Content uploaded by Hippolyte Lefebvre
Author content
All content in this area was uploaded by Hippolyte Lefebvre on Apr 21, 2024
Content may be subject to copyright.
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 1
RETHINKING DATA GOVERNANCE:
A VIABLE SYSTEM MODEL
Completed Research Paper
Hippolyte Lefebvre, Faculty of Business and Economics (HEC), University of Lausanne,
Lausanne, Switzerland, hippolyte.lefebvre@unil.ch
Christine Legner, Faculty of Business and Economics (HEC), University of Lausanne,
Lausanne, Switzerland, christine.legner@unil.ch
Abstract
Data governance is a prerequisite for organizations wanting to harness the strategic potential of data.
Although the conceptual foundations of data governance have reached a sound level of clarity, research
still does not explain how data governance unfolds in large and complex organizations. To address this
gap, we introduce the Viable System Model as theoretical lens and examine data governance at five
multinational companies with varied organizational structures. We find that data governance
orchestrates data practices on multiple, interconnected levels, through sub-systems. The interplay
between these sub-systems facilitates the establishment of a dynamic balance, enabling (1) the
delineation of responsibilities, distinguishing between global and local data governance that
orchestrates data practices, and (2) the implementation of data practices at the operational level that
simultaneously emphasize control and foster innovation. Our research contributes to rethinking data
governance and addresses previous calls for research that accounts for its dynamic nature in practice.
Keywords: Data governance, Data practice, Viable System Model, Systems thinking.
1 Introduction
Successful organizations recognize the strategic potential of data for sustainable competitive advantage
(Jones, 2019) and its vital role in creating business value, such as cost efficiency or better market
positioning (Günther et al., 2022). A prerequisite for unlocking the potential of data is data governance,
i.e., the specification of “a cross-functional framework for managing data as a strategic enterprise
asset” (Abraham, Schneider and vom Brocke, 2019, p. 425). Grover et al. (2018) even argue that
“without appropriate organizational structures and governance frameworks in place, it is impossible to
collect and analyze data across an enterprise and deliver insights to where they are most needed” (p.
417). Data governance has long been concerned with the quality and protection of data assets and the
adherence to regulatory requirements (Weber, Otto and Österle, 2009; Otto, 2011). Today, data is at the
heart of value creation in enterprises, resulting in data governance having the dual purpose of
simultaneously balancing control and innovation (Vial, 2023).
Data governance research has mainly focused on clarifying the basic understanding and defining the
scope and overall framework of data governance (Khatri and Brown, 2010; Abraham, Schneider and
vom Brocke, 2019). Building on IT governance literature, it conceptualizes data governance as an
ensemble of mechanisms (Tallon, Ramirez and Short, 2013; Abraham, Schneider and vom Brocke,
2019; Vial, 2023) encompassing structural mechanisms (e.g., roles, responsibilities, locus of decision
making), procedural mechanisms (e.g., processes, monitoring), and relational mechanisms (e.g.,
communication, training). While the foundations of data governance are increasingly clear, criticism has
emerged from practice claiming that data governance cannot be viewed only “as series of mechanisms
implemented in organizations, at the expense of understanding the process of governing data” (Vial,
2023, p. 6). Concretely, research still mainly lists what to do and does not explain how to do data
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 2
governance, i.e., data governance in practice (Alhassan, Sammon and Daly, 2016; Aaltonen, Alaimo
and Kallinikos, 2021; Vial, 2023). Moreover, given global firms' complex organizational structures,
establishing data governance for them remains a challenge (Otto, 2011). In order to be effective, data
governance must reach many different parts of an organization and shape the situated data practices
through which data acquires its value (Parmiggiani and Grisot, 2020). Federated data governance
models, which combine global and local data governance responsibilities, have been proposed as a
solution in rolling out data governance in accordance with the primary organizational structure (Grover
et al. 2018; King 1983). However, so far, no link has been established for understanding how data
governance mechanisms materialize at local and global levels. Further, the rather static view of data
governance mechanisms does not properly explain the dynamic nature of data governance which must
evolve in symbiosis with strategy and operations (Benfeldt, Persson and Madsen, 2020). As markets,
regulations, and organizational culture are continuously evolving, data governance obviously has to
adapt (Tallon, Ramirez and Short, 2013; Abraham, Schneider and vom Brocke, 2019).
In such a context, we ask the following research question (RQ):
RQ: How does data governance unfold in multinational companies?
In our study, we apply systems thinking to data governance and use the Viable System Model (VSM)
as theoretical lens. The VSM explains a system’s viability, i.e., its ability to maintain its existence in a
changing environment (Beer, 1985), and it has been used to explain IT governance setups (Peppard,
2005; Huygh and De Haes, 2019). Our study is embedded in a collaborative practice research
(Mathiassen, 2002), with 17 multinational companies. It is informed by insights from nine focus groups,
as well as in-depth case studies. To understand how governance mechanisms are implemented in large
and complex organizations, we analyzed the cases of five companies that have developed global and
local data governance responsibilities. Our findings reveal that data governance orchestrates data
practices on multiple, interconnected levels, through sub-systems. The interactions between data
practices happening at operational, governance, and strategic levels make it possible to establish an
appropriate balance that mediates (1) between global and local data governance, and (2) between data
governance activities that seek control on the one hand and innovation on the other. Overall, closing this
research gap advances the academic understanding of federated governance, paving the way for a new
angle in investigating data practices at strategic, governance, and operational levels. Our research offers
practitioners guidelines on how to set up a data governance framework that aligns with their overall
strategy and organizational structure.
In the remainder of the paper, we first give information on prior data governance literature and highlight
the research gap. Second, we motivate the relevance of systems thinking and the applicability of VSM
as a theoretical lens. Next, we present our methodology, and finally, we summarize and discuss our
findings, and also provide an outlook on future research.
2 Theoretical Background
2.1 Data governance
Data governance is seen as a framework describing cross-functional efforts for maximizing the value of
data as strategic enterprise assets and ensuring the compliant and strategic use of data (Tallon, Ramirez
and Short, 2013; Abraham, Schneider and vom Brocke, 2019). It thus fosters the contribution data makes
to achieving organization goals and generally aims to improve firm performance (Mikalef et al., 2020).
Data governance is shaped by both external environmental antecedents, such as legal and regulatory,
industry, or regional conditions, and internal ones, such as business strategy, corporate culture, or
organizational structure (Baijens, Huygh and Helms, 2021; Tallon, Ramirez and Short, 2013).
To set up data governance, firms should clearly identify its scope along three dimensions (see Figure 1).
First, organizational scope refers to “expansiveness of data governance”(Abraham, Schneider and vom
Brocke, 2019, p. 430), which can be intra-organizational or inter-organizational. Second, firms define
the data scope and identify the relevant data objects, data types, and data domains to prioritize for data
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 3
governance. For instance, master or transactional data objects are usually governed first, but other big
data-related types such as media data and sensors can come in scope later to support new data
applications (Abraham, Schneider and vom Brocke, 2019; Fadler, Lefebvre and Legner, 2021). Third,
the “depth” of the data governance program is defined by its domain scope, i.e., the different data
decision domains, such as data quality, data security, data architecture, data lifecycle, metadata, data
storage, and infrastructure (Khatri and Brown, 2010; Abraham, Schneider and vom Brocke, 2019).
