Quick Takeaways

A talented team can still struggle. The key is to identify what prevents people’s knowledge from becoming good work.

This article describes four common problems:

• Missing expertise: Nobody available has the knowledge needed for the task.

• Hard-to-reach expertise: Someone knows the answer, but the team cannot find or reach them when needed.

• Contributions that do not fit together: People do their individual jobs well, but their work does not combine into a useful result.

• Weak review and correction: The team struggles to question advice, spot mistakes, or act on valid concerns.

Different problems need different solutions. Missing knowledge may require training or specialist help. Failed handoffs may need clearer responsibilities and better coordination. Ignored warnings may require stronger review practices and a safer environment for speaking up.

Better teamwork can help in two ways: it can help specialists combine their strengths, or allow qualified colleagues and reliable routines to cover some individual weaknesses. It cannot reliably replace essential knowledge that nobody has.

Leadership matters because it helps identify and address the actual obstacle. Leaders also need sufficient authority and resources to make those changes.

The practical lesson: Before hiring another expert or adding another meeting, ask what is blocking the team’s performance. Then check whether the proposed solution changes that problem.

These are research-informed explanations and testable predictions, rather than a proven formula that works for every team.

Abstract

Why does adding expertise improve some teams while leaving others little better off? This conceptual article develops a constraint-location account of how task-relevant expertise becomes a collective resource. It distinguishes four mechanisms that can limit its contribution: availability, accessibility, integration, and evaluation. Leadership contributes by providing functions matched to the constraining mechanism, within organizational limits on authority and resources. The account separates complementarity, in which coordination increases the marginal value of expertise, from compensation, in which routines or assistance reduce dependence on a particular member's expertise. These patterns depend on knowledge configuration, task interdependence, substitutability, and essential competence requirements rather than following directly from a constraint label. A critical synthesis connects human capital emergence and functional leadership theory with meta-analyses, controlled interventions, and research on expert mobility and talent concentration. Five propositions specify intervention matching, conditional expertise returns, incentive alignment, portability, and adaptation. The article explains how to distinguish the constraints, maps them to competing interaction predictions, and proposes experiments and longitudinal studies that can challenge the account. Its contribution is a diagnostic extension of established theory: identifying which intervention should improve a team's use of expertise, through which mechanism, and under which task conditions.

Keywords: team expertise; human capital emergence; functional leadership; coordination; task interdependence; team performance

1 Introduction

Organizations frequently confront a practical puzzle: a team appears to possess the required expertise but repeatedly fails to deliver. Hiring another accomplished specialist may improve its performance. It may also leave the underlying problem untouched if existing knowledge is inaccessible, contributions do not fit together, or members cannot challenge consequential errors. The decision requires an account of where performance is constrained. It cannot be resolved by assuming that exceptional individuals always travel with their past achievements, or that effective teamwork can replace missing technical competence.

The theoretical foundations for this problem are well established. Knowledge integration is central to knowledge-based accounts of the firm (Grant, 1996). Human capital emergence theory explains how individual knowledge, skills, abilities, and other characteristics become a collective resource through task and social enabling conditions (Ployhart & Moliterno, 2011). Team research likewise distinguishes member inputs from behavioral processes, emergent states, and outcomes that feed into subsequent performance episodes (Ilgen et al., 2005). The proposition that collective performance depends on more than individual capability is therefore a starting point, rather than a sufficient contribution.

Existing reviews also directly connect leadership with the emergence of human capital resources. Eckardt et al. (2021) examined 132 empirical articles linking leadership to task and social enabling conditions. Ray et al. (2023) reviewed 194 articles in the broader human capital resource literature. These contributions make it untenable to portray strategic human resource scholarship as uniformly committed to accumulating stars, or to claim novelty merely by bringing leadership and talent into the same model. The remaining challenge addressed here is more specific: how should the location of a team's performance constraint change the expected relationship between expertise and coordination?

A broad claim that coordination raises the return to expertise conceals two different possibilities. Coordination may enable specialists to combine contributions that are individually insufficient, making expertise and coordination complementary. Alternatively, routines, backup behavior, or shared tools may reduce dependence on the least capable member, allowing coordination to compensate for a particular capability deficit. Both patterns are compatible with beneficial teamwork. They imply different interaction effects, different staffing decisions, and different interventions. A theory that predicts only a positive interaction cannot adequately distinguish them.

This article explains how constraint location changes the expected return to expertise and coordination. The unit of analysis is the team performing an identifiable task episode; organizational outcomes are considered separately. Sections 2–4 establish the scope, constructs, and evidence. Section 5 develops five propositions and maps the constraints to complementarity and compensation; Section 6 describes discriminating tests. Section 7 considers managerial implications and limits, and Section 8 concludes.

2 Scope and approach to the literature

This article is a conceptual synthesis and theory elaboration. It reports no new participant data, systematic review, or meta-analysis. Literature was located through targeted web searches of scholarly titles and topic combinations, inspection of publisher and institutional records, and reference tracing from major reviews. Searches addressed human capital emergence, functional and shared leadership, team composition, interdependence, information sharing, team cognition, psychological safety, teamwork interventions, and star mobility. Targeted updates sought relevant studies published through October 21, 2026. No claim is made to exhaustive coverage of subscription databases, unpublished studies, or non-English research.

Source selection followed theoretical relevance and evidential leverage. Conceptual contributions establish the relationships already recognized in the field. Meta-analyses inform the consistency and heterogeneity of associations. Controlled interventions provide more direct evidence that changing team practices can affect outcomes. Field mobility and personnel studies inform contextual dependence while retaining the limitations of observational identification. Where findings conflict, the disagreement is retained rather than resolved by selecting the result most favorable to the argument. Publisher records and institutional repositories were used to check central claims and bibliographic details; numerical claims are limited to information accessible in those sources.

