Ghosh’s point was not to judge whether that is a fair assessment, but to argue that the emerging trend of predictive analysis could represent a major shift in how HR is perceived and managed — namely, less art, more science:
As more and more money is being spent in the areas of developing and acquiring talented people, organizations are no longer content with fuzzy ideas about their ‘Return on Investment’ (ROI). Forget return, some large organizations want to predict how people will behave before they spend their money on them. This is causing a few pioneering firms to look at data analytics and predictive analytics. Starbucks, Limited Brands, and Best Buy, can precisely identify the value of a 0.1 per cent increase in engagement among employees at a particular store. At Best Buy, for example, that value is more than $100,000 in the store’s annual operating income.
Certainly the metrics of HR, and the bigger question of how a department’s results can be measured, has been an issue that has received more and more scrutiny over the last several years. Some companies already say they are seeing results from the implementation of predictive analytics. In the article “Predictive Analytics Improves Staffing and Retention,” Dr. Jac of Human Capital Source writes about UnitedHealth Group (UHG) and its foray into the field.
Managing 3,200 hospitals and some 340,000 doctors, UHG is one of the largest health-care companies in the U.S. Looking for a more effective talent management strategy, UHG implemented predictive analytics into its day-to-day HR operations, and the company reports that it’s had a significant impact. Writes Dr. Jac:
By using predictive analytics and having metrics available on a global scale, UHG is now better equipped to improve the quality of hire and staff retention worldwide. ‘Quality of hire is important for us. Having access to this data allows us to predict workers who are a better fit for the organization and become more productive quicker. We are now able to leverage our data and make stronger business decisions,’ says [International Recruitment Operations manager Michelle] Fernando. To improve HR staffing processes, HR developed a central reporting portal for all international reports. This enables staff to leverage predictive metrics and access a single set of global indicators…For dashboard reports, each metric has a description and a benchmark goal, and all metrics have a link to the detailed report where the measure originated. This enables management to better determine staffing costs, staffing cycle times, and overall productivity.
As these new models flourish, however, some HR experts make the point that not everything of value in HR can be quantified — no matter how sophisticated the analytics. For instance, though she’s an advocate of HR metrics, “HR bartender” Sharlyn Lauby wrote an interesting piece about the limitations of ROI when it comes to training issues:
I listened to a speaker recently who challenged my thoughts about the importance of calculating return on investment. His thought was alignment and impact of training are more important than ROI. My first reaction was to call bravo sierra on this one but after noodling it over, maybe he’s right. The function of training and development, like human resources or any other department, should be aligned with the organization. Ultimately, this means training programs need to be aligned to an organizational goal. If they are, then the results should have a positive impact on the business.
Perhaps the real issue is when and how to use data analysis, whether ROI or predictive analytics, effectively. In anarticle on HR metrics, Where Great Workplaces Start poses four questions that need to be considered when applying metrics:
What metrics are most important to the organization?
What data needs to be gathered or tracked to calculate these metrics?
How will the data be analyzed and benchmarked?
How will the analysis be used for action planning, development/improvement, and problem-solving?
Whether an organization is using newer or more more traditional models, such questions will always be a good place to start.