Figure 1. Conceptual framework for data governance by Abraham et al. (2019).
Three mechanisms—structural, procedural, and relational—constitute the core of data governance,
drawing from established IT governance frameworks (Tallon, Ramirez and Short, 2013; Abraham,
Schneider and vom Brocke, 2019; Vial, 2023). These mechanisms should be combined and not
addressed separately for maximum efficiency (Tallon, Ramirez and Short, 2013), and can typically be
bundled into archetypes aligned on the strategic context and scope for data (Fadler, Lefebvre and Legner,
2021).
Structural mechanisms focus on specifying roles (e.g., data owner, data steward) and responsibilities in
line with the organizational structure, and on allocating decision-making authority. This entails defining
where the different data teams are positioned and their reporting lines (Otto, 2011). The literature
differentiates between centralized, decentralized, and federated data governance designs (Brown, 1999;
Sambamurthy and Zmud, 1999; Weber, Otto and Österle, 2009). A strict centralized data governance
model implies that a central data unit has global authority and responsibility regarding data. Such a
model is convenient for company-wide control, efficiency, and reliability in the (re)utilization of data
assets because it leverages lateral organizational capabilities between units. However, it decreases local
units’ flexibility and capacity to innovate (Velu, Madnick and van Alstyne, 2013; Grover et al., 2018).
Conversely, in a fully decentralized model, business units hold local responsibility for their data, each
with their respective governance principles which enable rapid adaptation to changing requirements
(Velu, Madnick and van Alstyne, 2013). In this model, the lacking standardization leads to coordination
challenges, compliance concerns, data quality issues, limited collaboration, and complex data access
management. Federated (also called hybrid, or Hub-Spoke) models combine the two forms in a global
hub responsible for enterprise-wide standards, policies, methods, and tools, with business units as spokes
taking care of responsibilities closer to the relevant data operations (e.g., data creation, data quality, data
maintenance) (King, 1983; Grover et al., 2018). While offering numerous benefits such as greater local
autonomy, faster issue resolution, and improved agility, a federated model generally requires better
coordination mechanisms and acknowledged data ownership by respective business units (Velu,
Madnick and van Alstyne, 2013).
The procedural and relational mechanisms instantiate the structural mechanisms. Procedural
mechanisms describe decision-making related to data activities and processes, and thereby “emphasize
the operational means that are put in place to ensure compliance with governance principles”(Vial,
2023, p. 4). These include data strategy; policies, standards, and procedures; contractual agreements;
performance measurement; compliance monitoring; and issue management (Abraham, Schneider and
vom Brocke, 2019). Relational mechanisms ensure alignment, collaboration, and knowledge sharing
between stakeholders. To expand the reach and understanding of data governance principles, these
mechanisms usually comprise both formal (e.g., working groups, collaboration platform, training
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 4
events) and informal (e.g., job rotation, corporate events, communities) means of coordination
(Abraham, Schneider and vom Brocke, 2019). For instance, communities of practice foster knowledge
sharing and data literacy among both data experts and non-experts (Lefebvre and Legner, 2022).
The above view on data governance has attracted criticism because the governance mechanisms do not
explain data governance in practice (Alhassan, Sammon and Daly, 2016; Aaltonen, Alaimo and
Kallinikos, 2021; Vial, 2023). Recent research suggests “a shift from data governance as a matter of
asset management to data governance as a matter of work practice” because data governance is enacted
as part of local actors’ sense-making processes, such as during data curation tasks (Parmiggiani and
Grisot, 2020, p. 3). Therefore, firms naturally evolve toward federated data governance that
accommodates both global and local needs (Benfeldt, Persson and Madsen, 2020), thus pragmatically
reflecting the organizational complexity of the organization, specifically in multinational companies
(Velu, Madnick and van Alstyne, 2013; Khatri and Brown, 2010). This shift is also reflected in the
emerging data mesh paradigm which emphasizes data management responsibilities close to data creators
because they know the context the best (Machado, Costa and Santos, 2021). Further, data governance
should be addressed as a “dynamic element that is implemented and should evolve in conjunction with
strategy and operations” to maintain its dual purpose of balancing control and data-driven innovation
(Vial, 2023, p. 9). However, the literature neither explains how data governance responds to growing
operational needs (e.g., data requests in business) nor clarifies data governance’s role in assimilating
strategic decisions. This gap calls for further investigation of how data governance unfolds in practice.
2.2 A systems thinking approach to address data governance in practice
We argue that systems thinking, and especially the VSM, offers a promising lens to study data
governance as a system dynamically shaped by antecedents and composed of a set of interrelated sub-
systems. The VSM introduces the concept of viability, suggesting that a system is able to remain
functional despite a dynamic and fluctuating environment (Beer, 1985). It provides a framework for
describing organizations and how they process information between different entities, including internal
departments, external partners, and the broader environment which represents surrounding external
factors that could influence the system (see Figure 2). This framework emphasizes the continuous
interactions and information exchanges (symbolized by the arrows between each element), both critical
aspects of organizational decision-making, adaptation, and innovation.
To achieve viability, the VSM posits self-organizing systems as composed of five sufficient
interconnected sub-systems (Systems 1 to 5) that each have a role in maintaining the viability of the
system (Beer, 1985), i.e., all sub-systems must be active and continuously exchange information:
• System 1 represents the Operations element of the VSM. As system-in-focus, it describes the
different local operative units that execute the necessary tasks (i.e., work practices) that maintain
the entire system’s purpose. These operative units are typically embedded in the organization's
primary structure and have their own local management. They can communicate with one another.
• Systems 2 to 5 – coordination, control, intelligence, policy – together form the Management element
of the VSM, which acts as meta-system determining System 1. Thereby, they ensure smooth
operation delivery (e.g., scheduling, strategic planning).
By applying the VSM as theoretical lens we can gain a thorough understanding of how data governance
practices are arranged to assimilate and accommodate changes (e.g., in data scope). This lens also
illustrates how data governance is embedded in the organizational structure. This approach has been
employed to investigate IT governance (e.g., Peppard (2005), Huygh & De Haes (2019)) and, more
recently, to examine analytics governance, which emphasizes the contextualized output of data
utilization (Baijens, Huygh and Helms, 2021). The latter authors notably argue that analytics governance
is part of a meta-system for the totality of data analytics activities (e.g., data analytics projects).
However, data use depends on input data and, consequently, on data governance practices (Aaltonen,
Alaimo and Kallinikos, 2021; Legner, Pentek and Otto, 2020). Thus, we argue for data governance – as
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 5
a separate VSM – because “the actual work tasks carried out by individuals to curate and set up the
data are typically downplayed” (Parmiggiani, Østerlie and Almklov, 2022, p. 139).
Figure 2. Structure and relationships in the Viable System Model (simplified representation
based on Beer (1985)).
