BEST PRACTICES IN IMPLEMENTING A PROJECT MANAGEMENT OFFICE: A SYSTEMATIC REVIEW OF THE LITERATURE

Highlights: Project management office (PMO) is structurally configured in a particular way to adapt to the peculiarities of each organization and its strategic objectives, in order to promote project management practices. The purpose of this article is to explore the best practices for project management office implementation. A systematic literature review was conducted using 104 documents published between 2000 and 2018. The research allowed the identification of PMO’s data, such as functions, models, best practices in implementation, challenges to implementation, and success factors. Goal: The purpose of this article is to explore the best practices for project management office implementation. Design/Methodology/Approach: A systematic literature review was conducted using 104 documents published between 2000 and 2018. Results: The research allowed the identification of PMO’s data, such as functions, models, best practices in implementation, challenges to implementation, and success factors. Limitations of the investigation: Other factors related to PMO, such as the implementation phases, maturity models, process groups, and organizational variables that affect PMO. Practical implications: It is observed that there are relevant issues in PMO implementation structuring that are not consolidated, making it difficult for organizations to base their implementation on the available theoretical frameworks. Originality/value: As a result, it became evident that there is a lack of standardization of those characteristics related to the PMO; and that the so-called “best practices” require more academic studies before they can be established.


INTRODUCTION
Starting in the mid-1990s, companies began to face the challenges of managing several projects simultaneously. It is precisely in that scenario that the Project Management Office (PMO) emerges, aiming to provide methods, tools and techniques for the proper management of projects (Czekay, 2012). Studies by Dai and Wells (2004) show that, since 1994, PMOs have started to become popular and their implementation in organizations has increased significantly since then. The PMO is defined by the PMBOK Guide (PMI, 2017, p. 48) as "an organizational structure that standardizes project-related governance processes, and facilitates the sharing of resources, methodologies, tools, and techniques." PMO can be implemented to make more efficient use of resources, reduce the risk of project failures and increase project success rate (Kutsch et al., 2015), and to restructure processes, departments and projects (Correia et al., 2018). The PMO implementation also enables process integration to companies, as well as informing top management of their project portfolio status, besides seeking to compete in the market through successful projects (Czekay, 2012). In the implementation of the PMO, one of the main objectives is the implementation of the best practices, and the repetition of projects allows the learning of new lessons and the establishment of the best practices, leading to a more efficient delivery of projects (Philbin, 2016).
Notwithstanding the benefits of PMO implementation, there is still no common understanding on what drives the success of such organizational structure (Spalek, 2013), which still faces a lack of acknowledgment of its contribution, repeatedly forcing PMOs to justify their existence (Kutsch et al., 2015). There is no empirical evidence that the PMO is associated with an organizational competency of project management and the PMO is questioned about its value within an organization (Khalema et al., 2015). In addition, there is no consensus method to define the importance of a PMO (van der Linde and Steyn, 2016). Moreover, due to their unstable nature, PMOs do not deliver the expected benefits in the long run (Bredillet et al., 2018).
In the face of the gap between benefits resulting from the implementation of a PMO and the perceived value by the organization to which it belongs, the work here presented carries out a systematic review of the literature on best practices in PMO implementation by striving to answer three research questions: 1) What practices for implementing a PMO are addressed in the literature?
2) According to the literature, what are the best practices for implementing a PMO?
3) What analysis does literature carry out regarding the implementation of a PMO?
Thus, after a systematic review, it was possible to identify the functions, models, best practices in implementation, implementation challenges, and PMO success factors. Thereby, in view of the singularities of the organizations that implement the PMO, it is verified that the organizational structure presents considerable variability in its characteristics. And because of those particularities, it is unreasonable to establish "best practices" for such different organizations.