The synthesis develops an explanatory framework rather than estimating the literature's average effect. Overlapping meta-analyses are not independent replications, and effect sizes from selection, training, and goal-setting studies concern different estimands. They cannot rank the importance of expertise and coordination. The propositions are theoretical deductions informed by this evidence; the review does not provide a formal risk-of-bias assessment or pooled comparison.

3 Constructs and the location of performance constraints

3.1 Expertise and its composition

Expertise refers here to task-relevant knowledge and skills assessed independently of the focal team's outcome. It is narrower than general talent and distinct from reputation. A star's visibility, social capital, and sustained performance may affect access to resources as well as reflect capability (Call et al., 2015). Consequently, star status cannot be treated as a clean measure of technical expertise. Prior success also incorporates the support and opportunities available in the previous setting.

Within the broader human capital emergence framework, expertise is the task-relevant knowledge and skill input examined here; an emergent team human capital resource is the collective capacity to use members' capabilities for the team's purposes (Ployhart & Moliterno, 2011). The present account therefore addresses one component of human capital emergence rather than treating expertise as synonymous with its broader set of individual attributes. Expertise configuration describes what members know; accessibility, integration, and evaluation describe conditions and processes through which that knowledge can contribute collectively. Performance remains a separate outcome, so the presence of a collective resource is not inferred solely from success.

Team expertise is a configuration rather than a single universally appropriate average. A mean may describe broadly substitutable contributions. A minimum may matter when every member must meet a safety or technical requirement. A maximum may be relevant when one demonstrably correct solution can resolve the task. Complementary specialties require measures of coverage and fit. Bell's (2007) meta-analysis shows that relationships between team composition and performance vary with both the operationalization of composition and study setting. Researchers should select the aggregation rule from the task structure before inspecting which rule predicts performance best.

Individual selection research remains relevant but answers a different question. Sackett et al. (2022) revised estimates of selection validity after examining overcorrection for range restriction. This revision concerns the accuracy of particular predictor–criterion relationships; it does not show that expertise has become less important or that a team should substitute process investment for technical selection. A general mental ability test, a structured interview, accumulated specialist knowledge, and a count of stars are not interchangeable measures of the input considered here.

3.2 Four analytically distinct constraints

An availability constraint exists when the team lacks knowledge or skill necessary for the task. Additional discussion cannot reliably supply expertise that no member or accessible external source possesses. Training, recruitment, consultation, or task redesign may be needed. This is a substantive competence requirement, not a statistical claim that output becomes zero below an arbitrary score.

An accessibility constraint exists when relevant expertise is present but cannot be located or mobilized when required. Members may misunderstand who knows what, withhold information, or be unable to obtain assistance across organizational boundaries. Transactive memory research is pertinent because it concerns specialization, credibility, and coordination around distributed knowledge (Lewis, 2003; Ren & Argote, 2011). Accessibility should therefore be assessed through retrieval accuracy and actual requests for expertise, rather than inferred from a global satisfaction rating.

An integration constraint exists when accessible contributions fail to combine into a workable product or decision. Members may use incompatible assumptions, misunderstand dependencies, or deliver sound components in the wrong sequence. Shared task representations and behavioral coordination can address this problem. Meta-analytic evidence links team cognition with behavioral processes and performance (DeChurch & Mesmer-Magnus, 2010), and information sharing with performance and knowledge integration (Mesmer-Magnus & DeChurch, 2009). Neither finding implies that more communication is always useful. Information must be relevant, understood, and incorporated.

An evaluation constraint exists when the team cannot scrutinize a contribution or correct an emerging error. Relevant information may circulate without changing a decision because status outweighs evidential quality or disagreement carries interpersonal costs. Psychological safety concerns the perceived interpersonal risk of speaking, seeking help, and acknowledging mistakes (Edmondson, 1999). It differs from technical competence, accuracy, agreement, and freedom from performance standards. Reviews support its relevance to learning and other work outcomes while documenting a literature broader than any single causal pathway (Edmondson & Bransby, 2023; Frazier et al., 2017).

Evaluation also requires more than interpersonal permission to disagree. Voice research concerns whether members communicate concerns or suggestions; error-management research concerns practices for detecting, discussing, analyzing, and correcting errors (Morrison, 2023; van Dyck et al., 2005). These literatures support separating the expression of a concern from its technical assessment and the implementation of a correction. A team can be psychologically safe yet accept an incorrect claim, or hear a valid objection without revising its decision. Relevant indicators therefore include detection of independently verifiable errors, the evidential quality of review, and whether justified corrections are implemented. The organizational-level associations reported by van Dyck et al. do not establish a team-level causal effect or validate this category. No single established construct spans speaking up, competent assessment, and implemented correction. Evaluation is therefore the most provisional of the four categories and requires evidence that it is distinguishable from integration and that its measures predict correction beyond speaking-up frequency.

These categories are analytical distinctions, not a proposed validated scale or a claim that failure always follows a fixed sequence. Constraints may coexist and reinforce one another. A team unable to evaluate a specialist's recommendation may incorrectly diagnose an availability problem and hire another specialist. Conversely, apparently poor communication may reflect uncertainty because essential knowledge is absent. The framework requires independent diagnosis of the constraint; classifying every unsuccessful team as deficient in integration would make the argument circular.

