Today data collection is easy — maybe too easy.
The list of possible data sources for an insurer is expanding, everything from post-accident on-board automobile data to social media communications among claimants and more traditional sources like email, medical records and police reports. Add to that long list the low cost of storage and the possibility of a data tsunami is almost inevitable since there is little motivation not to gather every bit of data possible.
The downside is that data on its own has little value unless effectively managed. If a company’s technology capabilities fail to keep pace with how quickly it is creating data, that haystack will keep growing while those needles keep getting smaller.
Insurance companies have always analyzed data, perhaps to a greater degree than organizations in any other industry. After all, accurate actuarial tables — the underpinning of the industry — are themselves an astute and early example of data analysis. Overall, however, the industry has not yet applied this in-depth use of analytics to broader, real-time operational issues.
SAS recommends considering eight key issues that can have an impact on data management. For some companies, this may necessitate a more comprehensive review of how their data governance policies and data management technologies stack up against current and future business needs.
• Big data: There is a growing variety, volume and complexity of data being used by companies, including many sources of unstructured data. With regard to insurance, this could involve anything from emails to adjuster notes.
• Changing patterns of data consumption: People used to consume data primarily at work, accessing it from a networked desktop. Now, there are internet-connected tablets, smartphones and more. When, where and how data is used has changed dramatically, particularly in the past four or five years.
• More complex data management requirements: Since data is coming from so many different sources and systems, users or applications may require data migration; others may need data consolidation to get an enterprise-wide view. There are also more complex needs, including greater integration with external data sources such as vendor and customer data.
• Data demand from more applications: The data management framework must support front-office applications such as customer relationship management (CRM) and back-office applications such as Policy Management Systems.
• Pressure for faster turnaround: Batch processing is a thing of the past. Organizations are looking for real-time or near-real time data processing.
• Additional deployment options: The location of IT data architecture is no longer limited to being on site within corporate walls. Many companies are considering cloud solutions, which could be public, private or hybrids.
• Different needs for different roles: To gain more value from data, companies are looking to get that data into more and more hands. Everyone from those with limited technical skills to IT professionals and quantitative specialists are accessing and analyzing corporate data. As such, an organization’s information management strategy must reflect this.
Most of the aforementioned considerations, to some degree, are ongoing issues for Canadian insurers. The result has been a shift from a more traditional and tactical data management mindset to a broader, more strategic information management mindset.
In How to Manage Your Data as a Strategic Information Asset, a paper published by SAS earlier this year, beyond the eight key considerations is information management. This is defined as the confluence of the following three important capabilities:
• Data management: managing and governing data from a unified platform, including data integration, data quality, data governance and master data management, with the ability to access any type of data source across the enterprise.
• Analytics management: managing a portfolio of analytic models in a systematic way — including model development, testing, deployment and monitoring — and using the results of those models as new information assets.
• Decision management: embedding information and analytical results directly into business applications or processes at the point of decision, and supporting a feedback loop as decision outcomes are cycled back into the process.
So what does this mean for Canadian insurers?
Today’s leading companies look for data management to go beyond data integration, quality and consistency issues to address more project needs and use cases — everything from front-end CRM to back-end claims systems.
In the insurance industry, one area of renewed data management focus should be how to better understand broker performance. The industry has a pretty firm grasp of how much business revenue it receives from a broker, but to better understand the true value of the relationship and the broker’s performance, more data is needed.
The insurer needs to know not only what was sold and the sort of claims that came through, but also the types of people being insured.
Like retail’s desire to capture a 360-degree view of its customers, the insurance industry, through better data management, has the opportunity to get a better view of its customer — the broker. The necessary information is already being collected by existing systems; what needs to change is the insurer perspective on data management to allow for collecting information in a way that specifically addresses how to better understand broker performance and value.
Analytics goes beyond building a model designed to pull new information from data. Predictive models help companies manage risk, something found at the core of an insurance company’s success.
An analytics model can certainly be developed to solve a specific business problem, but this approach may not be ideal. Far better is to create competing models, employing different techniques, to tackle complex problems and business issues.
The ability to create competing models is increasingly important as insurers look to broaden their offerings, delving into more complicated insurance scenarios. Were one to examine the complexity of insuring a pipeline or an oil tanker, one would see a dramatic increase in the number of variables now being taken into account to properly ascertain risk. The more data that is available in a model, the greater the likelihood of improved accuracy.
But as the number of data sources grow and problems become more complex, companies may feel pressured to take shortcuts. Analyzing all the data takes too long; analyzing a slice of the data (especially in risk assessment scenarios) can become a dangerous business proposition.
One solution is to take the data, distribute it among many computers and have all the processes run in memory. The results are staggering — processes that may have taken hours or days now take seconds or minutes.
That said, a successful company needs to create an environment to manage these models — call it the analytics model factory — which are then embedded in business processes and monitored. When this occurs, it is easier to document models and collaborate across the many facets of a company.
By feeding the models back into the decision-making process, they can be continuously evaluated with regard to whether or not they are adding business value.
At the end of the day, data only has value when it can be used to drive strategic and accurate business decisions. Decision management is about taking information or analysis and embedding it into everyday business processes.
In the insurance industry, this could be adjusting actuarial tables to take into account any combination and permutation of variables (and doing so quickly enough to produce valuable results) or assessing the possibility of a claim being fraudulent based on the most up-to-date social data available.
Much of the industry’s focus is on policies and claims — the pillars of the industry — but there is a smaller and important aspect of the industry that occasionally gets overlooked in the decision management process: risk management as it pertains to investment.
For many decades insurers relied on returns in investment in the 5% to 8% range. For the foreseeable future, it looks like rates of return on investment will remain low.
In addition, with a number of social and fiscal problems continuing around the world, investment itself will likely remain risky. As such, it is important for insurers to reduce investment risk as much as possible by employing better data management. This, in turn, promises to lead to better decision management.
It appears that companies are increasingly realizing that to avoid data tsunamis and gain value from the almost endless sources of available data, they may need re-evaluate how they manage and drive value from their data.