Literature review
According to PMI (2017, p. 4) PMI (2017, p. 4), "project is a temporary endeavor undertaken to create a unique product, service, or result". Due to the complexity, the indispensability of managing, in an integrated way, the different disciplines present in the projects, such as scope, cost, time, and risk, for example, and constant cultural, technological, political, economic, and social changes, it is increasingly necessary to carry out effective project management. To Kerzner (2011, p. 3), project management "is the planning, organization, direction, and control of company resources for a relatively short-term objective that has been established to complete specific goals and objectives".
With increasing market competitiveness, characterized by increased competition and innovation rates of products and services, organizations have been dealing with increasingly numerous and strategically important projects. Therefore, in order to increase both the number and the strategic importance of the projects, many organizations have implemented the PMO (Hobbs et al., 2008), which has the role of helping organizations to plan, implement and monitor projects so their goals can be achieved (Ferreira et al., 2017). In addition, according to Andersen et al. (2007), PMOs have been established by organizations to take on responsibilities and coordinate functions and activities related to the project.
In their study based on the literary review, Spelta and Albertin (2012) present a summary of reasons for either or not creating PMOs and affirm that there are studies in the literature that show improvements in success rates in project management through the PMO. However, other studies indicate the opposite, even not recommending the creation of PMO in certain cases (Spelta and Albertin, 2012). Table 1 presents the reasons for and against PMO implementation.

Method
Literature review is usually an initial step in a research. It allows the researcher to map previously developed and existing expertise in an area . In this sense, the Systematic Bibliographic Review (RBS) is presented as the main technical procedure for carrying out the research (Costa and Toledo, 2016). RBS is a specific research methodology, formally developed to survey and evaluate available evidence related to a particular research topic, which is a specific problem, topic, area or phenomenon of interest (Biolchini et al., 2005;Brereton et al., 2007). Systematic review occurs by formulating a question clearly, using systematic and explicit methods to identify, select, and critically evaluate relevant research. In addition, the data from these studies are collected, analyzed and included in the review (Moher et al., 2015). In this sense, RBS is composed of three main phases: planning, execution and analysis of results . In the planning phase, the objectives of the review and its protocol are defined (Costa and Toledo, 2016). The stage of execution involves the initial identification, selection and evaluation of the studies according to the criteria established in the previous phase . In the third and last phase, the data of the selected studies are analyzed and synthesized (Biolchini et al., 2005).
In order to be more specific and operational, this research was based on the approach that subdivides RBS into five phases: problem formulation, data collection, data evaluation, data analysis and interpretation, and conclusion and presentation (Biolchini et al., 2005).

Problem Formulation
This stage refers to what type of evidence should be included in the review; then it is necessary to create definitions that can determine studies that are relevant and irrelevant to the specific subject under investigation (Biolchini et al., 2005). Specifying the research questions is the most critical element in the systematic review, because it is at this stage that the data to be extracted in the primary study is determined (Brereton et al., 2007). Therefore, the main objective of the review was to identify works published in different languages that address the best practices in the implementation of a PMO, based on the PMBOK guidelines, in public and/or private organizations, without limiting the sectors in which they are inserted and regardless of size. The research questions were presented in the Introduction section.

Data Collection
In this step, one determines which procedures will be established to find relevant evidence defined in the previous step, including the determination of the sources that can provide potentially important studies to include in the research (Biolchini et al., 2005). Therefore, the first step in bibliographic searching is the selection of the database according to the research protocol (Thomé et al., 2016).
The most widespread databases that are frequently used for searching the literature are Web of Science (WOS) and Scopus. However, Scopus covers a superior number of journals of recent articles (Aghaei Chadegani et al., 2013). Scielo and Scopus databases contain more works related to the subjects studied, including those related to production engineering, management, and administration (Costa and Toledo, 2016); therefore, these two bases were used in this research.
In addition, the keywords and their synonyms have been defined: PMO, project management office, implantation, implementation, performance, output, impact, best practice, good practice, management practice. The searches were conducted in May 2018.
At this stage, 252 papers, written in different languages, with the terms or part of the terms searched in the titles, abstracts or keywords were found. Of this total, 228 publications were found after initial research in the Scopus database, identifying 111 duplicate publications, leaving only 117 papers. In the Scielo database, the surveys generated 24 publications as results, 11 of which were duplicates, leaving only 13 publications. At a different moment, the searches of both databases were related. And of a total of 130 publications, five were identified in both databases, remaining 125 works. Table 2 summarizes the data collection from the systematic review of the literature.