3.2.1 Why these four mechanisms

The four mechanisms are derived by asking what must occur for distributed knowledge to affect a collective result. If no feasible source possesses the required knowledge, the team faces an availability constraint. Possession alone is insufficient: failure to retrieve or mobilize existing expertise generates an accessibility constraint. Once obtained, distinct contributions must fit the task's dependencies; failure of that combination generates an integration constraint. Compatible contributions can still be wrong, so the team also needs procedures for scrutinizing claims and correcting demonstrated errors; failure of that scrutiny generates an evaluation constraint. These requirements distinguish possession, mobilization, combination, and scrutiny rather than merely assigning four names to poor performance. Collapsing retrieval into combination would obscure whether a contribution was obtained at all; collapsing scrutiny into combination would obscure whether compatible contributions were technically sound. Causes that disrupt the same operation, such as a faulty expertise directory and a blocked request, remain within one category unless they imply a distinct mechanism.

The categories form a diagnostic heuristic, not an exhaustive taxonomy or a proposed reflective latent-variable model. Motivation, equipment, incentives, and authority can affect several mechanisms or constrain performance independently. The requirements are logically distinguishable but do not prescribe a fixed temporal sequence: teams evaluate contributions while integrating them, and correction can require renewed access to expertise. The unit of diagnosis is a defined task episode and an observable failure, rather than a global label applied to the team.

Ambiguous cases require evidence about both the failed operation and the relevant counterfactual. An incompatible interface is an integration failure when individually sound components were not coordinated; it is also an evaluation failure if a review procedure failed to detect a demonstrable defect. Researchers should code both when both mechanisms are observed, retain uncertainty when records cannot distinguish them, and test whether the proposed remedy changes its intended intermediate outcome. An upstream shortfall is a sensible first diagnostic check when downstream work depends on it, but priority follows evidence about which constraint remains binding under a feasible intervention; a downstream failure can independently block performance. Resolving access should not restore performance if indispensable expertise is still absent, while resolving one of several coexisting constraints may improve a process without improving final output. This possibility qualifies Proposition 1: its contrast assumes the targeted constraint is consequential and other binding constraints are held comparable.

3.3 Leadership as functional provision

Leadership is defined by the provision of task and relational functions that a team needs, rather than by formal position alone. Functional leadership theory permits these functions to originate inside or outside the team, and from designated or emergent leaders (Morgeson et al., 2010). The framework therefore does not presume that a supervisor personally coordinates every exchange. Shared leadership may also support performance, although its measurement and form matter (D'Innocenzo et al., 2016).

For the present argument, relevant functions include securing missing resources, establishing priorities and interfaces, connecting members with expertise, and enabling informed challenge. These functions are distinguishable from their consequences. A leader's invitation to question a decision is a behavior; psychological safety is an emergent perception; the subsequent correction of an error is a process outcome. Separating these variables prevents leadership from becoming an all-purpose label for everything that successful teams do.

The task environment also constrains leadership. A manager cannot necessarily change staffing, reward systems, equipment, or deadlines. This limits both causal attribution and practical responsibility. Organizational investment can enable team functions, and organizational incentives can undermine them. A credible account of leadership must include that possibility rather than treating every failure as a local managerial choice.

4 What the evidence establishes

4.1 Team processes and the effects of intervention

Team process research distinguishes transition activities such as planning from action activities such as monitoring, and from interpersonal activities that span both phases (Marks et al., 2001). LePine et al. (2008) found positive meta-analytic relationships between teamwork processes and team effectiveness criteria. These relationships support attention to how members work together, but observational associations cannot establish whether stronger processes caused better performance, resulted from it, or shared an antecedent with it.

Controlled intervention evidence strengthens the case that team functioning is modifiable. McEwan et al. (2017) synthesized 72 controlled interventions reported in 51 articles. For team performance, the full set of 50 interventions yielded d = 0.919; after five outlying interventions were removed, the estimate was d = 0.582, with a 95% confidence interval of 0.47 to 0.69. The difference matters: a single headline effect obscures sensitivity to extreme estimates. The remaining heterogeneity also cautions against treating training as a uniform treatment. These results support the effectiveness of some teamwork interventions, not a universal causal effect of leadership or a demonstrated interaction with expertise.

Lines et al. (2021) synthesized 24 randomized trials or experiments involving 4,339 participants. Team reflexivity interventions had an overall effect of g = .549 on performance outcomes. Effects were larger when outcomes were self-reported than when assessed objectively or by researchers. This evidence supports structured opportunities to examine objectives and working methods, while making independent outcome measurement particularly important. Neither review establishes that an intervention works by improving accessibility, integration, or evaluation specifically. Distinguishing those mechanisms requires additional measurement or experimental contrasts.

4.2 Goals and interdependence

Group goal setting has direct experimental support. Kleingeld et al. (2011) reported that specific difficult goals outperformed nonspecific goals, with d = 0.80 across 23 effect sizes. Their analyses also differentiated individual goals directed toward personal maximization from goals directed toward contribution to group performance. The relatively small evidence bases for some individual-goal contrasts make the exact magnitudes unsuitable as general forecasts. The useful inference is that a goal's level and content matter, not simply whether it is challenging.

Task interdependence concerns how much members rely on one another to complete the work. Outcome interdependence concerns the extent to which their consequences or rewards are linked. Courtright et al. (2015) distinguish these structural forms and their associated mechanisms. A team may have tightly coupled work and individually competitive rewards, or shared rewards for largely separable work. Treating these configurations as the same kind of interdependence prevents a clear test of alignment. A goal intervention should therefore be evaluated against the actual dependencies of the task and the incentives members face.

4.3 Contextual dependence and the portability of expertise

Studies of mobility challenge the assumption that observed individual performance is entirely portable. Groysberg et al. (2008) found that star analysts' post-move performance depended on the capabilities of the destination firm and whether colleagues moved with them. Huckman and Pisano (2006) examined cardiac surgery and showed that individual performance had an important organization-specific component. These settings offer valuable evidence about contextual dependence, but neither establishes the fraction of every worker's expertise that belongs to the organization.

