INTRODUCTION PERSONALIZED EXPERIENCE!
We have all enjoyed the activity of shopping in stores or online in the past. Even before the introduction of IoT or e-commerce, the process of shopping was simple. It started with identifying the product that you want, visiting the store where it may be available, finding the product and purchase the same at the checkout counter. The companies then employed ‘Help’ to assist customers through the long aisle of products and services to help them find what they are looking for. These helpers were then trained on the products and made to cross-sell and up-sell to the customer. This was probably the first-ever recommendation model used in the consumer market.
Insurance companies have created business rules to identify frauds, but the fraudsters are smart enough to exploit the gaps in these rules (mostly static) by manipulating information. When a system or software is made to behave like the store ‘Helper’ to find the products and services that might be of interest to the customer, we call it the Recommendation Engine. Now fast-forward to the 21st century where you are shopping online on Amazon and every time you add an item to your shopping cart; a popup window will recommend you buy one more product. There is a higher probability that you will end up buying the recommended product too.
HOW DOES A RECOMMENDATION ENGINE WORK?LEVERAGING CUSTOMER DATA
Typically, a recommendation engine involves 4 phases.
Data gathering
The data could be explicit or implicit data. Explicit data is gathered from the customer’s input like ratings and feedback on the product/services. Whereas, implicit data is the order/cancellation/return history, page views, etc. This data is created for every user that visits the webpage.
Data Storage
The more data you can gather, the better will be the results of the recommendation model. Depending on the type of data that will be used, different types of storage need to be implemented such as NoSQL Database, standard SQL database, etc.
Data Analysis
The data gathered needs to be analyzed for further processing. This process is called analyzing the data to segregate into categories and to bring out meaningful results. The analysis could be done by real-time, near real-time and batch analysis.
Data Filtering
Data filtering is basically segregating data to get relevant records which is necessary to provide recommendations to the end-user. This can be achieved by using different algorithms that best suits the recommendation engine. These could be Content-based, Cluster or Collaborative filtering.
The example shown of processing 1000 Claims is illustrative in nature:
RECOMMENDATION ENGINE FOR INSURANCELEVERAGING ON DATA
It has become indispensable for insurance companies to develop robust & effective strategies to proactively combat mounting fraud problem. The insurance sector has found a lot of value by building a recommendation engine that directly offers a product to its customers. The flexibility to manage changes to recommendations such as rule-based filtering is a very important feature that insurance companies seek. Examples of recommendation engine services include Agent Sales Enquiry and Web Recommendations.