A Working Taxonomy of Executive Soft Skills

Pull the scorecard from your last senior search and find the line that says something like Communication: 4/5. Now try to answer, from the file alone, a simple question: four out of five at what? At presenting to a board? At listening? At delivering a termination with dignity? At making a complex strategy legible to a sales force? Those are different capacities, held by different people, and the "4" is the average of an apple, a violin, and a Tuesday.

I once watched two experienced interviewers debrief the same candidate with full confidence and opposite scores on "leadership." One had seen a commanding presenter who owned the room; the other had seen a man who, on the evidence of his own stories, had never developed a successor and lost every peer negotiation he described. Both were right. The word was wrong. Or rather, the word was doing what bucket-words do: letting each rater fill it with a private meaning and then disagree, sincerely, about nothing.

This is the practical problem with "soft skills," and it is why this chapter exists. You cannot interview for a bucket. You can only interview for something defined tightly enough that two trained raters, watching the same evidence, would score it the same way. This chapter turns the bucket into eight things you can actually assess, and it does so under evidence, because our first attempt at the list didn't survive contact with the research, and the corrections are instructive.

The rules a usable taxonomy must obey

Four design rules, each earned by a failure mode you have seen in real scorecards.

Each cluster must be behaviorally definable — statable as observable actions in episodes, not as auras. If you cannot write what a 2 and a 4 look like in evidence terms, you have a compliment, not a construct.

Clusters must be distinct enough to rate separately. Here is the finding that reshaped our list: when raters are asked to score constructs that overlap heavily in real behavior, communication and influence and conflict handling, the ratings collapse into each other. The technical term is construct contamination; the practical result is that three scores carry one impression wearing three hats, and the scorecard's apparent thoroughness is an illusion. Fewer, cleaner clusters beat many muddy ones.

Clusters must be level-appropriate. The leadership-skills research has a finding with a wonderful name, the strataplex: skill requirements don't just grow with organizational level, they change shape. "Managing upward" for a plant manager means handling a boss; for a CFO it means governance, meaning boards and owners and auditors and dual accountability. An executive taxonomy anchored in middle-management behaviors measures the job the candidate left, not the one you are filling.

Each cluster must name its method. Some of these capacities yield to behavioral interviewing. At least one, the evidence insists, does not, and a taxonomy that pretends one method fits all is a question bank waiting to fail.

The eight clusters

Here is the list this book runs on, validated and restructured against the major competency frameworks, Korn Ferry's Leadership Architect and SHL's Great Eight and Hogan's domains and the academic taxonomies from Katz through Yukl, and against the criterion evidence per cluster.

1. Strategic and business acumen. Conceptual and financial grasp of the enterprise: reading the environment, framing the long game, connecting decisions to value. It heads the list and is not in this Part of the book, deliberately. The evidence (Chapter 6 made the case) says this cluster is assessed through track-record forensics and simulation, not behavioral interviewing; it lives in Part II with the methods that can actually see it. It appears here so the taxonomy is complete and its absence from your interview guides is a design decision, not an oversight.

2. Execution and accountability. Driving results, building operating cadence, holding senior people to commitments: the task-oriented core with the strongest operational validity in the leadership-behavior literature, and, per Chapter 1's Kaplan findings, the factor that actually predicts outcomes while charm gets hired. Interviewable, verifiable, reference-checkable: the workhorse cluster.

3. Building and leading senior teams. Not managing people in general, but building executive teams: choosing, developing, and occasionally exiting senior leaders; creating team learning; multiplying rather than hoarding. Grounded in the transformational-leadership evidence; sits behind more derailments than any scorecard admits, because inheriting a team hides the skill for a year.

4. Navigating stakeholders and conflict. The big merger, and the taxonomy's main surgical act: what began in our draft as three clusters, communication and lateral influence and conflict-and-negotiation, is one cluster here, because in executive behavior they are one performance. An executive influencing a peer coalition is communicating is managing conflict; asking raters to score them separately manufactured the contamination problem the design rules forbid. The cluster covers executive storytelling, persuasion across matrixed boundaries without authority, and entering conflict productively rather than avoiding or inflaming it.

5. Board governance and upward management. The strataplex-adjusted cluster: dual accountability, downward to the enterprise and upward to boards and owners; institutional reporting; candor under hierarchy; the choreography of dissent. Chapter 12 treats it in full as this Part's specimen, both because the failure data of Chapter 1 point straight at it and because it sits on the seam between the registers: the skills of board work are capability; the disposition under its pressures is identity.

6. Change and organizational transformation. Envisioning change, building readiness, metabolizing resistance, judging pace. Supported by time-lagged studies rather than folklore; distinct from execution (cluster 2) the way rebuilding a ship differs from sailing it hard.

7. Decision-making and judgment. Framing decisions under uncertainty, calibrating speed against quality, updating on contrary evidence without either rigidity or capitulation. It borders cognitive capability, and the taxonomy keeps it here because at executive level the differentiating behavior is decision conduct, not raw processing. And it carries a method flag the evidence would not let us soften: this cluster is assessed by simulation and structured judgment tasks, not by behavioral interview alone. Asking someone to describe their decision-making retrieves their theory of themselves; Chapter 14 shows how to watch the real thing instead.

