Telemarketing is a cost-effective instrument that offers products or services to customers through two-way and one-to-one communications over the phone. This research will use a real data collected from a Portuguese retail bank dataset during (May/ 2008 to June/ 2013) period, in order to sell long term deposits. The dataset comprises a total of 20 variables in a total of 41188 telephone contacts, and a binary (“yes” or “no”) response variable that describes whether or not the customer will purchase the long-term deposit. Logistic regression and Naïve base models are built for benchmarking. Based on the results, the predictive performance of the multilayer perceptron artificial neural networks is better than the other models.