The relevant distinction is between portable technical knowledge and the conditions required to deploy it. A specialist may retain the same knowledge after moving while losing familiar collaborators, trusted information channels, complementary assets, or locally learned routines. Because moving is selective, performance changes may also reflect assignment difficulty, client sorting, or employer choices. Evidence of a mobility penalty therefore supports investigating disrupted relationships and assets; it does not identify a coordination mechanism without further observation.

4.4 Disputed talent concentration effects

The claim that too much talent harms teams remains unsettled. Swaab et al. (2014) reported evidence of detrimental talent concentration in interdependent sports. Gula et al. (2021) challenged the basketball result and the inference drawn from quadratic specifications. Long et al. (2024) found no general harmful concentration effect in European club football. Their study covered six seasons in seven leagues, amounting to 42 league-seasons and 780 team-season observations; this should not be mistaken for 42 consecutive years of data.

Recent evidence also prevents dismissing the issue as closed. Loignon et al. (2025) linked the value of star talent to passing-network configurations in professional soccer. Their results direct attention toward how talent and interaction structure jointly relate to performance. They do not establish that one network structure is optimal across work settings, or that talented colleagues generally undermine one another. Taken together, the studies support explicit comparisons among linear benefits, diminishing returns, and negative marginal returns, with attention to measurement and context.

4.5 A boundary beyond the focal team

A team can retain valuable expertise while preventing the organization from deploying it where it is most useful. Haegele (2026) documents talent hoarding using personnel records and survey evidence from a large firm. Quasi-random exposure to hoarding is used to examine inhibited internal applications. The study identifies an incentive problem that cannot be reduced to the quality of interaction within the manager's existing team. It shows why local performance and organizational allocation must remain separate outcomes. Its findings do not, by themselves, estimate the net organization-wide cost of all retention decisions.

Table 1 summarizes the inferential role of the main evidence streams. Their convergence supports studying the deployment of expertise; it does not establish the proposed framework as an empirically validated model.

Table 1  Evidence streams and inferential limits

5 A conditional account of expertise deployment

5.1 The location of the constraint changes the intervention

Suppose a product team repeatedly ships components that pass specialist review but fail when combined. If independent assessment confirms adequate expertise and process observation reveals incompatible interfaces, another specialist may add less value than an intervention that establishes common specifications and tests cross-component dependencies. If the same team lacks knowledge of a newly required security protocol, interface clarification cannot replace training or specialist support. These examples illustrate theoretical contrasts; they are not observations from an original study.

The underlying claim is one of intervention specificity. An intervention should alter the mechanism responsible for the performance constraint. General encouragement, greater meeting frequency, and higher selection thresholds are unlikely to be equally effective across availability, accessibility, integration, and evaluation problems. Leadership contributes by identifying and addressing the constraint, but its effect depends on whether it has the authority and resources to do so. This develops the functional perspective into a comparison among matched and mismatched interventions.

The feasible response also depends on decision rights. A line manager who identifies missing expertise may lack authority to recruit or alter staffing. Functional provision may then require escalating the resource request, seeking authorized specialist consultation or contracting, reallocating work to a qualified unit, or narrowing or deferring the task. These options remain subject to organizational approval and capacity; they are not universally available workarounds. Reallocation changes where expertise is supplied or what the task requires, rather than showing that coordination creates absent knowledge. Where no feasible response meets an essential competence requirement, the constraint remains unresolved and responsibility for changing resources may lie above the team.

Proposition 1. With task demands, baseline expertise, and other binding constraints held comparable, an intervention targeting an independently identified, performance-limiting accessibility, integration, or evaluation constraint will improve team performance more than an equally intensive intervention targeting a different constraint. When essential expertise is absent, process-focused interventions alone will have smaller effects than interventions that supply or develop that expertise.

A direct test requires diagnosing the constraint before treatment assignment and defining equal intensity through contact time, resources, and implementation support. The proposition would be weakened if generic interventions performed equally well across independently classified constraints, or if performance improved without change in the targeted mechanism. Such findings could favor a general motivation explanation over the proposed specificity account.

5.2 Complementarity and compensation are different predictions

Complementarity exists when better coordination increases the performance return to additional expertise. It is plausible when specialists hold different pieces of necessary knowledge and those pieces can be combined only through reciprocal adjustment. Improvements in accessibility and integration allow additional expertise to affect the joint output. Greater interdependence may strengthen this relationship because unused specialist knowledge carries larger consequences for the rest of the team.

Compensation exists when coordination reduces dependence on a particular member's capability. Standardized checks, backup behavior, or task redistribution may protect performance against a limited deficit in a substitutable role. Here coordination can benefit the team while flattening the relationship between the relevant expertise measure and performance. Compensation has limits. A team cannot be assumed to compensate for a missing non-substitutable specialty, and repeated reliance on a few experts may create workload and continuity problems.

This distinction prevents a common logical error. A larger contribution of process to performance does not imply a smaller contribution of expertise. In a complementary system, both can become more consequential together. Conversely, a negative expertise-by-coordination interaction does not necessarily mean that coordination is harmful; it may indicate successful compensation. The interpretation depends on the task, the composition measure, and the observed mechanism.

Proposition 2. Above the task's essential competence requirements, coordination will increase the marginal return to complementary expertise when success requires combining distinct specialist contributions. Where member contributions are substitutable and routines or backup can cover limited deficits, coordination will instead reduce performance dependence on the expertise of the covered member or role.

