PILLAR 02 · CEO Succession Architecture

The board's single most consequential decision should not rest on a Nine-Box Grid, a closed-loop black book, and four interview transcripts.

CEO succession is the highest-leverage fiduciary act a board performs. A mishandled transition destroys median 18–34% of equity value within twenty-four months, invites Section 220 books-and-records demands, and erodes proxy-advisor support for the entire incumbent board. We architect succession as continuous, quantitative, jurisdiction-aware infrastructure — not as an episodic relationship-driven retainer.

1,847,000
Assessments processed
4,412
Board engagements
73%
Derailment risk reduction
9.4×
Faster time-to-shortlist
01 · The failure mode we replace

SHREK-era succession is a relationship product wearing methodology costume.

Spencer Stuart, Heidrick & Struggles, Russell Reynolds, Egon Zehnder, and Korn Ferry built their succession practices on artefacts that pre-date modern data infrastructure. The output looks credible. The mechanism does not survive a deposition.

Legacy Artefact

Nine-Box Grid

Two ordinal axes (performance, potential) collapsed onto a 3×3 lattice by HR consensus. Zero predictive validity beyond eighteen months. Indefensible under Caremark, Marchand, or APRA CPS 510 scrutiny.

Legacy Artefact

Closed-Loop Black Book

Recycled relationship inventory of ~600 candidates per partner. Geographic skew toward London/NYC/Sydney. Excludes sub-CXO, cross-industry, and emerging-market profiles by design.

Legacy Artefact

Annual Self-Evaluation

Anonymous Likert survey distributed once per cycle. Captures politics, not capability. Cannot detect skills drift between AGMs or attribute risk-oversight gaps to named directors.

Legacy Artefact

Retainer-Funded Slate of Four

Economics force convergence on a small, conflict-free, network-adjacent slate. Counter-factual pool is never constructed. The board sees what the retainer permits.

Legacy Artefact

Opinion-Only Board Report

Narrative prose with no source citations, no model class disclosure, no z-scores, no audit hash. Cannot be re-run, stress-tested, or defended in a proxy fight.

Legacy Artefact

Episodic Engagement Cadence

Succession reviewed at 12–24 month intervals. Bench liquidity is unpriced between cycles. The board is structurally surprised by every unplanned CEO loss.

02 · Succession Architecture Framework

Eight phases. Continuous. Auditable. Jurisdiction-aware.

Each phase has named data inputs, a disclosed model class, a board-readable output, and a forensic audit trail. The framework runs on subscription, not on retainer, and is re-priced every fourteen days against changing telemetry.

01

Mandate Definition

Inputs
Strategy memo, 5-year capital plan, regulator correspondence, activist letters.
Model class
Structured NLP extraction; chair-validated mandate spec.
Output
Mandate Charter: 14–22 weighted criteria, signed by chair and NomCo.
02

Regulatory Skills Matrix Calibration

Inputs
ASX CGC 4th ed., UK CGC, SEC cyber rules, EU CSDDD, MAS, OCC, APRA CPS 510/511.
Model class
Jurisdictional skills drift detector against current board composition.
Output
Skills-gap heatmap; named deficiencies the next CEO must close.
03

Incumbent Stress Mapping

Inputs
Earnings-call linguistic telemetry, regulator filings, internal cadence data.
Model class
Cognitive load capacity model; derailment vector probability surface.
Output
Incumbent stress index, tenure-adjusted; planned-departure window estimate.
04

Internal Bench Telemetry

Inputs
Direct-report observation cadence, scope changes, P&L delegation depth, 9-quarter performance series.
Model class
Behavioural telemetry stack; counterfactual scope simulation.
Output
Ranked internal bench with z-scores across 7 cognitive dimensions.
05

External Liquidity Scan

Inputs
Continuous global scrape across 36 jurisdictions; 2.4M+ tracked profiles; refresh < 14 days.
Model class
Algorithmic shortlist construction; bias vector sweep.
Output
External slate of 18–32 named profiles with mandate-fit scoring.
06

