Innovations in data science are enabling the transformation of HR.
By Jeff Mike, James Guszcza, and Kathi Enderes
Underneath buzzwords like “disruption” and “digital transformation” lie some important truths for HR leadership. There is no denying that powerful technologies aimed at individual consumers have changed the game. The best of these technologies deliver compelling, personalized experiences to customers through digital platforms, smartphones, and increasingly, augmented and virtual reality. As a result, they have created a demand for similar personalization of work experiences and workplace applications.
These consumer and workforce technologies generate a tidal wave of data. This data is so valuable that the American Bankers Association has started to refer to it as “a currency of the modern economy.” Of course, one of the first implications of this influx of data is the protection of it and the privacy of the individuals who generate it. By now, the public debate over data privacy and the implementation of standards like the EU’s General Data Protection Regulation (GDPR) have made data security a priority for HR, IT, and corporate executives around the globe.
All this data raises another question: How can it be used to generate value? The answer lies in data science. The standard benchmarking and employee surveys HR has relied upon for decades help measure HR activity and understand the workforce, but only provide a snapshot of what’s going on at a given time. At worst, this approach provides out-of-context, static information for managing the workforce and the business in a dynamic environment. Sensing what’s going on in the workforce in real time is crucial for change. Fortunately, data science offers the promise of predicting and shaping behavior in this environment, enhancing the productivity and health of the workforce while driving key business and financial outcomes.
HR leadership must consistently look for innovations to realize the potential of data science, with tools that enable productivity by sensing the external and internal factors that are occurring in their workforce. According to Bersin’s High-Impact Human Resources research, the highest-performing HR teams are “pioneering and personalized.” Pioneering indicates that high performers are carving their own paths into the future based on their organization’s particular market, strategy, maturity, and culture. Personalized means that high-performing HR is treating the workforce like customers—sensing and segmenting the behaviors and mind-sets of the workforce, then designing offerings for maximum personalization and impact.
A key finding of the research report is that high-impact HR teams are partnering with other business units to understand what’s taking place, what’s driving employee engagement, and which tools and processes are enabling success. For example, HR can partner with IT and finance to create a more consistent employee experience with organizational support functions. Increasingly, these high-impact HR organizations are collaborating with product, marketing, customer service, and sales teams to focus their specialized expertise on the organization’s talent. In this way, HR should work closely with data scientists to better sense, understand, predict, and shape behavior of the workforce to achieve organizational strategies.
Data Science for HR
Data science is awash in its own buzz: Data is the new oil; artificial intelligence is the new electricity; and machine learning tools will democratize data science. Certainly, the availability of big data and machine learning tools give good reason for enthusiasm. People continually leave behind trails of data as they go about their digitally mediated lives, both on and off the job. All of this data can be stored and processed more cheaply than ever before and used to make all manner of predictions. Not to mention, easily available, user-friendly tools empower data scientists in ways they could only dream of not long ago. All of this makes for a fertile environment for data science innovation in HR and beyond.
Powerful HR applications abound. Email and calendar metadata enables HR to go beyond organizational charts by piecing together network maps to better understand flows of information and collaboration. Wearables and sensors are capable of measuring how well people sleep, whether they are getting enough exercise, and their levels of stress or engagement throughout the day. Social media can be scraped for clues about employee engagement. Continuous pulse survey and collective intelligence tools enable HR professionals to crowdsource ideas to better understand and improve employee engagement. Also, algorithms can be built to predict high-performing recruits, match people to jobs, estimate time away for injured workers, and model attrition. These applications only scratch the surface of what’s possible.
The above possibilities reflect conventional discussions of data science in HR. They focus on polling the workforce and leveraging data sources, analytical tools, and laundry lists of point solutions. This is a good start but it is important not to equate people analytic sophistication with specific types of quantitative technical sophistication, such as information processing or machine learning.
