| 4 | Clean and idiomatic, easy to extend |
4. Testing and verification
| Score | Looks like |
|---|---|
| 1 | Said "done" without testing |
| 2 | Tested only the given example |
| 3 | Dry-ran a normal case and at least one edge case |
| 4 | Found and fixed a bug through your own testing |
5. Communication
| Score | Looks like |
|---|---|
| 1 | Long silences; interviewer had to ask what you were doing |
| 2 | Talked sometimes, mostly when asked |
| 3 | Narrated plan and key decisions; checked in at least once |
| 4 | Easy to follow throughout; used hints smoothly |
6. Composure and time
| Score | Looks like |
|---|---|
| 1 | Froze or ran out of time with nothing working |
| 2 | Recovered from a stall, but slowly |
| 3 | Kept a steady pace; finished with a few minutes to test |
| 4 | Handled a wrong turn or follow-up calmly and on time |
How to score honestly
Scoring yourself is hard because memory is generous. Three ways to make it more accurate:
- Record the session. Audio is enough. Score while listening back, not from memory.
- Score within an hour. Details fade fast.
- Use evidence, not feelings. For each score, write one line of evidence: "Asked about duplicates at 2:30" or "Silent from 12:00 to 16:00."
If a partner ran the mock, have them score you separately, then compare. Large gaps show where your self-perception is off.
A sample scorecard
| Dimension | Score | Evidence |
|---|---|---|
| Problem understanding | 3 | Restated, asked about empty input and size |
| Approach and reasoning | 2 | Went straight to hash map; no brute force or complexity |
| Code quality | 3 | Working and readable |
| Testing and verification | 1 | Said "done" right after coding |
| Communication | 2 | Quiet while coding the main loop |
| Composure and time | 3 | Finished with five minutes left |
| Total | 14 / 24 |
The total is less useful than the lowest scores. Here, testing (1) and approach (2) are the obvious focus for next week.
Turn scores into a weekly plan
- After each mock, highlight your lowest one or two dimensions.
- Pick one concrete habit for the weakest: for testing, "dry-run two cases before saying done."
- Repeat that habit in every practice problem until the next mock.
- Track scores over time in a simple table.
| Week | Underst. | Approach | Code | Testing | Comm. | Composure |
|---|---|---|---|---|---|---|
| 1 | 3 | 2 | 3 | 1 | 2 | 3 |
| 2 | 3 | 3 | 3 | 2 | 2 | 3 |
| 3 | 3 | 3 | 3 | 3 | 3 | 3 |
Aim for steady 3s across the board before chasing 4s. A single 1, like never testing, can be the thing an interviewer remembers from the round.
Adapting the rubric to other rounds
- System design: replace code quality with "high-level design" and testing with "tradeoffs and failure modes."
- Behavioral: use story structure, specificity of your actions, results, and self-reflection.
- Debugging rounds: replace approach with "hypothesis forming" and testing with "isolating the bug."
Keep the 1–4 scale and evidence lines. The habit of scoring transfers.
Common self-scoring mistakes
- Scoring only "solved or not"
- Giving yourself 3s by default to feel better
- Changing the rubric every week so progress cannot be compared
- Working on all six dimensions at once
- Ignoring composure because it feels less technical
Soft practice note
A rubric works best when every mock is recorded and timed the same way. If you want a Chrome side panel to run consistent, timed practice sessions you can score afterward, Luna Interview is built for calm, structured rehearsal. It does not grade you for an employer, and it is not a way around interview rules.
Day-of checklist
- [ ] Review your last scorecard's two weakest dimensions
- [ ] One habit per weak dimension, written on a sticky note
- [ ] Restate and clarify before coding (dimension 1)
- [ ] Brute force plus complexity out loud (dimension 2)
- [ ] Dry-run two cases before saying done (dimension 4)
- [ ] Check in at least once (dimension 5)
Solving the problem is one part of the score. A rubric shows you the other five, and which one to fix next.
Rehearse with an AI interview copilot, then browse commonly reported interview questions.