A Bibliometric Review of the Business Platforming Literature: Theoretical Cornerstones and Research Trends

Scholarly interest in business platforms has risen consistently in the past 10 years. As a pervasive topic in the business literature, business platforms are addressed not only in the management field but also in domains, such as economics, strategy, marketing, engineering, and information technology. We provide a review of the literature on business platforms using a bibliometric methodology. In doing so, we extend the scope of research by including the closely related topic of business ecosystems. Bibliometric techniques allow consideration of a broad selection of papers with a consequent reduction in bias. We start with a selection of 1914 publications from 2009 to 2022. First, we identify the theoretical cornerstones employing cocitation analysis. This technique allows us to determine the fundamental literature, including documents that date back to previous decades. Then, we employ bibliographic coupling to determine the most recent research trends, limiting the span to the 2018–2022 period. Finally, we propose a research agenda based on the identified gaps in current research trends.


I. INTRODUCTION
I N RECENT years, scholarly interest in platform business has risen considerably.The investigation of platforms and related subjects began in the 1990s [1].Several key underlying concepts of business platforms have been introduced since then-for example, network externalities [2], business ecosystems [3], and, more recently, multisided markets [4].
The concept of platform lends itself to different interpretations.Parker et al. [5] defined a platform as a business based on enabling value-creating interactions between external producers and consumers.Recognizing the term's ambiguity, Cusumano et al. [6] identified platform businesses as entities that allow individuals and organizations to interact with the potential to generate increasing, nonlinear value.In short, we can define a platform business as a venture that creates value primarily by allowing different sets of actors to interact, profiting from the The authors are with the Entrepreneurship and Technology Management Section, University of Twente, 7522NH Enschede, The Netherlands (e-mail: m.didomenico@utwente.nl; e.hofman@utwente.nl;h.schiele@utwente.nl).
Color versions of one or more figures in this article are available at https://doi.org/10.1109/TEM.2023.3240300.
Digital Object Identifier 10.1109/TEM.2023.3240300presence of network externalities, and employing some form of modular architecture.In determining the scope of this article, we adopt this definition as a guideline for including a publication in this article, or for excluding a particular research area.In itself, that is not sufficient to determine the boundaries.Therefore, additional criteria are applied, drawn from those criteria adopted by other reviews in the field.In particular, we focus on the literature on platforms as business models.Consequently, the scope of this article fits, for the most part, into the digital setting because most firms with a platform business model operate online.Therefore, we exclude the line of research on platforms as products.Despite the affinities with platform business models and some degree of overlap in the respective fields, we decided to omit this research area.Moreover, from preliminary research, we found that the concepts of business and digital ecosystems are closely related to those of platform business models because the coordination of multiple firms is central to the functioning of platforms (a process referred to, in recent research, as "orchestration").For this reason, we include that stream of literature in the review.As mentioned earlier, contributions in the field have increased, and several scholars have responded by reorganizing the available knowledge.The work of Thomas et al. [7] and McIntyre and Srinivasan [8] are among the most recent efforts.Of late, a systematic review of platform competition has been undertaken by Rietveld and Schilling [9].In addition, several reviews have been performed on more specific topics exploring certain aspects related to platform business.We refer to these in the relevant sections of the article.
All these reviews have been undertaken using traditional methods.Unquestionably, these contributions have been crucial in organizing the research.At the same time, due to objective constraints, traditional reviews have restricted the scope of the analysis and the sources considered so that an appropriate number of publications can be properly evaluated.We consider it beneficial to adopt a more comprehensive methodology to provide a clear mapping of the research.This is particularly true for the latest lines of research, which would be at risk of being omitted by traditional reviewing approaches.For this reason, we apply bibliometric methods to provide a review of platform business literature at large, including the related topic of business ecosystems.The latest review of this type-by Facin et al. [10]-was conducted more than 6 years prior to our work in this field.Since then, hundreds of articles have been published.
Bibliometric methods help to increase objectivity by employing quantitative techniques [11].In this way, it is possible to reduce the bias from interpretation by researchers, which is often considered the major risk in traditional reviews [12].This methodology allows us to organize a vast number of pieces This work is licensed under a Creative Commons Attribution 4.0 License.For more information, see https://creativecommons.org/licenses/by/4.0/This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.
coherently and to help researchers to catalog them.Using conventional review techniques, it would be virtually impossible to include all relevant publications.In any case, we employ bibliographic methods as a first step in giving sense to the vast body of literature because, on their own, they are largely insufficient.We remedy this shortcoming in two ways.First, we perform text analysis on the entire sample to better identify the themes of the clusters.Second, for each cluster, we methodically select a subset of articles to study in depth, performing a traditional review of every cluster identified.Using this approach, which is consistent with traditional literature reviews, we analyzed over 300 publications.Our methodology is described in detail in the following section.
In this article, we pursue three objectives.First, we determine the theoretical foundations of the literature available on business platforms and ecosystems-that is, the "cornerstones" on which the field is based [13].We do this by applying cocitation analysis to identify both the seminal pieces and more recently established studies derived from them.Second, we identify the research trends on platform business, employing bibliographic coupling.This technique takes into consideration the similarity between bibliographies of recently published pieces and classifies them accordingly.In this way, we can identify the latest lines of research that are currently shaping the field.We connect these with the cornerstones from which they originate, employing cross-reference analysis of the bibliographies.Third, we propose opportunities for future article.These are derived from the gaps revealed by the current research trends that were identified in the previous section.
This article is structured as follows.In the next section, we present the methodology and the sampling procedure.Section III identifies the cornerstones from the literature.Section IV contains the research trends.In Section V, we deal with future research streams in the field.Finally, Section VI discusses the key findings and the limitations of this article.

