CHALLENGES IN ENTERPRISE ARCHITECTURE MANAGEMENT: OVERVIEW AND FUTURE RESEARCH

How to cite this paper: Brée, T., & Karger, E. (2022). Challenges in enterprise EA communication and support, EA programming, EA implementation, and EA governance — which are investigated separately. This article’s result is a comprehensive overview of research in the field of EAM challenges. Additionally, interdependencies between the dimensions are assumed. Furthermore, an outlook on future research opportunities from an organizational, corporate governance, project, and technical perspective is provided. Conceptualization T.B. and Validation E.K.; Formal Analysis T.B.; Investigation — T.B.; Resources — T.B.; Data Curation T.B.; Writing — Original Draft — T.B.; Writing — Review & Editing — E.K.; Visualization T.B.; Supervision E.K.; Administration — T.B. Declaration of conflicting interests: The Authors declare that there is no conflict of interest. Acknowledgements: We acknowledge support by the Open Access Publication Fund of the University of Duisburg Essen.


INTRODUCTION
In the current age of digitalization, the business world is changing rapidly and dramatically due to the usage of new information technologies (Urbach & Ahlemann, 2016). In order to stay competitive, it is a decisive factor to adapt fast to the new and steadily changing environment (Ahlemann, Stettiner, Messerschmidt, & Legner, 2012). However, enterprises are complex systems consisting of interconnected social, technical, and infrastructural elements. To be successful in this age, it is important for organizations to understand, engineer, and manage the complexity of the enterprise (Nightingale & Rhodes, 2004). A suitable approach to understand the entire organizational system, the interrelations, and to manage its complexity, is enterprise architecture management (EAM) (Lange & Mendling, 2011).
EAM is a well-investigated area in information system research. The management of enterprise architecture (EA) is a complex task due to the relationships and interactions between the elements of an EA (Jugel & Schweda, 2014). An EA represents the structure of the business its IT landscapes and their interrelations. It provides descriptions that focus on a domain-aspect (e.g., business or IT infrastructure) and time-aspect (e.g., as-is or to-be architecture) (Tamm, Seddon, Shanks, & Reynolds, 2011). Thereby, an EA can consist of thousands of different business applications ). Stakeholders have different information needs of an EA that range from only knowing the used applications, the interdependencies between objects, up to an overview of the entire EA (Rehring, Greulich, Bredenfeld, & Ahlemann, 2019b).
Despite this complexity, EA is important for business innovation, information technology success, and technology adoption (Stecher, Pohl, & Turowski, 2020). Information technology (IT) is increasingly important, and many organizations rely on IT to be able to realize their objectives and activities (Batyashe & Iyamu, 2016). In the context of digital transformation, governing IT successfully is, therefore, an important aspect. The corporate board should thus pay attention to, for example, data governance ; Jagals, Karger, & Ahlemann, 2021) as well as information systems governance . By providing an overview and a well-founded basis for decisionmaking in enterprises, EA can help companies with decisions about, for instance, reduced costs and complexity, an increased business and process flexibility, or an improved business-IT alignment (Tamm et al., 2011).
This well-founded decision-making offers rational arguments about EA (van der Linden & van Zee, 2015). However, several researchers described a low usage of EAs in the context of decision-making in organizations (Hiekkanen et al., 2013;Löhe & Legner, 2014). This has its reasons, for instance, in the understanding and quality of the visualization of EA models Löhe & Legner, 2014). Regardless of its relevance, many companies are still facing numerous challenges when implementing EAM. For example, architects of an EA struggle with the rapid changes and the documentation of these tasks due to inadequate tool support (Kleehaus & Matthes, 2021). Another example is employees of organizations criticizing a supporting lack from the EA based on missing resources like time or know-how (Uludag, Kleehaus, Reiter, & Matthes, 2019).
