In 1998, Frank Schmidt and John Hunter published a meta-analysis synthesising 85 years of research on the predictors of job performance. The findings were uncomfortable for most organisations’ hiring and succession practices. Unstructured interviews — the format used in the vast majority of executive selection decisions — showed a validity coefficient of .38, meaning they explained roughly 14% of the variance in subsequent job performance. Structured interviews did better at .51. Work sample tests were strongest at .54. The variable with the weakest predictive validity of all: years of experience, at .18. The factor that most organisations weight most heavily in succession decisions — track record and tenure in relevant roles — was the least predictive of performance in the new role.

This is not an obscure finding. The Schmidt and Hunter paper is among the most cited in industrial-organisational psychology. The evidence it synthesised has been available for decades, updated in subsequent meta-analyses, and consistently confirmed. The talent identification and succession practices of most large organisations have not changed in proportion to the strength of the evidence. The gap between what the research recommends and what organisations actually do in leadership selection represents one of the most durable examples of evidence-resistant practice in the management field.

The Performance-in-Current-Role Fallacy

The most pervasive error in succession planning is selecting for demonstrated performance in the current role rather than predicted performance in the future one. The logic seems sound: the person who has excelled at what they have been doing is the most proven candidate for the more senior position. The research finding that undermines it is Laurence Peter’s principle, empirically supported in subsequent research by Alan Benson and colleagues: organisations that promote their best performers into management roles routinely select for the wrong competencies, because the competencies that produce excellent individual performance are not the same as those that produce excellent leadership of other people’s performance.

The gap is larger at senior levels. The executive who was exceptional as a senior director succeeded in a context where their personal cognitive and technical contribution was the primary value they created. The role of CEO, divisional president, or chief of staff requires creating value primarily through the quality of the systems, culture, and people established and developed — a fundamentally different cognitive and interpersonal task. The prior role success was genuine. It was not evidence of fitness for the new role. And the succession process that selected them based primarily on that prior success was using a valid signal to predict the wrong outcome.

Noise in Talent Judgment

Daniel Kahneman, Olivier Sibony, and Cass Sunstein’s research on noise in human judgment documented a finding with direct implications for executive succession: the same candidate assessed by different interviewers receives substantially different evaluations, even when the interviewers are using the same criteria and have been trained to the same standard. Kahneman and colleagues estimated that roughly half of the variance in interview-based assessments reflects noise — the idiosyncratic factors of who happened to interview the candidate, on what day, in what sequence, and in what mood — rather than any signal about the candidate’s actual capabilities.

The noise problem is worsened by the specific social dynamics of executive succession processes. Senior leaders assessing candidates for senior roles are subject to similarity bias — the tendency to favour candidates who resemble themselves in background, communication style, and implicit values — and to the halo effect, in which a candidate’s strong performance on one visible dimension causes assessors to assume strong performance on unobserved dimensions. Both biases are stronger when the assessment is unstructured and when the assessors are operating under time pressure, which is precisely the condition under which most senior succession decisions are made.

What Actually Predicts Senior Leadership Performance

Schmidt and Hunter’s meta-analysis found that the strongest predictors of job performance were cognitive ability tests combined with structured interviews, and work sample tests. For senior executive roles, the practical translation of “work sample” is something most organisations do not build into their succession process: assessments that require the candidate to actually perform the cognitive and interpersonal tasks the senior role requires, rather than describing how they would approach those tasks in a retrospective interview. Strategic analysis of ambiguous information under time pressure. Receiving genuinely critical feedback and demonstrating what happens to their receptivity and reasoning. Facilitating a high-stakes group decision process in real time.

Morgan McCall’s research on the development of high-potential executives found a consistent pattern: the experiences that most developed executive capability — crucible assignments, significant stretch roles, situations requiring genuine recovery from failure — were also the experiences that most revealed genuine fitness for senior roles. The executive who had navigated a turnaround, managed a significant change initiative against substantial resistance, or rebuilt a team after significant disruption had a work sample available for assessment that the succession process could examine. The executive whose career had been a sequence of successful stewardship of already-performing businesses had a narrower experiential base from which the demands of the senior role could be predicted.