II. METHODS AND DATA
Our research procedure, as outlined in Fig. 1, was adapted from Zupic and Čater [11] and covers five steps: 1) definition of the research design; 2) data collection; 3) classification and visualization; 4) reading of the articles; and 5) interpretation of results.We decided to use bibliometric methods, which help to classify the available knowledge using objective, quantitative techniques [14].They were introduced in the 1950s [14], but it has only been possible recently to use their full potential, thanks to advances in software [11].To classify and visualize our data, we relied on two complementary bibliometric techniques: cocitation analysis and bibliographic coupling.To perform such analysis, we decide to rely on VOSviewer, mainly for two reasons: the possibility of performing both techniques, and its widespread acceptance as provided by its usage in other published bibliometric reviews.
These two techniques help us to identify the main areas of established research in a field (cornerstones) and its latest developments (research trends), respectively.The first method used, cocitation analysis, was introduced by Small [16] and has become the most used bibliometric technique over the years [17], [18].As visualized in Fig. 2, it measures the similarity between documents by defining the frequency by which they are cited together [11].In other words, for two publications to be cocited, there must be a third publication that cites both.The fundamental assumption of cocitation analysis is that the more Fig. 1.Research procedure, adapted from Zupic and Čater [11] and Obradović et al. [15].Fig. 2. Representation of cocitation analysis, adapted from van Oorschot et al. [13], to obtain the cornerstones.Colored circles represent the knowledge base-namely, articles included in the sample-and arrows indicate citations.Therefore, cocitation analysis identifies sources not included in the sample, either from the studied timespan (2009-2022) or from previous years.two articles are cited together, the higher the probability that their content is related [11].Thus, the strength of the cocitation relation is defined by the number of publications in which publications are cocited [18].Following previously adopted terminology [13], we label these clusters as cornerstones.These are the publications that are at the core of the research in the field.
Our second method is bibliographic coupling, which was introduced by Kessler [19].This technique can be considered the opposite of cocitation analysis [18].It measures the number of references shared by two publications: the larger it is, the This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.stronger the bibliographic coupling relation between two publications [18].In other words, it estimates the degree to which two bibliographies overlap [18].Therefore, bibliographic coupling is a static measure that does not require citations to accumulate.For this reason, it is particularly suitable for identifying emerging fields, even those of relatively small size [11], [17].Fig. 3 depicts the working bibliographic coupling.