3 Methodology
3.1 Research design
Considering our research question (How does data governance unfold in multinational companies?) and
our theoretical proposition (that data governance in multinational companies can be observed through
the VSM lens), we follow a qualitative research design (Dubé and Paré, 2003). Our study spanned the
period from September 2020 to November 2023. It was embedded in a collaborative practice study
(Mathiassen, 2002) and informed by insights from focus groups of 17 multinational companies, as
shown in Figure 3. To further deepen our analysis, we conducted five in-depth case studies (Yin, 2018).
Figure 3. Overview of the research design.
System 5: Policy
Environment
System 4: Intelligence
System 3: Control
System 2:
Coordination
Local
operations
Local
operations
Local
operations
Local
Management
Local
Management
Local
Management
System 3*:
Audit
Management
(Meta-system)
System 1 as ensemble of
all local Operations
Future
env.
Local
env.
Local
env.
Local
env.
Multiple case studyCollaborative practice research
Objective:
Deep
-
dive into how data governance
mechanisms unfold in multinational
organizations
Objective:
Benchmark data governance
approaches in multinational
organizations
Activities:
•
2x90-min semi-structured
interviews with data experts from
each of the five case companies
•
Analyze additional
documentation (e.g., data strategy
document, detailed role model
description)
•
Develop individual case
documentation and review
•
Do a cross-
case comparison based
on the reference model
Activities:
•
Review the current state of data
governance research
•
9x90-min focus groups with 34
data governance experts from 17
companies to discuss their current
approach, challenges
•Develop a reference model for data
governance and map individual
cases onto the reference model
Outcome:
•
Case documentation
•
Common patterns in data
governance setups
Outcomes:
Reference model for data
governance and theoretical
integration into the VSM
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 6
3.2 Collaborative practice research
In our collaborative practice research, we partnered with 17 companies seeking to benchmark their data
governance approaches. We organized nine 90-min focus groups with 34 high-profile data experts,
where participants provided an overview of their data governance approach, as well as describing its
evolution over time, which gave all participants a first understanding of their data governance
mechanisms. Besides the focus groups, we undertook research activities to review the literature on data
governance and develop a reference model as basis for the benchmarking study. This study was used to
map and compare individual companies’ governance approaches. Using purposeful sampling (Patton,
1990), we identified five companies’ data governance approaches for the subsequent case study analysis
(see Table 1 and section 3.3). Our interactions with the five case companies informed the subsequent
focus groups iteratively. The final focus group consisted of 22 data executives from the 17 companies
who discussed the findings, i.e., the reference model and the benchmarking study with illustrations from
the five cases.
3.3 Case studies
To be able to generalize a VSM, we opted for multiple cases as this supports better analytical
generalization (Yin, 2018). We selected companies with diverse characteristics regarding their industry,
the goal and scope of their data governance, and different organizational structures influencing the
design of global and local data governance teams. The case companies had implemented federated data
governance design decisions, e.g., they had complete role and process models at global and local levels.
Case,
Industry
Revenue/
Employees
Key
informant
Data governance’s goal
and scope
Global data governance
Local data governance
ManufCo
Automotive
manufacturing
$1B–$50B/
~90,000
VP Data &
Analytics
Governance
Enterprise-wide data
governance on 44 data
domains to stimulate data
use in all business units
and address all strategic
areas of digitalization.
Data and analytics
governance team (13
people) reporting to the
Chief Digitalization
Officer.
Data and analytics
coordination in each of
the 12 organizational
areas, i.e., functions,
divisions, regions (100
people for data
management).
BeautyCo
Adhesives &
Beauty
products
$1B–$50B/
~20,000
Director
Master Data
& Product
Lifecycle
Enterprise-wide master
data governance on two
domains (products,
finance) to improve
operational processes and
to improve value
generation from data.
Master data team (35
people) split between
business (supply chain,
finance) and IT with
respective reporting lines.
Three regional data hubs
close to the markets and
overseeing data lifecycle
in different countries (25
people).
PharmaCo
Pharma,
Chemicals
$1B–$50B/
~100,000
Head of Data
Framework &
Stewardship
Enterprise-wide
analytics-driven data
governance supporting
Analytics & AI
innovation, enablement,
and solutions.
Data Framework and
Stewardship (30 people)
in the Data & AI
Competence center
reporting to Global
Digital Services.
20 divisional digital
offices with about 200
data stewards.
EnergyCo
Energy
$100B-
$500B/
~100,000
Chief Data
Officer
Enterprise-wide data
governance on 16 data
domains defined
according to the business
model to drive data use
into operational
processes.
Small chief data office
focusing on data
foundation (5 people).
35 Chief Data Officers
allocated to divisions
with a small team each
(70 data architects in
total).
SoftCo
Software
$1B-$50B/
~110,000
VP - Head of
Data
Management
Enterprise-wide master
data governance on two
business-critical data
domains (products and
customers) to improve
operational processes.
Intelligent data
management (IDM) team
(98 people) in the Chief
Data Office reporting to
COO.
Three regional hubs (20
people in Europe, APAC,
South America),
Outsourced (80 people in
India).
Table 1. Cases overview.
To gain in-depth insight on the five companies’ federated data governance approaches, we conducted
semi-structured interviews with key informants who had been mandated to oversee enterprise-wide data
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 7
governance in the case companies. We selected only interviewees who had worked at the company for
an extensive period (>3 years), who knew the history of data governance initiatives, and had experienced
the issues and challenges associated with implementing data governance, such as involving business
stakeholders across different regions and divisions or assigning roles and responsibilities. We designed
our interview questionnaire to capture the strategic context and scope for data at the company, and we
complemented it with questions that address the three generic data governance mechanisms (see Table
2). Two researchers conducted the interviews via MS Teams video conferencing. Each lasted, on
average, 90 minutes as planned. The interviews were recorded and documented according to a pre-filled
template structured around guiding questions. After the interviews, we asked the informants to review
our interview reports and to confirm our documentation (e.g., key statements), and to address remaining
questions. The continuous interaction within the focus groups raised additional data requests, which we
addressed in follow-up discussions. After each interview, we performed an additional search for
secondary materials that could add to our documentation (e.g., a data strategy document, a detailed role
model, the structure of the primary organization), sometimes guided by the expert himself, e.g., to look
something up on the company website. To ensure construct validity and reliability of our findings, we
triangulated our interview data with further documentation (e.g., company presentations) that we had
collected during our research program or from public sources (e.g. presentations at practitioner
conferences, annual reports). The final set of data allowed us to obtain granular and complete details on
each data governance approach covering all three governance mechanisms. Overall, we obtained a rich
case study database built on a chain of evidence composed of primary and secondary data.
Protocol areas
Guiding questions
Strategic
context and
scope
Strategic
context
What are the drivers for data and analytics in the company? Do you have a data and/or analytics strategy?
If yes, since when and what is its focus? What business value and benefits do data and analytics create?
What are your top five data projects?
Scope
Which data domains do you distinguish? How do you define them? Which data types are established or
emerging? Which data and analytics products do you deliver?
Governance
framework
Structural
What organizational form has been chosen (e.g., line function, shared service)? Is the global
team/department part of the primary organization and, if so, where is it located in the organizational
structure? What are the responsibilities, headcount, structure, and composition of data and analytics
teams? Are there any boards and committees for data and analytics? What is their role?