In recent research, different literature reviews and bibliometric studies within the field of EA were published. For example, Simon, Fischbach, and Schoder (2013) provided a comprehensive overview of the literature about EA. As a more recent example, Gampfer, Jürgens, Müller, and Buchkremer (2018) conducted a holistic and systematic literature review (SLR) and presented a historical overview of research about EA. Research has also described numerous difficulties in the implementation of EAM tasks, already. To the best of our knowledge, however, there is still no overview or literature review with a clear focus on the identified challenges related to EAM. This lack of research is surprising, as the area of EAM is an important field for practice and many challenges seem to remain unsolved. Furthermore, a lot of interdependencies exist between the different EAM tasks and challenges. Therefore, a view covering the area of EAM as a whole is necessary to successfully address existing challenges. This review wants to close this research gap and aims to structure the research field of EAM challenges, identify open research directions, and thus serve as a platform for future research (Paul & Criado, 2020 The rest of this paper is structured as follows. In the next section, the theoretical foundations of EA and EAM are described to ensure a common understanding.
The third section describes the applied research method, namely an iterative SLR that was conducted to identify the tasks of EAM as well as their challenges. The results of the literature review about EAM tasks and their challenges are outlined in the findings section. In Section 5, the contributions of the research are presented, as well as an outlook on future research opportunities, based on this paper's results. This article ends with a conclusion that includes a short summary of the results.

THEORETICAL FOUNDATIONS
Enterprises struggle with the complex information technology environment and their frequent changes Winter, Legner, & Fischbach, 2014). However, for a successful enterprise, it is important to understand, develop, and manage this complexity (Nightingale & Rhodes, 2004). EA aims to address these problems by trying to give a holistic overview and norms. The discipline of EA has been introduced in the 1980s and has since then evolved into a popular practice with the goal to manage information systems and align them with business needs (Gampfer et al., 2018). EA is said to have many different advantages and benefits. Examples include, among others, a reduction of complexity of business-IT alignment and reduced costs and risks when realizing projects (Foorthuis, van Steenbergen, Brinkkemper, & Bruls, 2016).
In general terms, an EA can be considered as a framework to ensure the consistency of organizational objects, policies, IT objectives, and the decision-making process related to the information technology systems (Alonso, Verdún, & Caro, 2010). Thereby, architectural decision-making is influenced by stakeholders and their communication, organizational culture, and governance mechanism (Roos & Mentz, 2018). In research, however, many different definitions of EA exist (Rahimi, Gøtze, & Møller, 2017). The difference among these definitions is based on their diverse understanding of architecture and their enterprise scope. Based on their understanding of architecture, the definitions can be grouped into different categories: 1) description of an enterprise, 2) inherent structure of an enterprise, 3) inherent structure and management approach, 4) principles for guiding enterprise design, and 5) management approach for guiding enterprise design. Because this paper focuses on the challenges of EAM tasks, we rely on the definition of Lankhorst (2017). He defines EA as "a coherent whole of principles, methods, and models that are used in the design and realization of the enterprise's organizational structure, business processes, information systems, and infrastructure" (Lankhorst, 2017, p. 3).
According to Fischer, Aier, and Winter (2007), there are three main goals of EA: First, the current as-is architecture needs to be documented and communicated. Second, EA has to support the to-be architecture's design. And third, projects that aim to transform the as-is into the to-be architecture have to be implemented and realized. Recent research mentioned at least part of these goals (Rehring et al., 2019b; Zhang, Chen, & Luo, 2018). These goals are supported by EA models that create transparency, measurability, and consistency. In addition, a critical success factor for EA is the effective collaboration and engagement between different EA stakeholders (Kurnia, Kotusev, & Dilnutt, 2020). EA stakeholders are employees who have an origin in a business or IT area, who are affected by EA. Examples of EA stakeholders are the top management level, project teams, and architects (Kurnia, Kotusev, Shanks, Dilnutt, & Milton, 2021).
