Efficient protein secretion is very important in biotechnology as it provides active and stable enzymes, which are an essential prerequisite for successful biocatalysis. Therefore, optimizing enzyme-producing bacterial strains is a major challenge in the field of biotechnology and protein production. In this study, the Gram-positive model bacterium Bacillus subtilis was optimized for heterologous protein secretion using a novel approach. Two lipolytic enzymes, cutinase from Fusarium solani pisi and a cytoplasmatic esterase of metagenomic origin, were chosen as reporters for heterologous protein secretion. In a systematic screening approach, all naturally occurring (non-lipoprotein) Sec-type signal peptides (SPs) from B. subtilis were characterized for their potential in heterologous protein secretion. Surprisingly, optimal SPs in cutinase secretion were inefficient in esterase secretion and vice versa, indicating the importance of an optimal fit between the SP and the respective mature part of the desired secretion target proteins. These results highlight the need for individually optimal signal peptides for every heterologous secretion target. Therefore, the SP library generated in this study represents a powerful tool for secretion optimization in Gram-positive expression hosts.