Predictive Analytics Times
Predictive Analytics Times
The Art of Data Science
 With much of the latest discussion...
Wise Practitioner – Predictive Analytics Interview Series: Ashley Walsh at LeanTaas
 In anticipation of her upcoming conference...
Twelve Hot Deep Learning Applications Featured at Deep Learning World
  For today’s leading deep learning...
New Book: Stephen Few’s “Big Data, Big Dupe” Smackdown
 Five years ago, in 2013, two...

1 month ago
Wise Practitioner – Predictive Analytics Interview Series: Haig R. Nalbantian at Mercer


In anticipation of his upcoming conference co-presentation, What Millennial Employees Actually Value: Lessons from Predictive Modeling, at Predictive Analytics World for Business Las Vegas, June 3-7, 2018, we asked Haig R. Nalbantian, Senior Partner, Co-leader Mercer Workforce Sciences Institute at Mercer, a few questions about his work in predictive analytics.

Q: In your work with predictive analytics, what behavior or outcome do your models predict?

A: Our Workforce Sciences team has been applying advanced analytics to inform workforce management decisions since the early 90s. We work with two broad classes of models: the first, Internal Labor Market (ILM) Analysis® is focused on identifying, quantifying and prioritizing the drivers of critical workforce outcomes such as retention, advancement, performance and pay. The second, Business Impact Modeling®, is aimed at identifying, quantifying and prioritizing the workforce drivers of important business outcomes, such as unit profitability, productivity, customer retention and growth. We statistically estimate all or some of these models depending on the nature of the client engagement.

For instance, in a workforce planning assignment, we use ILM Analysis® to determine if and to what extent the organization is on course to secure the workforce required to achieve business goals. Statistically modeling the drivers of an organization’s talent flows and rewards, we are able not only to project what the organization’s workforce will look like across a range of dimensions – for example, demographics, job families, experience, etc. – but also determine which talent management policies and practices have the biggest impact in helping achieve the outcomes desired. ILM Analysis® delivers insights that reflect primarily the supply side of an organization’s internal labor market.

We use Business Impact Modeling® to help inform the demand side of workforce management. Specifically, by statistically modeling the running record of business performance, we are able to identify which workforce characteristics and management practices drive business value. Results from such models help the organization determine what its workforce should look like and how it should be managing that workforce to derive greater value from its investments in human capital. These results can then be compared to results from the ILM Analysis® to help identify and measure likely gaps and determine how best to close them. In this way, strategic workforce planning is built on a solid empirical foundation and is not purely an exercise is speculation or worse.

In other instances, we deploy our predictive modeling methods to address more discrete issues such as excessive unwanted turnover, the quality and health of the talent pipeline, the effectiveness of performance management, pay equity, workforce productivity, among others.

Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?

A: As consultants we deliver diagnostic work and associated recommendations to client organizations. For many years, our own company and multiple sister companies have routinely applied these analytics to guide their workforce strategies and management. The work has had significant impact.

As an example, in one of the organizations, an ILM Analysis® demonstrated serious problems in the use and management of internal talent mobility, whether across functions, geographies or business segments. Specifically, those experiencing lateral moves of these kinds didn’t fare well. All else being equal, they were less likely to receive high ratings and often less likely to advance in their careers – certainly no more likely than their less mobile counterparts. And they were significantly more likely to leave.

Frankly, this is a very common finding in our ILM work with client organizations. Getting mobility right seems to be a very hard thing to do. But for this company the problems observed with mobility were particularly noteworthy because enhanced mobility was a central component of the company’s expressed talent strategy. Leadership had documented evidence from the results of Business Impact Modeling® that breadth of experience in client teams was a prime driver of revenue growth. As a lynchpin of the company’s business strategy was to capitalize on synergies across business segments – for example, creating cross-segment client sales and delivery teams – having more employees with direct experience in multiple service lines would be particularly useful for achieving their strategic goals. Similarly, to support global growth, having more employees with experience working in different geographic regions was especially important. Yet, the diagnostics suggested the company was failing to achieve either of these outcomes. Moreover, given the hard costs associated with the company’s investment in talent mobility, the failure to nurture and secure its mobile talent indicated a low or even negative ROI. Something had to be done.