The two clauses concern different task structures and potentially different composition measures. They should not be tested by averaging all expertise into a single score and interpreting whichever interaction emerges as confirmation. A demanding test manipulates or independently assesses complementarity and substitutability, prespecifies the expertise measure for each, and observes whether coordination changes knowledge use or actual backup. A uniformly additive pattern would challenge the proposed distinction.

Constraint location and the expertise-by-coordination relationship answer different questions. The former identifies the operation preventing effective deployment; the latter describes how an intervention changes performance's dependence on a prespecified expertise measure. No constraint has an invariant interaction sign. Table 2 uses one hypothetical product team to show how task structure and the focal expertise measure determine the prediction. These are theoretical illustrations, not empirical findings.

Table 2  Constraint mechanisms and conditional expertise returns

Note. The interaction sign is not determined by the constraint label: amplification or flattening depends on task structure, knowledge distribution, and the focal expertise measure. Compensation reduces dependence on a specified member or role and requires a qualified substitute or effective routine; it does not eliminate essential knowledge requirements. Supplying knowledge changes an input and is not, by itself, evidence of an expertise-by-coordination interaction. With several binding constraints, improving one mechanism may not change performance.

5.3 Alignment changes the willingness to deploy expertise

Knowing how to coordinate does not guarantee a willingness to contribute. A member may recognize a colleague's need yet decline assistance if individual targets penalize the time spent helping. Conversely, shared rewards need not improve technical performance when contributions are largely independent or when accountability is weak. Goal clarity, reward structure, and task dependencies therefore require separate treatment.

For interdependent work, outcome alignment is expected to facilitate the use of coordination arrangements. This is a proposed interaction between structural conditions and intervention effectiveness, informed by the distinct literatures on goal setting and interdependence. It should not be attributed as an already demonstrated moderator in the goal-setting meta-analysis. Alignment also needs an explicit boundary: success for the team may involve behavior that harms another unit or the organization.

Proposition 3. In highly task-interdependent work, interventions intended to improve expertise accessibility or integration will produce larger performance gains when goals and rewards recognize contribution to the shared output than when they reward competing individual output. This advantage will be smaller in work with largely separable contributions.

Evidence against this proposition would include comparable intervention effects under misaligned rewards, provided members perceive the reward difference and the study has adequate precision. Researchers should also test whether pressure, rather than alignment, explains any apparent gains. Objective quality, assistance behavior, workload, and error concealment can distinguish improved collective contribution from intensified effort or suppressed reporting.

5.4 Portability depends on what must be rebuilt

Expertise portability is often evaluated as a binary attribute of a person or occupation. The constraint account instead predicts different recovery paths. If knowledge remains technically applicable and destination resources are adequate, a move that primarily disrupts accessibility and integration should be more recoverable through structured relationship building than a move into a setting without necessary complementary assets. Moving with collaborators can preserve some relationships but may also import routines poorly suited to the destination.

Proposition 4. Following an expert's move, recovery in performance will be associated with restoration of accurate expertise connections and task interfaces beyond the contribution of retained technical knowledge. Integration interventions will have weaker effects when the destination lacks essential complementary assets.

This proposition extends the portability question from whether performance travels to which conditions must be reconstituted. It remains untested as a complete mechanism. A credible study measures technical capability before and after the move, documents destination resources and assignments, and observes relationships repeatedly. If recovery occurs without restoration of the predicted connections, or is fully explained by assignment changes, the proposed mechanism loses support.

5.5 Coordination arrangements must fit changing work

Coordination can impose costs through delay, duplicated review, interrupted concentration, or excessive dependence on a central expert. The relevant variable is the fit of coordination arrangements to dependencies, not the sheer frequency of communication. A routine that helps a new team may become unnecessary after members develop shared knowledge. The same routine may become useful again when membership, technology, or task requirements change. Episodic and dynamic team frameworks provide the foundation for these expectations (Marks et al., 2001; Mathieu et al., 2019).

Proposition 5. When task dependencies change, teams that update expertise connections and coordination arrangements to the new dependencies will recover performance more quickly than teams that preserve the prior arrangements. In stable, well-learned work, additional coordination requirements that do not address an identified constraint will yield smaller benefits and may impose measurable time costs.

The proposition concerns adaptation to a documented change, rather than a universal inverted-U relationship. It could be challenged by finding that fixed routines perform equally well across dependency changes, or that additional coordination reliably benefits already well-matched teams. Outcomes should include both quality and time so that faster output is not mistaken for better overall performance. Figure 1 summarizes the account without implying that the proposed paths have been causally established.

Figure 1  Constraints on the collective use of expertise

Note. Dashed arrows indicate proposed relationships, not established causal effects. Constraints may coexist and recur across task episodes; the boxes do not prescribe a temporal sequence. Outcomes feed into later expertise, relationships, and arrangements. Table 2 separately maps constraints to conditional complementarity and compensation predictions.

6 Research designs that could discriminate among the explanations

6.1 Experiments on constraint matching

A useful first study would create tasks with independently manipulated constraints. Teams could receive sufficient or insufficient technical information, accurate or inaccurate expertise directories, compatible or incompatible component specifications, and evaluation procedures that permit or impede scrutiny. Manipulations must avoid making every constraint simply a harder version of the same task. Independent pilot work should show that each manipulation changes the intended constraint with limited effects on the others.

After baseline diagnosis, teams would receive a targeted or equally intensive alternative intervention. The design should include an untreated or usual-practice condition where feasible, because comparing two interventions alone cannot establish whether either improves performance. Outcomes would include blinded ratings of solution quality, objective errors, completion time, and direct measures of retrieval, integration, or correction. Prespecified contrasts would assess matching rather than treating any improvement as validation of the whole framework.

