This research introduces two different methods that are Levy Flight and Hooke and Jeeves to the Bees Algorithm with the aim of improving the convergence speed and its robustness. Both methods are incorporated to the Bees Algorithm at neighbourhood search of the elite bees since that particular locations are the most promising area during optimisation process. Each Bees Algorithm and the newly incorporated method with thirteen different parameter settings are subjected to fifteen different benchmark test functions. These benchmark test functions are represented with different characteristics in terms of its differentiability, separability, scaleability, and modality. Bees Algorithm with Levy-flight method incorporated to the local search performs excellent result for 13 out of 15 functions against standard Bees Algorithm in terms of its success rate and convergence speed in which it is validated by the statistical T test. As a matter of fact, the new method indicates better robustness for 13 functions in terms of achieving good result for solving different types of optimisation problems. For the Bees Algorithm with Hooke and Jeeves method, the new approach reaches a relatively better performance compared with standard Bees Algorithm in which one parameter excels at reaching optimum solution for most of the test functions.