Reading candidate scores and cheat flags
How Naqwa separates CV match from the interview score, scores every answer 0–10 with a reason, tags weak or not-genuine answers, and shows integrity flags with counts, times, and questions.
Naqwa gives recruiters more than a gut feeling. Two separate scores, per-answer reasons, and a timeline of integrity events help you prioritise candidates and spot issues early — especially valuable for regulated employers in Qatar and the GCC who need documented hiring decisions.
Two different scores: CV match and interview score
Naqwa shows two numbers, and they measure different things:
- CV match — how well the uploaded CV fits your job description (skills, experience, keywords). It is calculated before the interview starts.
- Interview score — how well the candidate actually answered your interview questions. It is built from the scores of each answer.
A strong CV with a weak interview is common. Keeping the two scores apart stops a polished CV from hiding a poor screening session.
How the interview score is built
Every answer is scored from 0 to 10 with a short reason you can read. The interview score is the average of those answer scores, multiplied by 10. Real, first-hand examples with specific actions and measurable results score highest.
Some answers are capped no matter how confident they sound:
- Empty or silent answers — scored 0
- Very short answers — capped at 2/10
- Off-topic answers — capped at 2/10
- Examples the candidate admits are fictional — capped at 3/10
- Hypothetical or generic answers with no real example — capped at 4/10
When an answer is vague, very short, or made up, the AI interviewer asks a follow-up such as "Can you give a real example, with numbers?" A good follow-up answer recovers part of the score. Follow-ups do not use up one of the interview questions.
Open Review & decide to see the breakdown table — each question, its score, the reason, and any flags — and the recommendation, which can never be more positive than the interview score allows.
Answer tags in the transcript
Weak answers are marked directly in the transcript so you can find them quickly:
- Weak — empty, very short, generic, or off-topic
- Not genuine — the candidate said the example was fictional, or only described what they would do
- Text not typed — the answer appeared without being typed (copy and paste is blocked), for example injected or dictated text
- Possibly AI-written — the answer reads like generic AI output
"Text not typed" and "Possibly AI-written" are separate signals: one is about how the text arrived, the other about how it reads. Both are prompts to look closer, not proof.
Voice liveness check
When voice answers are enabled, one question must be answered out loud. Candidates test their microphone before the interview starts. If a recording is silent or too short to understand, the candidate is asked to re-record. After repeated silent attempts, the answer is scored 0 and flagged as Liveness check failed, and the overall interview score is capped.
Integrity timeline
Naqwa records integrity events with a count, the time, and the question that was on screen — for example "Tab switch ×2 — Q3, 2:31 AM & 2:34 AM". Events include:
- Tab or window switches
- Blocked copy, paste, and screenshot attempts
- Webcam events when a webcam is required — face not visible, looking away, multiple faces, camera lost
- Page reloads, connection drops, and long inactivity — shown as information, not as cheating
- Name mismatch — the CV name differs from the name the candidate entered
Flags never reject candidates automatically. They highlight cases worth a closer look — particularly for roles involving financial access, customer data, or safety-critical work common in Gulf industries.
How to act on flags
- Open the candidate in Review & decide and read the answers that are tagged or scored low.
- Check the timeline: a single short tab switch on mobile is different from repeated exits during one question.
- For name mismatches, verify legal name vs. preferred name — common in multicultural GCC workforces.
- Record your decision (shortlist, hold, or reject) with notes for audit trails.
Compliance tip for Qatar employers
Keep structured notes on why candidates were shortlisted or rejected. Naqwa's per-answer reasons and integrity timeline support fair, consistent screening — but your HR policy should define how automated signals combine with human judgment.