A. Data Collection
To collect input data for the bibliometric analysis, we followed the procedure outlined by Zupic and Čater [11].First, we performed a query on the Web of Science database using the following keywords: platform, e-commerce, two-sided market, multisided market, module, modularity, complementor, network effect, network externality, winner take all, system market, and installed base.For this selection, we considered, among others, keywords from previous literature reviews on the subject and insights from domain experts.We ensured that all possible variations of these terms were included.A keyword had to appear either in the source's title, the keywords provided by the publisher, or the abstract.We confined the query to the macrocategories of economics, management, business, and operations research management.Furthermore, we included only peer-reviewed papers (including review articles), and we set the time span from 2009 to the present.More detail on the performed query, and a comparison with other reviews, is available in Appendix I.This query produced 8456 articles from 816 sources.
Second, we excluded sources with an impact factor lower than 1.5 (as provided by Web of Science and Scopus) and sources with a score lower than 2 in the Academic Journal Guide 2021 published by the Chartered Association of Business Schools.We then manually evaluated each abstract to remove false positives.In doing so, we applied the following criteria, guided by the scope of our research described above.In the "modularity" area, we included only articles on modularity in the digital context and, therefore, excluded product design modularity.In addition to articles related to innovation, technological, and ICT ecosystems, we included articles on the entrepreneurial ecosystem.
Finally, regarding the broad "platform" concept, we included all articles related to digital platforms, platform business, and platforms adopting a business-economical perspective.Platforms intended as organizational software or procedures used as an internal tool for the firm were excluded.We finished with 1914 articles from 222 sources, of which 302 from 30

B. Classification and Visualization
The two basic components of science mapping are classification and visualization [14].There are three main approaches to visualizing bibliometric networks: distance-, graph-, and timeline-based [18].We selected the distance-based approach, in which the nodes (in our case, the publications) are positioned in such a way that the distance between the two of them indicates the strength of their relationship [18].We applied the visualization of similarities (VOS) approach introduced by Van Eck and Waltman [20].This combines classification and visualization [20] and provides satisfactory results when applied to medium-sized databases [21].
The software constructs a map following three steps [22].First, a similarity matrix is created, which is a normalized co-occurrence matrix.It does so by measuring the similarity s ij between two items i and j, which is the association strength by two articles, computed as c ij w i w j where c ij is the number of co-occurrences of items i and j; and w i and w j are either the total number of occurrences of items i and j (in cocitation analysis) or the total number of co-occurrences of these items (in bibliographic coupling).Second, the VOS mapping technique is applied, based on the similarity matrix, so that each pair of items, i and j, reflect their similarity s ij as accurately as possible.Finally, the graphic representation of the map is optimized to make the outline easier to interpret.For a detailed discussion of the mapping and clustering approach of VOSviewer, we refer to Van Eck and Waltman [20] and [22], and Waltman et al. [23].
Regarding cocitation analysis, the 1914 articles in the sample provided 79 827 (nonunique) references that were either articles or other publications, such as books or book chapters.For inclusion in the analysis, we set a threshold of at least 5 citations resulting in 3248 references.An increase in the sampled number is expected because, by employing cocitation analysis, This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.
we can enlarge the base of our review.That is because most of the seminal publications on which research is based originate before the timespan of the collected sources.In the analysis, we included the 1000 publications sharing the strongest cocitation links, expanding our selection, and ended up including 2444 publications ranging from 1969 to 2022.The analysis resulted in six clusters.While interpreting the result, we merged two clusters, so to obtain five cornerstones.
For bibliographic coupling, we considered only the articles published in the 2018-2022 period, since the method requires a shorter timespan [11].Similarly with cocitation analysis, we considered the 1000 publications with the strongest coupling links out of 1376 preselected.Given the goal to detect research trends, we did not use a minimum threshold of the number of article citations.The analysis resulted in seven clusters.To facilitate the interpretation, after reading the articles, we decided to merge several clusters resulting in four main research trends.

