In anticipation of his upcoming Predictive Analytics World for Workforce conference presentation, Too Late? Too Soon? Using Predictive Modeling to Gauge the Timing and Consequence of Retirement Choice, we interviewed Haig Nalbantian, Senior Partner, Well-Known Authority on Human Capital Measurement and Management at Mercer. View the Q-and-A below to see how Haig Nalbantian has incorporated predictive analytics into the workforce of Mercer. Also, glimpse what’s in store for the new PAW Workforce conference.
Q: How is a specific line of business / business unit using your predictive decisions? How is your product deployed into operations?
A: Since the early nineties, our Workforce Sciences team has been a pioneer in the development and application of advanced workforce analytics to inform comprehensive workforce strategies and/or to facilitate problem solving. We have worked primarily with CHROs, Heads of Talent and Rewards, as well as top D&I Leaders to support workforce decisions and advance the practice of evidence-based management in their organizations. We have also delivered solutions to COOs, CFOs and/or Business line leaders to address pressing workforce issues, including those that cross over into risk management.
An area of application relating to operations concerns the issue of optimal spans of control. Traditional approaches rely on benchmark data to help determine if a given organization has opportunities to economize on middle management, streamline operations and hierarchy and raise spans. But these approaches fail to account for the impact of middle managers and their positions in the hierarchy on workforce productivity, career development and employee retention. Our predictive modeling tools help organizations bring a workforce perspective to the question of optimal spans, helping to avoid the frequent unintended, negative consequences that emerge from relying on operational benchmark data alone.
Q: If HR were 100% ready and the data were available, what would your boldest data science creations do?
A: In the best of cases, all larger organizations and major employers would practice evidence-based workforce management. They would bring an investment “mindset” to human capital. For many organizations, workforce expenditures are the single biggest investment they make but the one they know least about. That is untenable in today’s economy where human capital management is a singularly important, demonstrable driver of organization value. There are simply too many considerations to navigate using intuition, philosophy, benchmarks or so-called best practice comparisons alone.
Good analytics helps identify and prioritize critical action areas and helps leaders focus on high-yield actions and investments. We are living in the age of analytics and “big data” – the data and analysis methods are now in place to transition HR into an asset management function. In fact, many organizations have embraced evidence-based management and are moving quickly to build “in house” workforce analytics functions. I believe the move in this direction is inexorable and predictive modeling and big data algorithms will soon become the foundational tools of modern HR. Our boldest data science creations do just that: they transform the mindset of HR and business leaders with respect to human capital and enable them to tackle the challenge of managing a complex system of practices and influences that actually create their workforces.
Q: When do you think businesses will be ready for “black box” workforce predictive methods, such as Random Forests or Neural Networks?
A: Frankly, I am wary of “black-box” solutions and am concerned that some businesses are relying too much on them already. In my judgment, such methods are useful for discrete functions or transactional decisions but not for more significant strategy making. For example, while they may be of high value for recruiters seeking to fill high volume jobs, they are unlikely to be the basis for creating an effective recruiting strategy. Strategy making requires understanding the relationships among observable data in a way that permits development of a story-line that can compel action. In my view, when it comes to workforce strategies, explanatory analytics that transparently estimate the strength and robustness of associations between variables of interest trump black-box models based on purported predictive validity.
Q: Do you have suggestions for data scientists trying to explain the complexity of their work, to those solving workforce challenges?
A: I have spent a career battling against a mindset that rejects “complexity” even if that complexity will demonstrably improve results. For example, the rush to so-called “pay-for-performance” regimes has created unexpected problems in many organizations because they failed to develop performance measures that properly differentiate between actual employee performance and what we call performance risk – in a word, between performance and luck. Sometimes you need complex measures to get to the “truth” but the received wisdom in HR and, frankly, in the C-Suite, is strongly biased towards “simplicity.” So that’s how you ended up with so many organizations using plain vanilla stock options rather than more complex but more effective “indexed options” to be a prime vehicle for executive incentive compensation, leading to costly misallocations of performance risk and inherently non-economic incentive structures.
