Data-driven approach to predict policy cancellation and improve customer retention in the insurance industry

There is no promising success for insurers without customer satisfaction. Economic downturns and competition create increasing obstacles for the insurance industry. In the age of AI, companies rapidly adopt emerging advanced technologies to optimise their growth strategies. This thesis examines how data analysis and machine learning can help in policy cancellation prediction. The goal is to provide insurers with actionable insights for creating proactive retention practices and winning customer loyalty. This study analyses over 7,300 policy records from a real insurtech startup that serves as a digital intermediary between insurance companies and clients. The startup provides a simplified and smooth experience to get the right coverage. The data was collected from the insurance provider dashboard, payment platform and customer relationship management system. So, payment history, policy and business information were examined to catch cancellation trends. The exploratory analysis shows that the monthly payment plan has a much higher cancellation rate. Nearly 80% of all cancelled policies are on a monthly plan instead of an annual payment. This trend suggests that monthlies are a key area for retention improvement. To address customer churn, machine learning techniques such as logistic regression, decision trees, ensemble methods, and neural networks were applied to predict policy cancellations. The methodology included creating workflows using the data analytics platform KNIME which easily preprocess, learns from historical data and predicts on test data. These models demonstrated high performance, achieving an accuracy rate of over 85%. Additionally, a software library for machine learning, Tensorflow, was used to apply advanced deep learning techniques. After model evaluation and interpretability analysis, the overall accuracy rates suggest strong application potential for identifying customers at risk. The thesis also delved into the reasons behind cancellations. The results show that factors such as payment failures (over 1,100 cases) and business closures lead to terminating their insurance. Small businesses with low revenue and high monthly fees may be more likely to cancel. This points to the need for more flexible policies or adaptive payment options. The results offer a starting point for developing customer-centric retention strategies that meet today’s client expectations.