One of the most important issues in today's banking sector is that they want to add new customers to their bodies and want to keep their current customers. For this, banks spend a lot of money. Because of the methods they use, they either have to keep information flow to all their customers or they have to focus on their customers, whose traditional methods of probability leave. In this article, the probability of bank customers left by artificial neural networks is estimated. Presently, with the improvement of technology, a growing number of banks, holding existing clients for banks and combining new clients into their systems have earned significance. As the Bank's efficiency, it is essential to define the clients with the contingency of dropping within existing clients. The client pool generated by the classical methods utilized leads to the introduction of activities on a major number of groups for the bank. This outcomes in higher expenses for banks. The main purpose of this paper is to lay the foundation for further research for precision of who will the bank as a customer. The findings of our study with artificial neural networks have described a minimal and more compressed group as clients who are likely to leave. In this way, it is foreseen that the likely costs of banks will minimize. As a result of this study, the most accurate estimation was obtained by educating artificial neural networks with the most accurate values.