Financial services sector insight and analysis on what really matters in the world of financial services from Pinsent Masons.
Insurers are aware of the transformative effect big data is having on their industry. Whether used to combat fraud, better understand customers or more accurately price premiums, big data technologies are enabling smarter, more reliable decision making.
But insurers are also aware of the challenges they face in extracting value from data. Data silos continue to persist within their organisations, and communication channels between the risk and the sales and marketing departments are in many cases inadequate. Data silos outside the organisation restrict the effectiveness of analytics tools already available.
Data is being shared amongst insurers, but they face difficulties in incentivising cross-industry sharing arrangements. While an insurer may see real value in obtaining a bank's view of its customer, the bank may be less inclined to share.
Often the first discussion an organisation has when it decides to adopt a data sharing arrangement is about the potential legal and regulatory risk that the arrangement may create. But could legal and regulatory fears be unduly restricting the extent to which insurers benefit from using and sharing data? Below we take a look at some of the ways in which insurers are currently using big data and consider whether legal and regulatory fears are holding the industry back from better capitalising on this ever-increasing resource.
Big data and fraud
Insurers need to have trust and confidence in the data they acquire about customers and potential customers. The richer the data, the more likely will they be able to separate fraudulent applications and claims from genuine ones.
Big data analytics technologies can help insurers achieve higher levels of trust in the data which they acquire. Lee Brooke-Pearce, who focuses on digital transformations in the insurance sector at Capgemini Consulting, recently suggested to us that in 2014 and 2015 insurers will use big data technologies to reach new levels of sophistication in fraud detection. Rather than focussing on the 'rearview window', these technologies will better equip insurers to engage in more proactive intelligence gathering to a greater degree that they are currently doing so, he said.
But fraud detection processes enabled by better analytics technologies are only as valuable as the amount and quality of the data which they process. Well aware of this, insurers are also placing greater reliance on the data sharing arrangements with each other.
The legal basis for sharing data in the context of fraud detection derives from the Data Protection Act (DPA) which permits disclosure of personal data where necessary to prevent or detect crime. But the scope of this sharing power remains uncertain, and as the Chartered Institute of Insurance (CII) has suggested, the volume of data sought and the quality of requests made by insurers are being questioned.
As Out-Law reported some time ago, the CII is preparing a protocol that it proposes should govern data sharing arrangements between insurers. The CII's views and its proposal has its critics, but the bottom line is that insurers need to have clarity about the limits that should apply to data sharing arrangements in order to avoid regulatory action and reputational damage.
Big data, the customer experience and more accurate pricing
But fraud detection and reduction of claims liability may not be the primary interest for insurers looking to capitalise on the big data opportunities. Of more interest is the potential for big data analytics to improve underwriting processes and the overall customer experience.
For a long time, fixed datasets - such as demographics, purchase histories and retention information - have guided insurers' insights into their customers. But in order to increase accuracy in predicting future behaviour and in segmenting customer bases, fixed forms of data can be blunt. As Experian recently said in a recent BBC report "The way we've done insurance now compared to what we can do is sloppy".
Unlike fixed forms of data residing in traditional relational databases, unstructured data streaming from sources as varied as social media, telematic-enabled devices and mobile phones reveal current behaviours and current relationships. When these are matched and compared and contrasted with past behaviours and relationships, insurers get a better understanding of their customers. The more data insurers have and the quicker they use that data, the more likely they will be able to gauge future behaviour and more accurately segment their audience and price their products.
The increased volume of data available also improves the customer experience, for example by simplifying the pre-approval process. In the life insurance context for instance, a potential customer may be asked to answer over 40 questions, undertake a medical exam and perform various other tasks in order for his or her health risk to be determined. A predictive analytics model that matches social, health and data held perhaps by third parties against over 50,000 past underwriting decisions reduces the length of time required for the sales process to be completed.
But matching data for these purposes is very different to the use of it to detect fraud, as insurers cannot rely on a specific power to share. Insurers will need to become more involved in regulatory reform discussions taking place on data sharing related issues and be prepared for further rule changes in the future if they are to take advantage of future developments made possible by big data.