Procedural
Which data management processes have you established? Which steps/tasks are taken over by the global/
local data organization? Which analytics processes have you established? Which steps/tasks are taken
over by the global/local data organization? How do you monitor data governance progress and success?
Which metrics do you use and how do you report them?
Relational
How do you align and collaborate with business stakeholders? How do you align and collaborate with IT
stakeholders? How do you align and collaborate between data and analytics? Which data/analytics
communities exist? How do you engage with them?
Table 2. Semi-structured interview protocol.
In analyzing our data, we applied abductive reasoning because it allows for embedding empirical
findings into an existing theoretical model (Ketokivi and Mantere, 2010). This approach facilitated
theorization through a detailed examination of the data by employing inductive coding for categorizing
interview data and deductive coding for incorporating the VSM perspective. Figure 4 presents the coding
process and illustrates the data analysis process with illustrative quotes from one of the cases. First,
using inductive coding, the same two researchers labelled the statements following a bottom-up
approach to derive open codes (Gioia, Corley and Hamilton, 2013). Next, they identified relationships,
connections, and patterns between open codes, thus bringing a more comprehensive understanding of
the underlying concepts. This led them to a set of axial codes reflecting data practices. Last, they used
selective coding to derive core themes that describe clusters of these practices. They then used deductive
coding to apply the VSM lens. They focused their analysis on assigning data practices for each of the
five sub-systems so that they could clarify how the practices are distributed at various levels in the
organization. Eventually, they obtained the grouping of the data practices into larger themes that map
onto VSM sub-systems.
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 8
Figure 4. Data analysis process exemplified with quotes from ManufCo case.
We did our cross-case analysis in the form of a comparative analysis of the five cases. A cross-case
analysis is particularly relevant to this research as it supports the aggregation, simplification, and
generalization of complex cases (Miles, Huberman and Saldaña, 2014). For this, we leveraged a granular
understanding of each data governance approach. We searched for differences and commonalities
between cases by iteratively searching for similarities between codes. From the emerging patterns, we
were able to generalize a VSM for data governance by reviewing common data practices (axial codes)
and core themes (selective codes) necessary to describe each of the five sub-systems.
4 A Viable System for Data Governance
From our cross-case analysis, we theorized a VSM for data governance that addresses both global and
local data governance activities. We find that data governance should occur at multiple, interconnected
levels, i.e., in sub-systems (see Table 3): S2, S3, S4, and S5 together form a metasystem of data practices
performed in operational units (S1). While S2, S3, and S3* represent the data governance layer (i.e., the
data governance teams) and orchestrate data practices, S4 and S5 form the strategy layer (through boards
and committees) and shape data governance practices. In the following sections, in describing the
different systems, we exemplify the VSM with examples and quotes from our cases (e.g., regional data
governance at BeautyCo, divisional data offices at EnergyCo). We also clarify the notion of recursion
in the VSM using ManufCo’s federated data governance as a critical case (Yin, 2018).
4.1 Operations – Perform data practices
System 1 has a set of operative units which are typically business functions that embed data in their
work practices. These units provide data to their members and to other units, and consume data provided
by their members or by other units, for instance in creating dashboards, reports, and increasingly also
feeding advanced analytics/machine learning models. Two key data practices enact data provisioning,
namely data creation and data curation. Data creation involves the intentional and systematic
generation of data through various processes, for instance if the account manager in a regional sales
team creates a customer record. Data curation involves the deliberate and systematic maintenance of
data throughout its lifecycle to ensure that the data is processed in compliance with regulations and is
fit for purpose (data quality). As EnergyCo stated, “No-one owns the data lifecycle other than the data
domains themselves.” To support operative units, all five cases use shared service centers that handle a
part of the data curation tasks, as shown at SoftCo: “We have a team called ‘data operations’ that
executes data processes. For this, we have a three-level classical shared services setup. We have a
follow-the-sun approach with two regional teams in Prague and Manilla, it’s about 20 people. We also
consider a third offshore team in Brazil. We have a first level outsourced to a consulting company in
India, which works with ticketing. There we have another 80 people. It seems really big but actually this
is where we provide data maintenance services for all market units at the firm worldwide.” Business
Open codes
Empirical evidence
(Triangulated data)
Inductive coding: retrieve practices from literature and uncover new practices by encoding data into first and second order categories
(Gioia et al., 2013)
transformed into
Axial codes
(Data practice)
aggregated into
Data standards and guidelines
Data domain guidelines
Data quality monitoring
Deductive coding: apply VSM lens to
map data governance practices onto the
VSM’s five sub-systems (Beer, 1985)
Selective codes
(Core theme)
Data practices
oversight
mapped onto
“We provide guidelines on how to
manage a data domain, we have a
template too.”
“We have 35 KPIs for data quality
and we monitor them at domain
level; then we aggregate in a
corporate-wide maturity index.”
“It’s always the same thing. We
define standards that apply to all
domains, otherwise domains do
it.”
Performance management
Data standards System 3 (Control)
Deductive codes
(VSM’s subsystems)
mapped onto
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 9
functions are also responsible for addressing data consumption requests and should ensure that data
quality follows both standards and consumer expectations. Hence, operative units take ownership of
their data and manage data accessibility and data sharing in accordance with data access rights, as
articulated by EnergyCo: “We try to make data discoverable for possible usage through our data
catalog, Collibra. We have started working with the business to define the key curated data products
that we would like to see in place.” Data usage practice implies that business units use the data for
operational and analytics purposes (e.g., in analyzing the data to create a sales forecast). Data can be
consumed within the business functions or by outside units that need it to perform their own data analysis
or to enrich their own business function’s data.
Systems
Theory (Beer, 1985)
Description
Key data practices
Layer
System-
in-focus
S1
Describes the different
operative units that execute
the tasks expected to fulfil the
system’s purpose.
Represents all business units where data
practices are embedded in work
practices and performed by providers
and consumers of data.
• Data creation
• Data curation
• Data usage
Operations:
Perform data
practices
Meta-
system
S2
Handles coordination and
communication across the
different S1s, especially
during disturbances affecting
the VSM (e.g., environmental
fluctuations).
Ensures coordination between data
governance teams by assigning data
roles and responsibilities and
distributing the latest governance
principles to the entire network. It also
provides data management support,
training, and data applications to data
providers and consumers.
• Definition of data
roles and
responsibilities
• Data enablement
• Data management
support
• Data documentation
and architecture
• Data applications
specification
Governance:
Orchestrate
data practices
S3
Oversees the activities of the
system-in-focus (S1) through
“day-to-day management” to
ensure the smooth delivery of
data operations against
strategic goals.
Oversees all data practices in the
system-in-focus (S1) and ensures that
they are performed in line with strategic
goals and according to standards and
guidelines (e.g., for data collection,
storage, use, documentation). Monitors
the execution of the data strategy and
provides periodic strategic reporting.
• Definition of data
standards and
guidelines
• Performance
monitoring and
improvement
S3*
Complements System 3 act as
a compliance system of
operative unit (S1).
Performs data-related audits of operative
units to ensure compliance with laws,
regulations, and standards.