Data evaluation
In the third stage, qualitative criteria are applied to delimit studies that can be considered valid from those that should be considered invalid. At this stage, the guidelines for the extraction of information from the primary research reports are determined (Biolchini et al., 2005).
In order to extract relevant data for systematic review purposes, a spreadsheet was created with the publications selected in the previous stage, containing: title, name of the author (s), year of publication, abstract, keywords, publication source, type of document, original language, number of times the publication was cited, affiliation of the author (s), country of publication, type of publication, publication approach, type of organization addressed at work, and industry sector covered in the publication. Darling and Whitty (2016) carried out an extensive bibliographical review of the academic and non-academic literature in English. As a result, they claim that definitions for describing PMO have evolved over time. Although the earliest reference to a project office refers to the improvement of agriculture in the UK in the early nineteenth century, the first edition of the PMBOK does not mention the PMO and, only in the 2 nd edition of the PMBOK, published in 2000, the theme was addressed in two phrases (Darling and Whitty, 2016), stating that PMO exists in a variety of forms and has a variety of functions (PMI, 2000). Facing this fact, the year 2000 was defined as a milestone for this research. In this second phase, of the 125-remaining works, publications prior to the year 2000 were excluded, 9 in total, leaving 116 works.
After the inspectional reading on these 116 publications (Adler and Van Doren, 1972), 12 were excluded. Of these, three are publications by the Federal Register, an official US government newspaper that contains routine publications and public notifications of government agencies. Two are honorable mentions published in the same edition of a magazine, and do not meet the scientific criteria stricto sensu, as stated by the editors of the magazine. The remainders do not correspond to the universe under study. Thus, 104 works remained.

Data analysis, interpretation, conclusion and presentation
As mentioned in this section, the adopted approach divides RBS into five phases (Biolchini et al., 2005), and the last two steps of the method are Data Analysis and Interpretation, and Conclusion and Presentation. The fourth phase will be presented in the Discussion section, subdivided into two subsections, with the following titles: Quantitative Synthesis and Best Practices Synthesis, respectively. The fifth and final step will be presented in the Conclusion section.

Discussion
In this secti on, the fourth step of the systemati c review addressed, and the analysis and interpretati on of data are presented. In this phase, the procedures to be applied for the collected data are defi ned, so that the synthesis of valid studies becomes a relevant point, allowing generalizati ons to be made about the subject addressed (Biolchini et al., 2005). Although there is no universal recipe for the analysis phase, data such as years of publicati on, periodicals, authors, and study characteristi cs, relevant to the synthesis, are common elements in the systemati c review of the qualitati ve and quanti tati ve literature (Thomé et al., 2016).

Quantitative synthesis
In this subsecti on, the quanti tati ve summaries of the 104 documents selected are presented. Of this sample, 55 are arti cles, 34 are conference papers, six are chapters of books, four are reviews, two are notes, one is a lett er, one is a short survey, and one is an arti cle that was accepted by a journal and is available as an online version and has not yet been made available in print. Regarding t he languages of the documents, 90, 86.54%, are in English, 10 publicati ons were writt en in Portuguese, representi ng 9.61% of the sample, two are in Spanish, one document was originally writt en in Bosnian and one in German.
Regarding the genres of publicati ons, 78 are empirical and 26 are theoreti cal. According to their approaches, 56 publicati ons are qualitati ve and 48 qualitati ve-quanti tati ve. As to the classifi cati on of publicati ons according to their respecti ve country, it was adopted as a criterion that the country of publicati on is the country of the insti tuti on to which the authors of the works are associated. This way, a publicati on with more than one author associated to insti tuti ons from diff erent countries will be classifi ed as related to more than one country. Thus, despite the sample of 104 papers, 121 countries are related, since three publicati ons are allocated to three diff erent countries, and 11 publicati ons are att ributed to two diff erent countries. The United States, Canada, Brazil, Australia, and the United Kingdom stand out for having 24, 19, 18, eight and six papers, respecti vely, corresponding to 61.98% of the sample. Germany, China, Iran, and Sweden have four published works each.  Number 4, 2019, pp. 448-463 DOI: 10.20985/1980-5160.2019.v14n4.1580 maining 20 are theoretical publications in which this type of categorization of organizations is not applicable.
The latest quantitative analysis portrays the industry sectors covered in the publications. Of the total of 104 works, the most relevant sectors are information technology (IT), technology, health and construction, which have, respectively, 12, eight, seven and five works based on these sectors, corresponding to 30.77% of the sample. In addition, 16 papers were based on more than one sector of the industry and are therefore classified as diverse. Another 20 publications have a strictly theoretical approach, in which this classification by sectors is not applicable. In 12 papers, the authors do not specify in which industry sector they base their studies.

