Scoring & feedback
How we score every interview.
No black box. Every answer is scored across real dimensions, each explained below with an example — including a real answer from a candidate’s live session, mistakes and all.
✦
Every interview is generated fresh — nobody gets the same questions twice
Every question is written new for your session, never pulled from a fixed bank. No two candidates get an identical interview, and sitting a second one won’t repeat the first. There’s nothing to leak, share or memorise in advance.
The core five — every role question, every time
These five mark every answer to your role-specific questions and are weighted into one percentage: Relevance 25%, Accuracy 25%, Clarity 20%, Depth 20%, Confidence 10%. Your headline score is the average across those questions.
Relevance25% weight
Does the answer actually address what was asked, and match what the role genuinely requires — not just talk in the right general direction?
ExampleAsked how to keep a transactional endpoint consistent if a downstream step fails, an answer about general code readability would score low — however well written, it isn’t answering what was actually asked.
Accuracy25% weight
Is what the candidate actually said true? This applies to every role we interview for, not just technical ones — a confident, clearly delivered answer built on a wrong fact still scores low, because a real interviewer would catch it too.
Example“HTML is the fastest programming language, and it was created in 2023” is said fluently and confidently — but it’s wrong on every count, and scores accordingly low. Confidence in a false claim is worse than no claim at all.
Clarity20% weight
Is the answer well-structured and easy to follow, or does the interviewer have to untangle it to find the actual point?
Example“I’d wrap the write and the message in one transaction, and roll back if either fails” scores high — one clean sentence, the point is immediately obvious.
Depth20% weight
Are there specific examples, named tools, real numbers or genuine outcomes — or generalities that could apply to almost any answer?
Example“I’d make it robust” has no depth. “I’d use an outbox table so the message only publishes after the transaction commits” does — it names the actual mechanism.
Confidence10% weight
Does the language sound assured, or is it hedged with “maybe”, “I think”, “kind of” — even when the underlying idea is correct?
Example“I think maybe we’d sort of use a transaction, possibly?” scores low even if a transaction genuinely is right — the hedging undercuts it.
A real, unedited scored answer
“CivicForge Solutions asks you to expose an ASP.NET Core endpoint that creates a case and publishes a notification, but the database write and message publication must not leave the system inconsistent if either operation fails. How would you design the endpoint and surrounding components?”
“well I would definitely design them around a single transaction and roll back if there was an error”
Relevance15%
Accuracy90%
Clarity70%
Depth5%
Confidence85%
Accuracy, clarity and confidence all score well — nothing said was wrong, and it’s a short, assured sentence — but relevance and depth are honest about the real gap: the transaction alone doesn’t guarantee the message publishes, and the answer names no mechanism for that. Accuracy and depth are deliberately separate: this answer is entirely true, just incomplete.
Bonus points — added on top
Bonus points lift your headline score, but they’re never averaged into it — so an off day on one of them can’t drag down how you did on the questions that actually test the role. Your score can never go above 100%.
Character questions — ownership, execution & proactivenessup to +10 each
Amina (HR) asks open-ended character questions in every interview — one probing ownership and follow-through, one probing genuine forward planning. The wording is generated fresh each session, so you won’t get the same pair twice. They sit on top of your role questions, never in place of any of them.
What earns the pointsOwnership — did you personally drive or decide it, not just describe what “we” did? Execution — did you see it through to a real outcome? Proactiveness — genuine, specific planning, not a rehearsed non-answer. “Our team shipped a big project” shows neither; “I noticed the process was broken, proposed the fix, got buy-in and shipped it three weeks later” shows both.
Bonus round — a live knowledge check+10 per correct answer
Partway through most interviews, a surprise multiple-choice question fires — hard and specific to your role. Get it right and you earn 10 bonus points. Get it wrong and you receive the full explanation on the spot, then move straight back into the interview — no penalty.
Ask the interviewers+5 points
Near the end of about half of interviews, Amina asks whether you have any questions for the panel — the moment many candidates regret staying silent. Take it: we’ll suggest a strong question to ask, save it to your Question Bank, and award 5 bonus points for engaging.
If you skip a question or ask for the answer
Pass on a question, or ask us to tell you the answer, and that question scores zero. The answer is shown so you learn it, but it isn’t marked as an attempt — which is the honest way to see where you really stand.
Your outcome — pass, keep on file, or not this time
Your final percentage, bonuses included, gives one of three outcomes. The bar rises with the level you choose, and an interview in a specific company’s style uses a level that matches that company’s bar.
| Level | Pass | Keep on file | Not this time |
|---|
| Beginner | 60%+ | 45–59% | below 45% |
| Standard | 65%+ | 50–64% | below 50% |
| Pro | 70%+ | 55–69% | below 55% |
| Expert | 90%+ | 75–89% | below 75% |
Keep on file means a real employer would keep your CV (or name) on file for future roles. A pass earns a certificate you can download and share.
See it happen in your own interview.
Every dimension above, scored live, on a role you actually name.
Start practicing →
