The substantial efforts kept by many companies in retaining customers has become more challengeable. In this aspect, forecasting a customer to churn in the upcoming future is an enormously prevailing job for many advertising crews. Present market suffers much-sophisticated charges in bidding to handle new-fangled customers than to hold prevailing ones. As a consequence, ample exploration has been capitalized to innovative traditions of recognizing individual customers who may consume a high hazard of agitating. Nevertheless, purchaser preservation pains consume huge volumes of an organizational resource. In accordance with these matters, the succeeding generation of churn administration ought to concentrate on down truth. A diverse technique of churn prediction has been established to fulfill the above requirements. The work that carried out in this paper is to develop a model that can predict better accuracy of churn. We used the existing models of SVM and kernel function of it and collaborated with the artificial neural networks in prediction. The results proved to be outstanding when both the models integrated with each other. Thus, most of the widespread innovations that have been distinguished in the writing for the extension of a customer churn administration stages which are using the soft computing techniques such as Neural networks, SVM etc.