Best Practices Synthesis
As previously described in section Data evaluation, after the inspectional reading (Adler and Van Doren, 1972), 104 papers were selected for quantitative synthesis. At this same reading stage for qualitative synthesis, 19 publications were selected by the authors because their studies stood out in the presentation of PMO best practices, including, but not limited to, the definition and identification of functions, services, typologies, models, and challenges related to the im- plementation of PMOs, in addition to PMO's best practices and success factors. In the next step, an analytical reading (Adler and Van Doren, 1972) was carried out on these 19 documents. From this total, five did not present the qualitative syntheses of the object of this research, leaving 14 publications, presented in Table 3. In these 14 final publications, the authors performed the syntopical reading (Adler and Van Doren, 1972) and, through them, the functions, models, implementation best practices, implementation challenges and PMO success factors were synthesized; they are presented in the following subsections.
Based on the suggestion of Moher et al. (2015), a flowchart adapted from the PRISMA (preferred reporting items for systematic reviews and meta-analyses) protocol, which exposes the document selection used in this research, is presented in Figure 2. This recommendation aims to help the authors to perform better reporting of systematic reviews and meta-analyzes (Moher et al., 2015).

PMO Functions or Services
Functional characteristics are PMO's functions and duties that are expected to be performed in an organization (Jalal and Koosha, 2015). There is a wide variety of options, both in the form and in the functions performed by the PMO (Andersen et al., 2007;Aubry et al., 2010;Darling and Whitty, 2016;Desouza and Evaristo, 2006;Desta et al., 2006;Fernandes et al., 2018;Hobbs and Aubry, 2007;Jalal and Koosha, 2015;Kutsch et al., 2015;Monteiro et al., 2016;Singh et al., 2009;Spalek, 2013). Such expected functions and practices differ as much as the organizations to which the PMO belongs (Hobbs and Aubry, 2007;Darling and Whitty, 2016). However, Andersen et al. (2007) state that even in PMOs with different structures certain characteristics, responsibilities, and main tasks are very similar. Szalay et al. (2017), in their exploratory study, present the typical PMO services. However, the authors emphasize that, in addition to these initial services, there are others that need to be investigated. In an exploratory and descrip-  14, Number 4, 2019, pp. 448-463 DOI: 10.20985/1980-5160.2019.v14n4.1580 tive study of PMOs, Desouza and Evaristo (2006) propose to segment the PMO functions into three levels: operational, tactical, and strategic. Despite this proposed subdivision, knowledge management is one of the main functions at all levels (Desouza and Evaristo, 2006). Andersen et al. (2007) presented the functions of successful PMOs by studying best practices for establishing, developing, and implementing PMOs.
In his studies, Hill (2004) proposes a PMO division in five stages, with the Basic PMO, stage 2, responsible for establishing a viable project management environment, including, but not limited to, the implementation of the 20 proposed PMO functions. Although this list contains 20 functions, it is emphasized that a PMO is unlikely to implement all these functions and that adaptations and adjustments will be necessary (Hill, 2004).
By focusing on variables in construction industry organizations related to PMO characteristics, based on the literature review, Jalal and Koosha (2015) identified and classified PMO functions. Desta et al. (2006) also identified PMO functions. In the empirical study conducted by Dai and Wells (2004), the establishment and use of PMOs were investigated over two years, and after reviewing the literature, the authors synthesized the PMO functions. Martins et al. (2005), in their empirical study, highlight the main functions of the PMO under implementation in a telecommunication company. Wang and Liu (2010) pointed out the main functions of the PMO when analyzing the management of multiple projects in enterprises of real estate companies. It is worth mentioning that, although not in the studied sample, the PMO functions presented by the PMI (2013a cited in Darling and Whitty, 2016;Spalek, 2013) and IPMA (2006cited in Spalek, 2013 stand out for having been referenced in the analyzed studies; hence their inclusion by the authors of this research. The PMO functions cited and their respective authors are presented in Table 4. Although it is the oldest publication, Hill (2004) presents the most complete PMO list of functions, grouped in five blocks: practice management; infrastructure management; resource integration; technical support; and business alignment. As expected, the PMO functions presented by the different authors have great variability among them, mainly in relation to the level of detail. However, some functions are present in most studies: developing project management methodology, managing portfolios, and ensuring project quality.