In the first years after the introduction of EA, research focused on understanding how to define different building blocks of EA and their dependencies. However, in recent years, an increasing amount of research appeared that focused on the management of EA, referred to as EAM (Gampfer et al., 2018). EAM has become an appropriate approach to manage and understand the complexity and change of an EA (Lapalme et al., 2016). An EAM supports an organization to improve its business performance by creating architectural transparency, a documented architectural vision, and the definition of clear architectural principles and guidelines (Ahlemann et al., 2012). EAM provides different tools and methods for the establishment, maintenance, and development of an EA from an integrated and holistic view (Aier, Gleichauf, & Winter, 2011;Simon, Fischbach, & Schoder, 2014). It aims to help stakeholders in cases of analyzing the as-is architecture and planning and defining requirements for to-be architectures independently of their background (Farwick et al., 2010). In order to fulfill the goals of EA, this paper is based on the definition of Ahlemann et al. (2012) that includes four different characteristics of an EAM. These characteristics describe EAM as a management philosophy, organizational function, methodology, and culture. Ahlemann et al. (2012) define EAM as "a management practice that establishes, maintains and use a coherent set of guidelines, architecture principles and governance regimes that provide direction for and practical help with the design and the development of an enterprise's architecture in order to achieve its vision and strategy" (p. 20). To fulfill the goals of an EA, the tasks of EAM can be grouped into the six dimensions: EA documentation, EA planning, EA programming, EA implementation, EA communication and support, and EA governance (Schmidt & Buxmann, 2011). These dimensions and their corresponding challenges that were identified in the literature are described in the findings section.

RESEARCH DESIGN
This paper is based on a SLR to identify 1) the different tasks of EAM and 2) the challenges of these tasks. A SLR may be used to evaluate and interpret "all available research to a particular research question, topic area or phenomenon of interest" (Kitchenham & Charters, 2007, p. 7). To answer the previously defined research questions, a SLR, therefore, was found to be a meaningful research method. We followed the guidelines of Kitchenham and Charters (2007) for conducting this research. For the complete research, we conducted a two-folded iterative SLR that is visualized in Figure 1. In the first iteration, we aimed to get an overview of the different tasks of EAM. In a second iteration, we aimed to search for challenges of the different identified EAM tasks. We conducted separate literature searches for each EAM task to not miss any important publication. This second iteration is based on the results, namely the identified dimensions of EAM tasks, of the first iteration. Therefore, seven SLRs were conducted that are independent of each other. We used Scopus as our primary database. Scopus is a meta-database that covers searches in several scientific journals and references (Paul & Criado, 2020). To make sure to not miss any relevant publication, we additionally searched in the Association for Information Systems (AIS) Electronic Library and IEEE Xplore since these are databases covering relevant research within computer science and information systems. Finally, we did a search in the Senior Scholars' Basket of Journals (https://aisnet.org/page/SeniorScholarBasket).
After we defined the research questions, we derived the first SLR's search string of it. Besides the term "enterprise architecture management" and "EAM", it contains the terms "task", "goal", and "objective" to identify publications that describe tasks directly, and tasks that are needed to fulfill the goals of EAM. In the second iteration, we built each search string based on a combination of the identified task dimensions and the term "challenge". The following Table 1 gives an overview of the search strings of each literature review. ("EA documentation" OR "enterprise architecture documentation") AND challenge EA planning ("EA planning" OR "enterprise architecture planning") AND challenge EA programming ("EA programming" OR "enterprise architecture programming") AND challenge EA implementation ("EA implementation" OR "enterprise architecture implementation") AND challenge EA communication and support (("EA support" OR "enterprise architecture support") AND challenge) OR ("EA communication" OR "enterprise architecture communication")) AND challenge EA governance ("EA governance" OR "enterprise architecture governance") AND challenge To get a final sample in each of the literature reviews, we conducted three steps of eliminations to select the identified studies. First, we removed publications based on their title, abstract, and keywords. If the publication does not fit the topic of research by reading the title, abstract, and keywords, it was eliminated from the sample. Secondly, we read the full text of each remaining publication. If a publication was relevant for the results of the respective iteration, it was added to the final sample. For being relevant, the publication should describe at least one EAM task (first iteration) or at least one challenge of an EAM task dimension (second iteration). After this selection process, we conducted an additional backward search in each literature review. Table 2 gives an overview of the elimination criteria, the process of it, and the number of the final sample of each literature review. For summarizing and getting an overview of the EAM tasks and their challenges, we used the approach of the concept matrix based on Webster and Watson (2002). It helps by structuring the results and enabling a discussion for each matrix. This research provides a basis to identify and structure the EAM tasks and their challenges.

