In the 17th century, London’s coffee houses were the marketplaces of merchants. Edward Lloyd’s coffee house specialized in marine information and when the rise of trade led to a demand for cargo and ship insurance, Lloyd’s coffee house started its marine insurance business. Back then calculating and spreading risk was the core concept of the insurance industry. It still is. Edward Lloyd’s coffee house has grown through the centuries to become the Lloyd’s insurance company we know. Today, Lloyd’s uses big data analytics to understand emerging liability risks.
Smooth data integration from multiple sources and the ability to process, combine and analyze all that data helps insurers calculate and identify risk, understand customers and target new ones, detect and predict fraud, or see which claims may turn out to be very expensive. Advanced analytics and machine learning also give insurance companies the tools to mine text, analyze real-time unstructured data, or build predictive risk models to better calculate potential losses or profits on a policy.
The Big Data Challenge in Insurance Industry
Still, insurers are embracing big data analytics at a slower pace compared to other industries. A Bain & Company 2015 survey of some 90 global insurance companies showed that one in three life insurers and one in five property & casualty insurers do not use big data analytics in any business department — be it sales, marketing, claims or underwriting. The insurers in the survey, however, plan to spend more on analytics in the coming 3-5 years, and the annual spending growth is expected to reach 24 percent for life insurance and 27 percent for P&C insurers. However, many companies do not have precise plans what to do with the data they collect, Bain & Company said in its survey.
In Comes Linked Data to Create Value out of Disparate Datasets
One smart and scalable approach to processing, combining and analyzing data from many different sources is the use of Linked Data which, combined with a semantic graph database, for example, discovers relationships between names, concepts, and entities. The Linked Data approach makes data interlinked and semantically rich, extracting meaning with the use of machines and eliminating the human subjectivity factor in assessing insurance risk.
In health insurance, for instance, Linked Data integration helps organizations automatically discover expert-proven medical information in sources approved by medical regulation authorities. Thanks to semantic technology and machine learning, the information discovered can later be reused as facts in order to create medical expert validation rules.
Linked Data & Analytics for Insurance Risk Assessment
Semantic technology and linked data help create insights in risk assessment by linking data from most varied sources. According to Capgemini), an insurer needs three types of data for a comprehensive risk assessment: core data, functional data, and enriched data. Core data includes names, addresses, legal expenses insurance, and corporate hierarchy. Functional data is the policy, claims, and billing; and enriched data encompasses location, geospatial, political, economic, social media, and financial data.
Linked Data integration provides links between all those heterogeneous sources to assess and predict whether a given insurance policy would be profitable for the company and how much premium it should charge from customers so that it can retain them and attract new ones.
Targeting diverse groups of customers becomes easier when insurance companies have the tools to analyze real-time or near real-time unstructured data from social media, IoT, mobile, and telematics. In auto insurance, telematics is already extensively used. Vehicle monitoring systems and geospatial data help insurers offer customized premiums to drivers based on individual driving data. In this way, the insurance company gets to know customers and price risk better, while the customer ‘safe driver’, on the other hand, saves money and improves driving safety, when offered a lower premium for a car insurance policy.
Like any other billion-dollar industry, insurance claims are susceptible to fraud. According to the FBI , insurance fraud, excluding health insurance, costs the industry in the US more than $40 billion annually. The National Insurance Crime Bureau has estimated that insurance fraud is the second most costly white-collar crime in the US behind tax evasion.
In flagging potential frauds, Linked Data helps smooth data integration from multiple data sources, including geospatial data, social media, and historical data. Linked Data and semantic technology ensure names and locations from free-flowing text are interlinked with datasets, thanks to the ability to identify and disambiguate between Paris, France and Paris, Texas, for example. Having the links between data, insurers see patterns and detect suspicious customer claim history and/or behavior.
The ability to have all proprietary data interconnected with external data and enriched with meaning gives insurers the power to predict when fraud is most likely to take place and adopt proactive approaches to preventing it.
The Advanced Analytics Approach to Predictive Models
Lloyd’s has grown from providing information at the 17th-century coffee house to adopting predictive analytics to understand liability risks. The insurer has teamed up with liability catastrophe modeling company Praedicat Inc to try to reduce the uncertainty in identifying liability risks as much as possible.
Many other insurance companies have taken to using big data and expect predictive modeling to help them in all areas of the business, according to the 2015 Predictive Modeling and Big Data Survey by global advisory Willis Towers Watson. As much as 74 percent of the property & casualty insurers that are lagging behind in embracing big data said that the data warehouses constraints and access to data are their major roadblocks. Another 63 percent of respondents identified challenges in data integration as the primary roadblock, the survey showed.
So, finding a way to extract meaning and value out of data is the first step towards building risk models. Linked Data and semantic technologies have the power to enrich data and pave the way to advanced analytics. Insurers with the insight and foresight to use linked data in risk assessment, customer experience, and fraud detection are set to cut costs, increase return on investment, and focus on raising profit margins.