Leveraging Data

Predictive analytics can leverage various business intelligence tools to create predictive models that allow them to more effectively and more accurately make business decisions. Predictive analytics solutions try to address various business issues, including claims fraud, underwriting and customer segmentation.

For instance, claims fraud detection is an area where insurers are increasingly investing in new technologies. These technologies typically run in real time upon data entry, using predictive analytics to identify the likelihood of fraud.

Another business area is price optimization. Price optimization is a method that helps insurers focus on key business questions to refine their pricing decisions using specific technologies.

It brings other dimensions into the pricing calculation, balancing profit and sales volume while applying customer behaviour and competitive analysis to maximize profit. It leverages analytical tools and large amounts of data about consumer behaviour to help insurers refine the price component that is a critical driver of consumer behaviour.

To better understand customers, insurers also use text and sentiment analysis. This technique consists of analyzing customer sentiment through voice pattern analysis and natural language used in mail, web-chat and short message service (SMS), to determine when a customer is either content or frustrated with the service being delivered.

This technology is often delivered as part of a contact centre solution where high volumes of calls, e-mails and web-chats are being transacted. Typically, both the call centre agent and supervisor are able to analyze contact records to either intervene or redirect a difficult conversation.

THE IMPORTANCE OF DATA

A crucial element to make the most from modern and advanced predictive analytics systems is the ability of insurers to leverage not only their internal data sources, but also external data. Celent has surveyed the insurance industry in 2013 to better understand its attitude to data (detailed in the report, Perceptions and Misconceptions of Big Data in Insurance, released in April 2013).

Two interesting observations can be drawn from survey results: 

A minority of insurers are currently using external data: Approximately 30% of surveyed insurers are already using data from data markets and data aggregators, and from public sources. Less than one insurer for five use data from open government schemes, and this proportion drops to one for 10 for other types of data sources, such as private customer data that are willingly shared, data from customer-owned devices and information from social networks. Therefore, it appears that insurers have not really captured the importance of mixing data from different sources to the benefit of their activities.

Insurers understand the importance of external customer data sources: About half of the respondents felt that private customer data willingly shared and data from customer-owned devices will be a key differentiator in the next two to three years. This proportion is much lower (approximately 20%) for data supplied by public institutions and government. It is also interesting that only 20% of respondents believe that data from social networks is going to be a key differentiator in the future. Celent takes the view that there is a “fear” factor here – insurers are not sure about what is allowed and not allowed by the regulator when it comes to exploiting private data from open sources on the Internet. The distinction between data willingly (implying knowingly) shared by customers is often the same data as that shared on social networks, but the permission to use that is explicit.

The “data source” issue – the types of data used and where data sources sit – is a question insurers have been faced with only recently. Indeed, the growing amount of data sources that are now openly available to insurers via Internet sites, social networks and media offer both an opportunity and a threat.

Insurers understand there is potential value to leveraging external data sources in different domains of their business. One of the major problems, though, is finding an efficient approach to include these data sources in their daily analysis.

USING PREDICTIVE ANALYTICS TO IMPROVE BUSINESS FUNDAMENTALS

In mature markets such as North America and Western Europe, including the United Kingdom, it is difficult for insurers to improve their core business profitability only through new customer acquisition. Therefore, insurers have to implement actions to improve their combined ratios. To do so, they increasingly invest in advanced predictive analytics systems in areas that have a direct impact on their technical results.

Optimizing policy pricing: With modern pricing optimization techniques, insurers want to leverage data around customer behaviour, their competitive landscape and elasticity of demand to complement their traditional pricing approach, which is mainly based on expected claims and cost to generate and administer a policy. To include these new parameters in their pricing strategies, insurance companies need to be able to run complex algorithms frequently and, therefore, they need enhanced calculation performance. The value is driven by combining modern pricing techniques with Big Data (or perhaps Fast Data) infrastructure to improve their reactiveness to specific customers’ needs and market conditions.

Improve understanding of customers: With the soft insurance market, insurance companies need to have a better understanding of their existing and potential customers. Internal data helps them identify single-product customers and potential upsell opportunities. Big Data, modern customer relationship management (CRM) tools and predictive analytics allow insurers to find pockets of growth in a tough market.

Enhance risk analysis: Underwriting remains one of the most important core processes performed by insurance companies. Many insurers have neglected this process and marketed products that were insufficient and inadequate to cover certain types of risks. Either these products were deemed to bring new market shares or the risks involved were just poorly evaluated. Nowadays, insurers want to improve and enhance underwriting and they understand they need to dedicate resources and effort to better leverage data in this domain.

Mitigate fraud: In many mature insurance markets, especially in general insurance, loss ratios have experienced deteriorations as fraud becomes more organized and structured. There is a growing investment in modern claims fraud detection systems, for instance witnessing the willingness of insurers to get appropriate weapons to fight claims fraud. It is Celent’s view that fraud mitigation tools can offer strong value to insurers when they are coupled with Big Data infrastructure, with a number of out-of-the-box solutions already implemented on Big Data techniques and technology.

PREDICTIVE ANALYTICS VENDORS

It is important that insurance companies consider investment in predictive analytics. In a recently published report, Predictive Analytics in Insurance: 2014 IT Vendor Spectrum, Celent reviewed IT vendors and their offerings in predictive analytics for the insurance area.

This report profiles 12 IT vendors and 16 predictive analytics systems and is not restricted to a specific geography. For each vendor, its solution offering options, customer base, data sources supported, functionality, predictive analytics and modelling capabilities and technology, and implementation and costs are described.

Although the list of vendors profiled is not exhaustive, it provides experience of implementations in the insurance industry.