Three CEOs in Seven Years. Same Board. Same Assessment Process. Same Outcome.

Three_CEOs_Seven_Years_Assessment_Prediction_Gaconnet_2026.pdf

Download PDF • 135KB

If the Assessment Keeps Missing the Same Thing, the Problem Is Not the CEO. The Problem Is What the Assessment Measures.

Don L. Gaconnet, CSE III

Founder & Principal Investigator, LifePillar Institute for Structural Identity Sciences

ORCID: 0009-0001-6174-8384 · SSRN Author ID: 7657314

June 2026

Same board. Same executive search firm. Same behavioral assessment methodology. Same personality tests and structured interviews. Three CEOs in seven years. Each one passed the assessment. Each one produced strong results in the first year. Each one degraded in the second. Each one was replaced.

The board is not making bad selections. The board is selecting well within the constraints of what the assessment measures. The candidates are strong. Their track records are real. Their capabilities are genuine. The personality profiles accurately describe their behavioral tendencies. The structured interviews capture how they think about leadership challenges. The references confirm their history.

None of this predicts whether the CEO can structurally sustain the performance under the specific load the role will impose over the hold period.

The board member staring at their third replacement in seven years is beginning to ask a different question. Not "how do we find a better CEO." The question is: "Is there a different kind of assessment — not a better interview, but a different instrument entirely — that measures the thing the current process keeps missing?"

What the Current Assessment Measures

Every behavioral assessment instrument in the executive evaluation pipeline — personality inventories, structured interviews, behavioral simulations, 360-degree feedback, reference checks, and the recently introduced AI-enhanced platforms — measures the same substrate: the executive's observable output.

What the executive says about themselves. How the executive responds to questions. What the executive does in a simulated scenario. How the executive's colleagues describe the executive's behavior. What the executive's references recall about past performance.

This substrate is real. It is measurable. It is informative. Behavioral assessment produces accurate information about how the executive presents, communicates, thinks about challenges, relates to teams, and performs under observation.

It is also the wrong substrate for the question the board is asking.

The board's question is not "how does this person present." The board's question is "can this person carry the weight of what we are about to put on them — and sustain it for the three to five years the deal requires."

That is a structural capacity question. Behavioral assessment does not answer structural capacity questions. It answers behavioral presentation questions. The two are different measurements of different substrates producing different information.

A structural engineer does not assess a building's load-bearing capacity by observing the building's appearance. The building can look excellent — well-maintained facade, clean lines, no visible damage — while the structural system beneath the surface carries loads that are approaching or exceeding its design capacity. The appearance predicts nothing about the structural trajectory. The structural assessment reads the system directly.

Why Behavioral Assessment Cannot Predict the Year-Two Failure

The industry's own data documents the prediction failure. 97% of PE firms use formal assessment tools or external interviewers to evaluate portfolio company CEOs (AlixPartners, 2026). 65% still replace the CEO during the hold. Year two is the spike.

Near-total adoption of assessment. Majority replacement despite assessment. The assessment was conducted. The CEO was cleared. The failure occurred.

The structural reason the behavioral assessment cannot predict the year-two failure: the behavioral assessment reads the performance layer, and the performance layer holds through year one. The executive who will fail at year two performs at full capacity through year one — because the executive has the skills, the experience, and the behavioral competencies the assessment measured. The assessment was accurate. About the wrong thing.

The year-two failure is not a performance failure. It is a structural capacity failure. The system sustaining the performance has been carrying a load that exceeds its structural capacity. The performance is maintained by allocating structural resources — cognitive, emotional, relational, somatic — to the performance layer at the expense of the system's own sustainability. The performance holds. The system degrades. By year two, the system can no longer sustain the allocation. The performance breaks.

From the outside, this appears sudden. From the structural perspective, it was visible from the beginning — in the structural state of the system, not in the behavioral output of the system. No behavioral assessment, regardless of sophistication, could have predicted it because the prediction requires reading a substrate the behavioral assessment does not access.

The Population Where Prediction Matters Most Is the Population Where Self-Report Fails Most

The inverse reliability finding (Gaconnet, 2026; SSRN 7657314; DOI: 10.17605/OSF.IO/MVYZT) establishes the structural trap: self-report accuracy degrades as structural severity increases. The executives under the most structural load — the executives where accurate assessment carries the highest stakes — are the executives whose self-report is most structurally wrong.

In a 10,000-case Monte Carlo simulation of the near-capacity executive population, 81.4% misidentify the domain where their structural failure lives. 73.0% minimize the depth. At the deepest structural severities, depth minimization exceeds 94%.

The executive who will fail at year two is the executive whose structural load is highest. That executive produces the most polished, most confident, most convincing assessment performance — because the system has allocated maximum resources to the performance layer. The behavioral assessment reads a high-performing executive because the executive IS performing at a high level. The performance is real. The structural cost of the performance is invisible to the assessment.

