This paper addresses the Clustered Orienteering Problem, a recent variant of the Orienteering Problem. In this variant, customers are grouped into subsets called clusters. A profit is assigned to each cluster and is collected only if all customers belonging to the cluster are served. The objective is to visit the customers of a subset of clusters in order to maximize the total collected profit with respect to a travel time limit. Our solution method is based on the order first-cluster second approach. It incorporates a split procedure that converts a giant tour into an optimal solution. Experiments conducted on benchmark instances show that our algorithm outperforms the existing methods in the literature. Actually, we have found the best known solution for 916 instances from 924 with strict improvement of 82 instances.