The exploration of unknown environments is an important task for an autonomous robot. When exploring an unknown environment, robots face a common trade-off between visiting already mapped areas or exploring new areas. This can be done by using a planning stage in conjunction with the SLAM algorithm. This is normally called integrated exploration. In this paper, we propose a novel integrated exploration strategy based on the information potential of a frontier. By analyzing a region around a frontier, we are able to select the best frontier to explore in terms of exploration gain. We performed a series of experiments, simulated and real, comparing our approach with entropy, a common metric for information-based methods in the literature. Results show that our approach can explore the environment faster than using entropy.