Additional diagnostic work revealed that mobility was failing both because the wrong talent was being moved and because clear accountability for mobile talent was lacking. Those on the move were falling through the cracks. No one “owned” them. The units and leaders “exporting” such talent no longer had a vested interest in these employees; those “importing” such talent had less incentive to nurture them as well, at least as compared to their incumbent employees who were likely to remain with them longer-term. The structure of incentive compensation also worked against the interests of mobile talent. The company had increased its reliance on variable pay over base pay, with significant short-term incentive payoffs keyed directly to individual and unit results. Operating under that pay structure, unit leaders had every incentive to hoard their best talent.  And those receiving such talent had little incentive to invest in colleagues whose duration on their teams was limited at best. In effect, the very structure of performance accountability in rewards conflicted with accountability for the firm’s mobile workforce.

Under these circumstances, it made more sense for this client company to embrace its mobile talent as “corporate assets,” whose value extends well beyond their immediate deployment. In face of this classic economic “externalities” issue – where value accrues beyond the unit directly employing the individual whereas they bear all the cost – it becomes imperative to move decision-making and associated investments to a level where the benefit-cost relationship can be more effectively aligned. Company leadership understood the argument and determined to act on the evidence that the current approach to mobility simply was not working. First, they determined to re-orient mobility decisions from their current reactive or “opportunistic” orientation to a more deliberate talent development approach. Second, and pursuant to that shift in orientation, they introduced a formal talent rotation program, where a select group of up-and-coming talent was intentionally rotated through different parts of the business and geographies to meet specific corporate objectives. This talent serves as a corporate or at least regional asset. Though utilized locally for different types of project or sales activity, they are not charged to the local business or geography. As such, there are incentives not only to accept them but to utilize them in a way that optimizes their skills and experience.

Subsequent ILM modeling showed significant changes in outcomes associated with mobility. First, the composition of the mobile workforce was changing as more early-career employees were moving as part of their planned development. This did not preclude opportunities for more experience employees to make moves later in career as fit their needs and company requirements. It just meant that there was more organizational focus on using mobility for development and strategic business purposes. Secondly, there was no longer a success deficit associated with being mobile. Employees making such moves were at least as likely as their less mobile colleagues to receive high ratings or be promoted. And, importantly, they were no longer more likely to turn over. The bleeding of the company’s mobile talent ceased. That, in itself, helped improve the ROI.

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: Our Business Impact Modeling® approach, as described above, speaks directly to ROI. Deploying the micro-economic construct of a “production function,” we statistically estimate the relationship between a variety of workforce characteristics and practices with measures of business performance. Reliance on longitudinal data strengthens our ability to identify causal relationships as does the application of learnings from the research literature that helps formulate and test clearly defined hypotheses. First, applying Business Impact Modeling®, we are able to determine how much of the variance of business performance across units and over time is explained by the people side of operations, as opposed to say, capitalization, plant/facility vintage or technology, customer characteristics and mix, seasonality or other random, environmental factors. So, in one large hospital system, we found that over 60% of the variation in workforce productivity – as measured by value added per employee – across hospitals was accounted for by human capital factors. On the other hand, in a large retail organization, we found that 62% of the variation in store profitability was explained by store characteristics relating to location, physical plant, customer density and income, among other things. That still left plenty of room for workforce management to affect performance. Being able to get a high-level picture of the relative importance of human capital and a potential upper bound of impact can be extremely useful for capturing the attention of business leaders and prioritizing policy actions.

Secondly, this kind of modeling can identify and quantify specific aspects of human capital management that matter most. So, in the hospital system, staffing ratios turned out to be hugely important. Just optimizing use of part-timers would raise workforce productivity by an amount equal to about 3% of revenue. In the Department store chain, optimizing headcount or employee “density” per square foot of retail space turned out to be of primary significance, in essence, reflecting efficient utilization of physical plant. Securing longer service among local managers and supervisors mattered a lot as well; less so for front-line associates. In other retail organizations with which we have worked the exact opposite has been found to be true regarding tenure of front-line workers. Hence, it is very important for organizations to undertake this kind of diagnostic work on itself.

