Exide Life Insurance leverages its in-house analytics tools for effective underwriting and to reduce fraudulent claims.
Working for digital transformation, analytics plays a big role in transforming business processes. Ashwin B, COO, Exide Life Insurance shares his experiences for the implementation of Analytics in his organization.
Q: What kind of business issues did you face, which analytics helped you to address?A: I would broadly classify the challenges into four clear buckets:
- Risk identification and mitigation
- Improvement in persistency
- Revenue generation
- Improvement in operational efficiency
We have embedded and leveraged analytics over the last few years to address issues in these four broad categories. The objective has been to support the business outcomes and increase operational efficiency, while reducing the amount of risk.
More specifically on the risk identification and mitigation side, we have undertaken several initiatives. The first is where we have been able to build analytical models, which support our underwriting process, to make it more effective in terms of onboarding the customer.
When we onboard new customers, we create an underwriting risk profile of every customer, which helps us predict the risk of early/fraudulent claims. This allows us to identify and classify customers into high, medium and low risk category, taking into consideration several parameters including the product type, the profile of the sales person/channel, the sourcing location etc.
Depending of the risk category, various interventions/due diligence are put in place to ensure that the customers who are on boarded meet the desired profile. For example, if it is high risk category customer, we have a higher level of due diligence for accepting the risk, but for a customer in a low risk category we can proceed with straight through processing.
We also introduced the concept of house holding, which has helped to identify the cross linkages between customers and provide a comprehensive view of all policies a customer is associated with across the organization. We have been able to effectively identify people across the household irrespective of whether it is mentioned in the proposal form or not, by using certain common parameters which are available across different clients. For example if the customer applies for a new policy but happens to be a nominee in another policy, we are able to identify and link them under a single house hold. So, when we underwrite the policy we are able to make a qualitative assessment on the basis of the total risk cover; premium to income ratio based on the house hold rather than do it on an individual policy basis.
Another common issue is of non-disclosure of existing policies by customers at the time of purchase of a new policy. To address this, over the last few months, we have been able to leverage the initiative undertaken by the IIB (Insurance Information Bureau of India) to develop the ‘Quest’ database. We have integrated this as the part of onboarding process. This helps identify customers who have multiple policies across other insurance companies and thereby minimize the risk of non-disclosure.
Q: Do you have any product or in-house analytics product for this?
A: We have built the analytics capabilities in-house, and the analytics team works on tools like SAAS, R and Python. There is no external product that we are using. We find it more efficient and effective to build analytical models in-house as we are able to tailor the same to meet our specific requirements.
Q: Once you have them deployed what kind of benefits have you accrued?
A: The deployment in risk identification and mitigation gets us two benefits. First is by identifying the high-risk category policies we are able to do a higher level of due diligence and accept those policies which meet the underwriting requirements or identify policies with potential issues and reject them upfront. The second is that the low risk category policies can get issued faster through straight-through-processing.
Q: Did you face any challenges when analytics became a part of your operations?
A: There are a couple of challenges, certainly. One is, of course for analytics you need as many data points as you can get. The data we can work with is limited to what is available to us from customers which comes along with the proposal form and any additional information which is provided by the sales team. There is also some additional information, which we now get from the IIB industry data base. So availability of the information of the customer is one limitation that we have, which we have to work with.
The second is that, since analytics is a relatively new field, there will be some acceptability issues and skepticism at the time of development and deployment of any new model. There is a need to ensure a buy in from the internal stakeholders as they are not 100 percent sure if they can really trust the models, and if they change their business processes basis the analytical models would that lead to problems later! Further if there is an added layer of higher due diligence required, it could have an impact on the issuance as well as time taken, since more information is needed about the customer prior to accepting the risk. So while there were challenges in implementing these models, we were able to work our ways through and successfully demonstrate the merits of these interventions. This has ensured that we have the requisite buy in from the stakeholders and in fact it is now fully embedded within our onboarding process.As far as improvement in pis concerned, which is the second portion of the business, we need to have a different approach from what we do for risk mitigation at the new business stage. We have developed a persistency model, which helps us to ascertain the likelihood of a customer paying the renewal premiums right at the onboarding stage. We run this model once again prior to the time of renewal payment due to be able predict which customer is more likely to pay and within what time frame. Basis this, we can focus our efforts on customers differently depending on the propensity to pay. So, wherever customers are less likely to lapse, we can do a light touch reminder, and focus more on customers who are less likely to pay. The benefit from this particular model is it helps to manage the entire renewal process more effectively, thereby improving efficiency as well as maximize the renewal collection and enable increase the in force base of our customers.Subscribe to Moneycontrol Pro and gain access to curated markets data, trading recommendations, equity analysis, investment ideas, insights from market gurus and much more. Get Moneycontrol PRO for 1 year at price of 3 months at 289. Use code FREEDOM.