FINDINGS
This section is divided into two parts. In the first subsection, the identified dimensions of EAM tasks are described.
In the second subsection, the challenges of each of the identified dimensions are explained.

EAM tasks
According to the framework of Schmidt and Buxmann (2011), the tasks of EAM can be categorized into six general dimensions. Schmidt and Buxmann (2011) group the six dimensions into strategic and operational tasks. The strategic tasks consist of the dimensions: EA documentation, EA planning, and EA programming. The operational tasks include EA implementation, EA communication and support, and EA governance. These six dimensions are the basis of the categorization of the tasks, and they are extended by tasks that are described by other authors. As our literature analysis shows, the categorization is complete, which is why it is used below to describe our results. Thereby, the literature review shows that tasks regarding the EA documentation dimensions are most mentioned in the literature (n = 12). In contrast, EA governance tasks are hardly ever considered in the literature (n = 4). EA documentation is needed to manage the complexity of the current EA. It provides a basis for the management of an EA and supports decision-making ( EA implementation is the first operational task dimension. Schmidt and Buxmann (2011) describe EA implementation as "the initiation and/or execution of system changes through the EAM function itself" (p. 174). It can include general system consolidation and the development of shared infrastructure components or reusable business services. Active EA implementation may result in improved IT flexibility and IT efficiency (Schmidt & Buxmann, 2011). After the implementation of a change, it is important to determine who will update the as-is architecture on time and keep the EA documentation consistent (Ahlemann et al., 2012). Therefore, the task of EA implementation is closely entangled with EA documentation.
EA communication and support is needed for an effective EA implementation as well as acceptance of EAM by the stakeholders. The last operational task dimension is EA governance. It describes the degree to which organizational EA-related guidelines and decisions are binding (Schmidt & Buxmann, 2011). This can be prescribed due to formal processes. It is important to check and approve projects by ensuring conformity due to to-be architectures (EA planning) and guidelines (EA programming) (Drews et al., 2017). A sufficient degree of EA governance, which may be achieved by formal reviews and approval processes in practice, is needed for an effective EA implementation (Schmidt & Buxmann, 2011). Further governance aspects are the formal mandate of EAM, and the centralization of and the governance mechanism for EAM-related decision-making (Lange et al., 2016).

Challenges of EAM tasks
All the previously described task dimensions struggle with different challenges to be carried out successfully. In this section, we describe the results of the SLR about the challenges of each EAM task. The challenges have a diverse origin. While challenges could be identified for the task dimensions of EA documentation, EA planning, EA implementation, EA communication and support, and EA governance, challenges regarding EA programming were not identified in the literature.
The challenges of EA documentation are quite variable. Until today, the documentation of EA models is often a task done manually ( Table 4 shows an overview of identified challenges within the field of EA documentation.  Although EA communication and support are important EAM tasks, they are faced with different challenges. According to Uludag et al. (2019), due to the limited capacity of enterprise architects, stakeholders may perceive a lack of support in their work. Additionally, they identified an overload of work and missing technical know-how. Moreover, during the architecting process challenges emerge in the internal communication. It is difficult to distribute messages across all relevant stakeholders of the organization and to mediate the corresponding purpose and value of EA (Chuang & van Loggerenberg, 2010). A lack of communication in processes has an impact on the EA implementation ability and the governance of an organization (Banaeianjahromi & Hekkala, 2019). Additionally, this lack is supported by a communication gap between stakeholders in the same organization based on the usage of the same terminologies in widely different ways . Also, the support of EA projects by management can be limited and hinder a successful EAM. As a result, a lack of management support can lead to a restriction on the needed resources (Banaeianjahromi & Smolander, 2019).