C. Reading and Interpretation
The last step is reading and interpreting the findings [11].To enhance the interpretation, for cocitation analysis, we studied the 15 most densely connected articles per cluster-that is to say, those with the most links to each other.In addition, we included the 10 most cited articles (that were not already among the most densely connected), provided that the strength of the links was above the first quartile.For bibliographic coupling, we selected the 25 most connected with each other.In the finish, the selection produced more than 300 articles.After reading these and manually scanning the abstracts of the others, the three authors independently labeled each cluster.After an in-depth discussion, they agreed on the final labeling of each cluster.This process was supported with text analysis to identify the most common expressions in each cluster.The interpretation results are described in Sections III and IV, regarding the cornerstones and the research trends, respectively.

III. THEORETICAL CORNERSTONES OF BUSINESS PLATFORMING RESEARCH BASED ON COCITATION ANALYSIS
In the cocitation analysis, we obtained five clusters, each representing a cornerstone of the platform business literature.We labeled them as follows: 1) the business ecosystem; 2) network effects and competition dynamics: modeling the multisided markets; 3) ecosystem governance, value creation, and value capture; 4) the service-dominant logic and the rise of the sharing economy; and 5) platform architecture: open innovation in a modular context.Table II provides a summary of the main themes discussed in each cornerstone.

A. Cornerstone 1: The Business Ecosystem
The concept of business ecosystem allows us to put aside the traditional view of firms competing in isolation and instead recognize the interdependencies among them.Such interdependence is often managed through platforms.The concept of business ecosystem was introduced by Moore [3].It deals with the fact that firms can no longer be considered members of a single industry but part of a broader multi-industry system.In that respect, they must both cooperate and compete to support new products, satisfy customer needs, and innovate [3].The term co-opetition-a concept that was introduced by Brandenburger and Nalebuff [24]-refers to the behavior that combines competition and cooperation.To prosper, firms need to pursue not only their particular interests but must take into consideration those of the entire ecosystem [25].To do so, firms need to develop platforms (i.e., services, tools, and technologies) that other firms can use to enhance their performance [25].Adner [26] defines the ecosystem construct as "the alignment structure of the multilateral set of partners that need to interact in order for a focal value proposition to materialize." This increased interrelation between firms creates opportunities for entrepreneurship, with new firms being developed as platforms, often by offering specialized services to users [27].Indeed, an ecosystem typically houses both well-established companies and new ventures, which may be sponsored by corporates or launched independently [28].
While this cornerstone is mainly devoted to describing the main characteristics of the business ecosystem paradigm, the following cornerstones will deal with governance, innovation dynamics, and organizational dynamics, respectively.