Data compliance
auditing
S4
Senses threats and
opportunities to the system by
scanning the environment.
Senses data opportunities (e.g., trends)
and threats (e.g., compliance) that could
impact the data organization.
Data threats and
opportunities sensing
Strategy:
Shape data
governance
practices
S5
Maintains the system’s
identity by describing the
system’s norms and purpose.
Provides strategic direction for the entire
data activities in alignment with
company strategy.
Data strategy
definition and
monitoring
Table 3. VSM sub-systems and their application to data governance.
4.2 Governance – Orchestrate data practices
System 2, taking care of coordination, is managed by the data governance team, be it at global or local
level. Its role is to communicate about data governance and to coordinate the network of data providers
and consumers (S1). Thereby, it ensures alignment at enterprise-wide level, be it between data providers
and data consumers within an operative unit (S1), or between several operative units (e.g., in data sharing
between customer and sales data domains). We identify five key data practices enacted by S2, which
are definition of data roles and responsibilities, data enablement, data management support, data
documentation and architecture, and data applications specification. Definition of data roles and
responsibilities is an established data governance practice that involves defining, assigning, and
communicating data-related roles and responsibilities, such as those of data stewards or data editors.
This practice also clarifies the role-players’ interaction and the collaboration models. For BeautyCo,
“the role definition is central, but the execution happens in regions. For this, we set up the regional hubs
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 10
and the roles have solid reporting lines to regional offices. But they also have a functional reporting
line to me.” Data enablement comprises an emerging set of data governance practices focused on
empowering individuals and teams to harness the full potential of data by providing the necessary tools
and skills. Typically, as the number of employees in data roles grows, increasing data literacy, for
instance through training programs, is a priority. Firms also initiate global data culture initiatives, as
EnergyCo explains: “We have a company-wide initiative called ‘The year of data,’ which is about
raising data awareness by showcasing three things: what you can do with data in general, where the
company stands and what it struggles with, and what can be done. We also follow up with a data mood
survey.” Executives at ManufCo drive data-driven culture with axioms such as “Data belongs to all
employees, and all can benefit from knowledge of the data”, “We acknowledge the value of data for the
company”, “We pay attention to error-free data and thereby guarantee a high level of customer
satisfaction.” However, due to the growing business ownership of data, data enablement must also
reach locally, as BeautyCo states: “Data enablement is central and regional. In the future, we want most
regional hub interactions to have local functions. For instance, our hub in Poland is quite active and
does a lot in this instance. They have built their own way of communicating with newsletters and so on.
They are very good. We are learning from them.” Data management support involves all data
governance practices aimed at coordinating business and project support (e.g., compliance with data
strategy, data needs), coordinating requirements with technical teams (e.g., data engineering), and
generally ensuring functional communication across the different units. Data documentation and
architecture practice involves systematically creating and updating comprehensive metadata
documentation. Thereby the organization creates transparency regarding its data. This is achieved by
designing and evolving the data architecture, and by how data is collected, stored, processed,
documented, and used. Data applications specification aims to define the supporting applications for
data provision and consumption. Applications with data governance in scope are typically related to
master data management (e.g., SAP MDG), data quality, and data cataloging. As PharmaCo explains:
“I am responsible mostly for the content part. Our task is to translate the technical data into meaningful
content. To make the data understandable and consumable for the entire organization, we manage the
company-wide data catalog, and along with our divisional stakeholders we are filling it in. We also use
a tool to build ontologies and knowledge graphs.” Governance practices around data applications are
performed in collaboration with IT (especially for the platform side). This involves defining the
functional requirements, change management, workflows, and UI modelling.
System 3, taking care of control, monitors all data practices in S1 and ensures that they are performed
in line with strategic goals and according to set standards and guidelines. At the interface of operations
and strategy, System 3 plays a pivotal role in standardizing data practices, as well as in strategy delivery
and reporting. It displays data governance practices identified as (i) definition of data standards and
guidelines, and (ii) performance monitoring and improvement. The definition of data standards and
guidelines involves creating a data governance framework, developing a local data ownership concept,
data process documentation, and data access rights. Control is typically exercised by both global and
local data governance teams, as ManufCo highlights: “Standards and guidelines mainly come from us
and are enriched in the specific domains. For instance, we do not give the guidelines for maintaining
payment conditions; this is the task of the finance data domain.” Performance monitoring and
improvement is an emerging data governance practice that pertains to the structured methods and tools
an organization employs to monitor, measure, and enhance data quality and data-related processes,
through, for instance, maturity assessments. While firms traditionally monitor data quality, they now
progressively integrate data consumption in their metrics framework (e.g., in the growing number of
data objects available on the data catalog at ManufCo and BeautyCo). At BeautyCo, “we measure the
increasing number of data objects on the data catalog. For success, we measure time-to-market in
regional hubs. We also monitor how the number of GTIN violations decreases.”
System 3*, the audit, complements System 3 by auditing data practices of operative units, thereby
ensuring that they agree with legal requirements, industry standards, internal policies, and data standards
and guidelines. It is mainly enacted through data compliance auditing practices which enforce adherence
to rules, regulations, and standards that govern the collection, storage, processing, and sharing of data.
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 11
For instance, at ManufCo, “data management is a mandatory, auditable process in the quality
management system. To support IT security and data protection, delicate data objects are flagged as
sensitive in the data model. Data domains that contain intellectual property are also closely monitored
to address potential risks and to initiate risk mitigation.”
4.3 Strategy – Shape data governance practices
System 4, related to intelligence, ensures that the whole system can adapt to disturbances by scanning
the environment to detect changes (e.g., new data trends, use cases) and by proposing mitigation plans.
It is mainly performed through the strategic practice of data threats and opportunities sensing and
involves actively monitoring, identifying, and responding to potential risks or beneficial situations in
the organization's data landscape. This proactive approach enables timely mitigation of threats, such as
to data security, and exploiting opportunities, such as new use cases for emerging technologies (e.g.,
Generative AI). A new local regulation can also impact the data activities, as raised by ManufCo: “Let’s
say we want to react to the EU data governance act. It will be discussed in the data council but due to
the effects on other enterprise areas we would also put it to the digital coordination council and to the
board. We also take it to the global risk and compliance committee, and to several other committees
that I am not going to list right now. So, it impacts way more than just data.” For this reason, and due
to the critical role System 4 plays in the system’s viability, companies might use ambassadors at
executive level to help with communication and collective acknowledgement. As SoftCo observed: “We
have a super senior executive coms person in our team. This is one of my biggest assets. Yes, I sit in the
organization at level three, but I communicate with everybody, including senior executives and the
board. This is sometimes challenging, especially if you want to discuss data topics at a business and
strategic level. The role is called ‘executive communications lead’ and helps us neutralize emotions and
politics that come with data topics at strategic level.”
