The performance inference gap in talent management: A critical integrative review

Talent management is frequently justified by its expected contribution to organizational performance, yet a reported performance relationship can represent different constructs, measurement sources, and levels of inference. This critical integrative review examines the correspondence between talent management exposures, organizational outcomes, and the designs used to connect them. A new public-source search on October 7, 2026, retrieved 4,899 OpenAlex records, all retained after DOI and Unicode-preserving title checks. A purposive, full-text case corpus was then assembled to contrast outcome definitions, information sources, analytical levels, temporal structures, and reporting credibility. The inventory is not presented as an exhaustively screened systematic review. The cases reveal several distinct inference problems: performance items that already attribute success to talent management; financial labels attached to managerial perceptions; retrospective comparisons presented as evidence across time; capability chains without formal indirect-effect estimates; and mediation coefficients whose causal interpretation exceeds the design. A temporally separated study and a qualitative skills-matching study provide useful contrasts while retaining their own identification limits. Reporting contradictions and probable dataset overlap further constrain some estimates. The review develops a framework that distinguishes construct correspondence, measurement independence, and causal identification, and specifies what evidence is needed to connect talent practices to realised organizational returns. Its contribution is a transparent audit of performance inference, rather than a representative effect estimate or a universal endorsement of talent management.