Training-hour indicators appear in human-capital sustainability reporting, but disclosure presence does not show which workforce groups are represented or whether information is consistent. Guided by stakeholder theory, this study examines disclosure depth, category granularity, worker visibility and two-year consistency in Malaysian listed palm-oil companies. Quantitative content analysis of public documents covers 85 company-years in 2023 and 2024, including 72 category-level observations and 31 companies with qualifying disclosures in both years. Relevant training information appeared in all 85 observations. Numerical hours appeared in 77 (90.6%), category-level values in 72 (84.7%) and separately named numerical worker-category hours in 44 (51.8%). The category-level observations contained 27 company-defined label patterns, with a median of four categories. In the paired subset, category structure was stable in 27 companies (87.1%), coded reporting-perimeter grouping in 28 (90.3%) and the hour unit in all 31. Five companies changed category structure or disclosure basis. Broad coverage therefore coexisted with heterogeneous classifications and incomplete numerical visibility of workers. This study contributes a disaggregated account of what a prescribed human-capital indicator makes observable. It does not assess stakeholder use, training effectiveness, employee experience or workforce outcomes.