System 5, dealing with policy, provides strategic direction for all data activities aligning with the
corporate strategy and business priorities. Strategic data practices introduced here revolve around the
enterprise-wide data strategy definition and monitoring customs and consist in planning,
implementing, and optimizing systematic approaches to create value from data. It also involves
identifying and assessing the data capabilities required to enable the business model. For instance, the
opportunities Industry 4.0 offers and the C-level's recognition of data’s strategic value led to ManufCo
updating their integrated data and analytics strategy in 2022. In fact, all cases had recently updated their
data strategies with a shift toward innovation and value creation from data. As PharmaCo explained:
“We are still working on our enterprise-wide data strategy. In such a big company, this is a long-term
project. Our team manages it because it is not about technology; it is about communities, change,
culture, this seamless data experience we want to bring. We also propose shifting to a kind of global
data office in combination with larger domain responsibilities.”
4.4 Federated data governance as recursive system
Consistent with existing literature, we note that global firms adopt a federated governance model, albeit
with various, sometimes subtle, distinctions. Using the VSM, we find that these federated data
governance practices unfold through several systems-in-focus (i.e., multiple System 1). Consequently,
global data governance practices can be distributed by being embedded, and often enriched, in local
systems which mirror the primary organization's existing regional, divisional, and functional structure.
This indicates a recursive logic with two (possibly more, depending on organizational structure)
systems-in-focus: (1) at level “n”, the totality of corporate data practices governed by global data
governance practices, and (2) at level “n+1”, local data practices governed by local data governance
practices. Above, in describing the different systems, we have already exemplified the recursion in the
VSM, giving various examples and quotes from our cases (e.g., the regional data governance at
BeautyCo, divisional data offices at EnergyCo). Following here, we clarify the notion of recursion in
the VSM by providing a vignette that illustrates ManufCo’s federated data governance approach.
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 12
Our analysis disclosed that ManufCo’s approach is the most advanced of the cases in that its data
operating model covers data governance practices at a global level and on a local level operates in
regions, divisions, and functions. The board’s publication of the so-called “digital agenda” was an
important contribution to securing the company’s long-term competitiveness, marking a paradigm shift
in the role of data, which now forms a “core component of value creation.” Consequently, ManufCO
launched in the project “Data Domain Management in all Data Areas,” working with the global data
governance team to remove bottlenecks and to establish a network of data roles spread globally (across
functions, divisions, and regions). This shift triggered an extension of the data scope from key master
data objects (such as suppliers and customers) to 44 data domains relevant for digitalization. Examples
of these domains are “R&D Engineering”, “Sales”, “Manufacturing Planning”, and “Finance
Accounting”.
ManufCo’s VSM displays a patent example of recursion, showing that most data practices enacted in
the five sub-systems are replicated at data domain level. Each data domain is itself a viable system that
strategically self-organizes, independently from other domains. Data domain governance controls and
coordinates domain data practices (e.g., data creation, usage, and maintenance in the data domain only),
and interacts with the local environment (e.g., correlating function, division, region, and outside world).
Figure 5 shows the role of each sub-system and highlights how key structural mechanisms (e.g., boards,
teams, roles) can be mapped onto them. Next, we describe the five sub-systems, thereby showing the
interplay between data strategy, data governance, and data operations.
Figure 5. Viable System Model for data governance at ManufCo.
System 5 is enacted through the Digital Transformation Council (DTC). It is composed of six company
board members who meet bi-annually to monitor the progress of the so-called “digital agenda,” which
is the digital arm of the company’s strategic goal to be the technology leader in the "mobility of
tomorrow." More specifically, the DTC aims to secure the company’s long-term competitiveness
through a paradigm shift in the role of data, which now forms a “core component of value creation.”
Concretely, having discussed data vision (e.g., data monetization), key drivers (e.g., data
democratization, data economy), and associated KPIs, the DTC formulates the data strategy.
System 4 is enacted through the Data Council, the organizational body responsible for the underlying
data-related activities, their prioritization, oversight and alignment, and possible implementation issues
(S4). It is composed of 20 members that include the head of data and analytics governance, all digital
transformation officers (one per division, function, and region), the head of compliance, the CIO, and
the head of enterprise architecture. During quarterly meetings they discuss topics such as how to react
S5: Digital transformation
council
Environment
S4: Data council
S3: Global data and analytics
governance team
Functional
data
operations
Functions (e.g.,
Finance, IT, HR)
Divisions (e.g., Auto
OEM, Industry)
Regions (APAC,
China, Europe,
Americas)
Future
env.
Local
env.
Local
env.
Local
env.
S3*: Internal
audit S2: Data and
analytics coordinators
Divisional
data
operations
Regional
data
operations
•All functions, divisions, and regions provide and consume data
according to data governance standards and guidelines.
•20 data domain managers allocated to functions, divisions, and regions
oversee, adapt, and instantiate global principles to control, monitor,
and strategically develop data management in 44 data domains.
•Data domains are also accountable for ensuring and reporting on data
quality in their domains to the global team so that data can be
consumed by others. For this, about 100 data stewards appointed in
domains support by checking data documentation and data quality.
•Data consumption can happen within a business unit or by requesting
data from other units with the help of data and analytics coordinators.
S1: Operations Governance Strategy
•The global data and analytics governance team (13 employees)
defines standards, guidelines, and rules that define how data should be
managed, by whom, and with which applications. It defines, monitors,
and reviews metrics that capture both operational and governance
performances (e.g., data maturity assessments).
•Local data governance responsibilities are delegated to 20 data and
analytics coordinators sitting in functional and divisional digital
transformation offices.
•Operational compliance with data governance is part of internal audit.
•The digital transformation council is composed of the company’s
board members who meet 6 times a year to define vision, mission, and
strategic targets for data, following the firm’s digitalization strategy.
•In the data council (20 employees), the head of data and analytics
governance, all digital transformation officers (one per division,
function, and region), the head of compliance, the CIO, and the head of
enterprise architecture meet 4 times a year to discuss data opportunities
and threats to the digitalization program.
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 13
to new regulations (e.g., the EU data act), or how data can support the different business processes in
creating business value (e.g., where to find trustworthy data, what count as dependencies, as key
vocabulary, as important security and privacy aspects, and as business processes’ requirements). The
data council also manages the data domains portfolio that currently includes 44 data domains. Eligibility
questions for opening new data domains typically include: Which business processes or other principles
would justify a new domain? Does the corresponding function or division generate its own data (e.g.,
specific data entries)? Would the domain be temporary or sustainable in the long term? Would data be
useful in all departments? Would the domain be cross-functional? Are there synergies with other
domains that could justify an integration/merger? Would setting up a regional satellite for this domain
be wise? Based on lean templates, each data domain’s profile is documented (e.g., in a description of its
content and data objects, sensitive data, relevant business processes).
System 3 is enacted through the global data and analytics governance team. Composed of 13 experts
who control all corporate data operations, the team provides the general data standards and guidelines
applicable to all domains, and it monitors various metrics to demonstrate progress on the data strategy,
such as data quality improvements, the data documentation rate, data tools use, and data literacy
assessment. The team also gets support from internal audits to assess various data domains’ compliance
with global standards and guidelines.
System 2 is enacted by 20 data and analytics coordinators who act as counterpart in division, functions,
and regions, who sit in the respective digital transformation offices. They communicate the global
standards provided by System 3 to all domains. This way, the entire network builds knowledge of the
data strategy, data roles and responsibilities, data processes, data applications, data model, and data
quality. Further, they provide data management support, for instance by coordinating data provisioning
and data consumption requests across operational units. This is facilitated by the “Data Domain Manager
Round Table” that enables cross-domain practice exchange.