Models or types of PMO
Model or type of PMO is usually an organizational structure that supports the company's business strategy and development, describing the logic of how PMOs act and deliver value to the organization (Monteiro et al., 2016). As complex as organizations themselves, the attempt to group specific project management structures that cover PMOs is very difficult, if not impossible, due to the significant differences between PMOs (Aubry et al., 2008;Hobbs and Aubry, 2007). Many models have been proposed and constructed around the typology of the PMOs (Monteiro et al., 2016;Szalay et al., 2017). However, these authors argue that because PMOs are structurally configured differently in organizations, it becomes difficult to find a standard way to typify them. Additionally, Monteiro et al. (2016), who have been dedicated to researching PMO typologies, have identified that PMOs are characterized by variation in name, structure, assumed roles, and perceived value, and that the most common typologies have three to five types of PMO models. Although these studies were not part of the sample of this research, it is worth noting the PMO models proposed by the PMI (2013b, cited in Fernandes et al., 2018;Monteiro et al., 2016;and Szalay et al., 2017), Crawford (2001, cited in Andersen et al., 2007and2010, cited in Monteiro et al., 2016), and Rad et al. (2002, cited in Andersen et al., 2007. The authors inserted such models in the present research. The PMO models cited and their respective authors are presented in Table 5. It should be noted that these summarized models corroborate the study of Monteiro et al. (2016), in which, after literature review, 47 PMO models were identified by 12 authors, and due to the similarities identified, they were reduced to 25 different types of PMO models.

Best practices in PMO implementation
The best practices in the implementation of a PMO generally contribute to the optimization of business processes and organizational results (Alves et al., 2013). PMI (2017) presents three PMO definitions; the Support PMO provides project best practices, besides providing templates, training and access to information and lessons learned from previous projects (PMI, 2017). In their studies, Abdi and Kaddoura (2011) point out that the objective of applying best practices is to achieve successful results. The sharing of best practices and the promotion of continuous improvement of processes are obtained from storing and managing the communication of lessons learned from projects (Morris, 2016).
In accordance with best practices, Desouza and Evaristo (2006) note that successful PMOs have very clear documents that prove their credibility and the lack of them can have serious consequences, such as a lack of clarity regarding the roles and responsibilities of the PMO. The description of such documents and their functions are shown in Table 6, which also summarizes the other best practices mentioned above, as well as their respective authors.
Despite the definitions and characteristics of best practices presented in this research, Darling and Whitty (2016) identified that PMO professionals refer to all book authors in the area as an academic community, even though the books read by such professionals are generally at a more basic level of knowledge and present the most focused solutions. The authors mentioned also found that few project managers have read scholarly papers with peer-reviewed research and that apparently there is no understanding regarding the difference between scientific research in the field of management and in a business book. By saying that they are following "best practices", PMO practitioners are actually adopting the practices described in business books and professional association guides (Darling and Whitty, 2016). Essentially, Hobbs and Aubry (2010) argue that best practices significantly require more studies for their establishment. Such conclusion is corroborated by this research, considering the few publications on the subject found in the literature review.

Challenges to the implementation of PMO
The empirical studies dedicated to the implementation of the PMO suggest that the establishment of this structure is a difficult challenge for most organizations and there is a high failure rate (Singh et al., 2009). These challenges have a potential negative impact associated with a higher probability of project failure (Salamah and Alnaji, 2014). Therefore, academic works usually explore PMO roles, functions and services, but tend not to highlight or discuss the tensions and challenges inherent in these roles (McKay et al., 2013). This information is corroborated by Singh et al. (2009), who argue that, although anecdotal evidence suggests that implementation of PMOs can be quite difficult, few studies are dedicated to addressing the challenges involved in the task and to how organizations can overcome them. Spalek (2013) was able to identify the challenges of PMOs that were closed more than a year after their creation and of PMOs that had been in operation for two years or more and were still operating. Desta et al. (2006) found in their research the challenges to establish and maintain PMO capacity. When researching an IT and software development organization, Salamah and Alnaji (2014) identified the main challenges of the PMO. It is important to highlight the study by Singh et al. (2009) in which the 13 main challenges in the implementation of PMO were identified and classified using the Delphi method. The list of these challenges indicated in each author's study is presented in Table 7, which demons-

