A question is posed on how a particular subassembly sequence is generated in randomized assembly. An extended design of mechanical conformational switches  is proposed that can encode several subassembly sequences. A particular subassembly sequence is generated due to conformational changes of parts during one-dimensional randomized assembly. The optimal subassembly sequence that maximizes the yield of a desired assembly can be found via genetic search over a space of parameterized conformational switch designs, rather than a space of subassembly sequences. The resulting switch design encodes the optimal subassembly sequence so that the desired assemblies are put together only in the optimal sequence. The results of genetic search and rate equation analyses reveal that the optimal subassembly sequence depends on the initial concentration of parts and the defect probabilities during randomized assembly. The results indicate that abundant parts and parts with high defect probabilities should be assembled earlier rather than later.