The Physiological Dimension of Talent Assessment

Talent assessment errors have a physiological component that the assessment literature has not fully engaged with. The assessors who evaluate senior candidates are, typically, executives who are themselves operating under the allostatic load conditions described elsewhere — chronically elevated cortisol, reduced HRV, the attenuated prefrontal function that sustained stress produces. The specific cognitive capacities that chronic stress impairs — sustained attention, perspective-taking, resistance to availability and confirmation bias — are exactly the capacities that high-quality talent assessment requires. The assessment that would need to resist the halo effect, correct for similarity bias, and evaluate a candidate’s work sample performance across multiple dimensions simultaneously is the kind of effortful, deliberate processing that a depleted prefrontal cortex performs poorly.

The succession process that produces the most reliable outcomes is not simply the process with better tools — though tools matter. It is the process in which the assessors are in the cognitive condition to use good tools well. And that condition is a function of the physiological state the assessors bring to the assessment, which is a function of the same recovery and regulation practices that the SEAM protocol addresses in the assessed executives themselves. The talent read and the talent being read are both physiological states. Both can be improved. Four slots available monthly. Apply here.

Frequently Asked Questions

If unstructured interviews are so unreliable, why do organisations continue to rely on them?

The persistence of unstructured interviews in the face of decades of evidence against their validity reflects a combination of factors that are individually understandable and collectively costly. Interviewers who use unstructured formats consistently rate their own assessment accuracy highly — the subjective experience of conducting a wide-ranging conversation with a candidate and forming a strong impression feels like valid assessment even when the research shows it largely isn’t. The results of an unstructured interview are also more narratively satisfying than those of a structured one: the interviewer can tell a coherent story about why they believe the candidate is or is not right for the role, which feels more useful and more defensible than a validity coefficient. And the structured interview, work sample, and cognitive ability testing approaches that the research supports require more design investment and create more explicit accountability for prediction quality — both of which are barriers to adoption in organisations where succession decisions are treated as leadership prerogatives rather than prediction problems.

How does the Peter Principle actually manifest in senior leadership roles?

Benson and colleagues’ empirical study used sales performance data from a large firm to test whether organisations systematically promoted their best individual performers into management roles regardless of management aptitude. They found that they did — and that the promoted individual’s sales performance was the strongest predictor of their likelihood of being promoted into management, while being negatively correlated with subsequent management performance. The pattern in senior leadership roles is structurally similar but harder to document because the outcomes are more complex, the time horizons are longer, and the counterfactual is harder to establish. What the longitudinal research on leadership tenure and organisational performance consistently finds is that the variance in outcomes attributable to individual leaders is real and substantial, and that the selection processes organisations use do not reliably identify the leaders who will produce the better outcomes — which is the prediction equivalent of the Peter Principle operating at the senior level.

What would a research-aligned executive succession process actually look like?

The structural changes the research supports are three. First, structured assessment against the specific competency profile of the future role — not the current one — with consistent criteria applied consistently across candidates. Second, the inclusion of work sample assessments that require candidates to perform representative tasks of the senior role rather than describing their approach to such tasks retrospectively. Third, explicit debiasing processes that separate the gathering of candidate evidence from its evaluation, and that use multiple independent assessors whose individual judgments are aggregated before discussion rather than subjected to the group influence dynamic that produces convergence around the first strong view expressed. None of these are expensive or exotic. All require the organisation to treat succession as a prediction problem with measurable outcomes rather than as an exercise of leadership judgment — which is the cultural shift that the research supports but that most organisations have not made.

How does SEAM engage with an executive’s talent judgment capacity?

The Clarity Index assesses the executive’s self-reported confidence in talent judgment and their typical process for making significant people decisions — specifically the degree to which their process is structured, the degree to which they seek disconfirming information about candidates they have formed positive initial impressions of, and the degree to which their talent judgments, in retrospect, have proved accurate. Most executives in the assessment identify at least one significant talent misread in their recent past — a hire or promotion that did not deliver what the assessment predicted — and can identify in retrospect the cognitive or procedural error that produced it. The 90-day protocol includes a specific track on evidence-based people decision practices: the process changes that reduce noise and bias, and the physiological state changes that improve the prefrontal function that careful, outside-view talent assessment requires.