Interviewing for Hard Skills

Every senior candidate's first answer is true the way a press release is true. Watch one question descend through its layers, a CFO candidate, asked about the refinancing on her CV:

Layer zero, the prompt: "Walk me through the 2023 refinancing." — "It was a challenging environment, but we built strong relationships with the banks, aligned the board early, and closed a structure that gave the group real headroom." True, polished, and useless: it could describe any refinancing by anyone, ever.

One layer down: "Take me to the hardest week of it." — "Probably late March. Our lead bank's credit committee balked at the covenant package, and we had a bond maturity clock running." Now there is a scene.

Two layers down: "What did you actually do, that week, yourself?" — "I flew to Frankfurt without the advisors. I offered the security package our own board had told me to hold back — I judged we'd lose the syndicate otherwise — and then I had to go back and tell the chairman I'd exceeded my mandate."

Three layers down: "What would the chairman say about that conversation?" — a pause — "He'd say he was furious for a day and grateful for a year. You can ask him."

Four exchanges. The press release became a decision, the decision acquired a scene, the scene acquired ownership, and the ownership acquired a witness. Nothing about the candidate changed between layer zero and layer three. What changed was the interviewing, and that is this chapter: the machinery, promised since Chapter 2 put the structured interview at the top of the validity table and borrowed by every chapter since, for descending from what candidates say to what they did.

Episodes, not hypotheticals

First, the design decision underneath everything: at senior level, ask about the past, not the hypothetical. The comparative evidence is clear and gets clearer as jobs get complex. Past-behavior questions ("take me to a time...") hold their validity for high-complexity roles where situational questions ("what would you do if...") fade. The reason connects to Chapter 14's hazard: a hypothetical retrieves the candidate's theory, polished and espoused and costless, while an episode retrieves a commitment to facts that have witnesses, dates, and consequences. A senior candidate can theorize brilliantly about refinancings; only an episode contains a Frankfurt flight, an exceeded mandate, and a furious chairman who can be called. Hypotheticals keep one honest use in this book, Chapter 6's simulations, where the "hypothetical" is enacted under observation rather than narrated, but in the interview chair the rule is flat: episodes carry evidence; hypotheticals carry eloquence.

The components that carry the validity

"Structured interview" suffers from its own name, which suggests a script read at a hostage. The research decomposed structure into its components and found the validity concentrated in a handful, which is liberating news, because the handful is compatible with a conversation a chairman would enjoy:

Same evidence targets for every candidate — the demands list from Chapter 3, converted into the same core stimuli, so candidates are compared on the same terrain rather than on wherever each conversation wandered. Anchored rating, per answer — evidence scored against Chapter 16's anchors while the words are exact, not reconstructed from glow at the end. Planned probing — and read this one carefully, because it is where the caricature dies: the research does not forbid follow-ups; it forbids improvised, unequal follow-ups. Probes designed in advance — the same descent available to every candidate — are structure, not a violation of it; they are, in fact, where this book's whole method lives. And consistent interviewers with separated missions — Chapter 4's chairs, Chapter 19's architecture.

What turns out to be dispensable: the frozen script, the ban on warmth, the identical robotic wording. What is not dispensable is one discipline this book adapts rather than adopts: prior-information management. The research finding is that reviewing scores and others' impressions before interviewing contaminates judgment; the interview becomes confirmation. At executive level you cannot feign ignorance of the CV, and shouldn't, because the hooks of Chapter 3 require knowing the claims. The workable rule: the interviewer enters knowing the demands list and the claims to be verified, never the other interviewers' scores, never the psychometrics, never the client's early favorite, and the fourth layer below exists precisely to discipline the confirmation risk that the necessary knowledge creates.

The four-layer architecture

The book's master template, assembled from the components the evidence blesses. For each demand on Chapter 3's list, build one tree:

Layer 1 — the standardized stimulus. One prompt, identical for every candidate, engineered to select a diagnostic episode rather than a favorite story: specific ("take me to..."), bounded ("a moment when..."), and pointed at the demand's hard case — the endangered commitment (Ch. 9), the inherited team (Ch. 10), the peer with every reason to say no (Ch. 11), the fog-bound decision (Ch. 14). The prompt's craft is in what it makes impossible: a well-built stimulus cannot be answered with a press release.