8. Self-management and global distance. The second merger: internal regulation, meaning composure and feedback receipt and ego management and resilience, joined with the metacognitive and motivational capacities of leading across cultures and distance (the cultural-intelligence evidence base). They merged because the research locates them in the same intrapersonal domain: the executive who regulates herself well is running the same machinery when she adapts across a nine-time-zone leadership team. The construct lives here; the method for assessing across cultures is large enough to claim its own chapter (17).

What happened to emotional intelligence

Readers will notice a famous phrase missing from the list. That is deliberate, and the reasoning is worth one paragraph because clients ask.

The evidence for "EI" as a named, standalone, assessable cluster is a mess the field itself acknowledges: competing models that barely correlate, measures that range from ability tests to relabeled personality inventories, and a meta-analytic literature whose best reading, the cascading model, says the predictive substance of EI decomposes into things we already assess: emotion perception feeding regulation, showing up behaviorally in clusters 4 and 8. So the taxonomy embeds the substance and retires the label. Nothing real about a candidate escapes; only a marketing category does. If a client's framework requires the term, map it: perception and use of emotion in influence goes to cluster 4; self-regulation under pressure goes to cluster 8. The evidence transfers; the halo does not.

How Part III works from here

Chapters 9 through 15 take clusters 2 through 8, one each, in a fixed anatomy of the same seven panels every time, so the book works as a reference when a live search has you in one cluster at 11 p.m.: what it is and what it predicts (with the evidence grade stated); how it fakes — the rehearsed version you will actually hear, because every senior candidate has one; the probe architecture — Chapter 5's four layers applied to this terrain; what shows in the room versus what only references can see; red flags with their innocent explanations — the anti-halo discipline cuts both ways; the reference question that triangulates it; and scorecard anchors — what a 2, a 3, and a 4 look like written in evidence. Chapter 12 runs the full anatomy at its deepest; treat it as the pattern.

Two cross-cutting disciplines govern all seven chapters. Every anchor is written function-first, "surfaces difficult information upward effectively" rather than "openly challenges the boss in the meeting", because Chapter 17 will show how culturally thick anchors quietly score conformity to one region's style. And every cluster keeps the seam with Part IV visible: capability chapters assess what a person can do; where a cluster's terrain touches who the person is, cluster 5 above all, the chapter says so and hands the thread forward rather than pretending one method covers both.

Where the rules run out

The honesty paragraph. A taxonomy is a map, and every map is a set of decisions: other defensible eight-cluster lists exist, and the major commercial frameworks slice the same behavioral territory into 8, 20, or 38 pieces for reasons that are partly scientific and partly product design. The criterion evidence is genuinely uneven across our eight, strongest for execution-oriented and team-leadership behavior, thinner and more inferential for change leadership and the global-distance half of cluster 8, and much framework validation is publisher-produced, which this book treats as informative rather than independent. Finally, clusters are lenses, not parts of a person: a real candidate's behavior in a real episode will cross three clusters in a sentence, and the scorecard exists to discipline attention, not to dismember human beings. Hold the map lightly; hold the evidence standard tight.

Instrument: the cross-walk table

For translating between this book's clusters and the frameworks your clients already use, so a search never stalls on vocabulary.

Note the near-empty cell in row 5: the major frameworks genuinely underweight board governance and upward management, middle-management DNA showing through enterprise products. It is not a coincidence that the same row is where Chapter 1's failure data concentrate, and it is the reason the next specimen chapter is that one.

Notes and sources

Evidence grades: [M] meta-analytic/systematic; [L] peer-reviewed primary; [S] publisher/practitioner, treated as claims; [T] flagged inference.

• Framework structures: Korn Ferry Leadership Architect technical materials; Bartram, "The Great Eight Competencies" (Journal of Applied Psychology, 2005); Hogan domain mappings; Katz (1955); Mintzberg (1973); Yukl's hierarchical taxonomy (2012). [L/S]
• Strataplex: Mumford, Campion & Morgeson, "The Leadership Skills Strataplex" (The Leadership Quarterly, 2007). [L]
• Construct contamination and rater confusion under overlapping dimensions: assessment-center dimensions-versus-exercises literature (Ch. 6 sources); competency-modeling critiques. [L/T]
• Criterion evidence per cluster: leadership-behavior meta-analyses (Judge and colleagues on transformational/transactional; DeRue et al.); change-leadership time-lagged studies; influence-tactics research. [M/L]
• EI decomposition: Joseph & Newman, "Emotional Intelligence: An Integrative Meta-Analysis and Cascading Model" (Journal of Applied Psychology, 2010); measurement-model critiques. [M]
• Cultural intelligence: Earley & Ang (2003); Schlaegel, Richter & Taras meta-analysis (Journal of World Business, 2021). [M]
• Decision-making's simulation requirement: DR-5 synthesis with the simulation evidence of Chapter 6's sources (Lievens & Patterson, 2011). [M/T]