How this book works

This book began with a CFO who was almost impossible not to hire. He was gone fourteen months later. This is a working book: built to be read once and then used for years. Four design decisions run through every chapter, and knowing them now will make the rest go faster.

The two registers. Evidence about a senior candidate comes in two kinds. Capability is what a person can do, and it leaves traces: results, artifacts, verifiable episodes, observable performance. Identity is who the person is such that the doing happens: motivation, values, relation to authority, what a possible future means to them. It leaves different traces, and reluctantly. One word on what the term does not mean, because the cover cannot carry the caveat: identity here is never demographic identity, never personality typing, never diagnosis; it names the candidate's recurring motivations and non-negotiables as they bear on one particular organizational world, and nothing else. The two registers require different evidence, collected by different methods. The book's central claim, argued from the failure data in Chapter 1, is that most executive hiring fails because it runs one process for a two-register problem. Part I states the claim and the evidence standard; Parts II and III build the capability machinery (the technical half, then the behavioral half); Part IV builds the identity method; Part V assembles both into a judgment someone owns. An appendix maps the legitimate and illegitimate uses of AI in the workflow.

Two doctrines to watch for. The book makes two arguments you will not find elsewhere, and they carry much of its practical value. Chapter 4 argues that subject-matter expertise in the interviewer's chair, the former CFO interviewing CFO candidates, is a documented liability past a threshold of functional literacy. The expertise which actually assesses well is a different craft: outcome-corrected assessment experience in a market. Chapter 7 argues that candidate-supplied references are a door, not a destination: the two-level method treats named referees as gateways to the referees they nominate, and moves the evidentiary weight to that second ring. The first ring vouches; the second ring testifies.

The repeated anatomy. The seven behavioral-cluster chapters (9–15) share a fixed structure: what it is and what it predicts; how it fakes; the probe architecture; what shows in the room versus what only references see; red flags with their innocent explanations; the reference question; scorecard anchors. The structure is fixed so the book works as a reference when a live search has you in one cluster at eleven at night. Chapter 12 runs the anatomy fullest and deepest; treat it as the pattern.

The honesty paragraph. Every chapter ends with a section titled Where the rules run out: what the method cannot see, where the evidence is thin, what remains inference. These are not disclaimers, and the Coda explains what they were pointing at all along. Relatedly, every evidence claim in the book carries a grade: [M] meta-analytic or systematic; [L] peer-reviewed primary research; [S] credible practitioner, publisher, or regulatory material, treated as claims; [A] audited claim, a widely cited figure whose provenance was checked rather than taken on reputation, kept only for what survives the audit; [T] this book's own synthesis from practice, flagged as such and awaiting the audit that Chapter 25 builds. A book that scores candidates for evidence discipline should submit to the same scoring; the grades are the mechanism.

The chapters cite one another constantly, because the method is a system: the interview plants hooks the references verify; the psychometrics map where the conversations dig; the scorecard feeds the arithmetic; the arithmetic serves the signed recommendation; the recommendation feeds the log; and the log, at its anniversaries, corrects everything upstream. Read straight through the first time. After that, the index below takes you to the instrument you need.

Start here on your next search

The system above was built over years and does not have to be adopted in a weekend. If the architecture looks heavier than your next search can carry, it is, and the honest sequence is this. Five practices deliver most of the method's early value, cost almost nothing, and require nobody's permission:

One. Before meeting any candidate, write one page on the world: what this role must actually do, tolerate, and answer to, and what must be true in twenty-four months for the hire to have worked. Score everything that follows against that page and refuse to let it drift (Chapter 3).

Two. Score independently, in writing, before anyone speaks in the debrief, and test every claim that reaches the file against the four tests: specific, owned, verifiable, relevant. Adjudicate disagreement; never average it (Chapters 2 and 16).

Three. In every interview, convert "we" to "I" and plant verification hooks: what did the document say, who was in the room, may I ask them. Then run the reference calls against the hooks you planted, not against a generic script (Chapters 5 and 7).

Four. Hold one decisional-evidence conversation per finalist: two or three real career forks, what was chosen, what was paid, and offer the emerging pattern back to the candidate for the counterexample that breaks it (Chapter 20).

Five. Write the recommendation as five elements: this person, for this organization, under these conditions, for these reasons, with these uncertainties, and sign it with a name. Date your predictions. Open the file again at twelve months (Chapters 24 and 25).

That is a functioning miniature of the whole book, and a team that runs it for three searches will know, from its own log, which of the remaining instruments to adopt next. For mature teams the full architecture is the destination: the mandatory core is the demands page, the scorecard with its confidence marks, the two-layer shortlist, and the signed memo; the situational instruments (the enactment exercises, the psychometric integration, the cross-cultural protocol, the second-ring method in its full legal frame) are engaged when the mandate warrants them. Nothing in this book asks you to believe it. It asks you to run it and check.

The instruments

Every chapter closes with a working instrument. In order of appearance:

A note on the research beneath the practice

This book stands on a research corpus published alongside it: on judgment, on what machine fluency is and is not, and on why a candidate can never be reduced to a profile. None of it is required to use anything here. For the reader who wants the why, all of it is at kitalent.com/research. Where the book adopts instruments from the practitioner tradition, the sources are credited in place: the chronological career history, the scorecard spine, the success-profile-before-names discipline.

A note on the cases

Every case in this book comes from real practice, and no case in this book is reportage. Identifying details have been altered throughout; several examples combine elements of more than one search or candidate into a single account; dialogue is reconstructed from notes and memory, not recordings. The disguise protects candidates and clients alike, and it is itself part of the method's ethics: the same confidentiality this book demands of an assessment process governs the book about it. No inference should be drawn about any identifiable person, company, or mandate, and any resemblance beyond the pattern being taught is unintended.