Every completed screening lands in the review queue at Candidates → Screening, scored 0–100 with a recommendation attached. This guide covers how to read a result, when to trust the score, and what to do when it's borderline — or missing.
| Tab | What's in it |
|---|
| Pending | Scored responses waiting for your decision |
| Disqualified | Candidates knocked out by a hard disqualifier — reviewable, and overridable if the knockout was too blunt |
Disqualification is never final by itself: the exact question that fired is flagged on the response, and you can override and advance the candidate if you disagree.
- Per-question weights. Each question carries a weight; critical questions move the overall 0–100 score more than nice-to-haves. Weights come from your template or the AI-generated questionnaire and are editable in the builder.
- Hard disqualifiers. Questions configured with acceptable answers (e.g. right to work = yes) bypass scoring entirely — a failing or missing answer disqualifies, fails closed, and shows which question fired.
- Soft tags. Answers can apply tags to the candidate record (e.g. a work-pattern or seniority tag) for filtering later, without affecting the score.
| Score | Recommendation | Suggested action |
|---|
| ≥ 70 | Advance | Move forward — check any listed concerns first |
| 50–69 | Review | Read the answers; often one probe response decides it |
| < 50 | Reject | Critical mismatches — skim before declining |
For the 50–69 band:
- Open the response detail panel and read the actual answers — especially replies to push-back probes, where thin candidates stay thin and good ones add substance
- Check the concerns the AI listed alongside the score
- Weigh the non-negotiable answers (already verified) against the soft gaps
- Still torn? A five-minute call resolves more borderline screens than any re-read
If the AI couldn't produce a valid score you'll see an amber scoring unavailable panel instead of a real number — never a silent fake score. From there you can:
- Click Retry AI Scoring to re-run it immediately
- Read the raw answers and decide manually — the candidate's submission is fully intact
- Do nothing: a background job retries failed scorings automatically
The queue learns from your decisions. When you disqualify 3 or more responses on the same role's questionnaire within 30 days, PlacementFlow analyzes the pattern — which disqualifiers keep firing, what keeps slipping through — and regenerates the screening questions so the same mismatch gets filtered earlier, at the gate, before it costs anyone time.
- Scores responses against your weights and thresholds and drafts a recommendation
- Flags hard-disqualifier hits with the exact question that fired
- Regenerates questions after a pattern of disqualifications on a role
- Every advance, rejection, and override is your decision — recommendations never act on their own