B. Cornerstone 2 Network Effects and Competition Dynamics: Modeling the Multisided Markets
The second cornerstone deals with network effects and competition dynamics in platforms, with a focus on pricing issues.The concept of network externalities (or network effects) was introduced by Katz and Shapiro [2] and [29].In the context of a given network, the utility that a user derives from a product or service depends on the number of other users.Typical examples include the telephone, operating systems, and credit cards.Here, competition dynamics shift from the product to the network level.Expectations, coordination, and compatibility are the key issues to be addressed [2], [29].Although usually positive, network effects can also be negative.This happens when an excessive number of users on one side has a negative impact on users on the same or the other side.Typical examples are TV advertisements, negatively perceived by viewers [30], and too many app producers that might confuse smartphone users [31].
Network externalities are at the core of the concept of twosided markets, which was introduced by Rochet and Tirole [4], [32].These are defined as markets with network externalities in which two (or more) distinct sides interact through the presence of a common platform [4].The principal peculiarity of such markets has to do with facilitating interactions by pricing: platforms typically charge one side and subsidize the other [4].Examples include video games and shopping malls, in which consumers are considered on the subsidized side while the profit-making segment is located on the side occupied by software developers and shops, respectively.Moreover, technology advancement allows platforms to dynamically adjust pricing [33], [34], [35].
Multisided markets require specific strategies.Eisenmann et al. [36] have identified the three main challenges that platform providers face: pricing, winner-takes-all dynamics, and the threat of envelopment.In addition to the seminal work by Rochet and Tirole [4] and [32], various authors have contributed to pricing, including Caillaud and Jullien [37], Parker and Van Alstyne [38], Hagiu [39], Armstrong [40], Weyl [41], and Rysman [42].The winner-takes-all paradigm consists of the tendency of a single technology architecture to emerge as the dominant design [43].Indeed, that is often the case as shown by the multitude of so-called standard wars in which a dominant platform design emerges as competitors are driven out of the market [44], [45], [46].However, the winner-takes-all paradigm about platforms has come under scrutiny lately [47], [48].Finally, platform envelopment refers to the entry strategy consisting of a platform bundling its functionalities with those of a rival platform to leverage shared user relationships [49].

C. Cornerstone 3: Ecosystem Governance, Value Creation, and Value Capture
Publications on the third cornerstone describe the strategic dynamics of firms in a business ecosystem setting and, in particular, those aiming to obtain and maintain the dominant position of platform leaders.
A key premise is the issue of the appropriability of innovation.Indeed, Teece [50] underlines the risk of failure of the innovator to profit from its innovations, while other firms in the network might take advantage of those innovations.That leads to the dichotomy between value creation and value capture [51].
In fact, Iansiti and Levien [25] recognized that strategy can no longer focus on the internal operations of firms but has to adapt to the ecosystem dynamics in which the firm is embedded.Furthermore, the value-creation process can no longer be considered a single-firm achievement.Instead, successful platforms seek to increase the value of the entire system rather than suboptimizing the value of its parts by careful orchestration of their network of producers and consumers.Adner [52] underlined this point by introducing the concept of innovation ecosystem, arguing that, in an interdependent setting, companies can create value that is not possible to achieve individually.Moreover, due to the complexity of such innovation ecosystems, it is difficult to predict the outcome of such value propositions [53], [54].
In addition to new possibilities, this interdependence leads to a set of risks for the firm.Risks are related to initiative, interdependence, and integration [52].The concept of dynamic capabilities deals with them, as well as with value creation and capture dynamics [55].Dynamic capabilities refer to the ability of the firm to appropriately respond to change to foster its competitiveness [56], [57].Concerning strategy, the identified levers of platform leadership include scope (that is, the degree of ownership the leader is willing to maintain), product technology, relationships with external complementors, and internal organization [58], [59].In any event, commitment from the platform leader is pivotal to engaging complementors [60], [61].In summary, platform leaders should be aware of the dynamic interdependence of actors in a business ecosystem, and of the opportunities and threats.Therefore, they must adapt their strategy accordingly.

