In this paper, we focus on ways to optimize the manufacturing of products whereby the final product performance is measured by the aggregated performances of its components. Specifically for hard disk drives, the final recording capacity is the aggregate of the components’ achievable capacities. As is the case for hard disk drive, a drive normally has a targeted capacity such that if it falls below its target, it will be rejected or downgraded. Similarly, products achieving well above the target will still be sold at the target capacity. At both extremes, it results in loss of profits. Hence, the aim of this paper is to propose a novel selective assembly strategy which can improve profitability by reducing the variation of components in the final product assemblies and achieving the target performance. As our case study, the assembly of hard disk drive (HDD) is used to demonstrate its feasibility in real manufacturing settings. Two theorems, (1) discarding theorem and (2) binning theorem, are formulated to guide the selective assembly strategy. They provide the rules for discarding inferior components before assembly to ensure the quality of the starting population and the rules for selecting matching pairs of components to prevent producing over qualified product. The improvement in component utilization rate over random binning is guaranteed by the two theorems.