System 1 represents all operational data practices across functional, divisional, and regional units. Each
corresponding data domain takes ownership for creating, curating, and using their data or using other
domains’ data. Recursively, in each domain, data domain managers adapt global principles and define
their own data domain principles, i.e., they control, monitor, and strategically develop data management
in their data domain. These managers are also accountable to the global team for data quality in their
domains, ensuring the quality and reporting on it. For instance, the Finance Accounting domain gets
contributions from other domains, e.g., gaining inventory data that belongs to the Storage and Shipping
domain or costing data that belongs to the Sales and Marketing domain. The regional data domain
managers are responsible for coordinating the data domains in a given region, thus linking the data
domain manager to the operative business units (e.g., helping to define the access authorization concept
in compliance with local regulations). Data stewards support the data domain manager in documenting
data (e.g., metadata) and maintain data quality in each domain by integrating business knowledge in
data curation tasks. They are also responsible for responding to data users’ data access requests, in- or
outside the domain.
Since implementing their federated data governance model in 2021, ManufCo has observed substantial
improvements in business performance. The duration of both the “Initial Order” and the “Request for
Quotation” processes were significantly reduced thanks to data quality improvements. Further, the
recorded cycle time of the business partner approval process was reduced by 30%, the cycle time of
intercompany service requests by 45%, and ManufCo could achieve a striking 97,3% duration reduction
in all processes within the 24-hour timeframe agreed in the service level agreement. These results show
the critical role of data governance in enabling innovative local data practices.
5 Contribution, Discussion, and Implications
Although the foundations of data governance have reached a sound level of clarity, much of the research
to date remains conceptual and proposes generic, static mechanisms. This study is among the first to
focus on the implementation of data governance mechanisms and their adaptation in large and complex
organizations. Our results explain how data governance unfolds in practice in multinational companies
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 14
through a viable system composed of multiple, interconnected levels, i.e., sub-systems with their own
sets of data practices. The application of the VSM in this study demonstrates that data governance cannot
be seen only as a static framework that shapes structural, procedural, and relational mechanisms; rather,
it needs a dynamic framework that supports the expansion of data practices in all areas of the
organization. This is in line with Vial (2023, p. 6) who stresses that “the instantiation of this design in
practice is important to understand how an organization protects and leverages data for digital
innovation.” Overall, the use of the VSM supports a better understanding of such seemingly paradoxical
activities by explicating both the dynamics of control (e.g., data protection) and the dynamics of value
creation (e.g., from data use).
Our results confirm and extend prior research, arguing that global firms adopt federated (also called
hybrid) models for data governance (King, 1983; Grover et al., 2018). Through the lens of the VSM, we
show how companies thoughtfully merge and maintain global responsibilities, such as universal
standards, protocols, and methodologies, with local responsibilities that are uniquely tailored to
individual business units, including data quality monitoring and project execution. This model involves
transferring certain data governance responsibilities from the global data governance unit and assigning
new responsibilities to local roles in business (e.g., data steward). Data access is mainly managed by
business experts (i.e., data managers) themselves, following corporate policies set by the global data
team (System 3). This obliges the audit function (System 3*) to take on additional responsibilities that
will mitigate data management risks. Overall, while global data governance fosters uniform enterprise-
wide data management principles, standards, and methods, federated data governance practices favor
local business expertise. System 2 is then crucial for cross-functional projects and network enablement.
We find that data governance practices are enacted according to an organizational hierarchy, thus not at
the same level. The recursion highlighted in our VSM demonstrates that federated data governance is
enacted through a cascading system that assign data governance responsibilities across multiple hubs
typically aligned on the organization's primary structure (e.g., corporate, functional, regional). This
model further branches out through "spokes," representing the data creators and users within the
business, ensuring that governance reaches all levels of operation. Hence, unlike a hub-spoke model that
centralizes data governance responsibilities at a corporate level, hub-hub-spoke models, which can
embed more than just one recursion, offer numerous benefits such as greater local autonomy, use of
domain knowledge, faster issue resolution, and improved agility. For their respective sectors, each hub
sets strategic data objectives, defines data standards and guidelines, creates transparency on data, and
fosters data enablement. In return, a hub-hub-spoke model generally requires better coordination
mechanisms (e.g., a data council, data communities, local hub monitoring). However, coordination
mechanisms (System 2) generally “do not arise prior to coordinating but are constituted through
coordinating” and they typically follow a system’s disruption (Jarzabkowski, Lê and Feldman, 2012, p.
907). This highlights the pivotal role of environment sensing on both a corporate and on local levels to
continuously update data coordination mechanisms. For instance, the strategic need to develop new
analytics use cases (e.g., Generative AI) might enlarge the scope of data governance (e.g., extending to
new data types) and trigger an update on the role and board model. Future research could investigate
hub-hub-spoke models in greater detail, and especially how they unfold into different organizational
structures. In this regard, the study of global-local coupling in federated data governance systems could
be an interesting starting point, for example, by examining the impact of external turbulences based on
the responsiveness and specificity of the system in focus (Weick, 1976; Orton and Weick, 1990). This
avenue could investigate how to build modularity, the right level of redundancies, adaptability, and
resilience into federated models.
From an academic perspective, the VSM perspective paves the way for investigating data governance
from a new angle. It contributes to the previously neglected dynamic nature of data governance and
addresses the need to investigate data governance in practice (Vial, 2023). The insights developed in
this study further provide valuable guidance on how to design the organizational counterpart to technical
data mesh principles by showing, for instance, how different business units enact ownership of their
data. Besides data creator and data user roles, our study shows that data steward and data (domain)
owner roles, which are seldom clearly distinguished and are often misunderstood (Vial, 2023), are
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 15
essential to the execution of domain-level data governance practices thanks to their knowledge of the
business context. Future research could further investigate the interaction between the technical
architecture and the operating model for data governance, especially considering the difficulty of
knowledge integration and the data literacy gap between business and analytics teams (Kollwitz,
Mengual and Dinter, 2018; Someh et al., 2023). From a practical perspective, our findings support
decision makers in global firms to define, adapt, and implement data governance. They can leverage the
VSM to build their own federated data governance framework, that addresses both global and local
levels.
Since this study takes a new, systems theory approach to examining data governance, it is inherently
prone to first mover limitations, and we strongly encourage future research in this area. Beyond the
potential future research activities mentioned above, the understanding of data governance as a self-
organizing system could be further deepened. As this study focused mainly on elucidating the five sub-
systems, our findings also open avenues for further research into how antecedents affect data governance
as a system. For instance, researchers could investigate how different industries’ strategies and operating
environments impact the system’s viability. In such a case, certain principles from VSM theory, like
variety and transduction (Beer, 1985) with which this paper could not deal extensively, provide
interesting possibilities for refining the model.
References
Aaltonen, A., Alaimo, C. and Kallinikos, J. (2021) ‘The Making of Data Commodities: Data Analytics
as an Embedded Process’, Journal of Management Information Systems, 38(2), pp. 401–429.
