THE RISE OF DATA ANALYTICS is one of
the hallmarks of 21st-century business.
By the turn of the century, companies
had been accumulating data in various
transaction systems for several decades,
and many desired to analyze the data to
make better decisions. Their interest
intensified in the early 2000s as they
saw the great success of online firms
from Silicon Valley, many of which were
highly analytical.
In fact, during the mid-2000s, I
conducted research showing that some
companies were “competing on analytics”—
that is, emphasizing their analytical
capabilities as a key element of their
strategies—and that those companies
tended to outperform other firms in
their industries. Information about analytics
even made it into popular culture,
especially through books such as Moneyball,
which was also a successful movie.
Both depicted the way the Oakland A’s of
California built a winning baseball team
through targeted data analysis.
These market forces rapidly led to
a demand for analytical talent within
firms. The first wave primarily focused
on traditional analytical professionals
and web analytics experts. But by 2010,
online firms had started to realize they
needed a new version of the analytics
professional, someone who could analyze
the massive amounts of data they
had been collecting from online clickstreams,
social media, and the sensors
within internet-enabled devices (the
“Internet of Things”). They needed professionals
who not only had strong analytical
skills, but who also could push the
frontiers of the new technologies—many
of them open-source programs—that
were designed to store and structure
data. They needed data scientists.
Not only did Silicon Valley companies
hotly pursue data scientists, mainstream
businesses also saw the need to
bring them on board. In a 2011 report,
McKinsey Global Institute researchers
estimated that there would soon be a
shortage of up to 200,000 data scientists
in the U.S. alone—although there was
never great clarity about what it took
to become one.
Then, around 2015, companies
became enthused about the potential of
artificial intelligence (AI) to transform
business, and this excitement sparked a
new wave of analytics-related demand.
Silicon Valley companies such as Google
and Facebook again led the charge,
incorporating many AI capabilities into
their offerings, and other firms soon
followed. I have helped conduct surveys
that suggest that up to 25 percent of large
companies in the U.S. have multiple AI
projects underway. A few highly aggressive
firms have launched more than
100 projects, some of which are entering
the production deployment stage.
Not surprisingly, all of these firms are
seeking data scientists or AI engineers
who understand how to apply new technologies
such as deep learning to business
contexts. As before, these firms face
what appears to be a severe shortage of
individuals with this skill set. What kind of opportunity does this represent for
business schools?
WHERE WE STAND NOW
Business schools were swift to respond
to the first wave of demand for professionals
with analytics skills. Many
quickly designed master’s programs in
analytics by drawing from strengths
in established disciplines, including
statistics and operations research. They
also brought in faculty from marketing,
finance, and organizational behavior to
show how analytics can relate to those
specific business disciplines.
As a result of all this activity, master’s
degrees in business analytics currently
are among the most popular new
offerings in business schools. North
Carolina State University debuted
what might have been the first master’s
in advanced analytics in 2007; today,
according to estimates from AACSB
International, there are more than
400 analytics degree programs offered
by nearly 220 business schools worldwide.
An article on the Poets & Quants
website declared, “Without a doubt, the
business analytics master’s is the belle
of the specialty degree ball.”
But business schools have found it
somewhat more difficult to respond to
the second and third wave by creating
programs in data science and artificial
intelligence. Not only is data science relatively
new and difficult to define, it also
has a strong computer science orientation.
Unlike data analytics programs,
which primarily focus on analytical/statistical
methods, data science programs
teach students complex programming
methods for handling big data. For that
reason, many data science programs are offered by schools of engineering, information,
or computer science.
In fact, some universities have
multiple analytics programs residing in
various colleges within the institution.
For example, the University of Minnesota
and the University of California
San Diego offer master’s degrees in both
data science and business analytics. At
Minnesota, the MS in business analytics
is offered through the Carlson School of
Management, while the MS in data science
is run by the schools of engineering,
liberal arts, and public health.
The University of California, Berkeley,
has at least six different master’s
degrees that are oriented to analytics.
An MBA focus area in data science and
strategy is offered by the Haas School of
Business; the rest can be found within
other departments and colleges, such as
engineering, public health, and computer
science. To help students navigate the
many options, the university maintains
a datascience@berkeley webpage that
describes the “rich ecosystem of data
science-related research, teaching, and
communities across campus.”