EA governance deals with several challenges that are mostly based on the organizational structure. A lack of the organizational structure can have its cause in a missing central responsible EA unit (Banaeianjahromi & Smolander, 2019). Moreover, organizations often struggle with a lack of clarity regarding EA governance that could lead to unclear responsibilities (Cram, Brohman, & Gallupe, 2016;Lam, 2004). Also, these companies do face problems with the enforcement of EA governance rules (Lucke & Lechner, 2011) due to insufficient resources (Seppänen, Heikkila, & Liimatainen, 2009). Compliance with EA governance rules and principles is even more difficult due to external contract work and the usage of legacy systems because of the difficulties and costs to keep them conform with the as-is EA (Boh & Yellin, 2006;Espinosa, Armour, & Boh, 2010). In addition, also an insufficient understanding of EA can obstruct the implementation and work of EA governance (Ajer & Olsen, 2019; Bourmpoulias & Tarabanis, 2020).
Summarized, most challenges of EAM consist in tasks related to EA documentation, EA planning, and EA communication and support. Further, some challenges of EA implementation and EA governance are based on issues in other tasks. Overall, 44 different challenges of the six tasks dimensions were identified. Thereby, the most challenges are mentioned in tasks of EA implementation (n = 12). Whereas no EA-specific challenges of EA programming could be found in the literature. Table 5 gives an overview about the number of publications and number of challenges that were found for each dimension.

DISCUSSION
As shown in the results, we identified different challenges in the dimension of EA documentation, EA planning, EA communication and support, EA implementation, and EA governance. However, challenges for EA programming could not be found and need to be addressed in future research.
Additionally, based on the identified challenges, it should be investigated in the future how these challenges can be quantified and solved in the best way. As our analysis of the literature shows, there are several opportunities for future research. Table 6 presents an overview and categorization of open research in the field of EAM and EAM challenges. We used the framework of Schmidt and Buxmann (2011) to categorize the different EA tasks into six dimensions. This framework does not make a direct statement about the interdependencies between the dimensions. Based on the identified tasks and challenges, several interdependencies could be assumed. Figure 2 visualizes these interdependencies on the level of the six dimensions. It can be recognized that the dimensions of EA documentation, EA planning, EA implementation, and EA governance are faced with challenges like a lack of knowledge (Nowakowski et al., 2018), wrong or outdated information (Kleehaus & Matthes, 2021), or a lack of communication and support (Ajer & Olsen, 2018) that have their origin in the dimension of EA communication and support. Furthermore, the challenges of EA implementation are influenced by tasks of EA documentation, EA planning, and EA governance. EA implementation struggles with the difficulties of the link between EA and norms, values, and culture, and a lack of governance (EA governance) (Alwadain, 2020;Bui & Levy, 2017), and with a lack of documentation (EA documentation and EA planning) (Löhe & Legner, 2014). Because no EA-specific challenges of EA programming could be identified, it has no interdependencies with the other dimensions. Figure 2. Overview of the interdependencies between the EAM task dimensions However, the interdependencies between different EAM tasks and challenges can be complex. Therefore, it is necessary to deepen the research of potential interdependencies. This can be done based on different foci. First, the interdependencies could be analyzed in more detail due to their emergence in theory and practice. Future research can investigate how EAM-related challenges can be solved, whether solving a challenge has an impact on other tasks. For instance, better documentation could improve the implementation tasks of EA. This also involves the question of how influential the interdependencies between EA documentation and EA implementation are.
Besides future investigations about the challenges and the interdependencies between these challenges, there is an opportunity to conduct future research from an organizational, project, and technical perspective. From an organizational perspective, this research leaves open questions about the interdependencies between the identified challenges and the type, size, and operating industry of an organization. Furthermore, the responsibilities for solving and avoiding EAM-related challenges and in particular, the role of an organization's management is an interesting further research approach because a missing central EA unit (Banaeianjahromi & Smolander, 2019) and therefore, a lack of clarity about responsibilities is a major challenge in the dimension of EA governance (Cram et al., 2016;Lam, 2004).
