What Do the Database Marketing Spin-offs Mean to Market Research?

From Merit Job-Recruiting to Tracking Ad View on Web Pages

race track divider line
The Race Is On To Co-Opt Big Data Before The Regulations Catch Up. Ben Joossen © 2004 Stock.xchng.com

Work Force Science is not the old Taylorian stopwatch approach to getting the most out of employers. In fact, the new workforce science is a contrarian approach that eschews management theory and could well put the human resources departments out of work…save, perhaps, for the Benefits function.

It is human nature to consider the past behavior of potential workers as an important marker for future performance.

But, research conducted by workforce scientists has positioned them in alignment with the Securities and Exchange Commission (SEC) that famously provides the following required disclosures to fund investors: Past performance is not a guarantee of future performance. Flying in the face of conventional wisdom, workforce scientists argue that employers should not overly consider attributes such as job-hopping or periods of unemployment when making a hiring decision. And these workforce scientists can back up their argument with data. Lots of data. Big Data.

The workforce research conducted by big data scientists substantially underscores the strength of the relationship between quality supervisors and employee performance and tenure. A supervisor with strong communication skills and personal warmth has been found in a preponderance of research studies to carry more weight than the individual characteristics and work experience of the employees.

These findings turn managerial understandings upside down. Step-wise processes, which are employed in regression analysis, have been adapted for use in human resource guides on recruitment, hiring, and promotion. But these step-wise models are being questioned in statistical modeling – and in a plethora of applications, such as traditional management strategies.

Big data proponents say that these models don’t properly reflect uncertainty and that gut feel can’t be expected to fill the gaps.

Labor force-related market research indicates that if left to their own devices, managers (advised and, often, constrained by human resource departments) go badly awry. For example, managers tend to hire people who are like them in some important ways (gender, age, alumni status, team affiliation, recreational interests), all of which are essentially unrelated to job performance. What this can mean over time is that a firm can substantively skew its workforce toward a particular type of employee who is fundamentally a clone of his or her boss. While this situation contributes to greater comfort among employees, it does not guarantee that job performance will be better because of these similarities. In fact, the opposite can be true. High levels of homogeneity can result in a group think mentality that can be disastrous. Examples of failures of this type include the problem with the O-rings on the ill-fated Space Shuttle Challenger, the heavy investment in credit swaps in the 2008 fiscal meltdown, overconfidence of quants in their algorithms, and – for the historians – the Tulipmania of the 1600s.

Moreover, the numbers of potential employees that can be reviewed using big data techniques, in comparison to conventional human resource processes, is enormous. As Moneyball showed, all the digital activity of people can be collected at relatively low cost and that data mined for insights about skills, communication, and work attributes. Digital trails are constructed by phone calls, instant messaging, emails, webpage clicks, and written code. Digital natives, in particular, seem unconcerned about the consumer activity trails they leave behind. For firms in a hiring mode, these easy-pickings are a boon to recruitment and hiring decisions.

Gild is a start-up company that uses unstructured big data to automate the discovery of talented programmers. Examining the digital evidence of real-time participation in programming discussion groups and Open Source projects, Bu looking at their public code and social networking activity, Gild seeks to quantify what people can do and how they perform – often while simply following their own interests or chasing their own particular muses.

In a recent article in The New York Times, How Big Data Is Playing Recruiter for Specialized Workers, Matt Richtel wrote: People in Silicon Valley tend to embrace certain assumptions: Progress, efficiency, and speed are good. Technology can solve most things. Change is inevitable; disruption is not to be feared. And, maybe more than anything else, merit will prevail.

Kenny Mendes, head of recruiting at Box claims that Gild has consistently given us new candidates that we know are good, but wouldn’t have found elsewhere – the hidden talent, so to speak. Gild’s Vivienne Ming, a chief scientist at Gild, argues that Silicon Valley is not as merit-based as they tout themselves as being. Ming argues that Silicon Valley’s recruitment and hiring practices result in strongly talented, if somewhat maverick, people are misjudged and ignored to the degree that substantial numbers of great performers fall through the cracks.

Perhaps Gild makes the case, too, for the importance of qualitative data. Without the skepticism (a decidedly qualitative variable) of scientists like Ming and Gild founder, Luca Bonmassar, the traditional walls of the human resources silos would not have been breached. Come to think of it, Google's people analytics specialists say the company considers its people decisions to be as important as its product decisions. Google relies less on numbers and grades and degrees when hiring that it did in the firm's early days.