The growing digital data has increased companies’ knowledge of their customers and let companies sell new products to existing customers, which means cross-selling [1]. With cross-selling, companies can increase their profit easily since they already know the customer and have a relationship with them [2]. Some studies show that possible cross-sellings can be detected by using machine learning and deep learning models [3]. However, most of the studies cooperate with a company, and they use confidential data. Moreover, since cross-selling can be applied to any business field, the scope of the use cases differs. Therefore, these studies cannot conveniently be applied to a different use case. This project uses a dataset from a private insurance company with the aim of comparing traditional machine learning and deep learning model performances that predict cross-selling opportunities for life insurance. Instead of finding the possible cross-sellings, one of the goals was to test whether a deep learning model can outperform a traditional machine learning model with less feature engineering. In this study, it is shown that both the LightGBM model that we used as a traditional machine learning model and the LSTM based deep learning model can predict cross-sellings successfully. However, we spotted that the LightGBM model that was trained in this study gave better results with fewer computational resources than the LSTM model.