With the regulatory regime changing for insurance companies, the focus is on sufficient capital to cover risk. One of the key aspects of this new regime is to estimate the probability of Surrender, Lapsation and Claims. With a well specified model, the insurance company was able to operate with the optimal capital based on the risk its portfolio carries. The same model can also be used by marketing to reduce Lapsation and Surrender by identifying vulnerable customers. It can also be used to target a less risky customer by using the insights from the Claims model.
A logistic regression model was used to score the customers based on a number of variables that include demographics of the policy holders, environmental variables such as interest rates and performance of the stock exchange as well as behaviour data in terms of purchase and payment modes. A high level of significance was set to ensure accuracy of the model as it was to be used by a regulator, and easily available data was used to keep the model practical and usable.
The results were that 90% of the lapsers were in the top three deciles of customers organised according to their probability of lapse (anything closer to 1). This was a robust model that identified the vulnerable consumers and the insurance company could target this group and retain them thus reducing churn.
An Alco-Bev major wanted to understand the sizes of each of the categories of alcohol such as Whisky, Vodka and so on into the future for a plan strategy for a 5 year period.
The advantage of this methodology is that ROI of key digital channels can be compared and spends optimized in line with these insights, significantly improving overall efficiency.
The problem this CPG brand was faced with was a typical problem- Brand had presence across multiple connect points, paid media, owned media ( website, Facebook page etc) and earned media ( Twitter and other social mentions) but not knowing clearly what works across market clusters, and the interrelations between these connect points.