There’s one industry that might set an example for the rest of us – Major League Baseball (MLB). After years of resistance, MLB succumbed to the rule – perhaps the tyranny – of Baseball’s version of workforce analytics, “Sabermetrics.” Why? Because it works. Small market, low-budget clubs can now compete with large market, high-budget ones because they use data intelligently to determine what player attributes and field strategies drive wins and have developed new measures that more accurately gauge actual player performance. It is almost a miracle to me that Baseball now routinely trucks in these new, often complex performance measures that many front office executives, field managers, coaches, scouts, players, sports journalists do not really understand. They may get the general idea, but they don’t grasp the measurement methods behind them and certainly could not calculate these measures themselves, the way they could the simple conventional measures of old. But they accept them.
So what do decision makers in Baseball know that the rest of us don’t? Are their executives smarter than ours? Are their field managers and personnel leaders more astute? I don’t want to think that. Proof of value has helped a lot. And perhaps a more venturesome spirit envelops the culture of the game that has been seriously challenged by the growth in interest in competing sports. In any event, this is something I am following closely because there is much to learn from their experience.
We need to do the same in business generally. There is nothing wrong with complexity so long as the complexity significantly improves key outcomes. The onus is on us to prove that’s the case. We need to chip away at the wall of suspicion that too many in our field have erected. We have to stop being defensive and be confident and proud of where we’re taking workforce management. Making it a more scientific discipline is something that benefits management and labor alike.
Q: What is one specific way in which predictive analytics actively is driving decisions?
A: A classic example relates to unwanted turnover. In the past, the approach to understanding unwanted turnover was to rely on exit interviews. I don’t need to repeat why the results of such interviews may be flawed or misleading. I/O Psychologists have taught us a lot about this issue. As an economist, I would simply add one thought that could be a killer for exit interviews, namely, there is a fundamental difference between “before-the-fact” (“ex ante”) and “after-the-fact” (“ex post”) motivation for actions. Telling you why I left after I have found alternative employment (or leisure) says nothing about what motivated me to search or even be open to alternatives. The only thing I really care about as an employer is “ex ante” motivation. “Ex post,” it’s too late and what the leavers tell me then may speak little to what I need to do to change employee choices “ex ante.”
Fortunately, we have an answer. Using predictive modeling to identify and measure the antecedents of actual stay/quit decisions is most helpful in addressing this problem. While the stories “leavers” tell us may offer color that helps communicate the reasons for policy change, knowing what actually predicts behaviors provides the basis for defining what those policy changes should be in a way that actually improves retention. And good predictive models can be adapted to attach flight risk scores to individual employees, providing a new management tool to supervisors and line leaders. With so many in management taxed for time, can you imagine how helpful it is to be given effective ways to prioritize what to focus on and how to achieve results? We have done this kind of work with quite a few clients over the years. The record of results is compelling.
Q: How does business culture, including HR, need to evolve to accept the full promise of predictive workforce?
A: I think I largely answered this question above. Human Capital Management is simply too important to business success not to use the discipline of Workforce Sciences to inform decisions about the massive investments many organizations make in their workforces. Competing in a world of globalizing labor markets makes its reckless to make location, recruitment and selection decisions without hard data and the requisite interpretative capability. Would you make capital investment decisions that way? Of course not. Today, decisions about human capital require the same kind of disciplined support. It helps enormously if those in leadership positions within HR have some formation in quantitative methods. They don’t need to be econometricians, psychometricians, or data scientists, but they do need to feel comfortable with data and know enough to ask the right questions and detect the risks of confusing correlation with causation. Just as there is something called business acumen, there is something called quantitative acumen. A good HR leader today needs both. When they have it, they can lead the kind of culture change that challenges dismissive or cynical attitudes to the people side and earns their function the standing that managing such a large and key asset warrants on its face.
Q: Do you have specific business results you can report?
A: We have many results to share. I refer the reader to our book Play to Your Strengths (McGraw Hill, 2004) and to other publications like “How Fleet Bank Fought Employee Flight” (HBR, 2004); “Making Mobility Matter,” (HBR 2009). “A Big Data, Say/Do Approach to Culture and Climate: A Consulting Perspective, Oxford University Handbook on Climate and Culture (2014), to name a few. These publications review many examples of work with client organizations where predictive modeling identified significant opportunities and looming risks and led to policy actions that demonstrably improved results.
Don’t miss Haig’s conference presentation, Too Late? Too Soon? Using Predictive Modeling to Gauge the Timing and Consequence of Retirement Choice, at PAW Workforce, on Monday, April 4, 2016, from 11:30 am to 12:15 pm. Click here to register for attendance. USE CODE PATIMES16 for 15% off current prices (excludes workshops).