Analytic Talent: More Beatles, Less Unicorns

There has been much recent ado about hiring and retaining “good” data scientists, (whatever that means).  Even the emerging notion of what it is to be a “data scientist” remains in question.  Some would take issue with the term “scientist”, others the idea that this is a new activity at all.  Semantics aside, data management has become a fundamental part of doing business.  There are new roles for Chief Data Officer and Chief Analytics Officer at many organizations.  Analytic talent and tools are being applied across the continuum of what organizations do, from supply chain to customer research, to HR and workforce and more.  

Amidst all of this rush to “the new” is an evolving realization of what is working and what is not.  One recent survey of 300 IT department employees suggests that about 55 percent of analytic projects are abandoned. [1]  The reasons most cited were: a failure to understand the true scope of the project and a lack of knowledge of the business.  The first relates to asking good questions, the second to organizational strategy.

Part of this failure to may lay in the conceptualization of the field itself.  Many companies are rushing into an unknown driven more by hype than by measurable results.  The “we’ve gotta get us some of that analytics and data science” mentality has plagued many companies who shop for skills (R, SAS, Python, Hive) over more adaptable and creative problem solvers.  Others have suggested that the competent data scientist is the do-all, be-all, end-all utility player who can seamlessly walk from knowing what the question is that is being asked, what should be asked, understanding the computational back-end abilities, writing the code, doing the correct analysis, applying the appropriate scientific rigor, and presenting it in Ted-Talk like precision to the benefit of the bottom line. 

Unicorns to be sure. 

Now that might be an embellishment, and there are no doubt some individuals in the world who meet this description, but in our zeal for this new and emerging role or set of roles, we have forgotten some very important tenets.  People are still people and just like in the real world, they are different and bring different skills to the table.

This past summer (2014), we ran an experimental summer analytics institute where we placed students into fairly random teams.  The students were comprised of those from industry, PhD students, Masters trained students and some current undergraduates.  They were from fields as diverse as health care, statistics and math and history.  One team, comprised of a PhD student in history, a bachelors level environmental science graduate, and a graduating business student from the music industry tackled a perennial problem for Universities; retention-or success of the student in staying at the institution.  It was a question none of them had any exposure to other than their own school experiences.  On top of that, they had four weeks to complete it.  To make a long story short, they toiled with some messy and diverse student data, and slogged through to make some meaningful recommendations.  And they did.  One that struck many of us was that they found that there was one dorm, that if someone lived in as an enrolled freshmen and did not do anything extracurricularly but go to class, had a greater than 50% failure rate (of leaving the University by the end of year one).  To support this they had some great visualizations, heat maps, flow data and the like.  But what struck me was that when asked how they came to this conclusion, each pointed to the others.  It was Joe’s visualization skills, or it was Jim’s computations, or it was Marcy’s history context (names fictional).  And it was clear that they wouldn’t have gotten to the same point with any other group.  Jim, the music industry focused student, used analogies of the performance.  They were a group.  They were more The Beatles than Unicorns.  Take one out and the music isn’t the same. 

It’s all about the team. 

It’s not enough to hire good analytic talent if they are going to be thrown into the old operational systems.  Team dynamics are key to most organizational strategies, so why shouldn’t that hold true in Data Science and Analytics, which are embedded in an organizations’ core strategy.  It is not surprising that over 50% of projects fail due to poor strategy alignment. 

Good analysts are good problem solvers.  They are intellectually curious.  They are solving human problems that are multi-dimensional and rooted in everyday life.  Why does someone wander through the paint isle before purchasing the nails they needed?  Why do some people drive in circles for five minutes looking for “the” parking space while others take what’s there and walk further?  Why do some students attend one University only to transfer to another quickly while others stay and graduate?

The questions are people questions and the answers require teams with vast perspectives and good conceptual and problem solving skills.  Creating and nurturing successful analytic talent will be akin to promoting good teams and providing them with an environment to succeed (the focus of another discussion), but suffice to say that utilizing teams is not new to industry[2].  What has been more recent of focus is ensuring that there are good team dynamics that foster needed innovation, or what some call value congruence[3]. 

In short, those looking for analytic talent need to be thinking less like Grimm’s fairy tales, and more like Rick Rubin.

[1] InfoChimps 2013.  http://visual.ly/cios-big-data

[2] Lawler, E. E. I., Mohrman, S. A., & Ledford, G. E. J. (1995). Creating high performance organizations: Employee involvement and total quality management. San Francisco, CA: Jossey-Bass.

[3]   Mitchel, R. et. al.  2012. Perceived value congruence and team innovation.  Journal of Occupational and Organizational Psychology 85, 626–648