Insurance Retention Analytics
To design customer-specific treatment strategies that could optimally utilize all the available treatment channels thus maximize customer retention while saving costs.
Background and Challenges
Our client, a leading BPO company focusing on collection and retention services in the domestic market, was working on retention of consumers for a large insurance company. The insurance company had a variety of products with customers of varying attributes and profile, and wanted to increase the retention rate of existing customers.
The availability of customer demographic data and the various multimedia channels – SMS, IVR, and Outcall made it possible to design customer-behavior specific multimedia treatments rather than a one-for-all strategy that the insurance company was earlier used to.
However, there were a couple of constraints in this project:
- Campaign activity allowed for only two months post receive of data.
- Magnitude of difference in costs of the various multimedia channels.
To prepare behavioral data, all the customers were operated on the multimedia channels for a small but sufficient period of time. While doing this, no differentiation was made between the customers, and the responses to the various channels were duly recorded. This data was then coupled with the demographic data of the customers to produce a behavioral scorecard that predicts the paying propensity of the customers. The variables were carefully chosen using various statistical methods and the technique of logistic regression was used in building the mathematical model. The output was a ‘score’ value ranging between 0 and 1, and while half of the data was used in building the model the other part of the data validated it nicely.
The customers were then distributed in five buckets with each bucket having customers with similar attributes of expected payment behavior (Please refer to Illustration 1).
The responses of customers on the various multimedia channels (SMS/IVR/Outcall) were then studied for each of the five buckets. As the two particular channels – SMS and IVR involve an appreciable cost advantage over the outcall, their effects on payments were also studied for cases where these have been successful in attracting the customers without introducing them to the early stages of outcall efforts.
Not only the optimal number of effort allocation for each of the three channels was discovered, the optimal usage of SMS/IVR is also found available that could generate payment cases before an outcall attempt is made (Please refer to illustrations 3 and 4).
Now that the optimal effort is known, subsequent efforts can then be directed to ‘juicy but left out’ portion of the portfolio which can result into more paid cases (Please refer to illustration 2).
Results and Implementation
Using the score based prioritization, client was able to focus on the customers who had more likelihood of payment and thus was able to win a part of the portfolio quick and early. The optimal intensities allowed the company to target the consumers with optimal cost resulting in more retention at a lesser cost.