Bottom line: your strategy is your own, your products and services are your own, your culture and your workforce are your own, as is your local environment. How these interact with each other and play out in your own organization is something unique to you. In this age of big data and analytics, there is no excuse for not applying disciplined analytics to uncover the sources of human capital value in your organization. These insights can then inform unique and effective workforce strategies that become a prime and enduring source of advantage.

Q: What surprising discovery or insight have you unearthed in your data?

A: There are so many examples I could reference. For our purposes here, I would go back to the findings around mobility referenced above, as they are particularly striking. Many large, global firms invest a lot in mobility because they believe it is essential to the development of general managers and a vital driver of global growth. They build a significant infrastructure around mobility, particularly global mobility to facilitate the transactional side of such moves. Yet, few actually know if, where and how their investments in mobility are paying off. They rely on survey responses or informed opinion to gauge how they are doing. Frankly, relying on opinion alone just doesn’t cut it any more. Our ILM Modeling work offers a very sobering view about how mobility is playing out in organizations and what kind of return it is actually generating for employers.

This is not to throw cold water on mobility. It is to recognize that managing mobility is hard and organizations that pursue it as part of their talent strategy or even for operational reasons alone need to be far more deliberate about the way they structure their programs. They need to spend at least as much effort addressing the strategic considerations related to mobility as they do building out an infrastructure to execute the transactional side. Specifically, they should answer six key questions:

  • What role should mobility play in their overall talent strategy?
  • Who should be moving and how frequently?
  • Should these moves be “sponsored” by management or “spontaneous” – that is, initiated by requests from employees themselves?
  • What kind of internal moves are most beneficial? Cross functional? Cross business? Cross geography?
  • Who should take responsibility for the careers and management of mobile talent? (Should they be a corporate asset?)
  • Are incentives aligned to support mobility or do they encourage “local” hoarding of top talent?

Purely opportunistic, non-strategic mobility can work in organizations. It may not require lots of heavy duty analytics to execute successfully. But using mobility as part of a talent or leadership development strategy is a very different challenge. It requires careful planning, management and regular tracking if it is to succeed. (See Haig R. Nalbantian and Rick Guzzo, “Making Mobility Matter:  Moves that Develop a Leader.” Harvard Business Review, Volume 87, No. 3, March, 2009).

Our ILM modeling work makes that clear. Our experience suggests that at present, corporate functions responsible for mobility are simply not equipped to deliver on this charge. Strategic mobility seems to fail more often than not. The modeling work also demonstrates the remarkable power of good analytics to inform such strategic decisions and help secure greater ROI.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

My colleague, Tauseef Rahman, and I will be talking about research we have conducted synthesizing findings from turnover modeling across multiple organizations, with a particular focus on the behaviors of Millennials. Specifically, we are focused on identifying factors that most explain and predict the actual turnover behavior of Millennials as compared to their older counterparts. We will get quite specific about key drivers of Millennial turnover and how they differ from older employees.

The one take-away I would emphasize here is that our results are quite different from those commonly seen in surveys of Millennial populations. It turns out that what Millennials SAY they value in the employment relationship may be quite different from what they DO, in fact, value – that is, from what turnover modeling indicates is driving their actual behavior.

This is not the first time we see a divergence between what employees SAY and DO. In fact, that is a common experience in our work, across organizations, across different workforce segments and demographic groups.  This divergence could be particularly important from a policy standpoint because there is so much survey data out there about Millennial “values” and organizations seem to be paying a lot of attention to those findings. Perhaps they should focus more on behavioral evidence of the kind we bring to this question and start using analytics to better understand their own Millennial workforce. They may be quite surprised by what they find.   


Don’t miss Haig’s conference co-presentation, What Millennial Employees Actually Value: Lessons from Predictive Modeling on Tuesday, June 5, 2018 from 10:30 to 11:15 AM at Predictive Analytics World for Business Las Vegas, June 3-7, 2018. Click here to register to attend. Use Code PATIMES for 15% off current prices (excludes workshops).

By: Eric Siegel, Founder, Predictive Analytics World

Eric Siegel is the founder of Predictive Analytics World ( — the premier machine learning conference, with cross-vendor industry events in Las Vegas, Washington DC, London, Munich, and Berlin — and the author of the award-winning book, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die – Revised and Updated Edition, (Wiley, 2016).

Leave a Reply