Among the wide spectrum of business units that have an application for analytics, human resources (HR) might not stand out as an obvious area that benefits from data-driven decisions. But as HR increases its strategic presence at the executive table of influence, its need for analytical, fact-based approaches is on the rise.
Kristie Evans, founder of HR Logistics LLC, talks about how analytics, when used to manage company health care, results not only in a better quality of life for employees but also a reduction of health expenses. Such financial impact solidifies the strategic role for HR. We asked Evans a few questions on the matter – which is explored further in the article, Predictive Modeling – The Yellow Brick Road to Strategic HR, originally published by the International Association for Human Resource Information Management.
In your article, you go into great detail about applying predictive modeling to manage health care risk and expense. What other predictive modeling applications have you seen that allow HR to impact the organization’s bottom line?
Other examples of the uses of predictive modeling include analytics that compare turnover, tenure, and performance evaluations. This type of analytics can reveal the average tenure of an employee and – through a series of questions – the environment and attributes which contribute to an employee’s decision to terminate from an employer.
For example, a revealing correlation might be the onboarding processes and the percentage of employees that terminate if the onboarding process is weak or missing important inputs to support the employee’s integration into the organization. In this example, an organization could predict the percentage of turnover based on criteria and the potential expense of recruitment associated with that process if left unchanged.
Another example of predictive modeling is the use of assessment tools to “pre-assess” a candidate’s potential to be successful within a specific role within the organization. These tools are already providing a “predictive” look at the candidate skills and abilities by modeling their responses against the best scenario for success. My company, HR Logistics, is also working on a next generation of these tools which takes the predictive process a step further.
What is your estimate of the adoption rate for companies using predictive modeling to analyze health care data? Any idea on the growth rate?
My assessment of the market is very rudimentary as I have not conducted research that would give me a more dependable representation. But I do not think companies have embraced the use of predictive modeling of healthcare data to any great extent as yet. The surface has barely been scratched and most of the work utilizing this data is being done by vendors who sell this service.
I think this is has two drivers behind it – analyzing the data requires some specialized software and the ability to focus on data analysis which may be difficult to resource in house, and more importantly, private health information is highly sensitive and in a very protected state legally, therefore working with it would potentially expose an organization to litigation and increased risk that would far outweigh benefits of the analysis.
You mentioned that Pitney Bowes, a software and services company, and United Medical Services, a mobile medical solutions provider, agree that “it takes about two years to yield a positive ROI on the use of predictive modeling, at a minimum, and it is imperative to invest in a strategy, not just implement and expect results.” How have companies reacted to this expected time-to-results?
I think the extended time required for the ROI may be a deterrent to adoption of the strategy. It can be difficult for a conceptual ROI i.e., an ROI that is intangible for an extended period of time, to be embraced. Business changes rapidly and often, a company is shopping each year to negotiate the best healthcare premiums. In order for that 2-year timeline to yield results, the company becomes more committed to a healthcare provider because a partnership with the provider is necessary to get the data necessary for the analysis. It is almost impossible to receive data for a fully insured plan and data for self-insured plans requires a partnership between the organization and their healthcare provider.
In addition, plan benefits are often renegotiated each year which changes the configuration of the data. So it is important to choose data elements that can be tracked consistently over time to yield results that have a consistent correlation.
Last, I think organizations are often short sighted and want more immediate gratification for the effort. It can be a difficult sale when the benefits of the projects require such a long commitment of time and resources.