Electronic Journal of Management & System
Volume 14, Number 4, 2019, pp. 448-463 DOI: 10.20985/1980-5160.2019.v14n4.1580 459 PMO charter: essentially a documented roadmap that defines the key questions or issues to be addressed by the PMO as well as what it will deliver Desouza and Evaristo (2006) PMO policy: necessary to establish sufficient uniformity or management and to enable effective project portfolio management PMO methodology: should adress the business needs of the organization and provide project managers with a framework of tools, processes and metrics trates the lack of theoreti cal consensus on these challenges, thus evidencing the need for more empirical studies on this issue.

PMO's Factors of Success
Given the variety of structures and diff erences in terms of functi ons, size and applicati ons within organizati ons, the only criterion for success, unique to all PMOs, is that their structure is aligned with the organizati on's corporate culture (Desouza and Evaristo, 2006). Andersen et al. (2007) state that the success of the PMO is related to the assurance of PMO authority and also to the support of top management, as well as meeti ng the organizati on's true needs. Alves et al. (2013), when quoti ng Bullen and Rockart (1981), say that criti cal success factors (CSF) are some key acti vity areas, and through the favorable results of these factors, the projects achieve their objecti ves. However, the concept of best practi ce (discussed in secti on Best practi ces in PMO implementati on) is adopted to minimize the determinism of CSF (Alves et al., 2013).
Additi onally, Desta et al. (2006) identi fi ed that the main factors for PMO success were those that, when absent, contributed to failure. Andersen et al. (2007), through a benchmarking study, identi fi ed the most important factors for PMO success that should be emphasized or avoided. Alves et al. (2013) present in their research what factors of success are, and their applicati ons oriented to Project, to the PMO and to the value of the business. All the success factors here menti oned and their respecti ve authors are presented in Table 8. The variety found in the literature review evidences the theoreti cal non-uniformity on the issue.

CONCLUSION
The recent and progressive advance of services and acti viti es related to informati on age and its inherent intricacies, along with the fact that the beginning of the expansion of this sector occurred at the same ti me the PMO began to consolidate in the mid-1990s, have made the informati on technology into a relevant and broad fi eld for PMO deployment. Consequently, many studies have this industry sector as a research universe.
Regarding the fi rst research questi on, the study identi fi ed that, due to organizati ons' parti culariti es, which vary in strategy, processes, and available resources, the lack of standardizati on in defi ning the functi ons adopted by the PMO was evidenced, in the same way that it is diffi cult to establish generic PMO models, due to its structure characteristi c, which is to always seek strategic alignment with the organizati ons it belongs to.
In view of the second questi on, it was observed that, due to the variability in PMO characteristi cs and its host organizati ons, it is hardly plausible to establish "best practi ces" for such diff erent structures, with such peculiar objecti ves. It should be noted that part of the community of project management professionals, by using the term "best practi ces", is oft en adopti ng defi niti ons addressed in business books without academic criteria and rigor. Therefore, for the establishment of these "best practi ces", it is necessary to publish more academic studies on the subject. In relati on to the third and fi nal research questi on, the research showed, in a similar way, that there are also many success factors linked to the PMO, always varying according to the perspecti ve observed and initi al characteristi cs and objecti ves defi ned in PMO implementati on. Although the studies analyzed prove that PMOs sti ll have a high failure rate, it has also been observed that there are few empirical studies dedicated to understanding challenges related to PMO implementati on, as well as the proposal of alternati ves to overcome those challenges.
Despite such inability to standardize, this research was able to build a consolidati on of the main characteristi cs inherent to PMO implementati on menti oned in literature in the last 18 years, since the year 2000. However, it is observed that there are relevant issues in the PMO implementati on structuring that are not consolidated, making it diffi cult for organizati ons to base their implementati on on the available theoreti cal frameworks.
As suggesti ons for future research, it would be interesti ng to review literature on other factors related to PMO, such as the implementati on phases, maturity models, process groups, and organizati onal variables that aff ect PMO, emerging issues observed in conducti ng this research, in additi on to intensifying empirical studies on structuring, such as best practi ces and the challenges of PMO implementati on.