Derailment Vector Modelling

Inputs
Public conduct record, prior board exits, regulator interactions, litigation indices, social-graph anomaly signals.
Model class
Multivariate derailment risk vector; ensemble of 11 classifiers.
Output
Per-candidate derailment probability with confidence interval and feature attribution.
07

Scenario Simulation (3 / 12 / 36-month CEO loss)

Inputs
Bench liquidity state, regulator calendar, deal pipeline, activist exposure.
Model class
Monte-Carlo succession path engine; 25,000 trajectories per scenario.
Output
Tri-window contingency plan with interim CEO designation and equity-value variance band.
08

Continuous Reassessment

Inputs
Live telemetry refresh; chair queries; mandate drift events.
Model class
Subscription-grade reassessment cadence; chair co-pilot interface.
Output
Fortnightly succession dashboard; immutable audit hash for every revision.
03 · Capability surface

Nine capabilities a chair cannot reproduce with a relationship firm.

Behavioural Telemetry

Continuous capture of decision velocity, ambiguity tolerance, and stakeholder cohesion across the executive layer — not a one-off interview impression.

Derailment Risk Vector

Ensemble model assigning calibrated probability of public-record derailment over a 36-month horizon, with feature attribution per candidate.

Cognitive Load Capacity

Empirical ceiling estimate for concurrent strategic, regulatory, and operational decision streams under volatility.

Jurisdictional Skills Drift

Real-time delta between the board's skills matrix and active regulatory regimes across 36 jurisdictions.

Fiduciary Defensibility

Every recommendation hashed and timestamped; reproducible model run; deposition-ready feature ledger.

Algorithmic Shortlist Construction

Slate built by mandate-fit scoring against a 2.4M+ profile graph, not by partner network proximity.

Counter-Factual Pool

Mandatory shadow slate of qualified candidates the board would have missed; quantifies inclusion bias.

Continuous Reassessment Cadence

Fortnightly refresh of bench and external slate; succession is never ‘current’ for more than fourteen days.

Chair Co-Pilot Interface

Direct chair-grade query surface: ask the model who replaces the CEO on a Tuesday morning and receive an evidenced answer in seconds.

04 · Evidence

Derailment risk against strategic yield, plotted across the most recent 64 CEO-shortlisted candidates.

73%
Derailment risk reduction vs. SHREK-sourced slates across 4,412 engagements.
9.4×
Faster time-to-defensible-shortlist: 72 hours against 8–12 weeks.
180 bps
Median 24-month TSR uplift in placements scoring top-quartile on derailment vector.
41%
Counter-factual pool candidates rejected by SHREK slates that scored top-decile on mandate fit.
0.02ms
Engine latency per query against the 2.4M+ profile graph.
05 · Cost curve of a derailed CEO

Cumulative equity-value impairment, indexed, by months since a flawed CEO appointment.

The cost is non-linear. The first six months are absorbed by patience; the second six by analyst downgrades; months 18–30 by activist letters and regulator engagement; and the long tail by proxy-vote consequences for the entire incumbent board.

06 · Why the Nine-Box Grid is a fiduciary liability

Six structural failure modes that will not survive a Caremark inquiry.

Ordinal compression

Continuous performance distributions collapsed into nine cells. Information loss exceeds 80% against a calibrated z-score model.

Rater drift

Cell assignments performed by HR + line managers with no inter-rater reliability protocol. Test-retest agreement falls below κ=0.35 in audited samples.

No counterfactual

By construction, the grid contains only the people HR already named. Dark-horse, cross-industry, and sub-CXO candidates are mathematically excluded.

Static cadence

Cells are updated annually. Cognitive load, regulatory exposure, and decision velocity move on weekly timescales.

No model class disclosure

The grid is unfalsifiable. There is no published feature set, weighting scheme, or validation cohort to cross-examine.

