Real Time Analytics - Insurance Domain

Insurance businesses continue to look for ways to differentiate, given the highly competitive nature of the industry and limited differences in product features. Real-time analytics can provide new ways to meet customer expectations of a more personalized experience and faster decisions.

Digital channels of customer interaction (such as online channels) as well as conversations online (such as social media) have created new streams of real-time event data. High technology and online businesses continue to leverage real-time analytics to create personalized experiences for their customers. We believe that there is a clear opportunity to cross-pollinate these ideas in other industries. Insurance, being a data rich industry and a high customer life time value business, can gain immensely from real-time analytics. Here are some of the opportunities and challenges faced by insurance companies in the area of real-time analytics.

Real-time Analytics – What does it mean?

Here are a few scenarios where an insurance firm can deliver greater value to its customers by responding to events faster:

·         A prospective customer visits the website looking to get a quote. Real-time analytics can be used to predict the propensity of the customer to leave the site without applying for a quote. This in turn, can be used to trigger interventions (e.g. a free consultation).

·         It is well known that fast-tracking of claims improves customer satisfaction significantly. However, fast-tracking of claims can increase the risk of fraud. Real-time analytics can be used to reduce the risk of fraud, even while accelerating the speed of processing claims.

·         Some auto insurers are already collaborating with automobile firms to gather real-time information from vehicles on a continuous basis. With GPS-enabled telemetry devices in place, insurers can devise innovative policies where the premium could be thought of as a car gas-tank that is filled up at a station. Just as the actual consumption of gas changes dynamically based on a variety of conditions, the premium can be ‘consumed’ in real-time based on driving behavior – say, if one drives safely premium must be extended longer than when driving rashly.

Common thread in all these scenarios is the ability to process data in real time as it arrives, rather than storing and retrieving at a later time for analysis.

Real-time analytics requires taking decisions with very low latency / high speed to have the required impact in time. How much time is available for a decision depends on the situation. Arbitrage in markets need decisions in micro seconds, stopping credit card transactions for fraud before approval needs to happen in seconds and responding to visitor behavior on the website needs decisions in minutes. ROI calculations will trade off the value of this impact with the cost.

Real-time Analytics – How to make it happen?

Real-time analytics can take many functional forms. It can be pre-set deterministic rule based or can be a combination of rule based and analytic scoring models where a rule is applied based on the scoring done in real time. It can also be implemented in a way that the scoring model is updated frequently.

Decisions based on pre-set rules
In rule based analytics automated decisions are taken based on the pre-set business rules – in a deterministic manner. To solve the online customer loss problem we can set up a rule – Pop up a chat window offering help for any customer who spends 7 minutes on the web site and / or visited the agent locator page but did not place a request within 90 seconds.

Decisions based on pre-determined model
The classic case of pre-determined model would be capturing fraudulent claims in insurance sector. Based on a prior statistical or machine learning model the characteristics of a fraudulent claim are identified. As soon as the claim is raised on the system, it can be identified as a fraud based on the scores generated by a pre developed model – during every step of the claims process.

Decisions based on self-learning model
In certain scenarios there is a need to continuously improve and adapt the model, as the patterns of incoming events could be continuously changing. For insurance fraud modeling, a machine learning algorithm can be set up to separate fraud from non-fraud. In this approach, the self-learning model will be in continuous test and control mode, while refreshing the model when there is a statistically significant change in characteristics or on time based triggers. Real-time analysis of claims data for fraud will then be an inline process, continuously guiding handlers, adjustors and representatives in the right direction.

From a technology perspective, implementation of real-time analytics needs organizations to exploit advances in Big Data technologies, event bus architecture to create extendible yet loosely coupled systems and agent based models to simulate the impact of autonomous entities on the system as a whole.

In conclusion

Insurance businesses continue to look for ways to differentiate, given the highly competitive nature of the industry and limited differences in product features. Real-time analytics can provide new ways to meet customer expectations of a more personalized experience and faster decisions. Real-time analytics also strengthens risk management by integrating fraud assessment as a continuous inline process, rather than an offline review process.

Exploring all customer touch points to identify application of real-time analytics is often a good place to start for any organization. Over the next few months we intend to put forth specific ideas on implementing real-time Analytics in Insurance organizations.