Chapter 11: Writing Good Behavioral Questions and Running a Structured Interview

Why this matters

Everything covered so far has been written primarily from the candidate's side of the table. This chapter turns around: a poorly-designed interview process produces noisy, unreliable signal regardless of how well any individual interviewer means to evaluate fairly — and most of the fixes are structural, decided before any candidate ever walks in, not improvised in the room.

The core concept: structured beats unstructured, reliably

[BEGINNER] A structured interview asks every candidate for a given role the same, pre-planned set of core questions, evaluated against the same pre-defined criteria — as opposed to an unstructured interview, where each interviewer asks whatever comes to mind, in whatever order, evaluated against their own personal, unstated sense of what's good. Decades of hiring research consistently show structured interviews produce meaningfully better predictions of actual job performance than unstructured ones — not a marginal improvement, but one of the largest, most consistently replicated findings in personnel selection research.

[INTERVIEWER'S VIEW] The reason structure helps so much isn't mysterious: it removes a huge amount of interviewer-to-interviewer and candidate-to-candidate variance that has nothing to do with actual candidate quality. An unstructured interview where one candidate happens to get an easy, rapport-building question and another happens to get a harder, more probing one — purely because of interviewer mood or improvisation — produces scores that partly reflect luck of the draw, not comparative candidate strength.

Writing a good behavioral question, precisely

[SENIOR INSIGHT] A well-written behavioral question has three properties: it's phrased in the past tense, about a real instance (Chapter 1's core lesson, restated from the interviewer's side); it targets a specific, named category (Chapter 3) rather than being generically open-ended; and it's calibrated to the role's actual level — a question appropriate for evaluating a staff-level engineer's organizational influence is a poor fit for a new-grad loop, and vice versa.

Weak question: "Tell me about a challenge." (No category signal, no
                level calibration, invites an answer from ANY category
                the candidate happens to have prepared, making
                cross-candidate comparison unreliable.)

Strong question, new-grad level: "Tell me about a time you had to
                learn a new technology or tool quickly to complete
                a task." (Clear category -- learning agility/ambiguity
                -- calibrated to what a new grad is actually likely
                to have experienced.)

Strong question, senior/staff level: "Tell me about a time you had
                to influence a technical direction across multiple
                teams that didn't report to you." (Clear category --
                leadership/cross-functional influence -- calibrated
                to organizational scope a senior candidate should
                plausibly have encountered.)

Building a structured interview guide

[IN PRACTICE] A complete, structured behavioral interview guide for a given role typically includes:

## Structured interview guide template
1. A FIXED set of core questions (typically 3-5), one per category
   the role most needs to evaluate, decided BEFORE any candidate is
   interviewed -- not chosen live based on how the conversation is going.
2. A small set of PRE-WRITTEN follow-up prompts for each core question
   (e.g., "what was YOUR specific role," "what would you do differently"),
   ensuring probing depth is also consistent across candidates, not
   dependent on which follow-ups a specific interviewer happens to think of live.
3. A DEFINED scoring rubric per question (Chapter 12) that every
   interviewer on the panel uses, rather than each interviewer scoring
   against their own personal, unstated bar.
4. An assigned INTERVIEWER-TO-CATEGORY mapping across a multi-person
   loop, so different panel members cover different categories rather
   than several interviewers redundantly asking about the same one and
   leaving others uncovered entirely.

Why deviating from the structure is sometimes still appropriate

[SENIOR INSIGHT] Structure doesn't mean rigid, robotic adherence regardless of context — if a candidate's answer to a core question naturally and richly surfaces material relevant to a different category the interviewer was also responsible for covering, following that thread briefly is reasonable and often more efficient than artificially redirecting back to a scripted order. The actual discipline structure requires is: every candidate for the role gets the same core questions asked, at some point, evaluated against the same rubric — not that every interview transcript reads identically start to finish. This is the same calibration theme running throughout this entire book series: the goal (comparable, reliable signal) matters more than mechanical adherence to the specific means (a fixed script), as long as the actual goal is still being met.

Common mistakes

1. Running unstructured interviews and relying on individual interviewer judgment and improvisation, producing noisy signal that partly reflects luck of the draw rather than genuine, comparable candidate differences.
2. Writing generic, uncategorized questions ("tell me about a challenge") that invite an answer from whichever category the candidate happens to have most prepared, undermining reliable cross-candidate comparison.
3. Asking questions calibrated to the wrong level — a question that assumes organizational scope a new grad hasn't had the chance to develop, or one too basic to differentiate strong senior candidates from each other.
4. Treating "structured" as meaning "must follow the script rigidly regardless of what the conversation naturally reveals", missing that the actual requirement is comparable coverage and consistent scoring, not an identical, robotic transcript for every candidate.

Checklist: writing questions and structuring an interview

1. Write every core question in the past tense, about a real, specific instance, never as a hypothetical.
2. Target a specific, named category (Chapter 3) per question, and calibrate each question's assumed scope to the role's actual level.
3. Fix the core question set, follow-up prompts, and scoring rubric before the loop begins — not improvised live, and not varying candidate-to-candidate.
4. Assign categories across a multi-interviewer panel deliberately, ensuring full coverage without redundant overlap.

More sample questions on this theme (as an interviewer, drafting them)

1. Drafting a Conflict question for a mid-level engineer: "Tell me about a time you disagreed with a code review comment or a technical decision from a peer." — past tense, specific category, appropriately scoped to peer-level (not organizational) conflict.
2. Drafting an Ambiguity question for a new grad: "Tell me about a time you were given a task with unclear requirements." — appropriately scoped to task-level, not project- or org-level, ambiguity.
3. Drafting a Leadership question for a staff engineer: "Tell me about a time you drove a technical decision that multiple teams needed to align on." — appropriately scoped to cross-team organizational influence.
4. Drafting a Feedback question for any level: "Tell me about the most useful piece of feedback you've given a colleague." — works across levels, since giving specific, useful feedback is a skill relevant at every seniority.
5. Drafting a Mentorship question for a senior engineer: "Tell me about a time you helped a struggling teammate turn their performance around." — appropriately scoped to a senior's likely mentorship responsibility, more demanding than a general "helped someone" prompt.

Chapter summary / cheat sheet

• Structured interviews (fixed questions, fixed rubric, evaluated consistently) reliably outperform unstructured ones at predicting actual job performance — one of the most robust, consistently replicated findings in hiring research.
• A well-written behavioral question is past-tense, targets a specific named category, and is calibrated to the role's actual level.
• Fix the core question set, follow-up prompts, and rubric before the loop begins, and assign categories deliberately across a multi-interviewer panel.
• Structure requires comparable coverage and consistent scoring across candidates — not a rigid, identical, robotic transcript for every interview.

---