In recent years, the novel bacterial foraging optimization has been widely applied. However, in past studies, the process of bacterial foraging lacked guidance and the structure of the algorithm was inadequate, which resulted in a low convergence speed and a large number of parameters in the algorithm, thus reducing its search accuracy and speed. Additionally, researchers only improved the algorithm for complex situations, for which a comprehensive evaluation of its robustness could not be made. Here, to resolve these issues, two improved algorithms are proposed and compared comprehensively. Our algorithms are suitable for modeling the foraging process of organisms in nature: a small number of individuals with rich resources can attract other individuals to forage locally. First, we propose a decreasing composite function and gradient migration behavior and introduce the 80/20 rule. A few excellent individuals guide the population to migrate to the optimal solution and increase the convergence speed. Second, we introduce the renewal speed of particles and propose another composite function, and the biological characteristics of Escherichia coli are also introduced to achieve the screening of excellent individuals. Finally, we show the results of numerous experiments and comprehensively evaluate the applicability of the proposed organisms.