A second factorial study could vary expertise and coordination opportunities across tasks requiring complementary versus substitutable contributions. Random assignment of members or experimentally controlled information distributions would reduce selection problems. This design would provide a clearer test of Proposition 2 than correlating a team's global process rating with its average ability. Ethical and practical constraints mean that deliberate competence deficits may be more suitable for simulations than for safety-critical field work.

6.2 Longitudinal field measurement

Field studies should collect expertise measures before the focal performance episode and process measures during it. Technical assessments, work samples, or independently established qualifications are preferable to supervisor ratings that already incorporate team success. Expertise coverage should be distinguished from average proficiency and from tenure. Network observations can capture who requests, supplies, and uses knowledge; task records can capture handoff errors and rework. Global team surveys remain useful but cannot alone identify where a constraint occurs.

Perceptions of psychological safety require a defensible aggregation model. Researchers should examine within-team agreement and between-team variation before treating individual responses as a shared state. Low agreement may be theoretically informative if junior members face risks that senior members do not. The same discipline applies to role clarity and shared goals. A team average should not erase a subgroup whose expertise is systematically overlooked.

Repeated observations should align with meaningful task episodes and allow performance to affect later processes. Teams are nested within organizations, members may work in several teams, and membership may change. The statistical model must reflect these dependencies. Independent outcomes, multiple raters, and measurement separation can reduce shared-source bias, although temporal ordering alone does not establish causation. Prior performance, task difficulty, resources, team size, and membership change are important competing explanations whose roles should be specified before analysis.

6.3 Testing complementarity and compensation

The central test is whether coordination changes performance's dependence on independently assessed expertise under a defined task structure. A positive expertise-by-coordination contrast supports complementarity; a negative contrast supports compensation only when a measured backup or substitution mechanism covers the specified deficit. Neither sign establishes causality. Researchers should prespecify the focal expertise measure, distinguish these patterns from ceiling or floor effects, and measure the mechanism directly. Comparisons across task-interdependence conditions require sufficient independent teams and joint variation in the predictors. Appendix A provides an illustrative specification and design considerations; the full three-way model is not required for every test of the account.

6.4 Talent concentration and credible alternatives

A separate analysis is needed for claims that concentration becomes harmful. A negative quadratic coefficient does not establish that additional talent reduces performance within the observed range. Researchers should compare linear, monotonic nonlinear, and turning-point models; estimate marginal returns and uncertainty; and verify that both sides of any claimed turning point contain adequate observations. Diminishing positive returns must remain distinct from negative returns.

Talent concentration should also be separated from team size, wage budget, role redundancy, and prior success. Objective indicators can still embed team outcomes in the definition of talent, creating mechanical relationships. Prespecifying player or employee classifications from earlier periods can help, but introduces changes in form and assignment that also require attention. Cross-setting replications should assess mechanisms directly instead of assuming that a sports result demonstrates status conflict in workplace teams.

Table 3 links each proposition with a diagnostic observation and a result that would challenge it. These tests are intended to make the framework vulnerable to evidence. A pattern that can be explained only by repeatedly relabeling the constraint after outcomes are known would not count as support.

Table 3  Empirical contrasts and disconfirming results

7 Implications and limits

The practical implication is to diagnose before escalating either selection or coordination investment. A team with a genuine competence gap may need training or recruitment. A team with sufficient expertise but frequent failed handoffs may need clearer interfaces and dependency checks. If relevant knowledge is routinely ignored, another meeting may be less useful than changing how recommendations are evaluated. These are conditional recommendations derived from the framework, not a validated diagnostic instrument.

Managers can make the approach testable in ordinary work by stating the suspected constraint, identifying an observable mechanism, and choosing a bounded intervention with an independent outcome. For example, an expertise directory should improve accurate routing of technical questions; a handoff protocol should reduce rework caused by interface errors. If the mechanism does not change, a favorable output fluctuation is weak evidence that the diagnosis was correct. If the mechanism improves without a better outcome, the team may face another constraint or the intervention's cost may offset its benefit.

Talent acquisition decisions should distinguish technical capability from the relationships and assets supporting past performance. Integration plans should account for access to colleagues, systems, and decision rights. Assessment should also separate individual output from effects on collaborators. The present argument does not require a general claim that stars are harmful, or a numerical ranking of the value of avoiding harmful behavior versus hiring a high performer.

Organizational evaluation must extend beyond the current team. Retaining an expert may improve local delivery while limiting development or allocation elsewhere. The appropriate managerial objective therefore depends on organizational priorities and employee opportunities as well as team results. Haegele's (2026) study provides a concrete reason to examine internal mobility incentives alongside local team effectiveness, without assuming that every retention decision is hoarding.

Several limitations constrain the theoretical contribution. First, the constraint categories elaborate established emergence and team-process theories; their discriminant validity and diagnostic reliability remain to be demonstrated. Second, selection into teams, tasks, and leadership arrangements complicates field inference. Third, compensation may shift workload rather than remove a capability problem, so short-term output should be assessed alongside sustainability and member outcomes. Fourth, the evidence spans laboratory groups, organizational teams, financial analysts, surgeons, and sports teams. Mechanisms may travel across these settings more readily than effect sizes, but that transfer requires testing.

Finally, the framework does not establish the optimal allocation of a staffing budget. Recruitment, training, coordination, retention, and external consultation have different costs and time horizons. Even a well-estimated interaction does not tell an organization which investment has the largest net return. Addressing that decision requires credible counterfactual interventions and explicit accounting for implementation costs, opportunity costs, quality, and the time over which benefits persist.