Another important perspective that needs to be considered is the corporate governance perspective, which especially deals with the role of top management and the board of directors. On the one hand, the board of directors can benefit from valuable information which can serve as the foundation for decisions and strategic management.
Here, EAM can serve as a foundation that enables the generation of data and its provision as usable information to the board of directors. On the other hand, IT needs top-management support to successfully transform or improve the infrastructure and processes. For management to adequately support IT, some understanding of its importance, relevance, and function is required. Therefore, it is also important that the board of directors knows the tasks of EAM to a certain extent and is also involved in this.
To solve the identified challenges, organizations would probably carry out different projects. From this project perspective, in the future, the specific characteristics of projects that aim to solve EAMrelated challenges should investigate what guidelines for organizations could look like. For instance, this includes the composition of the project team and the workflows. Moreover, the competencies and responsibilities that were mentioned as a challenge before, if they are not defined, should be considered.
As the last perspective, future research can be considered from a technical perspective. Nowadays, disruptive technologies, such as artificial intelligence and blockchain, become one focus of research (Salah, Rehman, Nizamuddin, & Al-Fuqaha, 2019). This leads to the open question of how these and other disruptive technologies could be used in EAM, especially for the solving of EAM challenges. Furthermore, for technologies like artificial intelligence and big data, the availability of high-quality data is an important precondition. How this data can be generated and afterwards used are therefore two important areas to be investigated. Thereby, it should be taken into account that AI technologies can be used for supporting employees' work and replacing their work by automatizing systems and processes (Rao & Verweij, 2017). Possible EAM tasks dimensions that consider human work are EA documentation, EA planning, and EA communication and support. Furthermore, for using AI technologies, the availability of high-quality data is necessary to conduct analyses (Sturm & Peters, 2020). Hence, it is necessary to investigate how the analysis of data can help to solve the challenges and which data quality is needed for AI. Moreover, it is open if the current data quality of an EA is sufficient to use this technology, and how the needed data quality can be warranted.

CONCLUSION
Enterprises are complex systems with several social, technical, and infrastructural elements and components. Organizations need to be able to adapt to changes quickly to stay competitive in the digital age that is characterized by a fast-pacing environment. The management of the internal infrastructure is, therefore, an important factor for today's companies. Despite its relevance, however, EAM is still an area many companies are struggling with. Surprisingly, research lacks comprehensive overviews and reviews about the challenges of different EAM tasks. This article aimed to close this research gap by conducting a systematic literature review about EAM, its tasks, and the respective challenges. We wanted to answer the following two research questions: What are the tasks of enterprise architecture and which challenges do exist? What are future research directions and opportunities in the field of EAM challenges?
The field of EAM can be divided into six groups of tasks, namely EA documentation, EA planning, EA communication and support, EA programming, EA implementation, and EA governance. Our review of the literature identified several challenges for each of these groups except EA programming. In total, 44 challenges were identified, with the most challenges related to tasks within EA documentation, EA planning, and EA communication and support. For the field of EA programming, however, no challenges are described in the literature.
Our research is not without some limitations. First, the iterative structured literature review was conducted on a specific database. This may lead to the fact that publications were overlooked and not considered in this article. Furthermore, our research is based on six tasks dimensions. Therefore, EAM tasks that are not fit in one of the dimensions may not be included in this project. However, to the best of our knowledge, the six dimensions of the framework of Schmidt and Buxmann (2011) are still comprehensive and cover most of the EAM tasks. As a result, however, some specific EAM challenges that are not covered by the six dimensions of Schmidt and Buxmann (2011) might not be included in this article. Furthermore, we found that several interdependencies between the different EAM dimensions exist. It is likely that a closer investigation of these interdependencies might lead to insights into new challenges that have not been covered by literature, so far. This results in several open research fields within the field of EAM and EAM challenges. First, the investigation of interdependencies between different EAM tasks might lead to interesting insights. Also, EAM tasks and challenges can be further researched from an organizational, technical, and project perspective. Finally, also the role of the board of directors and how they can benefit, and benefit from, EAM should be addressed by future research.