Many P&C carriers say they will become more data-driven, and many of them have tried -- with mixed results. Such efforts can only succeed through a constant effort to employ data to make better pricing, underwriting and claims decisions.
A great data strategy transforms a company's data into valuable insights and top-line growth, but shouldn't be limited to internal data. Insurance carriers understand that external data is a must-have, and there are two compelling reasons. First, carriers think it will help their existing business processes and operations. Second, they see the value that leaders are realizing from external data.
Although a must-have, external data requires considerable effort. Leading carriers consume volumes of third-party data from multitudes of channels, and this data is most often highly fragmented. The system architecture may also need to be changed to accommodate the data, which can be challenging.
Most carriers don’t have a plan for using and valuing the external data. Procurement is usually driven by senior leadership, without a complete handle on the gaps in internal data that it can fill. They begin acquiring external data without even knowing the landscape and having a definite plan for what to do with the data. Insurance leaders must start with questions such as:
- Can external data assist you in getting a better understanding of your customers and prospects? Can it help in answering questions that were not answered before? Do we, for example, have the potential to better evaluate and comprehend property risk by using features generated from external data?
- Can external data also aid in discovering new operating models? For instance, can we use fitness tracker data to offer rewards and lower premiums on a health/life insurance plan, resulting in better underwriting and risk management? Similarly, can imagery data be used in property insurance to improve underwriting and cut down on the number of manual inspections?
- Can external data help in demystifying biases and provide granular details? Does adding external data into the training pool reduce the bias? For example, a data provider may provide access to crime data at the location level, allowing for a more accurate picture of risk at the local level rather than the ZIP code level.
- Can external data help provide directional insights in the decision-making process for an insurance carrier? While data precision is vital, it should not be the primary goal; however, the total data quality should not suffer significantly as a result of adding more data elements. What matters most are the directional insights that external data can bring when paired with internal data.
- Is the external data really providing value and worth the investment? A significant portion of the book of business should benefit from the usage of external data. For example, if a carrier is spread out across the Midwest and the external data is biased to, say, only Minnesota and Iowa, it is of minimal value to the carrier.
In this arms race to leverage external data to gain a competitive edge, carriers have made numerous attempts. However, these attempts need to either be successful in the first instance or fail fast. Failing fast is crucial, as it enables carriers to reflect back on what went wrong and can lead to experimentation that creates successful products.
Numerous things can go wrong, from data need assessment to data collection to data management and everything in between. Successful companies look at the process holistically and have a team that looks for data sources outside the company for a specific use case, as well as an integration team that helps with data acquisition, cleaning and integration within the company. Instead of viewing external data as a one-time purchase, successful carriers retain and maintain the data, as well as find and compare insights based on marrying external data to internal data.
We recommend that insurance carriers follow a three-pronged approach to undertake a more systematic adoption of external data in their offerings:
Look across lines of business
Starting with the end in mind will make it easier to ingest and integrate external data, as well as enable the necessary data architecture modernization. We've seen carriers use third-party data vendors to automate the underwriting pipeline, improving customer experience by providing initial quotes and reducing the standard time for issuance and binding. In another instance, we have seen property insurers leveraging vendors providing property attributes extracted from AI and computer vision and doing digital inspections for underwriting, reserving manual inspections for highly complex instances only.
See also: A Blueprint for Casualty 2.0
Analyze the impact of external data on the value chain
Not all functional areas of the value chain need to use external data. Upfront planning starts with an assessment of the existing data environment to determine how it can support ingestion, storage, integration and governance and ultimately the use of the external data.
This analysis will help in determining answers to the following questions:
- How many external data application programming interface (API) calls are being made across the different value chain functions?
- What kind of data parameters are expected to be passed via API calls across the different value chain functions? What is the expected format of the return objects and attributes?
- What is the utilization level of the external data attributes being pulled in across different value chain functions? (Complete/significant/partial)
- What is the underlying pricing structure of the external data vendor API? How does it affect costs across different value chain functions?
Once this impact analysis is complete, the data engineers can begin the needed data plumbing for different micro-services (i.e. data integration, feature engineering, AI/machine learning, business intelligence) to be developed for the respective value chain functions.
Set up efficient model deployment, data governance and monitoring
Insurers should develop sophisticated tech stacks for efficient model deployment and monitoring, which employ the latest analytics workbenches. Carriers should also ensure the data quality by constantly monitoring the incoming data, checking whether the source data have changed and understanding the drivers of any changes.
Strong governance standards should be in place to address data security and usage aspects of external data. A successful strategy is to constantly analyze the value of various external data sources in terms of predictability and accuracy, and to dismiss vendors that generate poor ROIs.
Conclusion
More data is better data if it aids in connecting the dots and expanding marketing opportunities and future growth. Minimizing risk and generating value with external data, on the other hand, necessitates a combination of creative problem solving, infrastructure and sound execution strategy.
Realizing the value of external data through data orchestration is a big responsibility and can overwhelm carriers. An effective approach is to begin by solving a well-defined problem and then use that success to generate momentum for expanding external-data efforts across the organization.