To illustrate, consider the challenge of using data to make better hiring decisions. It is increasingly practical to train complex machine learning algorithms on rich datasets to identify high-performing recruits. While valuable if done well, this alone does not constitute sophisticated people analytics—it merely begins the sensing process. This starts with recognizing the pitfalls of relying on unaided judgment about information derived from unstructured job interviews. Humans have a hard time weighing together five variables, much less 50 or 500; they are liable to make different decisions before lunch than after lunch; and they fall prey to “thinking fast” decision traps, such as halo effects, over-generalizing from personal experience, and unconscious bias. Unbiased algorithms can help overcome the limits of human judgment, as long as the data sets used to train them aren’t inherently biased themselves. Training machine learning algorithms on biased datasets and deploying them can therefore amplify, rather than mitigate, the cognitive biases.
An alternate way is to use data analytics to select job interview questions and to apply behavioral design principles to de-bias hiring environments. Such data-driven and psychologically informed interventions help take some of the human biases out of hiring. Tapping into the wisdom of crowds by having multiple interviewers follow the same structured procedure eliminates some of the noise from the process. For example, the noise created by one interviewer’s bad mood will tend to get averaged away by the uncorrelated judgments of the other interviewers. Not surprisingly, Bersin’s High Impact People Analytics study reports that organizations that use people insights well are 10 times more likely to see a significant positive outcome from that data on hiring.
This specific use case of data-driven hiring illustrates the more general point that people analytics goes beyond the routine application of machine learning techniques. It should be conceived broadly to incorporate methods from data science, the psychological and behavioral sciences, and the use of experiments to figure out what works. People analytics is ultimately about adopting a culture of evidence-based decision making and using the scientific method to enable better people-related decisions to design personalized work environments in which employees can perform at their best.
Beyond individual people data, HR can benefit from data science on the workforce level. Automation and smart machines will change the way work gets done, augmenting humans with machines and fundamentally redefining work. Without data and insights, designing the workforce of the future is a guess at best and misleading at worst. With data science, HR leaders can forecast complex scenarios that consider automation possibilities, a broader workforce continuum beyond on-balance sheet employees, and a skills pool potentially unencumbered by geography. This can provide needed direction for accessing evolving and needed skills, countering the (much lamented) skills gap through proactive actions ranging from traditional development (“reskilling the workforce”), partnership with academia, joint ventures, mergers and acquisitions, and even crowd work. The future of work is full of thoughtful, integrated solutions that make the best of available technology tools.
A Talent Mandate
Exponential change, powerful technologies, and emerging skillsets have created a talent mandate for HR leadership. Never before has human talent been such a differentiator and so critical to an organization’s market success, and never before has the right talent been so hard to find and keep. In this talent-constrained world, designing productive and compelling work and experiences for all workers has become a business necessity. In fact, according to Bersin, high-impact HR organizations are 3.5 times more likely to focus relentlessly on creating an engaging workforce experience when designing HR offerings as low-performing organizations.
HR leadership needs a new mind-set and new tools to take on these challenges and to thrive in a new paradigm for managing enterprise talent. Without targeted insights on people’s needs, expectations, performance, and skill sets, the tsunami of data available today remains untapped. To derive actionable insights and shape productive, healthy behaviors, all this data needs to be integrated, analyzed, visualized, and summarized.
Encouraging leadership to obtain a better sense of the workforce and the factors affecting it is a first step. Sensing is also key to helping reduce noise from these sources and unlock insights for creating business impact through people. The most efficient and effective way to build this capability in HR is to tear down the silos and work closely with data scientists and the rest of the business to understand and define the future of the enterprise, the future of the workforce, and the future of how work gets done.
Jeff Mike, EdD, is vice president and head of research ideation and Kathi Enderes, PhD, is vice president and talent and workforce research leader for Bersin™, Deloitte LLP. James Guszcza, PhD, is chief U.S. data scientist at Deloitte Consulting LLP.