No audit trail

Cell movements are not hashed, timestamped, or linked to evidence. A successor decision cannot be reconstructed in litigation.

07 · Dark Horse Discovery

The CEO the board would not have named, evidenced.

Across 4,412 engagements, 38% of placed CEOs were absent from the original internal slate and 41% were absent from the SHREK-sourced external slate. Dark-horse discovery is a deliberate methodology, not a happy accident.

Discovery vector

Cross-industry sweep

Mandate-fit scoring is industry-agnostic; a regulated-utility CEO mandate routinely surfaces qualified candidates from defence, insurance, and infrastructure investment.

Discovery vector

Cross-jurisdiction sweep

36-jurisdiction coverage with regulator-regime tagging; a Singapore-listed mandate can be filled defensibly from a Toronto-listed comparable.

Discovery vector

Sub-CXO layer scan

The richest seam: BU presidents, country heads, and divisional CFOs running >US$1B P&Ls with measured decision velocity but no plc visibility.

Discovery vector

Activist-aligned candidates

Profiles with documented turnaround telemetry the incumbent network systematically avoids surfacing.

Discovery vector

Regulator-fluent operators

Ex-regulators and ex-prudential supervisors with measured cognitive load capacity; under-priced by relationship firms.

Discovery vector

Adjacency operators

Founders of companies the target firm has acquired or attempted to acquire; mandate-fit often dominates on strategic yield.

08 · What you lose when CEO transition goes wrong

The hard ledger. Three categories of consequence the board carries personally.

Equity value
  • −18% to −34% median market-cap impairment within 24 months of a flawed appointment.
  • 180–420 bps cost of capital deterioration as credit analysts re-rate.
  • 2–5x increase in implied option volatility through the transition window.
  • Tier-1 institutional holders trim positions an average of 11% within two quarters.
Governance standing
  • ISS / Glass Lewis withhold recommendations against named NomCo members.
  • Activist letters with median 47-day arrival post-departure.
  • Section 220 books-and-records demands; APRA / OCC informal supervisory engagement.
  • Spillover director-resignations within four AGM cycles in 31% of cases.
Strategic optionality
  • M&A pipeline freezes through the transition window; targets re-price counterparty risk.
  • Top-200 talent attrition rises 14–22% as direct reports recalibrate.
  • Regulator licence renewals and capital-relief applications stall.
  • Strategy execution clocks reset by an average of seven quarters.
09 · The cost of defaulting to incumbents

What the board loses by staying with a SHREK firm.

DimensionExecutive SearchSHREK Incumbent
Time-to-defensible-shortlist72 hours8–12 weeks
Counter-factual poolMandatory; ratio disclosedNone
Model class disclosurePublished per phaseNot applicable
Reassessment cadenceFortnightlyAnnual / ad-hoc
Audit trailHashed, timestamped, deposition-readyOpinion notes
Jurisdictional coverage36 regimes, liveLondon / NYC / Sydney anchor
PricingSubscription + success-linked33% fixed retainer
Reusability of assessment IPAPI; remains with the boardSiloed; held by retainer firm
10 · Succession implication

Architecture, not events.

A board that runs CEO succession as architecture is never surprised. The bench is priced every fourteen days. The external slate is current every fourteen days. The chair can ask, on any Tuesday morning, who replaces the CEO at 09:00 and receive an evidenced answer with feature attribution, confidence interval, and audit hash before lunch.

A board that runs CEO succession as events is structurally surprised. The bench is unpriced between cycles. The external slate is constructed in eight weeks under crisis tempo. The chair receives a four-name memo from a retainer firm whose economic incentive is to close, not to be right. That asymmetry is the difference between defensible governance and the next class-action.

Initiate

Run CEO succession as infrastructure. Begin with a confidential architecture briefing.

Median first response: 4 hours. Median proposal turnaround: 48 hours. Confidentiality posture: jurisdiction-aware NDA by default.