An insurer with over $1B in Direct Written Premiums, was experiencing a steady increase in policy churn rates over time, as well as an increasing trend of lapsed policies.
A Latin American insurer with over $1B in Direct Written Premiums was experiencing a steady increase in policy churn rates over time. Despite churn models created by their internal data science team, the insurer began looking for external solutions to improve customer retention.The insurer enlisted our help to better understand:
Relativity6 set out to improve the insurer’s retention rates by using proprietary cross-sell, win-back and retention algorithms, as well as an insurance product recommendation engine.
The insurer provided Relativity6 with an internal customer dataset from their Personal Auto, Health andLife insurance segments. The datasets included policies, claims, transactions, customer information, and product descriptions. Relativity6 performed an exploratory data analysis, then pre-processed data sources, obtained data models and applied feature learning. We then applied proprietary machine learning algorithms to train win-back and retention models.
Purchase propensity predictions and product recommendations were then provided to the insurer in order to perform a phone campaign to cross sell and win back customers.A confusion matrix and derived metrics (precision, recall and accuracy) were used to measure results of the campaign.
Relativity6’s proprietary algorithms significantly outperformed previous statistical techniques used by the insurer. According to a confusion matrix of possible outcomes, the overall accuracy of the prediction models reached 90%.Qualitatively, the director of the insurer’s parent group's analytics department had this to say, "Relativity6 was quickly able to generate improved win-back and cross-sell results, which were impressively precise. Winning back customers and adding lines of insurance to existing customers will greatly impact our bottom line.”