Executive Briefing: Barriers to Implementing External Data Analytics in Property and Casualty Insurance
Insurance is a centuries-old industry built on the foundation of data collection and analysis to evaluate risk and reward. Industry practices have led insurers to obtain and review vast quantities of data on their customers, but have found this internal data collection wanting.The rise of social media, company loyalty programs and content marketing have created a boom in external data, both structured answers and unstructured input, that partners are willing to sell directly or provide access to.
The increased availability of information has expanded the use of data analysis for property and casualty insurers from product management to a customer-focused model that provides a backing for marketing and pricing as well as claims and underwriting. The trick for insurers is now in data selection and analysis. Market availability of external data is at its apex and the corresponding noise is at an all-time high, leaving insurers uncertain of data’s value before full analysis. To cut through the noise, insurers must improve their data management and analytics resources by integrating data and deﬁning risks across each department.
Insurers will also need to develop project plans that include the time and patience needed to process external data, potentially multiple times, as they learn which information proves useful to better understanding their customers.
With this data resting at an insurer’s ﬁngertips, the question is now: how can property and casualty insurers best beneﬁt from increasing the use of external data across their business operations and what stands in the way of proper analysis and execution?
To address the concern, this brieﬁng will show what gains property and casualty insurers are experiencing and where their integration efforts have slowed.
The Advantages of External Data Analytics in Property and Casualty Insurance
Many property and casualty insurers are sitting on large amounts of internal unstructured voice and text data in their underwriting, claims and marketing sectors that traditionally has been hard to analyze. Recent advancements in data analytics allow for the creation and adoption of unstructured data analysis tools by removing some reliance on legacy IT infrastructure. With technology like Hadoop and MapReduce, companies can start to move past some of their processing and storage needs and shift funding to cheaper, cloud-based options that scale well with need.
Unfortunately, many are ﬁnding that their internal datasets are not robust enough to develop a deep understanding of customers, their habits and potential predictors for claims processing.
To fully utilize this data, insurers must expand their collection to new avenues, including information in the public domain, collected user information from other industries such as retail and banking, and available unstructured content from shared digital resources including social media.
"Companies have become a little more outward focused because of external data, and use of this data is changing the way insurers view and interact with the external world," said Upendra Belhe, SVP and Chief Enterprise Business Analytics Scientist, Chubb Group of Insurance Companies.
Unstructured data analysis remains elusive, despite the base advantages of this analysis being the same as core beneﬁts of structured data analysis for insurers:
- Improved prediction models in terms of accuracy and usefulness
- A better understanding of customer habits and risks
- More effective marketing with the ability to judge customer reactions
- Improved tracking of insurance markets and the overall business health
- Development of new products and programs to capitalize on market changes
- Increased fraud detection
Research from IBM and The Economist (2012) that surveyed 1,168 executives (two thirds of which were C-level) across nine industries worldwide showed that companies that adopt analytics are already seeing gains over those that do not, such as:
- 2.5 times stock appreciation.
- 2 times larger EBITDA growth.
- 1.6 times higher revenue growth.
For claims, the chief reason that companies add in external data is "to improve on the accuracy of what you’re trying to predict," said Edward Vandenberg, Farmers Insurance’s Director of Advanced Analytics. "If a prediction model is only 4 to 5 percent better than a naive [pure guessing] prediction, but new, external data can help improve that by a couple of points across millions of transactions, that’s huge."
Improving the accuracy of data has potential advantages for every aspect of an insurer’s operations. As insurers proceed with the integration of external data sets, they must augment capabilities and culture to properly leverage new insights.
The addition of external data from social media and secondary sources, such as grocery store and credit card reward programs, can help develop an in-depth proﬁle of each customer beyond information they provide when seeking a policy.
Members of an insurer’s potential customer base are hunting around for the lowest-cost premiums, and the additional data they are willing to provide during that search can help an insurer best gauge the kind of rates to give.
Insurers that create customer segment proﬁles and pursue data around understanding when and where to meet their potential clients will beneﬁt in all of their customer-facing pursuits.
Not only is there a wealth of data available on potential cohorts, but much of this information is freely provided by the individuals themselves through various channels, reducing some privacy concerns as long as the data is used ethically.
Marketing and claims are the ﬁrst candidate areas that many insurers are starting to consider using social media data, even before underwriting and pricing, said Belhe.
Data analytics have created a marketing renaissance for insurers. Omnichannel distribution strategies are able to reach customers on the Web as well as their mobile devices, interacting with the customer where they prefer and where they are most likely to engage. Adding in external user data can provide insights into buying habits for a cohort, allowing marketing and technology officers to develop a single strategy to target potential customers where they plan to spend.
"There’s a good deal of beneﬁt to be gained in understanding not just who your best customers are, but where they are," said Janine Johnson, Director of Analytics at ISO. External data can allow insurers to determine "the optimal number of distribution locations" to meet demand, she said.
The data used to initially target individuals can be applied to accounts when they become customers, helping insurers measure successes and monitor customer satisfaction.
Ad-buying itself is shifting to an automated model, with analytics making display ad decisions based on pre-programmed preferences of the advertiser. The more knowledge an insurer has about its preferred clients, the better it can target ads based on demographics, location or online habits.
Already, one out of every six dollars in personal auto premiums is sold through direct response channels, according to the Independent Insurance Agents & Brokers of America in a February 2013 study that utilizes 2011 data.
Extra Fraud Detection
External data can help a company to better deﬁne its profitable customers for marketing efforts, while online policy issuance uses external data to provide checks and balance useful for fraud detection, said Janine Johnson, Director of Analytics at ISO Innovative Analytics.
Outside data can be used to conﬁrm information users provide for online price quotes and insurance applications, both verifying the data they supply and checking for risk factors users omit.
LexisNexis’s Managing Director of UK Insurance Dan Marshall has said that this under-reporting of true risk costs insurers £1.9 billion ($3.03 billion) in undetected fraud annually.
The speed of cloud systems has risen enough to allow insurers to automate a fact-checking process. As soon as a dataset is proven reliable in matching information to individuals, there is the chance to add in this support.
Price and Program Testing
Gaining insight into different customer data allows insurers to create in-depth proﬁles on customer purchasing segments, which can then be further tested through the use of variable pricing. This allows an insurer to determine the optimal pricing model by testing different price points across the same cohort.
Additional surveys and discussions can also help to reﬁne the price-to-value ratio, while securing information on previous insurance quotes or purchases can help insurers understand the price elasticity of demand. This allows an insurer to develop pricing and bundling strategies for each customer segment it covers.
Insurers can monitor consumer purchase habits and react to growth or changes in customer activity such as search traffic; areas where customers are more likely to buy can become key targets for product offers.
Maintaining external data collection, particularly around social media, will help the insurer judge the long-term viability and successes of both its pricing changes and loyalty offerings.
"Wherever or whenever there is an opportunity to commoditize the product, there is more possibility for using external data," said Belhe.