D. Cornerstone 4: The Service-Dominant Logic and the Rise of the Sharing Economy
The fourth cornerstone deals with the value cocreation process between platforms and users and the subsequent development of the sharing economy.Sharing-economy platforms base their business models on the exchange of services and the creation of content by users.
The service-dominant logic was introduced and developed by Vargo and Lusch [62] and [63].It deals with the transition from a marketing view focused on products to one addressing services.In fact, the value added from services to world GDP has risen from 54% in 1995 to 65% in 2017, with the incidence in OECD countries around 70% (Source: data.worldbank.org).This shift in the economy of developed countries-that is, from products to services-led to an acceleration in the innovation of services, notably with the development of service ecosystems empowered by modular IT technologies [64].Such service ecosystems are characterized by actor-to-actor (users and/or firms) networks, in which value is cocreated by participants in the network who share mutual values and logics [64].Service platforms, consisting of tangible and intangible modular components, allow participants to interact and facilitate the value cocreation process [64], [65].
New services and a network organization in which hierarchies are replaced by a crowd of participants are at the core of the concept of the sharing economy [66].This includes the development of peer-to-peer (P2P) activities thanks to intermediary platforms that enable users to collaborate, often without clear distinctions between full and casual employment and related issues [66].One of the most studied applications is related to e-commerce and, as its subset, social commerce [67], [68], [69], [70], [71], [72].Undoubtedly, the sharing economy paradigm is ubiquitous and involves most industries, especially if associated with This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.so-called "collaborative consumption" [73], [74], [75].Indeed, several authors find the naming of the sharing economy at times misleading, since it is often used to refer to the consumption of goods and the provision of services through the use of platforms [76].A comprehensive review of the platform in the sharing economy is provided by Sutherland and Jarrahi [77].Examples of platforms relying on the sharing economy, including social media, are countless-with Airbnb, Uber, and Facebook being among the most widely studied examples.They constitute a prominent research trend, as discussed in Section IV.

E. Cornerstone 5: Platform Architecture: Open Innovation in a Modular Context
"Open innovation" is a term that was coined less than two decades ago by Chesbrough [78].It denotes the shift from the classical view of innovation fostered by R&D activities to a more comprehensive outlook, considering innovation to, from, or with actors outside the firm.
The focus is primarily on other firms encompassed in an ecosystem with the leader firm in the central position.This requires the development of new business models and a common architecture to make collaboration possible and profitable [78], [79].The democratization of innovation is a related concept, introduced by Von Hippel [80].It recognizes the importance of the user base of a particular product or service in shaping its evolution, and the relevance of apparently counterintuitive strategies, such as free revealing.Free revealing is closely linked to the concept of open source, as opposed to the proprietary strategy [81].This is especially relevant in the IT context and consists of opening a proprietary platform to external contributors while maintaining a certain degree of control over it to make adequate returns [81].
Architecture is a fundamental concept in the business ecosystem environment.Indeed, modular design rules allow actors other than the platform owner to participate in the value-creation process [81], [82], [83].Modular design rules include a modular architecture, clear interface specifications, and testing standards that allow hierarchically independent ecosystem members to develop and supply complementary services, physical subsystems, or digital content that together deliver an overarching value proposition [84].
However, this engineering design perspective-which focuses on architectural organization and modularity-has traditionally been separated from the economic perspective, mainly centered on the concept of multisided markets [85].On the other hand, the two perspectives are undoubtedly linked and interdependent, and both are essential to fully understanding the phenomenon [85].The technological evolution of the platform requires carefully calibrated business decisions, and the two aspects should not be considered separately [1].
The concepts of architecture and modularity initially referred to physical products, apply to the software sphere, with the development of the internet-and mobile-based platforms [86].The influence of the ecosystem environment-that is to say, other firms collaborating (or competing) with the platform leader-is true of pivotal importance generally and, specifically, for the evolutionary process of the platform itself [86].Such connection between collaborating firms in the software realm is made possible by the application of specific application programming interfaces (APIs) [87].The firms that make the emergence and dominance of a platform possible-namely, the myriad of complementors, who are often small independent software or hardware companies-play a role that is of particular relevance [88], [89], [90].Indeed, as we shall discuss in the research trend section, a prominent part of the work of scholars in recent years refers to internet-based platforms, analyzing the relationships between the platform leader and the complementors interacting in the ecosystem.Inevitably, the degree of openness established is a key strategic decision for platform leaders [91].