Layer 2 — the planned probes. Four to six follow-ups, written in advance, that force the episode to cover the scoring anchors: sequence, options, actions, costs, others' roles. These are the rubric wearing conversational clothes — the candidate experiences curiosity; the file receives coverage.

Layer 3 — the verification probes. The conversion layer: from "we" to "I," from narrative to checkable claim, from story to witness. What did the document say? Who was in the room? What would the chairman say — can I ask him? Every answer is a hook planted for Chapter 7's calls, and, the double effect noted throughout Part III, the candidate hears the hooks going in, which changes the room's honesty in real time. Layer 3 is where interviewing stops being a genre of conversation and becomes a chain of evidence.

Layer 4 — the disconfirming probes. The interviewer's own bias countermeasure, mandatory because of the prior-information rule above: the deliberate search for the episode that cuts against the forming impression — the miss, the reversal, the failed coalition, the resister who was right, the lucky win. Part III built a signature pair for every cluster; the construction rule generalizes: whatever the first three layers are confirming, the fourth layer attacks. An interview without a Layer 4 is an audition with extra steps.

Construction, in one line each: start from the demand, not from a favorite question; write the anchors before the probes (you cannot probe toward standards you haven't set); pre-plan the descent three deep; and cap the tree at one stimulus, one descent, eight to twelve minutes, because the time economics are unforgiving and decisive: three or four episodes taken to layer three beat twelve questions answered at layer zero, every time, for the same reason four core samples beat forty photographs of the ground.

Running it at senior level

Executives resist schoolish formats, and the resistance is partly right: a fifty-question checklist would insult the candidate and the craft. The resolution the best practitioners converge on is covert-but-real structure: the candidate experiences a serious professional conversation, warm and responsive and senior, while the interviewer runs the trees underneath it, same stimuli, planned descents, anchored scoring, in whatever order the conversation makes natural. The structure is in the coverage and the scoring, not in the visible choreography. Three disciplines keep it real rather than merely covert: verbatim capture — key phrases in the candidate's words, because Chapter 16's evidence cells need quotations, not paraphrase; the silence tool — after a layer-2 probe, three seconds of patience routinely buys the sentence the rehearsal didn't include; and the talk-time audit from Chapter 4, run on yourself — if the recording would show you above twenty percent, you were being interviewed.

And the dual purpose, honestly: a senior interview must assess and attract, and the tension is smaller than feared, because the architecture itself attracts the candidates worth attracting. Strong operators consistently report, and tell their networks, that the deep-episode interview was the most serious conversation of the process; the candidates alienated by rigor were, on the record of Chapter 1's failure data, the ones the rigor exists to catch. Sell the role in the bookends, opening context and closing vision, and let the middle do its work. A process that sells for ninety minutes and assesses for none has chosen its outcome.

Targeting the hard skills

Everything above serves both registers; Part II's specific business is the technical half, and two rules from Chapter 4's chair map govern it. The deep technical dive is designed by the expert and scripted — chair 2 builds it, chair 3 runs it, anchored and bounded — because that is where domain knowledge becomes a precision instrument instead of a jargon duet. And recall is minimized: the target is never what the candidate knows (threshold, checkable, Google-adjacent) but episodes of technical judgment exercised in role, the architecture choice defended to a board, the accounting treatment fought over, the make-or-buy call that aged well or badly. The same four layers descend; only the terrain is technical.

Three specimen trees

CFO — demand: capital-structure judgment under stress. L1: "Take me to the financing you closed under the most pressure — where the clock and the market were both against you." L2: What were the live options at the worst point? What did each cost? Who disagreed — advisors, board, banks? What did you concede, and what did you refuse to? L3: What did your board paper recommend — the actual document? Who ran the syndicate on the other side — would they take my call? L4: "Now a financing decision you'd take back — structure, timing, or covenant you regret. What did it cost, and who bore it?"

