👥 HR & Recruiting Prompt Generator Free · No Signup

STAR Interview Scorecard Generator — Weighted Scores, Evidence Only

⚡ Quick Answer

A STAR interview scorecard turns each competency into behavioural questions (Situation, Task, Action, Result) with rating anchors an interviewer can score against evidence. This generator builds the full scorecard — weights validated to sum to 100%, observable anchors at 1, 3, and 5, and a debrief sheet — and forbids scoring anything but job evidence.

⚡ STAR Interview Scorecard Generator — Weighted Scores, Evidence Only

Configure your facts below to generate a production-ready, constraint-locked prompt.

100% Client-Side

What this generator does

Unstructured interviews feel rigorous and predict poorly, because every interviewer scores a different thing: one scores confidence, one scores similarity to themselves, one scores the quality of the small talk. A scorecard is the fix — the same competencies, the same questions, the same anchors, for every candidate in the process. The STAR method supplies the questions: instead of asking what a candidate would do, you ask what they did — the Situation, the Task they owned, the Action they took, the Result that followed — and you score the evidence in the answer.

This generator takes your competency list and weights and produces the complete interviewer pack:

  1. The scorecard table — Criterion | Weight | What good looks like | Concerns to probe | STAR questions (with follow-up probes) | Rating anchors at 1, 3, and 5 written as observable behaviours.
  2. The arithmetic, enforced — weights must sum to 100%. The generator checks this client-side before it builds anything; if your weights total 90 or 110, it tells you and waits. Interview tools that let weights drift produce totals nobody can compare across candidates.
  3. A debrief sheet — per-criterion evidence summary lines, the weighted total, and a hire / hold / pass threshold decided before the interviews, not negotiated after a favourite emerges.

Who it's for

How it works

  1. Choose 5–8 competencies, no more. Each must be provable in an interview answer. "Leadership" is not provable; "has resolved a disagreement between two teams with a decision both sides accepted" is.
  2. Weight by consequence. Ask: if this competency is weak, how badly does the hire fail? Weight follows that answer, and the total must be 100.
  3. Generate the scorecard. The template writes two STAR questions per competency with probes ("What was your specific contribution?" is the probe that separates team credit from personal evidence).
  4. Write the anchors as behaviours. A 5 is not "excellent leadership" — it's "describes a conflict they owned end-to-end, names the trade-off they chose, and the measurable result". The template refuses vague anchors.
  5. Calibrate before you go live. Score one mock or past interview as a panel and compare. Anchors that panellists interpret differently get rewritten now, not argued about in a debrief.
  6. Use the same scorecard for every candidate in the process. No per-candidate edits. That sameness is what makes the totals comparable — and defensible.

The prompt template

WORKED EXAMPLE / CODE 
You are an industrial-organisational assessment specialist building a structured interview scorecard.

ROLE BRIEF:
- Role and level: [title, level]
- Interview stage: [screen / panel / final]
- Competencies and weights: [competency — weight%, repeated for each; weights MUST sum to 100]
- Outcomes this hire must prove: [list]
- Evidence-based disqualifiers from the role brief: [list, or "none specified"]

STEP 1 — VALIDATE BEFORE BUILDING: Add up the weights. If they do not total exactly 100, stop and tell me the total and which competencies to adjust. Do not silently rebalance them.

STEP 2 — BUILD THE SCORECARD TABLE with these columns:
Criterion | Weight | Positive signals (observable) | Concerns to probe | STAR questions | Rating anchors

For EACH competency:
- Write 2 STAR questions that ask for a past, specific example ("Tell me about a time when…"), each with 2 follow-up probes that isolate the candidate's personal action from their team's.
- Rating anchors at 1, 3, and 5 ONLY, each written as observable behaviour in the answer — what the candidate describes doing and what resulted. Never personality adjectives.
- Positive signals and concerns must be things an interviewer can hear or fail to hear in an answer — nothing visible, nothing assumed.

STEP 3 — DEBRIEF SHEET: one evidence-summary line per criterion ("Evidence heard: …"), a weighted-total calculation, and the decision thresholds I set in advance: [hire at ≥ X / hold between / pass below — leave as fields if I didn't provide them].