Abraham, R., Schneider, J. and vom Brocke, J. (2019) ‘Data Governance: A Conceptual Framework,
Structured Review, and Research Agenda’, International Journal of Information Management, 49,
pp. 424–438.
Alhassan, I., Sammon, D. and Daly, M. (2016) ‘Data governance activities: an analysis of the literature’,
Journal of Decision Systems, 25(sup1), pp. 64–75.
Baijens, J., Huygh, T. and Helms, R. (2021) ‘Establishing and theorising data analytics governance: a
descriptive framework and a VSM-based view’, Journal of Business Analytics, pp. 1–22.
Beer, S. (1985) Diagnosing the system for organizations. Repr. Chichester, UK: Wiley (The managerial
cybernetics of organization).
Benfeldt, O., Persson, J.S. and Madsen, S. (2020) ‘Data Governance as a Collective Action Problem’,
Information Systems Frontiers, 22(2), pp. 299–313.
Brown, C.V. (1999) ‘Horizontal Mechanisms under Differing IS Organization Contexts’, MIS
Quarterly, 23(3), p. 421.
Dubé, L. and Paré, G. (2003) ‘Rigor in Information Systems Positivist Case Research: Current Practices,
Trends, and Recommendations’, MIS Quarterly, 27(4), pp. 597–636.
Fadler, M., Lefebvre, H. and Legner, C. (2021) ‘Data governance: from master data quality to data
monetization’, in. Proceedings of the 29th European Conference on Information Systems (ECIS), An
Online AIS Conference.
Gioia, D.A., Corley, K.G. and Hamilton, A.L. (2013) ‘Seeking Qualitative Rigor in Inductive Research:
Notes on the Gioia Methodology’, Organizational Research Methods, 16(1), pp. 15–31.
Grover, V. et al. (2018) ‘Creating Strategic Business Value from Big Data Analytics: A Research
Framework’, Journal of Management Information Systems, 35(2), pp. 388–423.
Günther, W.A. et al. (2022) ‘Resourcing with data: Unpacking the process of creating data-driven value
propositions’, The Journal of Strategic Information Systems, 31(4), p. 101744.
Huygh, T. and De Haes, S. (2019) ‘Investigating IT Governance through the Viable System Model’,
Information Systems Management, 36(2), pp. 168–192.
Jarzabkowski, P.A., Lê, J.K. and Feldman, M.S. (2012) ‘Toward a Theory of Coordinating: Creating
Coordinating Mechanisms in Practice’, Organization Science, 23(4), pp. 907–927.
Jones, M. (2019) ‘What We Talk about When We Talk about (Big) Data’, The Journal of Strategic
Information Systems, 28(1), pp. 3–16.
Data governance as viable system
Thirty-Second European Conference on Information Systems (ECIS 2024), Paphos, Cyprus 16
Ketokivi, M. and Mantere, S. (2010) ‘Two strategies for inductive reasoning in organizational research’,
Academy of Management Review, 35(2), pp. 315–333.
Khatri, V. and Brown, C.V. (2010) ‘Designing Data Governance’, Communication of the ACM, 53(1),
pp. 148–152.
King, J.L. (1983) ‘Centralized versus decentralized computing: organizational considerations and
management options’, ACM Computing Surveys, 15(4), pp. 319–349.
Kollwitz, C., Mengual, P.M. and Dinter, B. (2018) ‘Cross-Disciplinary Collaboration for Designing
Data-Driven Products and Services’, in In Proceedings of the Pre-ICIS SIGDSA Symposium on
Decision Analytics Connecting People, Data & Things. San Francisco, USA, p. 16.
Lefebvre, H. and Legner, C. (2022) ‘How Communities of Practice Enable Data Democratization Inside
the Enterprise’, in Proceedings of the 30th European Conference on Information Systems (ECIS).
Timisoara, Romania, p. 16.
Legner, C., Pentek, T. and Otto, B. (2020) ‘Accumulating Design Knowledge with Reference Models:
Insights from 12 Years’ Research into Data Management’, Journal of the Association for Information
Systems, 21(3), pp. 735–770.
Machado, I., Costa, C. and Santos, M.Y. (2021) ‘Data-Driven Information Systems: The Data Mesh
Paradigm Shift’, in Proceedings of the 29th International Conference On Information Systems
Development. Valencia, p. 6.
Mathiassen, L. (2002) ‘Collaborative Practice Research’, Information Technology & People, 15(4), pp.
321–345.
Mikalef, P. et al. (2020) ‘The role of information governance in big data analytics driven innovation’,
Information & Management, 57(7), p. 103361.
Miles, M.B., Huberman, A.M. and Saldaña, J. (2014) Qualitative Data Analysis: A Methods
Sourcebook. 3rd edn. SAGE Publications.
Orton, J.D. and Weick, K.E. (1990) ‘Loosely Coupled Systems: A Reconceptualization’, Academy of
Management Review, 15(2).
Otto, B. (2011) ‘Data Governance’, Business & Information Systems Engineering, 3(4), pp. 241–244.
Parmiggiani, E. and Grisot, M. (2020) ‘Data curation as governance practice’, Scandinavian Journal of
Information Systems, 32.
Parmiggiani, E., Østerlie, T. and Almklov, P.G. (2022) ‘In the Backrooms of Data Science’, Journal of
the Association for Information Systems, 23(1), pp. 139–164.
Patton, M.Q. (1990) Qualitative evaluation and research methods. SAGE Publications, inc.
Peppard, J. (2005) ‘The Application of the Viable Systems Model to Information Technology
Governance’, in In Proceedings of the 26th International Conference on Information Systems (ICIS).
Las Vegas, NV, USA.
Sambamurthy, V. and Zmud, R.W. (1999) ‘Arrangement for Information Technology Governance: A
Theory of Multiple Contingencies’, MIS Quarterly, 23(2), pp. 261–290.
Someh, I. et al. (2023) ‘Configuring Relationships between Analytics and Business Domain Groups for
Knowledge Integration’, Journal of the Association for Information Systems, 24(2), pp. 592–618.
Tallon, P.P., Ramirez, R.V. and Short, J.E. (2013) ‘The Information Artifact in IT Governance: Toward
a Theory of Information Governance’, Journal of Management Information Systems, 30(3), pp. 141–
178.
Velu, C.K., Madnick, S.E. and van Alstyne, M.W. (2013) ‘Centralizing Data Management with
Considerations of Uncertainty and Information-Based Flexibility’, Journal of Management
Information Systems, 30(3), pp. 179–212.
Vial, G. (2023) ‘Data governance and digital innovation: A translational account of practitioner issues
for IS research’, Information and Organization, 33(1), p. 100450.
Weber, K., Otto, B. and Österle, H. (2009) ‘One Size Does Not Fit All---A Contingency Approach to
Data Governance’, Journal of Data and Information Quality, 1(1), pp. 1–27.
Weick, K.E. (1976) ‘Educational Organizations as Loosely Coupled Systems’, Administrative Science
Quarterly, 21(1), pp. 1–19.
Yin, R.K. (2018) Case Study Research and Applications: Design and Methods. 6th edn. Los Angeles:
SAGE.