8 Conclusion

The contribution of expertise depends on which constraint a team must overcome. Essential knowledge must first be available; its collective value can then depend on whether it is accessible, integrated, and evaluated effectively. Leadership provides functions that can address these constraints, within the authority and resources of the setting. Coordination may amplify the value of complementary expertise or compensate for limited deficits in substitutable contributions. Those possibilities require different predictions and measurements.

The resulting research question is precise: which intervention changes the constraining mechanism, for which expertise configuration and task structure, over what period? Progress requires designs that distinguish mechanisms and identify where the account fails. If supported, this approach would replace broad prescriptions to hire better people or improve teamwork with explanations of which change helps a particular team and why.

References

Bell, S. T. (2007). Deep-level composition variables as predictors of team performance: A meta-analysis. Journal of Applied Psychology, 92(3), 595-615. https://doi.org/10.1037/0021-9010.92.3.595

Call, M. L., Nyberg, A. J., & Thatcher, S. M. B. (2015). Stargazing: An integrative conceptual review, theoretical reconciliation, and extension for star employee research. Journal of Applied Psychology, 100(3), 623-640. https://doi.org/10.1037/a0039100

Courtright, S. H., Thurgood, G. R., Stewart, G. L., & Pierotti, A. J. (2015). Structural interdependence in teams: An integrative framework and meta-analysis. Journal of Applied Psychology, 100(6), 1825–1846. https://doi.org/10.1037/apl0000027

D'Innocenzo, L., Mathieu, J. E., & Kukenberger, M. R. (2016). A meta-analysis of different forms of shared leadership–team performance relations. Journal of Management, 42(7), 1964–1991. https://doi.org/10.1177/0149206314525205

DeChurch, L. A., & Mesmer-Magnus, J. R. (2010). The cognitive underpinnings of effective teamwork: A meta-analysis. Journal of Applied Psychology, 95(1), 32-53. https://doi.org/10.1037/a0017328

Eckardt, R., Tsai, C. Y., Dionne, S. D., Dunne, D., Spain, S. M., Park, J. W., Cheong, M., Kim, J., Guo, J., Hao, C., & Kim, E. I. (2021). Human capital resource emergence and leadership. Journal of Organizational Behavior, 42(2), 269–295. https://doi.org/10.1002/job.2446

Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383. https://doi.org/10.2307/2666999

Edmondson, A. C., & Bransby, D. P. (2023). Psychological safety comes of age: Observed themes in an established literature. Annual Review of Organizational Psychology and Organizational Behavior, 10, 55-78. https://doi.org/10.1146/annurev-orgpsych-120920-055217

Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113-165. https://doi.org/10.1111/peps.12183

Grant, R. M. (1996). Toward a knowledge-based theory of the firm. Strategic Management Journal, 17(S2), 109–122. https://doi.org/10.1002/smj.4250171110

Groysberg, B., Lee, L.-E., & Nanda, A. (2008). Can they take it with them? The portability of star knowledge workers' performance. Management Science, 54(7), 1213-1230. https://doi.org/10.1287/mnsc.1070.0809

Gula, B., Vaci, N., Alexandrowicz, R. W., & Bilalić, M. (2021). Never too much: The benefit of talent to team performance in the National Basketball Association. Comment on Swaab, Schaerer, Anicich, Ronay, and Galinsky (2014). Psychological Science, 32(2), 301-304. https://doi.org/10.1177/0956797620960396

Haegele, I. (2026). Talent hoarding in organizations. American Economic Review, 116(8), 3110–3151. https://doi.org/10.1257/aer.20220264

Huckman, R. S., & Pisano, G. P. (2006). The firm specificity of individual performance: Evidence from cardiac surgery. Management Science, 52(4), 473-488. https://doi.org/10.1287/mnsc.1050.0464

Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D. (2005). Teams in organizations: From input-process-output models to IMOI models. Annual Review of Psychology, 56, 517-543. https://doi.org/10.1146/annurev.psych.56.091103.070250

Klein, A., & Moosbrugger, H. (2000). Maximum likelihood estimation of latent interaction effects with the LMS method. Psychometrika, 65(4), 457–474. https://doi.org/10.1007/BF02296338

Kleingeld, A., van Mierlo, H., & Arends, L. (2011). The effect of goal setting on group performance: A meta-analysis. Journal of Applied Psychology, 96(6), 1289-1304. https://doi.org/10.1037/a0024315

LePine, J. A., Piccolo, R. F., Jackson, C. L., Mathieu, J. E., & Saul, J. R. (2008). A meta-analysis of teamwork processes: Tests of a multidimensional model and relationships with team effectiveness criteria. Personnel Psychology, 61(2), 273-307. https://doi.org/10.1111/j.1744-6570.2008.00114.x

Lewis, K. (2003). Measuring transactive memory systems in the field: Scale development and validation. Journal of Applied Psychology, 88(4), 587-604. https://doi.org/10.1037/0021-9010.88.4.587

Lines, R. L. J., Pietsch, S., Crane, M., Ntoumanis, N., Temby, P., Graham, S., & Gucciardi, D. F. (2021). The effectiveness of team reflexivity interventions: A systematic review and meta-analysis of randomized controlled trials. Sport, Exercise, and Performance Psychology, 10(3), 438–473. https://doi.org/10.1037/spy0000251

Loignon, A., Fonti, F., Bagherzadeh, M., Gurca, A., & Shresta, S. (2025). When more is less: The role of social capital in managing talent in teams. Academy of Management Discoveries, 11(2), 255–284. https://doi.org/10.5465/amd.2023.0039

