Predictive analytics enables users to extract information from historical data and use it to predict future outcomes. One of the most effective ways the insurance industry is leveraging predictive analytics is to gain valuable insight to make effective employee staffing decisions.

Predictive analytics provides insurance executives with critical insights-not normally available to them via traditional methodologies-that can help improve the efficiency of operational issues, such as employee staffing.

Staffing decisions in the insurance domain tend to be demand-driven. For example, in a customer service support operation, an organization needs to make decisions around how many employees are required for different shifts. Factors such as volume of calls coming in and the types of calls being placed are key data points that help facilitate staffing decisions. This requirement extends to other areas dependent on volume, such as underwriting or claims administration.

Effective staffing decisions can be made if quality estimates regarding demand fluctuation are available. In the customer service call center context, this might involve predicting how call volume varies over the course of a day, day of the week, time of month or by month. Since the skill sets needed vary by call type-marketing versus claims, for example-this analysis would need to be performed separately for individual call categories.

A common staffing challenge organizations face is in predicting the numbers required. An insurance provider might require a quarterly projection of demand to make hiring decisions. However, weekly projections might be necessary to make decisions related to staff work schedules, and half-hourly level projections might be needed by shift managers to make the right decisions on a day-to-day basis.

The need for these numbers at different levels of aggregation poses some challenges to an organization. One could roll the data up to the highest level (quarterly, for example) of aggregation and perform analysis at that level. Lower-level numbers (weekly and half-hourly) would be obtained by using some splitting rules. This is commonly referred to as the top-down approach.

Alternatively, the forecast could be generated at the lowest level (half-an-hour slots, for our example) and higher-level numbers could be obtained by rolling up these numbers. This is called the bottom-up approach.

A top-down or bottom-up approach will always be less accurate at one of the levels. One could use independent forecasts for the different levels, but this will lead to internal inconsistencies in numbers. A hybrid approach can help resolve this problem. The hybrid approach will:

Generate lower-level forecasts using granular data
Generate top-level forecasts with aggregated data
Use the lower-level forecasts to generate proportions, which are then used to split the top-level number.

This approach is useful because it takes advantage of the higher accuracy of the top-level numbers, and then factors in some of the lower-level information by calculating proportions from lower-level forecasts. The use of predictive analytics to develop a hybrid forecasting approach will allow insurers to better manage staff schedules to deal with fluctuations in call volumes, and serve their customers better and in a more cost effective manner.

Elements of a Successful Paradigm

There are key factors to consider when designing a framework to include predictive analytics in the staffing process. A successful strategy involves a number of components, including having the right type of data available for analysis, access to the right tools and techniques, as well as a solid understanding of implementation capabilities and limitations.

Before any analysis can be performed, the data asset needs to be created. While the data elements are being put in place, one needs to start thinking about tools and techniques needed to perform the analysis. Organizations take different routes to accomplish this step. One approach is to build the analytical capability in-house by hiring trained modelers and acquiring the tools needed to perform the analysis. Another approach is to outsource this process to external vendors. Both methods have their merits.

Finally, a successful paradigm should account for all the constraints and needs of the implementation environment. One needs to be cognizant of these limitations upfront, and then design the analysis plan accordingly.

This article can also be found at

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access