IV. RESEARCH TRENDS ON BUSINESS PLATFORMING BASED
ON BIBLIOGRAPHIC COUPLING Concerning bibliographic coupling, we identified four research trends: 1) platforms in the sharing economy; 2) new roads to value in the evolving ecosystem; 3) entrepreneurial opportunities in the ecosystem context; and 4) platform strategies and ecosystem orchestration.Table III presents schematically the main themes of each research trend.

A. Research Trend 1: Platforms in the Sharing Economy
Publications in the first research trend deal with the role of platform businesses in the realm of the sharing economy.In their conceptualization, Eckhardt et al. [92] identified the key role of platforms (especially internet-based ones) as enablers of the efficient matching between providers and users.Therefore, they define the sharing economy as "a scalable socioeconomic system that employs technology-enabled platforms to provide users with temporary access to tangible and intangible resources that may be crowdsourced" [92].Three categories of platform are attracting specific attention from researchers: social media, P2P marketplaces, and crowdfunding services.
Through social media (or social networks), users can share information and media.They also constitute an important marketing channel for brands and firms to engage with their customers [93], [94], [95], [96], [97].Twitter [93], [98] and Facebook [99] are among the most studied.However, a long list of studies is available on less well-known platforms.A specific review of social media research has been recently provided by Kapoor et al. [100].
In the financial realm, crowdfunding is a form of financing that involves three main actors: the investors, the entrepreneurs seeking capital, and the platform that facilitates matchmaking between the two [107].P2P lending is a particular form of crowdfunding in which individual users aggregate their funds to finance loans requested by other users as well as businesses [108].Both crowdfunding and P2P lending have emerged in recent years as a topic of research interest.Being activities based on platforms, researchers investigate "typical" issues, such as platform openness [109], [110], [111] and user innovation [112].However, the peculiarity of the setting allows us to deepen other aspects, in particular, the odds of obtaining financing.Examples include the role of gender [113], [114], [115], [116], geographical factors [117], [118], sustainability issues [119], and the application of machine learning techniques [120], [121].
This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.

TABLE III SUMMARY OF THE RESEARCH TRENDS
Finally, some authors examine criticisms of the sharing economy, investigating critical aspects of the setting, particularly about conditions for workers, and the potential upsurge in inequality [122], [123], [124].

B. Research Trend 2: New Roads to Value in the Evolving Ecosystem
The second research trend integrates the traditional concept of the business ecosystem with the impact of new digital technologies driving change in platform ecosystems.In this context, platform businesses emerge and develop new ways to create and capture value, powered by the possibilities offered by such technical innovation.
The concept of ecosystem has grown notably in the past decade, allowing interdependent organizations to collaborate without explicit hierarchies [125].This is made possible by modular architecture [125], [126].In this framework, new digital technologies, such as APIs and developer toolkits, play an increasingly important role in the coordination of value creation by ecosystem members as they ensure seamless integration.A new trend in platform literature is research on understanding how platform architectures-ranging from perfectly modular to integral-facilitate or inhibit value creation by members of the platform ecosystem.In modular architectures, digital technologies including clearly specified APIs and developer toolkits act as boundary mechanisms facilitating independent members' participation in the value-creation process.As long as they adhere to the rules set by the platform leader, complimentary services and technological subsystems or software will seamlessly integrate into a well-functioning whole fulfilling their ecosystem's value proposition.
Indeed, technological progress and other types of innovation require firms to act dynamically in conditions of uncertainty to create new value propositions [53] in their complex "digital platform ecosystems" [127].Ambiguity is indeed a constant given that settings are increasingly complex and firms need to act rapidly to sustain change [127].In particular, the rise of digital technologies and platforms is important in shaping new ways to create and capture value [51].In shaping the ecosystem strategy, the right balance between cooperation and competition must be considered [128].The need to adapt over time to contingencies is increasingly apparent [128].
Moreover, interdependencies caused by digital technologies require firms to develop new business models and platform strategies to take advantage of the new possibilities of value creation and capture [129], [130], [131], [132].In their review, Gomes et al. [133] identified the transition from value capture in traditional business settings to value creation in innovation ecosystems.Indeed, the dichotomy between the two aspects of value creation and capture is at the core of this research trend.At the same time, traditional firms are repositioning themselves as platform players, seeking to gain a prominent role in the ecosystem [134].