Country Manager — demand: building a market through partners you don't control. L1: "Take me to your hardest distributor or partner problem — where the business ran through someone you couldn't instruct." L2: What did they want that you couldn't give? What leverage did you actually have? What did you change — terms, people, structure? How long did it take? L3: Who was the principal on their side — would they speak to me? What did headquarters know, and when? L4: "And a market where your approach didn't work — where you misread the channel or the culture. What were you carrying in that you shouldn't have been?" (Note the seam: that last probe is Chapter 17's misread-culture question wearing commercial clothes.)

CTO — demand: platform decisions with ten-year consequences. L1: "Take me to the build-versus-buy — or replatform-versus-patch — decision you're most accountable for." L2: What was the honest state of the estate? What did the losing option have going for it? Who in your own team opposed you? What did you promise the board, in numbers? L3: Is there an architecture decision record — a paper? Where is that platform now — and who runs it? L4: "And the technical debt you left behind somewhere — the shortcut that someone else paid for. What would your successor say you handed them?"

Red flags — in the interviewer

The anatomy inverts for this chapter, because at layer zero the interview's failure modes belong to your side of the table. The rescue prompt: the candidate stalls, and you helpfully supply the answer's shape, "so I imagine you escalated?", converting your hypothesis into their testimony. The leading descent: probes that telegraph the scored answer. The favorite question: the clever riddle or trademark stumper, beloved of senior interviewers and predictive of nothing but familiarity with the interviewer. The tour: letting a fascinating candidate walk you through terrain no demand requires, an hour of pleasure, zero cells filled. And the talk-time drift: measured, not estimated. Every one of these is invisible from inside; the correctives are the printed trees, the self-audit, and Chapter 25's habit of occasionally reviewing one's own recordings, which is as humbling as it sounds and twice as useful.

Instrument: the tree-builder

One form per demand, filled at chair-2 stage, printed for the room:

Demand (from the Ch. 3 list): ________ · Anchor set (what 2/3/4 look like in evidence — write first): ________ L1 stimulus (specific, bounded, aimed at the hard case): ________ L2 planned probes (4–6, covering the anchors): ________ L3 verification probes (documents, witnesses, hooks → Ch. 7): ________ L4 disconfirming probe (attacks whatever L1–L3 will tend to confirm): ________ Budget: 8–12 min · Scoring: per answer, verbatim evidence, F/H mark per Ch. 16.

Where the rules run out

The honesty paragraph. The structured-interview evidence is the strongest in this book and it was built, overwhelmingly, below the altitude where you will use it: executive-specific validation is thin, and the senior adaptations (covert-but-real structure, the dual-purpose management) are practitioner craft consistent with the evidence rather than tested by it. The past-behavior advantage at high complexity rests on a modest number of studies, solid but not vast. Planned probing's validity depends on the plan actually being followed; the method degrades to exactly the degree the trees stay in the drawer. And the deepest limit is the one Chapter 4 spent itself on: this machinery disciplines judgment; it does not replace the judgment that must still read what the machinery yields. The trees produce evidence. Someone still has to be worth handing it to, which is Chapter 25's problem, and the book's.

Notes and sources

Evidence grades: [M] meta-analytic/systematic; [L] peer-reviewed primary; [S] credible practitioner/survey; [T] flagged synthesis.

• Structured-interview validity and its components: Campion, Palmer & Campion, "A Review of Structure in the Selection Interview" (Personnel Psychology, 1997); Levashina, Hartwell, Morgeson & Campion, "The Structured Employment Interview" (Personnel Psychology, 2014) — including the planned-probing position. [M]
• Past-behavior vs. situational at high complexity: Pulakos & Schmitt (1995); Huffcutt and colleagues' comparative work. [L/M]
• Prior-information contamination: interview decision-process research within the structure literature. [L]
• Anchored rating and note-taking effects: components research above; Chapter 16's sources. [M]
• Interviewer failure modes (talk-time, leading, the illusion of the favorite question): Chapter 4's sources (Highhouse; Dana, Dawes & Peterson). [L]
• Senior-candidate reactions and the dual-purpose interview: applicant-reactions literature (McCarthy et al., 2017) plus practitioner adaptation, flagged. [L/S]
• The four-layer formalization, the tree-builder, the time economics as stated, the specimen trees: this book's synthesis from the evidence and practice. [T]