The board receives an assessment that says: strong candidate, excellent behavioral profile, high confidence in leadership capability. The board approves the CEO. By year two, the gap between the behavioral assessment and the structural reality becomes visible to everyone — the operating partner, the board, the team, everyone except the CEO, who is the last to see it because the mechanism preventing self-detection IS the mechanism worsening the condition.

What Predicts the Year-Two Outcome

Structural capacity under specified load predicts the year-two outcome. Not personality. Not behavioral tendency. Not track record. Not how the executive handles a simulated strategic inflection point. Whether the executive's system can sustain the weight the deal thesis imposes over the hold period.

This requires measuring a different substrate — the condition of the system producing the performance, not the performance itself.

Independent structural measurement reads this substrate directly. The instrument does not ask the executive to describe their state. It does not observe the executive performing under simulated pressure. It reads the structural condition of the system through independent channels that the executive's performance layer does not control.

The output is not a prediction based on behavioral extrapolation (if the executive behaved this way in the past, they will likely behave this way in the future). The output is a structural assessment of load-bearing capacity (the system is carrying X load in Y domain at Z depth, the structural capacity is Q, and the trajectory under continued load follows this path).

This is the structural difference between behavioral prediction and structural measurement. Behavioral prediction extrapolates from past performance. Structural measurement reads current capacity. The executive who performed brilliantly in the past and is structurally degraded in the present will be predicted to succeed by behavioral extrapolation and identified as at-risk by structural measurement.

The board that has replaced three CEOs in seven years does not need better behavioral prediction. The board needs structural measurement of a different substrate.

What the Board Should Ask the Assessment Provider

When the board evaluates an executive assessment methodology — whether traditional behavioral, AI-enhanced, or any alternative — one question determines whether the methodology can predict the year-two outcome:

Does this instrument read the structural state of the executive's system, or does it read the executive's behavioral output?

If the answer is behavioral output — regardless of how the output is observed (interview, simulation, AI analysis, 360-degree survey) — the instrument reads the presentation layer. It will produce an accurate behavioral profile. It will not predict whether the system producing the behavior can sustain it.

If the answer is structural state — an independent measurement of the system's condition, load, capacity, and trajectory — the instrument reads the substrate where the year-two failure originates. It will tell the board what no behavioral assessment can: whether the executive's system is building, maintaining, or degrading under the current load, and what the trajectory looks like over the hold period.

The board does not need to choose between the two. Behavioral assessment tells the board how the executive is likely to lead. Structural assessment tells the board whether the executive's system can sustain the way they lead under the specific load the deal imposes. The due diligence file needs both — the behavioral profile and the structural finding.

What the due diligence file currently contains is the behavioral profile alone. The structural finding is absent. The board is making a multi-million-dollar commitment on half the information the question requires.

The Pattern Breaks When the Measurement Changes

Three CEOs in seven years is not a selection problem. It is a measurement problem. The board selected well within the constraints of what the assessment measured. The assessment measured behavior. Behavior did not predict structural sustainability. The board replaced the CEO. The next assessment measured behavior. The next CEO's behavior was also strong. The next CEO's structural capacity was also insufficient. The pattern repeated.

The pattern breaks when the board measures the substrate where the failure actually lives. Structural measurement — cognitive due diligence — reads the executive's system directly and produces the finding the behavioral assessment cannot: the structural condition, the load distribution, the capacity margin, and the sustainability trajectory.

The board that adds structural measurement to its assessment process will not eliminate CEO turnover. It will identify — before the commitment is made — which candidates have the structural capacity to carry the deal thesis through the hold period and which candidates present well but carry structural loads that will surface at year two.

That is the difference between selecting on presentation and selecting on capacity. It is the difference between three CEOs in seven years and one CEO who carries the investment to exit.

References

AlixPartners. (2026). 11th Annual PE Leadership Survey. March 2026.

The Conference Board / Egon Zehnder. (2025). CEO Succession 2025. November 2025.

Gaconnet, D. L. (2026). The Recursive Reliability Effect. LifePillar Institute. SSRN 7657314. DOI: 10.17605/OSF.IO/MVYZT.

Heidrick & Struggles. (2026). Route to the Top US 2026. May 2026.

Russell Reynolds Associates. (2026). Global CEO Turnover Index 2025. February 2026.

Don L. Gaconnet, CSE III

Cognitive Systems Engineer III

Founder & Principal Investigator, LifePillar Institute for Structural Identity Sciences

ORCID: 0009-0001-6174-8384 · SSRN: 7657314

Institute: lifepillarinstitute.org/research · Practice: dongaconnet.com/writings

Lake Geneva, Wisconsin · don@lifepillar.org

Copyright © Don L. Gaconnet, June 2026. All rights reserved. The assessment instrument, its operational architecture, scoring methodology, and all associated protocols are proprietary trade secrets of Don L. Gaconnet and the LifePillar Institute for Structural Identity Sciences.