While data science degree programs
have begun to proliferate, programs
built around AI are still rare. Both
Northwestern University’s engineering
school and the University of Georgia’s
college of arts and sciences offer master’s
programs in AI, and both MIT and
Carnegie Mellon University’s computer
science schools have recently instituted
undergraduate programs in AI. Several
schools, mostly in Europe, also offer
master’s degrees in artificial intelligence—
these include the University of
Gothenburg, the Free University of Amsterdam,
and the Ecole Polytechnique in Paris. However, these programs are
either contained within the computer
science or engineering schools, or
they draw from multiple schools
within the university.
To my knowledge, no business
schools in the U.S. have degree programs
in AI. This is not surprising, given the
paucity of faculty with expertise in AI.
Technical faculty who are able to develop
AI programs often have been hired—
at very high compensation levels—by
private sector firms such as Google
and Facebook. Such faculty maintain
university affiliations, but often not
with business schools.
PRIMARY SKILLS
Because AI programs are so difficult to
put together, most business schools will
concentrate on programs for analysts
and data science professionals instead.
In my opinion, these professionals must
acquire four different types of expertise
to be competent in their jobs.
Quantitative and statistical skills
are the foundation of any analytics role.
Anyone holding such a job must be proficient
at general statistical analysis—at
least up to logistic regression analysis,
often conducted in fields such as medicine
and the social sciences to predict
binary outcomes based on one dependent
and one independent variable.
Such professionals also should be able to
analyze categorical data, probability and
statistical inference, optimization, and
basic experimental design.
If graduates want to be hired into
specific industries or business functions,
they also will need to master
other quantitative techniques, such as
lift analysis in marketing, stochastic
volatility analysis in finance, biometrics
in pharmaceuticals, and informatics in
healthcare fields. Some types of analysts—
those involved in “business intelligence”
or reporting work—may be able
to get by without substantial statistical knowledge, but this lack probably would
limit their careers.
Analysts also must know how to use
the software associated with their type
of analytical work, whether it is used to
build statistical models, generate visual
analytics, define decision-making rules,
conduct “what-if” analyses, or present
business dashboards. These tools once
were offered only as proprietary packages
from vendors, but are increasingly
open source today.
Data management skills are just as
important to analytical professionals as
statistical and mathematical expertise.
Such professionals spend the majority of
their time manipulating data—finding,
integrating, cleaning, matching, and so
on. And surveys suggest that the skill
most commonly sought by data scientist
employers is not expertise with a statistical
program, but rather with SQL—a
query language for data management.
Like analytics professionals, data scientists
also perform data management
tasks, but their tasks tend to be more
complex and often rely on languages
such as Python. Their jobs often involve
taking relatively less structured data like
text and images and creating the rows
and columns of numbers that are well
suited to statistical analysis.
The topic of data management also
encompasses data privacy, security, and
ethical issues. In fact, some master’s
programs in analytics, including one
from the University of Notre Dame, require
students to take a course in ethics.
Business knowledge and design
skills enable analysts to be more than
simple backroom statisticians. Analysts
need enough general business background
to understand the problems
and processes they are analyzing, so
they should be familiar with marketing,
finance, HR, and new product development.
They also should know how
analytics can be used to drive business
value, and they must have insight into the opportunities and challenges their
employers are facing. Many of these
skills and content domains are taught
in MBA programs, but most business
analytics degrees don’t require a lot of
traditional MBA courses.
Relationship and communication
skills are vital to the success of all
analytical projects. Analysts who can
advise, negotiate, and manage expectations
will be able to work effectively
with their business counterparts to
conceive, specify, pilot, and implement
analytical applications.
These skills are critical when analysts
must communicate the results of their
work. They will need to share best practices
with their bosses and colleagues,
while emphasizing the value of analytical
projects. Outside the business, they will
need to develop working relationships
with customers and suppliers. They also
might have to explain the role of analytics
in meeting regulatory requirements—
for instance, an analyst might use data to
help a utility company make a successful
bid to increase its rates for service.
Analysts who can “tell a story with
data” are highly prized. The ability to
communicate effectively about analytics
is the single most sought-after capability
among graduates from analytics and
business intelligence programs, according
to one survey of employers, and
some schools realize this. For instance,
Communicating with Data is a required
course within the new master of business
analytics program at the Massachusetts
Institute of Technology’s Sloan
School of Management.
In addition to mastering these four
sets of skills in the classroom, students
typically participate in internships or
practicums, during which they work
on real analytical problems for real
organizations. Students not only gain
experience in what it’s like to be analysts
or data scientists, they also get a chance
to meet potential recruiters.
For schools, the challenge is combining
all these necessary components into
a single degree program. It’s particularly
difficult to cover all the material because
the typical master’s in analytics is only
one year or three semesters long—hardly
enough time to adequately train students
who arrive at the program lacking
basic analytics skills. It is probably more
feasible for business schools to target
students who already have undergraduate
degrees in quantitative fields and
provide them with “finishing school”
skills in business and communication.
Another challenge schools face is
assembling faculty with expertise in all
the relevant areas. Some universities
draw on faculty members from multiple
schools to create their analytics
programs. For instance, the master’s
in advanced analytics at North Carolina
State, which was intended to be
cross-disciplinary from the beginning,
involves faculty from the statistics,
mathematics, bioinformatics, and computer
science departments—and even
the English department, because communicating
effectively about analytics
is so important. Such cross-disciplinary
programs can be excellent for student
learning, but are not always easy for
schools to maintain.
THE NEED FOR DIFFERENTIATION
There’s yet another challenge business
schools face as they design their analytics
offerings: how to set their programs
apart from those of their competitors.
Geography simply isn’t enough of a
differentiator, because many programs
are wholly or partially online, and others
are offered at multiple locations. For
instance, Notre Dame runs one MS in
business analytics program on its main
campus in South Bend, Indiana, aiming
it at students right out of undergraduate
programs—but it runs a second one in
downtown Chicago, where it targets
working professionals.
Instead of competing based on their
locations, schools commonly differentiate
themselves by the topics they focus
on in their programs. For instance, some
emphasize “applied analytics,” assuming
that students will need a mixture of
business and analytical expertise. Others
are more quantitatively intense or
give more weight to data science issues.
Going on the assumption that analytics
generalists won’t be enough for some
employers, other schools have focused
their programs on functional specialties
within the field. For instance, Pace
University’s Lubin School of Business
has an MS in customer intelligence and
analytics, American University’s Kogod
School of Business has an MS in human
resource analytics and management,
and Rutgers University offers a master’s
in supply chain analytics. Stevens
Institute of Technology and New York
University both offer master’s degrees
in financial analytics, while Babson
College’s MS in business analytics aligns
with the school’s entrepreneurship
orientation by focusing on the intersection
of analytics and entrepreneurship.
Washington University’s Olin Business
School offers multiple tracks in subject
areas such as healthcare, supply chain,
financial technology, and customer
analytics.
In the future, I expect that more
master’s programs will focus on specific
analytical methods and tools, particularly
AI, although specialist AI programs
may remain within schools of engineering
and computer science. At this
point, it’s not yet clear what the role of
business schools will be in the next wave
of analytics programs.
THE FUTURE OF ANALYTICS
What is clear is that technology is
continuing to reshape the competencies
that students will need to cultivate. Up
until this point, analysts have needed
the ability to select the right algorithm
for the problem and data at hand. For
example, if I’m trying to predict a binary
variable using several continuous variables,
I need to know that logistic regression
is the method of choice. Whatever
the algorithm used, I also need to know
the process of analysis—such as how to
choose independent and dependent variables,
select data, deal with missing data,
do the analysis, create some variable
transformations to improve the fit of the
model, and interpret the outcome.
Increasingly, however, these types
of expertise are not required of human
analysts. “Automated machine learning”
programs can determine which type
of algorithm is most appropriate; they
also can perform most of the process
described above with relatively little
intervention by the data scientist. In
other words, we are probably entering a
“post-algorithmic” era in which detailed
knowledge of analytical methods may no
longer be necessary.
If this trend continues, it’s likely that
the most valuable skills will be nontechnical
ones—understanding the business,
building relationships and trust, communicating
analytical outcomes, and
working in teams to convert analytics
into action—the very skills taught in
business school. If automated systems
can eventually do most of the analytical
heavy lifting, it may be that an MBA is the
perfect analytics degree of the future.
Thomas H. Davenport is President’s
Distinguished Professor of IT and Management
at Babson College in Wellesley,
Massachusetts, and a Digital Fellow at
the MIT Center for Digital Business in
Cambridge. He is also a senior advisor for
Deloitte Analytics.
Read a review of his latest book, The AI
Advantage.
This article originally appeared in BizEd's January/February 2019 issue. Please send questions, comments, or letters to the editor to bized.editors@aacsb.edu.
RELATED READING
Deriving Meaning from Data
AI at the B-School