Making Big Data small
The “Big data” phenomenon has gained enormous interest within the last couple of years. Big data refers to the vast and ever increasing amounts of digital and available data generated by government, companies, individual and devices registering more and more granular data, social media activity and the use of multimedia.
The Insurance paradox
The “Big data” phenomenon has gained enormous interest within the last couple of years (Figure 1). Big data refers to the vast and ever increasing amounts of digital and available data generated by government, companies, individual and devices registering more and more granular data, social media activity and the use of multimedia. As an example the “Internet of Things”, a term commonly used to describe devices, like smart phones, tablets and machinery connected to the Internet, is generating 50 petabyte of data on a daily basis. In 2009 the world produced more data than the previous 5000 years combined and 90% of available data has been produced within the last two years, so the scale of data we are talking about is huge.
Figure 1: “Big data” interest
Within insurance, big data has also received a lot of interest; however, thus far the adoption has been relatively slow. This is a paradox as the business of insurance is really about leveraging past data to infer something about the future, which big data is essentially about. Furthermore, embracing big data should be easier for insurers than many for other industries because they already spend more on IT than almost any other industry and employ people skilled in the field of statistics – two prerequisites for tackling something as complex as big data.
Master or disaster
Leveraging big data within insurance is a huge opportunity. With big data insurers can enrich their own data with external data for pricing, underwriting, claims handling, fraud etc. to improve customer service and advice, gain new consumer insights, price risks better, be more selective about risk, automate decision capabilities and improve profitability.
While big data poses a number of opportunities, these opportunities could also be threats if insurers fail to act in time. Big data coming from On-Board-Units in vehicles, also known as “telemetrics” is such an example. Cars equipped with on-board-units register all kinds of data about driving, e.g. location, speed etc, and feed this information back real-time for storage and analytics. This completely changes the understanding of driving behavior and how risks can be evaluated and priced. Right now some insurers are experimenting with this new technology, although primarily using the technology as a “Big Brother” device to gather data on mileage and location information if the vehicle gets stolen. However, if insurers use of this new data remains limited, chances are that car manufacturers or others will end owning and using this data – and as a consequence, gain a monopoly on information extremely relevant for the insurer. Telemetrics is only one example, the same story can be told for almost all physical objects, e.g. houses, machines, devices registering and transmitting data through the Internet of Things.
We are entering an Age of “Data Super-Abundance” with more data more easily available and a host of 3rd party data providers are emerging, providing data as a service, essentially commoditizing information1. This will changes industry dynamics: your competitor or new entrants will have access to the same data as you and the size of your portfolio and history of underwriting will matter less than your ability to analyze and integrate big data into your processes, IT systems and risk models.
Make it small
So why haven’t insurers embraced big data yet? According to a recent survey this is due to a number of softer factors like culture and skills2. Another key reason is the fact that insurers today struggle with their own internal “small” data – getting it integrated across systems and getting sensible information from it - leaving little room or desire to start struggling with external big data.
Big data can easily seem quite “Big” and hence very frightening to get started on. A lot of research has focused on the IT and analytics capabilities needed to work with vast volumes of data, e.g. grid computing, Hadoop etc., and many of the acclaimed leaders in the Big Data space are IT companies with astronomical scale and IT budgets like Google and Facebook. Insurers should not measure their progress within big data at this level, but rather think about how they can get started on a smaller scale and make this very operational to tap into some of the obvious benefits of big data.
Some examples of small and operational efforts with big data could be:
Sort out the internal small data issue:
- Establish one view of the customer across the company value chain and couple this with social media, e.g. let the underwriter or claim handler see LinkedIn or Facebook profiles on the customer to increase customer intimacy
- Provide telemetrics-based auto insurance and start gathering data on actual driving behavior and use this information to start experimenting with pricing and services for specific segments, e.g. fleet insurance
- Use geo codes to store objet location to allow for a visual representation of location, e.g. through Google maps, and to tap into geo-spatial data , e.g. risk zones, flood stats, crime stats etc through GIS services to make pricing more fine-tuned and the underwriting context more educated
- Partner with a security company so that the insurer gains access to when customers are and aren’t home. Based on this information, the insurer can provide different rates against burglary insurance depending on whether the customer is home or not
Insurers have to start thinking about how to leverage big data. Doing this requires a shift in mentality towards opening up systems and minds to use external data to improve decision making, raise the quality of operations and make pricing more sophisticated. Insurers are well-positioned to enter the age of data super-abundance: technology, data and analytics savvy and should start in a very operational manner. Big data can be a source of competitive advantage for the insurer. However, if insurers leave it to others then they risk missing out on a great opportunity and risk others getting in on the action instead.
1) Novarica Report (Novarica Market Navigator: Data Services for U.S. Insurers 2012(Q1)
2) Celent Report: How Big is Big Data? by Craig Beattie and Bob Meara, March 13, 2013