There are too many chefs in the science lab

Herding
cats
is not as hard as it sounds – if you have the right shepherd.
Sometimes, managing a data science team feels like herding cats. For some
reason, people tend to underestimate the creative capacity that some data
scientists have, which translates to a management challenge if you’re not
prepared. There’s another side to the coin.

Sometimes it seems like you’re giving your data science team
all the direction in the world, but they’re going nowhere fast (and burning
your cash while they’re at it). This is all indicative of a common problem I
see on analytic efforts. When running a strategy or initiative that involves big
data, it’s important to align your strategic objective with your leadership,
management, and approach to data science.

Leaders and managers and data
scientists, oh my!

Misalignment between leadership, management, and approach to
data science is a common mistake that’s easily fixed, if you understand some
fundamentals of why things get messed up. Let’s start with leadership versus
management. This whole topic is rife with confusion – even within management
community! So let’s clear things up.

In simple terms: leadership deals with change; whereas,
management deals with complexity. What leaders do and what managers do are two
seemingly dichotomous functions that drives people crazy. Leaders are
visionaries who keep their eye on the horizon; whereas, managers are planners
who keep their eye on the bottom line. Leaders motivate people with their
influence; whereas, managers direct and control people with their authority.
Good leaders establish an emotional connection with their followers, keep an
open mind, listen more than talk, and reduce boundaries to explore new
territory. Good managers maintain emotional distance, have an expert mind, talk
more than listen, and create boundaries to contain scope. As you can see, it’s
dangerous to assign the right people to the wrong job.

Next, let’s look at strategic objectives and your approach
to data science. When you develop a product or service to offer to your market,
you must decide whether your new offering will be competitive, distinctive, or
breakthrough. Competitive and distinctive offerings tend to be transitional:
you know where you want to go; you just need a good route to get there. These
efforts are best suited for quantitative data science.

Breakthrough offerings tend to be transformational:
you know you need to change, but you’re not yet clear on what the solution
looks like. These efforts are best suited for qualitative data science.
Like leaders and managers, qualitative and quantitative data scientists have
very different cultures (I bet you didn’t know that). Qualitative data
scientists value exploration and are okay with uncertainty and relatively
blunt, crude measurement devices. Quantitative data scientists value structure
and numerical precision, and often criticize qualitative approaches.


Also
read: Define roles clearly because big data success requires it


Aligning the stars

Team structure and development should then follow from your
strategic objective. If you’re shooting for a transitional offering (i.e.,
competitive or distinctive), then your data science team should favor
management.

Let’s say you’re way behind the curve on digital marketing
so you need to create a relationship offering (a service that you don’t charge
for) that’s on par with your key competitors to prevent churn (lost customers)
from the Generation Y market. Since you already know what your competitors are
doing, you have a clear idea of where you’re going with this offering, and you
have a tight but achievable timeline. To achieve this objective, build a team
of quantitative
data scientists that favors management. You still need leadership to keep
people motivated; however, the emphasis should be on planning, organizing,
directing, and controlling. Your quantitative data scientists will appreciate
the structure and clear objectives.

If however, you’re shooting to achieve a breakthrough,
transformational offering, then the team structure must be radically different.

Let’s say you’re already the market leader; however,
competitors are closing the gap and you need a breakthrough innovation to
sustain your market leadership. You’re witnessing the mega-trends – like social
journalism and the emergence of India as an economic power – shift like
tectonic plates, and you need to ride the ensuing tidal waves, but you’re
unsure of where they will take you. Putting a management team on this effort
would be a disaster – you need strong leadership. Assemble a team of qualitative data
scientists who are more comfortable with the upcoming uncertainty, and install
strong leadership. You will need some management to keep things under control;
however, the emphasis should be on team development, change leadership, and
flexibility to navigate unknown and often turbulent waters.

Bottom line

Strategic alignment problems are common, costly, and easy to
avoid once you understand a few underlying principles. You never want to be
devoid of analytic leadership or management; however, the emphasis of one over
the other depends on what you’re trying to achieve.

To make a strategic transition to a competitive or
distinctive offering; a management-centric, quantitative approach that
emphasizes planning, control, and precision would serve you best. However, to
make a strategic transformation to a breakthrough offering, a
leadership-centric, qualitative approach that emphasizes exploration and change
leadership is your best bet.

Now that you know how to line everything up properly, take
some time to examine your data science team structure and see if it fits your
strategic objective. If it currently feels like herding cats, maybe you just need
a cat-herder.

Also read: