You are on page 1of 10

2016 49th Hawaii International Conference on System Sciences

The Big Data Analytics Gold Rush:


A Research Framework for Coordination and Governance

J. Alberto Espinosa Frank Armour


Kogod School of Business Kogod School of Business
American University American University
alberto@american.edu fjarmour@gmail.com

Abstract The abundance of success stories in the popular


press [1, 2] has lead educational institutions and
Big data analytics (BDA) is rapidly becoming a organizations to embark in a rush to implement BDA
most critical activity for organizations. Examples of programs. But most recipes for BDA success center
how BDA enhances the competitive advantage of around building BDA capabilities [3, 4]. While there is
organizations abound in the literature, which has no doubt that having a sound strategy and practice in the
created a gold-rush-like atmosphere. The predictions area of BDA is critical for business decision making,
about shortages of specialized analytical labor has simply having such a BDA practice is insufficient for
caused every major university to launch big data, organizational success that is sustainable over time.
analytics or data science programs, and every major Studies are finding that as BDA continues to be
organization to adopt aggressive BDA initiatives. The embraced widely, the competitive advantage of
promise of enhanced competitiveness and sustainable individual organizations using BDA will erode and data
strategic advantage are appealing and virtually all management challenges will continue to become
major decisions need to be backed up with data. This overwhelming and costly [5]. We argue in this article
has substantial implications for organizations, which that BDA capabilities are necessary for BDA success,
need to align their business process architectures, data but these capabilities alone are insufficient, and that
architectures and data management practices to effective coordination and governance of BDA is
support BDA initiatives. But while most research in this necessary.
area has focused on either building BDA capabilities or
on the benefits that these capabilities bring to BDA is a complex endeavor because it sits at the
organization, we know little about how this effort can intersection of various disciplines, including: data
be most effectively coordinated and governed to ensure management, analytical processes and tools, functional
organizational success. In this research we develop a domains, data types and sources, and the business
research framework to study whether and how processes that BDA aims to support [6]. The various
coordination and governance influence BDA outcomes. activities in these areas are highly interdependent, thus
We build on the foundations of coordination theory and requiring a substantial amount of coordination, which is
information systems governance. about managing dependencies among the associated
task activities [7]. Furthermore, because data has
1. Introduction substantial implications on important things like:
privacy, ethics, responsible practices, legal aspects,
ownership, and intellectual property, among other
Organizations will find there is little value in just
things, big data analytics governance is of paramount
storing large amounts of data. Business value can only
importance to organizational success. To the best of our
be derived from data once it is captured, organized and
knowledge, this perspective has not been thoroughly
analyzed to discover and provide an understanding of
investigated in research.
relationships, trends, patterns and other insights that
enhance decision-making and problem-solving. The use
of use of analytics as a competitive differentiator in In this article we develop an initial research
selected industries is exploding in just about every framework to study BDA coordination and governance
discipline, marketing, sales, human resources, IT by integrating three different bodies of research: BDA,
coordination theory and IT governance. The integration
management, and finance. There are multiple studies
of these three research areas is very important because,
and articles in the popular press claiming that
while each area has been recognized as important in the
organization adopting data-driven decision making
achieve higher productivity gains. respective literature, the integration of the three provide
a powerful lens to understand what leads to the success

1530-1605/16 $31.00 © 2016 IEEE 1112


DOI 10.1109/HICSS.2016.141
of BDA initiatives. As far as we know, no attempts have many companies either continue to manage their
been made in the research literature to address these corporate data using traditional database technologies,
issues. This presents an excellent opportunity to develop or have data warehouses where the big data can be
a sound framework to guide further studies of enterprise downloaded and structured for further analysis. The
architecting. In the next 3 sections we discuss each of McKinsey Global Institute’s infamous study in 2011
these 3 perspectives in more detail. In section 5 we claims that the US faces a talent shortage of 140,000 to
discuss the BDA outcomes, leading to our proposed 190,000 professionals with deep analytical skills, and
research approach and framework described in section about 1.5 million managers and decision makers with
6, followed by a concluding discussion section. analytics skills [10]. Once again, reports like these have
created an analytics education and hiring frenzy and data
2. Big Data Analytics scientists is now considered the “sexiest” profession
[11]. The problem with this data science explosion is
Before we start we need to address some of the that there is a lot of variance in what students learn about
confusions we often hear about, especially with respect analytics today and how organizations do analytics,
to terminology used in reference to BDA. BDA can ranging from a stove piped approached led by one or a
mean different things to different people, therefore it is few talented data scientists, to a more organized
important that we define these terms more precisely. approach matching the strategic needs of the
BDA encompasses three related disciplines, contained organization. We posit that in order for analytics to be
in its name – big data, analytics and big data analytics. effective, it is necessary but insufficient to have
For the purposes of this research, we use the term BDA analytical talent and effective tools, and it is also
to refer to any combination of these three disciplines, necessary to have the appropriate coordination and
which we briefly discuss next. government mechanisms in place, aligned with the
strategic goals of the organization.
Big Data – Big data is about gathering, storing,
managing and accessing the necessary data for Big Data Analytics – While we use BDA in this paper
analytics. The term “big” is often used and abused to as a catch all term to include all big data and analytics
simply mean large volumes of data that are difficult to activity, BDA has a very specific meaning and it refers
manage with existing technologies. But this definition is to doing the analytics work directly in the big data
imprecise. It is more useful to view big data from the environment. This, of course, requires more technical
lens of the infamous 3 V’s of big data proposed cleverly talent and programming skills. Most organizations that
by Gartner’s Doug Laney [8]: volume – amount of data; engage in BDA need to analyze the data in real time and
velocity – speed at which the data arrives and how long often have software applications to run automated
it stays relevant; and variety – incompatible data formats predictive models. For the purposes of this research on
and non-aligned data structures. Big data operations BDA coordination and governance, the arguments we
present substantial challenges stemming from the 3 V’s made previously about big data and analytics hold for
and require a tight coordination of various activities, BDA also. Therefore, we posit that in order for BDA to
including: data ingestion and storage technologies; data be effective, the appropriate coordination and
management methods; enterprise data architecture; data governance mechanisms need to be in place and aligned
cleansing and validation; and data access and with the strategic goals of the organization.
ownership, among other things. Prior studies have
shown how easy it is to collect redundant or inconsistent The BDA Life Cycle – Analytics projects follow a well-
data [9]. We posit that in order for big data to be defined life cycle, which includes: (1) analytic problem
effective, it not only needs to be effectively managed, – the identification and formulation of a relevant
but the appropriate coordination and governance analytics question to answer, which requires sound
mechanisms need to be in place and aligned with the knowledge of the functional domain of analysis; (2) data
strategic goals of the organization. work – identification, gathering, manipulation, pre-
processing and preparation of the necessary internal and
Analytics – The Institute for Operations Research and external data for analysis; (3) modeling decisions –
Management Sciences (INFORMS), which is the selecting the appropriate collection of modeling
leading professional and academic organization in the methods for the analysis; (4) analysis – conducting the
analytics community, defines analytics as “the scientific actual analysis; and (5) reporting – articulation of key
process of transforming data into insight for making results from the analysis. In this study we argue that in
better decisions” – see https://www.informs.org/About- order for BDA to be successful, all related activities
INFORMS/What-is-Analytics. Not all analytics are need to be effectively coordinated.
conducted in big data environments. On the contrary,

1113
3. Coordination Theory analytical task they are coordinating—a process. In
contrast, when management finds a strong alignment
BDA is a multi-discipline team-based activity that between a particular analytic recommendation and the
brings together various perspectives, including: associated functional goals, the work can be said to have
technical, quantitative, analytical, functional and been successfully coordinated—an outcome. We now
strategic. Naturally, this will require agreement and discuss coordination processes and outcomes in more
coordinated action among many people and detail.
organizational units. As such, the process will be replete
with dependencies, which will need to be managed 3.2 Coordination Processes
effectively, thus the importance of incorporating
coordination theory into our research framework. In the The process of coordinating is an old problem in the
next subsections we discuss the basic concepts behind classic organizational literature. March and Simon
coordination theory. In later sections we integrate this discussed coordination processes in detail way back in
theory into our framework. the 1950’s [14] and argued that coordination can be
accomplished “mechanistically” or “organically”.
3.1 Foundations
x Mechanistic coordination, also referred to as task
Coordination theory defines coordination as “the programming or coordination by plan [13-15] is
management of dependencies among task activities” useful for task activities that are more routine and,
[12]. This definition is very useful because it implies therefore, can be more easily programmed. The
that if task activities can be carried out independently, implementation and use of mechanisms like
then coordination is less necessary or not necessary at schedules, specifications, plans, and procedures. It is
all for collective performance. Malone and colleagues important to note that we view some aspects of
[7] extended this concept in a more interdisciplinary governance as the formal rules for BDA activity and,
way by arguing that dependencies exist, not only among therefore, a form of mechanistic coordination.
people, but also among processes and resources. For Because of its importance, we discuss governance in
example, the work of a data scientist not only depends more detail in a dedicated section below.
on the work done by functional workers (e.g., marketers,
financial analyst), but also on the technology, tools and x Organic coordination, also referred to as coordination
methods employed to manage data and do the analytical through communication, by feedback or by mutual
work. adjustment [13-15] is more effective with uncertain
and non-routine task activities that cannot be
To better understand the nature of these coordinated mechanistically because conditions are
dependencies it helps to think in terms of the topology often changing. This communication can take place in
of dependencies formulated by Thomson [13], which a number of ways, including informal and
includes, in increasing order of complexity: (1) spontaneous interaction, formal and planned [16, 17],
independence—i.e., task activities can be carried out interpersonal or group interaction [15], and verbal or
individually without coordination; (2) pooled non-verbal interaction. Organic and mechanistic
dependency—i.e., task activities can still be carried out coordination are often referred to as “behavioral”
individually, but depend on and compete for the same coordination because they are based on what people
pool of resources (e.g., shared budget, equipment, “do” to coordinate. However, people can also
people); (3) sequential dependency—i.e., one task coordinate “cognitively” based on the alignment
activity depends on another, but not the other way knowledge and mental schemes. We discuss this next.
around, thus only the dependent activity needs to be
coordinated; and (4) reciprocal dependency or x Cognitive coordination, also referred to as implicit
interdependency—i.e., both activities are dependent on coordination, is based on collective or shared
each other, thus requiring tighter coordination. knowledge that colleagues have about one another
and about each other’s tasks [18-20], allowing them
One other distinction that is important to make is to coordinate their work based on spoken assumptions
that the term “coordination” can have dual about what others are likely to do [21]. Team
interpretation, depending on whether one is referring to cognition research suggests that when members of a
the process of coordinating—i.e., managing group share knowledge about each other’s task
dependencies—or coordination outcomes—i.e., extent activities they can better plan their own activities in
to which an activity was effectively coordinated. For coordination with the activities of others. Similarly,
example, when colleagues communicate about the when collaborators have knowledge about each other

1114
they can anticipate each other’s actions more x There is a “self-fueling” dynamic cycle in which
accurately and locate and access expertise more implicit coordination improves coordination
effectively. effectiveness, which in turn improves organic and
mechanistic coordination, which in turn makes
Implicit coordination through shared cognition implicit coordination stronger, thus creating a cycle.
comes in different flavors, but they are all based on The most successful organizations were able to
some knowledge overlap among collaborators, ignite this cycle and achieve continuous
including: shared task knowledge – the knowledge improvement over time.
that collaborators have about each other’s tasks [20];
shared mental models or shared schemas – the In sum, we posit that all three forms of coordination
alignment, similarity or overlap of individual – organic, mechanistic and cognitive – influence BDA
members’ knowledge connections and mental outcomes. We also posit that implicit coordination has a
schemas [22] about the task and their peers [23, 24] moderating effect on the effects of organic and
(e.g., shared vision, buy-in, etc.); common ground – mechanistic coordination on BDA outcomes, such that
mutual knowledge [25, 26] about the meaning of these effects are enhanced when implicit coordination is
terminologies and shared vocabulary used [27]; and stronger.
collective mind or collective sense making [28], a
most powerful form of group cognition for 3.3 Coordination Outcomes
coordination purposes because it requires “heedful
interrelating” so that individuals understand and take Coordination outcomes are the results of
into account how their actions will affect others in the coordination processes. Thus, coordination outcomes
group. represent the “state” of coordination, which need to be
distinguished from the acts of coordinating employed to
3.2.1 Empirical Evidence achieve that state. Coordination outcomes are easier to
identify when coordination is lacking because
Prior studies of coordination in large scale technical coordination problems are often very evident. Empirical
collaboration has found that in long-term, asynchronous studies have shown that there are three primary types of
tasks like enterprise architecting, the role of implicit coordination outcomes in technical tasks [31]: technical
coordination is critical to success because it provides the coordination – the various parts of the system work well
alignment between architects, IT staff and business together; temporal coordination – when the timely
stakeholders [29]. This alignment is the essence of large completion schedules of the various project component
complex tasks like enterprise architecture, whose goal is don’t interfere with the completion schedules of other
to provided standards and guidance to develop systems components; and process coordination – when the
efficiently, in support of key business processes and the established process for carrying out projects is adhered
execution of organizational strategies [30]. Because to (e.g., no priority conflicts, no incompatibility in
BDA is also a long-term, asynchronous, multi- approaches, no issues that require escalation, etc.).
functional technical task, we posit that such alignment Because BDA is not just a technical task, but one that
across the multiple function provided by implicit involves quantitative knowledge work and functional
coordination is also critical for BDA. These prior domain analysis, we need to map these outcomes more
findings, which are based on several case studies, can be closely to BDA activity. In this research, we propose to
succinctly summarized as follows: do this mapping as follows:

x Organic coordination is prevalent and necessary, but x Technical coordination – technology infrastructure,
it is insufficient and, furthermore, too much tools and methods are consistent across BDA
communication could be an indicator of problems problems in multiple domains, such that two similar
within the group. problems in different parts of the organization have
similar BDA approaches.
x Mechanistic coordination is not only necessary, but
it is most effective. It provides rules and procedural x Temporal coordination – solving analytical problems
discipline, which helps keep things coordinated. or answering analytics questions are time sensitive,
particularly in tight competitive environments. The
x Implicit coordination is critical because it not only various stages of the analytics life cycle described
influences coordination effectiveness directly, but it above need to be conducted in a timely manner such
also enhances organic and mechanistic coordination. that the analytics report is delivered in time to have an
impact.

1115
x Process coordination – as with other technical tasks, across the various IT activities, suggesting that there are
process coordination is about adhering to the two dimensions to governance: intellectual – mapping
established BDA process, such that there are no to the specific governance controls and compliance; and
conflicts of interest, re-work or hot escalation issues social – which are important for all parties involved in
resulting from the BDA work of the multiple people BDA to understand and commit to each other’s goals,
involved. One important aspect of process objectives and plans [37]. Others have also pointed to
coordination is the alignment between the goals of the the importance of having both, formal governance along
BDA activity with the goals of the organization. Any with more informal processes for collaboration [38].
BDA activity that does not contribute to the
organizational goals is out of alignment and can be Tallon [39] also suggests that an effective approach to
viewed as wasteful. corporate governance for big data is one that balances
three perspectives: value creation, risk exposure and
4. BDA Governance Literature Review cost. We argue that this is a very effective approach to
conceptualizing BDA governance, but we also argue
Governance has been defined as “the distribution of that it is the “A” in BDA that drives value creation.
IT decision-making rights and responsibilities among Analytics on big data enables effective prediction and
decision making practices that can translate into profits
enterprise stakeholders, and the procedures and
and long-term sustainable competitive advantage.
mechanisms for making and monitoring strategic
decisions regarding IT” [32, 33]. Consistently, BDA Tallon proposes three practice categories for big data
governance can be thought of as the approach that governance: structural (e.g., data ownership rights,
analytic based organizations use to define, prioritize and steering committees, etc.), operational (e.g., data
track analytic initiatives as well as to manage different retention, access rights, data protection, storage and
migration policies, etc.) and relational (e.g, awareness
types and categories of data related to analytics. BDA
governance represents the rules and controls that and education on data practices, communication, etc.).
participants must comply with when performing BDA We complement this view by proposing that these
work. In order to achieve proper coordination and practices need to be effectively coordinated for BDA to
integration of efforts, organizations must fundamentally be effective.
address issues related to IT governance [32-34] and
BDA is no exception. The academic literature on BDA With respect to the practitioner literature, according
governance is very limited, which is what motivates this to Gartner [40], not having a governance framework for
research. At the same time, there is a fair amount of BDA has substantial risks due to things like: insufficient
academic literature on IT governance and of practitioner use of analytics capabilities; loss of reputation due to
literature on BDA governance. Therefore, we lean on illegitimate or unethical work; incorrect decisions based
these two bodies of literature to develop our research on poorly understood models; failure to comply with
framework. regulatory requirements; and low quality or
misalignment of analytics solutions with organizational
With respect to the academic literature on IT needs. In the same report, Gartner suggested the
governance, Boh and Yellin [32] found that standards, following attributes of BDA governance: decision
coupled with the use of governance mechanisms were rights; business objectives; investments; policies;
beneficial for infrastructure standardization and procedures; measures; adherence; and behaviors. The
report also suggested four business objectives BDA
business integration in enterprise architecture [33, 35,
governance supports: (1) preservation and growth of
36], and stated that prior research on IT governance has
focused mainly on centralization versus decentralization shareholder value; (2) coherent strategy realization; (3)
of IT decision making, but argued that IT governance compliance and assurance; and (4) risk management.
goes beyond this, which is particularly relevant for BDA The first two relate to these BDA governance objectives:
viability, relevance, efficiency, consistency, alignment
because it permeates all aspects of business, thus
and appropriateness. The last two relate to: legitimacy,
requiring that we take a broader view of governance
integrity, security, privacy, fidelity and quality.
Other studies have also found the importance of
having a governance framework that specified rules and BDA governance can be decomposed into its main
roles to oversee the use of IT resources and approve two components: analytics governance (i.e., viability,
business cases for large IT initiatives was beneficial relevance, transparency, legitimacy, appropriateness)
and information (or data) governance (i.e., integrity,
[37]. Prior research has also suggested that in addition
fidelity, quality, security, privacy, retention). These
to rules, standards and controls, the informal and social
aspects are also important for alignment and integration components affect three constituent groups: analytics
producers, analytics consumers and corporate

1116
executives, to which we add one more component Thus, as with other forms of behavioral
related to big data – data administrators. coordination, we posit that BDA governance influences
BDA outcomes. We also posit that implicit coordination
Soares [41] also proposed a framework, which has a positive moderating effect on the effects of BDA
classifies big data governance along three dimensions: governance on BDA outcomes, such that this effect is
(1) function and industry (e.g., healthcare, retail, enhanced when implicit coordination is stronger. We do
customer service); (2) big data type (e.g., social media, not predict self-fueling effects of cognitive coordination
transactional, machine generated, etc.); and (3) on BDA outcomes because we do not have any
governance discipline (e.g., organization, metadata, data theoretical foundations for how cognitive coordination
quality, business process integration, information life may operate in BDA work, but we will explore the
cycle management, etc.). He also suggested a number of possibility of this effect in our initial qualitative study
policy areas that big data governance should cover, discussed in section 7.
including: data ownership, organization, data
architecture and modeling, data and metadata 5. BDA and Organizational Outcomes
management, data quality, and data security, among
others. When we incorporate the term “analytics” into Just about any traditional variable of organizational
BDA, we also need to consider governance policies for success and performance could be used to evaluate the
things like analytical methods, tools, processes, organizational impact of BDA. Stories about the
production and dissemination of analytics report, among organizational benefits of BDA appear in the news and
other things. And because BDA can only occur if the popular press daily and we constantly hear about the
processes are designed and implemented to capture and success stories. We don’t hear as much about failures or
produce the necessary data for BDA, policies must about less than desirable return on BDA investments.
include business process practices that support the We argue that in order to understand how BDA
generation and availability of these data [41]. influences organizational outcomes, we need to first
understand what drives BDA outcomes. As we have
Finally, recent studies of governance in enterprise argued in prior sections we posit that BDA outcomes
architecture have found that governance can be will be more substantial when the BDA effort is
proactive, post-hoc or ad-hoc [29]. Proactive effectively coordinated and governed. Without question,
governance is about establishing effective controls and isolated and stove piped BDA projects can have an
working in advance to implement practices to enhance impact, and a few talented individuals can make a big
the effectiveness of practices and reduce the likelihood difference in organizations. But we argue that in order
of encountering non-compliance issues, which is a most for these benefits to be sustainable over time and
effective mechanistic coordination practice. Post-hoc is provide strategic advantage to an organization, the BDA
mostly about compliance review, which is often viewed effort will require coordinated effort and effective
as a compliance issue and as a nuisance and an governance.
impediment, and therefore least effective for
coordination. Ad-hoc governance is the least effective Again, theorizing about organizational outcomes
and even detrimental because there is really no from BDA activities is beyond the scope of this
governance plan in place. research, but we begin by discussing BDA outcomes
from the perspective of capability maturity models. An
In sum, BDA is about establishing and following INFORMS survey of 230 participants shows that the
the rules of BDA work, which is often referred to concept of analytics maturity is very important, but most
respectively as control and compliance. Control is about of these organizations do not have a plan, model or
establishing the governance rules and compliance is mechanism to assess analytics maturity [42]. Prior
ensuring the adherence to these rules. But establishing studies of software engineering have used capability
controls is like establishing mechanistic coordination maturity models like the Software Engineering
plans and ensuring compliance is about evaluating if Institute’s CMM [43] successfully to characterize the
such plans were followed. Overall, we conclude that we level of sophistication of software organizations.
cannot divorce governance from coordination. On the Similar models have been developed for enterprise
contrary, governance controls and compliance are an architecture [30] and other technical endeavors. These
important aspect of coordination. As Tallon suggested, models are very useful because one can study which
BDA governance is important, but to be effective it coordination processes, governance practices and other
needs to carefully balance value creation, risk exposure activity lead to higher maturity and sophistication in a
and operational costs. given practice.

1117
To the best of our knowledge, while a few BDA analytics maturity (AM) model to the information
maturity models have been proposed, none have been evolution (IM) maturity of the organization based on
empirically tested to evaluate how well they Davis et al.’s Information Revolution [46], placing
differentiate among the various levels of BDA organizations in one of four quadrants depending on
proficiency in organizations. We expect that a number their maturity along these two dimensions: I (low AM
of BDA maturity models will be proposed in the not so and IM); II (low AM, high IM); III (high AM, low IM);
distant future. In fact, some models have already and IV (high AM and IM). Because our goal is to study
emerged. For example, INFORMS proposed its own the importance of coordination and governance of BDA,
analytics maturity model [44] based on twelve simple which involves both, information management and
components organized around three main capabilities: analytics capabilities, this two-dimensional model is the
organizational (i.e., people, leadership, measures and most suitable for our research.
processes); analytics (i.e., governance, roles/skills,
analytic services and analytic processes); and data and These stages are a step in the right direction and are
infrastructure (i.e., health, access, traceability, analytics very useful to help evaluate the effectiveness of best
architecture). One attractive thing about this maturity practices, coordination and governance in BDA, but
model is that it is organized about key capability areas – more work needs to be done to develop an empirically
organizational, analytic and infrastructure. However, tested BDA capability maturity model, based on specific
this model has not been empirically tested and it requires CMM-like key processes and capabilities that are
further elaboration. indicators of the various capability levels. Examples of
possible capabilities or key processes include things
Similarly, Davenport and Harris [1] proposed five like: data governance; having a chief data officer;
stages of development for analytics capabilities: (1) Pre- widespread data access; staff trained in analytics and
requisites to analytical competition – good transaction data science; standardized corporate tools for analysis;
data environment and operations exist at this stage; (2) big data storage capabilities; programming staff; having
Prove-It Detour – most organizations progress to stage an analytics department with key roles fully staffed;
two directly, but some need this detour to prove their having functional domain experts available to the
analytical capabilities; (3) Analytical Aspirations – this analytics team; and so on. The formulation of a big data
stage occurs when analytics gains executive support and analytics capability maturity model is beyond the scope
first major analytics projects are launched; (4) of this chapter, but we hope that this discussion
Analytical Company – this stage is achieved when the illustrates the importance of developing one.
organization has world-class analytical capabilities at
the enterprise level; and (5) Analytical Competitors – 6. Research Agenda
which happens when analytics go beyond being a
capability to become strategic for competitive The discussion we have presented above leads us to
advantage. Again, this is not exactly a capability formulate the research framework depicted in Figure 1.
maturity model, but a model outlines phases of This model is partly based on prior findings about
development of analytics sophistication. coordination and governance in enterprise architecture,
discussed above, modified to fit the nuances of BDA
Building on Davenport and Harris, Ferris from the work. We argue that this empirically-tested model
SAS Institute further developed this maturity model [45] extends to the BDA activity because both, enterprise
and proposed five stages: (1) Analytically Impaired – architecting and BDA their multiple similarities,
the organization has some data and interest in analytics; including: multi-disciplinary activity; long-term task;
(2) Localized Analytics – some analytics capabilities the work is done mostly asynchronously; multi-
and interest in specific functional areas; (3, 4 and 5) disciplinarily work; with potential strategic benefits to
same as Davenport and Harris. Despite the similarities, the organization; and requiring tight alignment across
Ferris takes the maturity model one step further by the various technical and business groups involved in
proposing a questionnaire of more than 140 items to BDA.
evaluate an organization’s analytical capability along
nine categories: (1) strategy; (2) management support; We propose the research model depicted in Figure
(3) culture; (4) technology; (5) data management; (6) 1 to guide our inquiry with an exploratory study. In this
analytics skills and competency; (7) decision making model we posit that organic and mechanistic
process; (8) execution; and (9) reporting and coordination influence BDA outcomes. We also posit
monitoring. that BDA governance, acting as a mechanistic
coordination activity, also influences BDA outcomes.
Again, the Ferris model has not been empirically We also posit that cognitive coordination has a main
validated, but it is an appealing model because it ties the

1118
effect on BDA outcomes because it helps align shared initial exploratory study because it will allow us to
goals and mindsets in one direction. We also posit a discover similarities and differences in BDA practices
positive moderating effect on the effect of organic and and approaches to coordination and governance, which
mechanistic coordination and BDA governance, such will help us further refine the research framework for the
that cognitive cognition makes both, organic and subsequent survey study. Grounded Theory can provide
mechanistic coordination more effective. In contrast to rich descriptions of the phenomena observed [47, 48].
prior findings about self-fueling effect in enterprise Grounded Theory has been widely as a qualitative
architecting, we do not argue for a self-fueling effect of method in information systems research [49, 50],
cognitive coordination on BDA outcomes, but suspect particularly when the study is exploratory, and the
that one may exist and will keep an open mind about this theoretical development of the topic is in its early stages
possibility. Finally, the effects of BDA capabilities on [50].
BDA outcomes and the effects of BDA outcomes on
organizational outcomes are outside the scope of this Our target sample population is composed of
research, but they are both included in the research multiple organizations in various sectors (e.g. private,
model for completeness. As articulated by the various government, institutional, etc.) with various levels of
proponents of BDA maturity models, BDA capabilities BDA maturity (per Ferris’ five stages of development).
will positively influence BDA outcomes. Similarly, For each organization in a sample, we plan to interview:
stronger BDA outcomes will lead to better a business executive; a chief data officer or equivalent;
organizational outcomes. a data administrator; a data scientist or analyst; and one
or two functional domain experts, depending on the size
The research framework presented in Figure 1 is of the organization. Each interview will be first
intentionally parsimonious because it is intended to transcribed verbatim and then analyzed as prescribed by
guide our qualitative inquiry on coordination and Grounded Theory using open coding to uncover
governance in BDA. Our future research plans has two recurring themes surfacing from participant accounts.
phases: (1) a qualitative study; followed by a (2) survey We will continue conducting interviews until theoretical
study. saturation is achieved (i.e., nothing new is learned from
subsequent interviews). We will then do axial coding of
6.1 Research Methods for Next Phases the interviews to uncover relationships among themes of
interest to our study. We will then do theoretical
6.2.1 Next Phase – Qualitative Study sampling to reconcile our findings with the extant
literature. In addition to uncovering themes associated
Because we know so little about coordination and with BDA coordination and governance, we hope to
governance in BDA, we plan to start our research with a obtain more nuanced insights into what makes up a
qualitative interview study. The study will be based on mature BDA organization, so our interview instruments
semi-structured interviews following a Grounded will have questions aimed at discovering key processes
Theory approach [47]. Grounded Theory is ideal for this

Analytics Information
Maturity Evolution
Maturity

BDA
Governance
BDA Organizational
Coordination Outcomes Outcomes
Processes
(Organic &
Mechanistic)

Cognitive
BDA
Coordination
Capabilities
(Technology, Tools,
Methods, Talent)

Figure 1: Research Framework for BDA Coordination and Governance

1119
and capabilities that are good indicators of strong BDA 4. Singh, B.l.-r.P., R. Mathur, and H.M. Srujana, Building a
capability. High Performance Big Data Analytics Organization.
Analytics Magazine, 2014.
6.2.2 Subsequent Phase – Survey Study Once we 5. Kiron, D., P.K. Prentice, and R.B. Ferguson, The Analytics
complete the qualitative study we plan to launch a Mandate. 2014, MIT Sloan Management Review and SAS.
survey study. We plan to use the qualitative study 6. Soares, S., The IBM Data Governance Unified Process:
findings and the extant literature to construct the study Driving Business Value with IBM Software and Best Practices.
variables to help us test the revised version of the model 2010, Ketchum ID: Mc Press.
presented in Figure 1. For the survey study we plan to 7. Malone, T. and K. Crowston, The Interdisciplinary Study
collect data from multiple organizations with of Coordination. ACM Computing Surveys, 1994. 26(1): p.
participants with similar roles to the ones utilized for the 87-119.
qualitative study above. 8. Laney, D., 3D Data Management: Controlling Data
Volume, Velocity, and Variety, in Application Delivery
Strategies, I.G. META Group, Editor. 2001: Stamford, CT.
7. Discussion
9. Espinosa, J.A., F. Armour, and W.F. Boh. Coordination in
Enterprise Architecting: An Interview Study. in 43rd. Hawaii
Our proposed research framework is by no means International Conference on System Sciences. 2010. Poipu,
complete, but we hope that it will spark much needed Kawai, Hawaii: IEEE.
discussion in this area. This framework has limitations 10. Manyika, J., et al., Big data: The Next Frontier for
because it has not been validated empirically and Innovation, Competition, and Productivity. 2011, McKinsey
because it only incorporates distance and time Global Institute.
boundaries. Further research is necessary to expand our 11. Davenport, T.H. and D.J. Patil, Data Scientist: The Sexiest
framework to include other global boundaries. Despite Job of the 21st Century, in Harvard Business Review. 2012.
this limitation, our proposed framework is soundly 12. Malone, T. and K. Crowston. What is Coordination Theory
grounded on well-established bodies of research—i.e., and How Can it Help Design Cooperative Work Systems. in
BDA, coordination theory, IS governance research and Computer Supported Collaborative Work. 1990. Los Angeles,
organizational outcomes. This framework is a first step CA: ACM Press.
to providing theoretical foundations for studies in this 13. Thompson, J., Organizations in Action. 1967, New York,
area. Our goal in this next phase of our research is to NY: McGraw-Hill.
discover the main coordination and governance 14. March, J. and H.A. Simon, Organizations. 1958, New
challenges confronted in BDA work and develop a more York: John Wiley and Sons.
nuanced understanding of the key challenges in this 15. Van de Ven, A.H., L.A. Delbecq, and R.J. Koenig,
work and the best practices to mitigate these challenges. Determinants of Coordination Modes Within Organizations.
We will use these findings to further refine and validate American Sociological Review, 1976. 41(2): p. 322-338.
our research model in Figure 1, so that we can conduct 16. Kraut, R.E. and L.A. Streeter, Coordination in Software
a more targeted survey study afterwards. Development. Communications of the ACM, 1995. 38(3): p.
69-81.
Our study will provide an important theoretical 17. Perry, D.E., N.A. Staudenmayer, and L.G. Votta, People,
contribution to BDA coordination and governance Organizations, and Process Improvement. IEEE Software,
because of the very sparse work in this area. It will also 1994. 11(4): p. 36-45.
provide important practical implications for managers to 18. Cannon-Bowers, J.A., E. Salas, and S. Converse, Shared
improve their BDA practices through effective Mental Models in Expert Team Decision-Making, in
coordination and governance. Individual and Group Decision-Making: Current Issues, J.
Castellan, Editor. 1993, Lawrence Erlbaum Associates, Inc.:
Hillsdale, NJ. p. 221-246.
8. References
19. Cannon-Bowers, J.A. and E. Salas, Reflections on Shared
1. Davenport, T.H. and J.G. Harris, Competing on Analytics: Cognition. Journal of Organizational Behavior, 2001. 22(2): p.
The New Science of Winning. 2007, Boston, MA: Harvard 195-202.
Business School Publishing Corporation. 20. Klimoski, R.J. and S. Mohammed, Team Mental Model:
2. McAfee, A. and E. Brynjolfsson, Big Data: The Construct or Metaphor. Journal of Management, 1994. 20(2):
Management Revolution, in Harvard Business Review. 2012, p. 403-437.
Harvard Business Publishing. 21. Wittenbaum, G.M. and G. Stasser, Management of
3. Brown, B., D. Court, and P. Willmott, Mobilizing Your C- Information in Small Groups, in What's Social about Social
Suite for Big-Data Analytics, in Insights & Publications. 2013, Cognition?, J.L. Nye and A.M. Brower, Editors. 1996, Sage
McKinsey Quarterly. Publications: Thousand Oaks, California. p. 3-27.
22. Rouse, W.B., J.A. Cannon-Bowers, and E. Salas, The Role
of Mental Models in Team Performance in Complex Systems.

1120
IEEE Transactions on Systems, Man & Cybernetics, 1992. 22: alignment in government. Information Technology & People,
p. 1296-1308. 2007. 20(2): p. 96.
23. Rentsch, J.R. and R.J. Hall, Members of Great Teams 38. Van der Raadt, B., et al., The relation between EA
Think Alike: A Model of the Effectiveness and Schema effectiveness and stakeholder satisfaction. Journal of Systems
Similarity Among Team Members, in Advances in and Software, 2010. 83(10): p. 1954-1969.
Interdisciplinary Studies of Work Teams: Theories of Self- 39. Tallon, P.P., Corporate Governance of Big Data:
Managing Work Teams (Vol. 1), M.M. Beyerlein and D.A. Perspectives on Value, Risk, and Cost. Computer, 2013. 46(6):
Johnson, Editors. 1994, JAI Press: Greenwich, CT. p. 223-261. p. 32-38.
24. Rentsch, J.R. and R.J. Klimoski, Why do Great Minds 40. Oestreich, T. Establish and Analytics Governance
Think Alike?: Antecedents of Team Member Schema Framework. in Garner Business Intelligence and Analytics
Agreement. Journal of Organizational Behavior, 2001. 22(2): Summit. 2015. Las Vegas.
p. 107-120.
41. Soares, S., Big Data Governance: An Emerging
25. Cramton, C.D., The Mutual Knowledge Problem and Its Imperative. 2013, Boise, ID: Mc Press.
Consequences for Dispersed Collaboration. Organization
Science, 2001. 12(3): p. 346-371. 42. New Survey from INFORMS Reveals Majority of
Organizations Do Not Have a Plan in Place to Assess the
26. Krauss, R. and S. Fussell, Mutual Knowledge and Maturity of Their Analytics Practices. 2014; Available from:
Communicative Effectiveness, in Intellectual Teamwork: https://www.informs.org/About-INFORMS/News-
Social and Technological Foundations of Cooperative Work, Room/Press-Releases/INFORMS-Analytics-Maturity-Model-
J. Galegher, R.E. Kraut, and C. Egido, Editors. 1990, Survey-San-Francisco.
Lawrence Erlbaum: Hillsdale, NJ. p. 111-146.
43. Humphrey, W.S., Characterizing the software process: A
27. Clark, H. and S. Brennan, Grounding in Communication, maturity framework. IEEE Software, 1988. 5(2): p. 73–79.
in Perspectives on Socially Shared Cognition, H. Clark and S.
Brennan, Editors. 1991, American Psychological Association. 44. Reitter, N. and B. List, Analytics Maturity Model, in
OR/MS Today. 2013, INFORMS.
p. 222-233.
28. Weick, K. and K. Roberts, Collective Mind in 45. Ferris, J., How to Compete on Analytics. 2008, The
Organizations: Heedful Interrelating on Flight Decks. Analytical Center of Excellence, SAS Institute Inc.
Administrative Science Quarterly, 1993. 38(3): p. 357-381. 46. Davis, J., G.J. Miller, and A. Russell, Information
Revolution: Using the Information Evolution Model to Grow
29. Espinosa, J.A., F. Armour, and W.F. Boh. The Role of
Group Cognition in Enterprise Architecting. in 44th. Hawaii Your Business. 2006, Hoboken, New Jersey: Wiley.
International Conference on System Sciences. 2011. Ko'loa, 47. Glaser, B.G. and A. Strauss, The Discovery of Grounded
Kauai: IEEE. Theory: Strategies for Qualitative Research. 1967,
Hawthorne, NY: Aldine de Gruyter.
30. Ross, J., P. Weil, and D. Robertson, Enterprise
Architecture As Strategy: Creating a Foundation for Business 48. Strauss, A. and J. Corbin, Basics of Qualitative Research:
Execution. 2006, Boston, Massachusetts: Harvard Business Techniques and Procedures for Developing Grounded Theory.
School Press. Second Edition ed. 1998, London: Sage Publications, Inc.
31. Espinosa, J.A., et al., Team Knowledge and Coordination 49. de Vreede, G.-J., N. Jones, and R.J. Mgaya, Exploring the
in Geographically Distributed Software Development. Journal Application and Acceptance of Group Support Systems in
of Management Information Systems, 2007. 24(1): p. 135- Africa. Journal of Management Information Systems, 1998-
169. 99. 15(3): p. 197-234.
32. Boh, W.F. and D. Yellin, Using enterprise architecture 50. Orlikowski, W., CASE Tools as Organizational Change:
standards in managing information technology. Journal of Investigating Incremental and Radical Changes in Systems
Management Information Systems, 2007. 23(3): p. 163-207. Development. MIS Quarterly, 1993. 19(3): p. 309-340.
33. Peterson, R., Crafting Information Technology
Governance. Information Systems Management, 2004. 21(4):
p. 7-22.
34. Brown, C.V., Horizontal Mechanisms Under Differing IS
Organization Contexts. MIS Quarterly, 1999. 23(3): p. 421–
454.
35. V., S. and R.W. Zmud, Arrangements for Information
Technology Governance: A theory of Multiple Contingencies.
MIS Quarterly, 1999. 23(2): p. 261–290.
36. V., S. and R.W. Zmud, The Organizing Logic for an
Enterprise’s IT Activities in the Digital Era: A Prognosis of
Practice and a Call for Research. Information Systems
Research, 2000. 11(2): p. 105–114.
37. Gregor, S., D. Hart, and N. Martin, Enterprise
architectures: enablers of business strategy and IS/IT

1121

You might also like