HARD RULES:
- Do not include protected characteristics or proxies. Do not score appearance, accent, family status, or cultural similarity. Tie scores to observable evidence.
- No question may ask about, and no anchor may reward or punish: age, marital or family status or plans, religion, nationality or origin, health or disability, or political views.
- Do not score "communication style" as fluency, accent, or polish. If the role genuinely requires a communication behaviour (e.g. presenting to executives), define it as the behaviour and its observable evidence.
- Do not include "culture fit" as a criterion. If I ask for it, convert it into the specific, observable work behaviours I describe — or flag it back to me as unscoreable.
- Every question must be answerable from work experience. No hypotheticals scored as evidence; if you include a hypothetical, label it "judgement probe — do not score as STAR evidence".
Best model for this: GPT-6.1 (table discipline and arithmetic). Use Claude Sonnet 5.5 when the competency list itself needs challenging and refining first.

Worked example

Filled inputs: Role: Senior Product Manager. Stage: panel. Competencies and weights: Product discovery 35%, Stakeholder management 25%, Data-informed decision-making 20%, Delivery ownership 20%. Outcome to prove: has owned a roadmap decision that traded off two executive stakeholders. Disqualifiers: cannot describe a decision they personally made (all "we"-answers).

Output excerpt (one row of the scorecard):

CriterionWeightPositive signalsConcernsSTAR questionsAnchors
Data-informed decision-making20%Names the metric, the baseline, and how the result was verified; describes a time data changed their mindMetrics quoted without baselines; analysis always done by "the data team""Tell me about a decision where the data contradicted your instinct. What did you do?" Probes: "Which metric, exactly?" / "What would you have shipped without the data?"1: Cannot name a specific metric or decision. 3: Names metric and decision; verification is anecdotal. 5: Names metric, baseline, decision, and a verified outcome; includes a case where data overruled them.

Debrief line (excerpt): "Weighted total: 4.1 / 5 — Hire threshold (≥ 4.0) met. Weakest evidence: Delivery ownership — both examples were team-attributed; panel to probe once more before offer."

Guardrails

Frequently Asked Questions

Practical answers covering model choice, workflow integration, and compliance rules.

The role and level, five to eight competencies, a weight for each with the weights summing to 100%, rating anchors at 1, 3, and 5 written as observable behaviours, evidence-based disqualifiers taken from the role brief, space for the interviewer's evidence notes per criterion, and the weighted total read against a hire / hold / pass threshold the panel agreed before interviews began. If any of those is missing, the scorecard is a form, not an assessment tool.

Weight by the cost of failure, not by how easy a competency is to interview for. Ask of each one: if this hire is weak here, what breaks, and how soon? The competency whose absence sinks the role in the first quarter takes the largest share; competencies that can be coached in six months take the smallest. Then check the arithmetic — the generator validates that your weights total exactly 100, because a scorecard at 90 or 110 produces totals that can't be compared between candidates.

Three mechanisms. It fixes the questions in advance, so candidates aren't graded on different evidence. It replaces impressions with anchored observation — the interviewer records what the candidate described doing, not how the answer felt. And it makes the decision auditable: in a debrief, a score must point at evidence on the sheet. Scorecards don't eliminate bias — the anchors themselves can smuggle it in, which is why the guardrails here forbid scoring accent, polish, appearance, or cultural similarity — but they move judgement from memory and affinity to recorded evidence.

Anything that is, or proxies for, a protected characteristic: appearance, accent, age signals, family status or plans, religion, nationality, health, and "cultural fit" or similarity to the interviewer. Also unscoreable: confidence as a performance (some excellent candidates interview quietly), and hypothetical answers presented as experience. If a criterion can't be tied to something the candidate observably did in a past role, it doesn't belong on the scorecard.

Yes — the same competencies, questions, and anchors for every candidate at a given stage. Interviewers can probe deeper where an answer opens a door, but the scored instrument stays identical; the moment panels improvise different criteria per candidate, the weighted totals stop being comparable and the process loses its defensibility. Different stages can use different scorecards — a screen weights communication of experience differently from a final panel — but within a stage, one instrument for all.

Internal links

---