Long, A. M., Graf, M., & Bilalić, M. (2024). Never too much—More talent in football (always) leads to more success. PLOS ONE, 19(2), e0290147. https://doi.org/10.1371/journal.pone.0290147

Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally based framework and taxonomy of team processes. Academy of Management Review, 26(3), 356-376. https://doi.org/10.5465/amr.2001.4845785

Mathieu, J. E., Gallagher, P. T., Domingo, M. A., & Klock, E. A. (2019). Embracing complexity: Reviewing the past decade of team effectiveness research. Annual Review of Organizational Psychology and Organizational Behavior, 6, 17–46. https://doi.org/10.1146/annurev-orgpsych-012218-015106

McEwan, D., Ruissen, G. R., Eys, M. A., Zumbo, B. D., & Beauchamp, M. R. (2017). The effectiveness of teamwork training on teamwork behaviors and team performance: A systematic review and meta-analysis of controlled interventions. PLOS ONE, 12(1), e0169604. https://doi.org/10.1371/journal.pone.0169604

Mesmer-Magnus, J. R., & DeChurch, L. A. (2009). Information sharing and team performance: A meta-analysis. Journal of Applied Psychology, 94(2), 535–546. https://doi.org/10.1037/a0013773

Morgeson, F. P., DeRue, D. S., & Karam, E. P. (2010). Leadership in teams: A functional approach to understanding leadership structures and processes. Journal of Management, 36(1), 5–39. https://doi.org/10.1177/0149206309347376

Morrison, E. W. (2023). Employee voice and silence: Taking stock a decade later. Annual Review of Organizational Psychology and Organizational Behavior, 10, 79–107. https://doi.org/10.1146/annurev-orgpsych-120920-054654

Ployhart, R. E., & Moliterno, T. P. (2011). Emergence of the human capital resource: A multilevel model. Academy of Management Review, 36(1), 127–150. https://doi.org/10.5465/amr.2009.0318

Ray, C., Essman, S., Nyberg, A. J., Ployhart, R. E., & Hale, D. (2023). Human capital resources: Reviewing the first decade and establishing a foundation for future research. Journal of Management, 49(1), 280–324. https://doi.org/10.1177/01492063221085912

Ren, Y., & Argote, L. (2011). Transactive memory systems 1985-2010: An integrative framework of key dimensions, antecedents, and consequences. Academy of Management Annals, 5(1), 189-229. https://doi.org/10.5465/19416520.2011.590300

Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. Journal of Applied Psychology, 107(11), 2040-2068. https://doi.org/10.1037/apl0000994

Swaab, R. I., Schaerer, M., Anicich, E. M., Ronay, R., & Galinsky, A. D. (2014). The too-much-talent effect: Team interdependence determines when more talent is too much or not enough. Psychological Science, 25(8), 1581-1591. https://doi.org/10.1177/0956797614537280

van Dyck, C., Frese, M., Baer, M., & Sonnentag, S. (2005). Organizational error management culture and its impact on performance: A two-study replication. Journal of Applied Psychology, 90(6), 1228–1240. https://doi.org/10.1037/0021-9010.90.6.1228

Appendix A  Illustrative interaction specification and design considerations

One illustrative specification for a single prespecified expertise measure is

Y(j,t+1) = α + βT T(j,t) + βC C(j,t) + βI I(j,t)

             + βTC TC(j,t) + βTI TI(j,t) + βCI CI(j,t)

             + βTCI TCI(j,t) + γX(j,t) + u(j) + τ(t) + ε(j,t+1)

Here Y is subsequent objective team performance; T is independently assessed expertise composition; C is an independently measured coordination practice or assigned intervention; I is task interdependence; X contains prespecified covariates; u(j) represents team heterogeneity; and τ(t) represents common period effects. TC, TI, CI, and TCI denote the corresponding products. All constituent terms are included. In a stable-team panel, team fixed effects would absorb time-invariant between-team predictors, so this model must be adapted to the actual sources of variation rather than applied mechanically.

The expertise-by-coordination contrast at a given interdependence level is βTC + βTCI I. A positive contrast is consistent with complementarity; a negative contrast can be consistent with compensation if the task and observed backup mechanism satisfy Proposition 2. Neither sign independently establishes causality. Goal and reward alignment requires additional prespecified terms or a separate factorial design; it is not tested by this equation. The equation also does not validate all five propositions, establish a single latent process construct, or imply that output is zero when any input is zero.

Interactions depend on scale and functional form. Ceiling effects can create apparent compensation, and floor effects can create apparent complementarity. Studies should report predicted outcomes over the observed range, use bounded-outcome models where appropriate, and examine substantively justified alternative forms. They should avoid relying solely on standardized product coefficients. Because interactions are often imprecise, design-specific power simulations should incorporate team clustering, measurement error, attrition, and the planned moderator distribution.

The three-way interaction is a demanding optional test, not a minimum requirement for evaluating the framework. Its estimation requires enough independent teams and sufficient joint variation in expertise, coordination, and interdependence; repeated observations within a few teams do not substitute for replication at the level of the relevant predictors. Sample requirements should follow simulation-based power and precision analyses for the planned design, rather than a universal sample-size rule. If that design is infeasible, researchers can test prespecified expertise-by-coordination contrasts within experimentally fixed task structures without claiming to estimate the three-way interaction. Centering continuous predictors at meaningful values can clarify lower-order coefficients, but it does not increase information or the power of the highest-order interaction in an otherwise equivalent complete model, and it need not reduce collinearity. Where constructs are measured by multiple indicators, latent moderated structural equations offer an approach to modeling latent interactions (Klein & Moosbrugger, 2000); their use requires a defensible measurement model and estimator-specific simulation checks, and is not a remedy for too few teams.