Conducting Interviews (Structured, Behavioral)

Scope

Covers

• Preparing and running structured interviews (screen + loop) with consistent criteria

• Behavioral interviewing mapped to competencies/values

• Getting to “substance over polish” (avoiding “confident but shallow” signal)

• Capturing evidence, scoring consistently, and writing a debrief-ready summary

When to use

• “Help me conduct interviews for a .”

• “Create an interview script / interview question set / scorecard for .”

• “Design an interview loop and structured rubric for .”

• “Improve interviewer consistency and reduce bias.”

When NOT to use

• You need to define the role outcomes or write the job description (use writing-job-descriptions first)

• You need legal/HR compliance guidance or to adjudicate complex employment risk (this skill is not legal advice)

• You need compensation/offer strategy or negotiation coaching

Inputs

Minimum required

• Role + level + function (e.g., “Senior PM”, “Engineering Manager”)

• Interview stage(s) to design/run (screen, hiring manager, panel, etc.) + duration(s)

• Evaluation criteria: 4–8 competencies/values to measure (or your existing rubric)

• Company/team context candidates should know (mission, what’s hard, why now)

• Candidate materials (resume/portfolio) + any areas to probe

Missing-info strategy

• Ask up to 5 questions from references/INTAKE.md.

• If criteria aren’t provided, propose a default criteria set and clearly label it as an assumption.

Outputs (deliverables)

Produce an Interview Execution Pack in Markdown (in-chat; or as files if requested):

• Interview plan (stage purpose, criteria, agenda, timeboxes)

• Question map (questions → competency/value → what good looks like → follow-up probes)

• Interviewer script (opening, transitions, probes, close)

• Notes + scorecard (rating anchors + evidence capture)

• Debrief summary template (evidence-based strengths/concerns + hire/no-hire signal + follow-ups)

• Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md

Expanded guidance: references/WORKFLOW.md

Workflow (7 steps)

1. Intake + define the stage

• Inputs: user request; references/INTAKE.md.

• Actions: Confirm role, stage(s), duration, and who else interviews. Identify must-measure criteria and any “must not” red flags.

• Outputs: Interview brief + assumptions/unknowns list.

• Checks: You can state the stage goal in one sentence (e.g., “screen for X; sell Y; decide Z”).

2. Lock evaluation criteria (don’t improvise later)

• Inputs: competencies/values; role context.

• Actions: Choose 4–8 criteria; define 1–2 “strong” and “weak” anchors per criterion. Ensure each criterion is observable via evidence.

• Outputs: Criteria table with anchors.

• Checks: Every criterion has a definition + evidence hints; no criterion is “vibe”.

3. Build the question map (behavioral first)

• Inputs: criteria table.

• Actions: Write 1–2 primary questions per criterion (behavioral: “tell me about a time…”). Add probes that force specifics (role, constraints, trade-offs, results, what you’d do differently). Add two global questions: “How did you prepare?” and “Why here?”

• Outputs: Question map table.

• Checks: Each question maps to exactly one primary criterion; no double-barreled questions.

4. Write the interviewer script (runbook)

• Inputs: question map; timeboxes.

• Actions: Assemble an interview flow: opening (set context + structure), question sequence, note-taking reminders, and a consistent close: “Is there anything else you want to make sure we covered?”

• Outputs: Interviewer script with timestamps.

• Checks: Script fits in time; includes “sell” moments appropriate to stage; includes candidate questions time.

5. Prepare for “substance over polish”

• Inputs: question map; candidate materials.

• Actions: Add “substance checks” for polished communicators (ask for concrete examples, counterfactuals, and specific decisions). Add “structure help” for less polished candidates (rephrase, clarify what’s being asked) without leading.

• Outputs: Substance-vs-delivery guardrails embedded in the script.

• Checks: The plan reduces false positives from confident delivery and false negatives from imperfect structure.

6. Score using evidence (immediately after)

• Inputs: notes; scorecard template.

• Actions: Fill the scorecard with evidence snippets before discussing with others. Rate each criterion with anchors. Write a 5–8 sentence evidence-based summary and list follow-up questions.

• Outputs: Completed notes + scorecard + summary.

• Checks: Every rating has supporting evidence; the overall recommendation is consistent with criterion ratings.

7. Debrief + quality gate + finalize pack

• Inputs: completed scorecard; debrief template.

• Actions: Produce the debrief-ready packet; run references/CHECKLISTS.md and score with references/RUBRIC.md. Include Risks/Open questions/Next steps.

• Outputs: Final Interview Execution Pack.

• Checks: Clear recommendation + uncertainty; fair process; next steps defined (additional interview, reference check, work sample, etc.).

Quality gate (required)

• Use references/CHECKLISTS.md and references/RUBRIC.md.

• Always include: Risks, Open questions, Next steps.

Examples

Example 1 (Screen): “Create a 30-minute phone screen for a Senior Product Manager. I want to evaluate product sense, execution, and collaboration. Output the Interview Execution Pack with a question map and scorecard.”

Expected: timeboxed script, behavioral questions, clear anchors, and a scorecard that captures evidence.

Example 2 (Loop): “Design a structured interview loop for a Staff Engineer, including a hiring manager interview and a cross-functional panel. Map questions to our values and include a debrief template.”

Expected: stage goals, consistent criteria across interviewers, and artifacts that make debriefs evidence-based.

Boundary example: “Just tell me if this candidate is good; I don’t have criteria or notes.”

Response: require criteria + evidence; propose default criteria and ask the user to paste notes or run a structured interview first.