The choice of a good clustering algorithm is vital in many tasks to optimize results. Nowadays, the most used algorithms use only one strategy to find and form the clusters of data, which can limit the effectiveness of the process. This paper presents a new approximation to clustering, called Essence-Based Clustering, that combines multiple strategies in a series of steps, allowing two levels of configuration of parameters, both for the whole algorithm and for each strategy used on its own. Experimental results in known data repositories show that this approach is well suited for solving clustering problems and it can do it with equivalent or better results than the current approaches.