C. Research Trend 3: Entrepreneurial Opportunities in the Ecosystem Context
The third research trend focuses on the entrepreneurial dynamics that take place within the ecosystem.Indeed, the "platformization" of the ecosystem not only affects already established firms-that perhaps attempt to become platform leadersbut also creates opportunities for new firms to emerge.
Indeed, the development of digital technologies in previous years has had a strong influence on the entrepreneurial process, and, here, further investigation is required to understand the dynamics of this "digital entrepreneurial ecosystem" [135], [136].This is particularly true of the risks faced by entrepreneurs in entering a platform-dominated ecosystem characterized by significant power asymmetries [137], [138].In fostering entrepreneurship, new actors gain importance and attract attention from scholars.In particular, research focuses on players, such as incubators and venture capitalists [139], [140], [141].A central role in many investigations is reserved for the growing importance of universities and research centers [136], [142], [143], [144].
Therefore, new perspectives are needed to understand the actions in this modified environment and their governance [141], [142], [145], [146].A comprehensive review specific to entrepreneurship in the platform context has been recently provided by Fernandes et al. [147].

D. Research Trend 4: Platform Strategies and Ecosystem Orchestration
The fourth research trend deals with the dynamics of platforms as ecosystem orchestrators, including strategic choices related to openness and the handling of network externalities.In particular, this line of research recognizes the necessity of platforms to deal with the increased complexity of their ecosystems, the reasons for their success, and the strategic role of complementors as innovators and sources of value.
One of the main topics in this research trend concerns the coordination efforts of complementors.This is particularly important given their increasing interdependence and the complexity of This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.platform ecosystems, enabled by more sophisticated technological solutions.Therefore, the platform owner must implement particular strategies taking into account the complementors and their specific served users, since the process of value capture and creation is dependent on them [55], [101], [148], [149], [150], [151], [152], [153], [154], [155], [156].
The engagement of complementors is closely related to another theme of research-that of the openness of the platform to complementors.This concerns the degree of freedom they can exercise, such as the opportunity to access the system and the level of autonomy they can enjoy once they have gained access.It consists therefore of a gradient of possibilities, with the tradeoff between closed platforms where the owner maintains substantial control and open platforms that benefit from innovative inputs from external parties [91], [157], [158].Moreover, the process of value capture and creation with network effects is part of this research trend, with the aim of better understanding their role and magnitude [159], [160], [161], [162], [163], and balancing the power dynamics in a platform ecosystem context [164].Multihoming refers both to the presence of the same complementor in different platform ecosystems and to the adoption by the same user of different platforms.Multihoming contemplates the presence of multiple platforms, and it is in contrast to the winner-takes-all paradigm-a phenomenon linked to the increased complexity of the environment.Indeed, it is a theme in this research trend that concerns both complementors and end users [163], [165], [166], [167], [168], [169].

E. Considerations on Trends in Platform Business Research
Naturally, research trends build on theoretical cornerstones.Fig. 4 below illustrates the connection between cornerstones and research trends through cross-citation analysis.The area of the circles is proportional to the share of citations from the corresponding cornerstones.Table IV also provides the most cited references in the research trends from the corresponding cornerstones.

TABLE IV MOST CITED CORNERSTONES' REFERENCES IN THE RESEARCH TRENDS
From the identified cornerstones and research trends, it follows that the platform research realm is far from being a static one and has evolved over the past three decades.
To gain additional insights, trending topics in platform research were also analyzed using keyword burst analysis with CiteSpace.A keyword burst signals the first appearance of a topic and its fast-increasing prominence in the platform literature.In effect, it signals new research hotspots and trends.Table V presents the identified keyword bursts, their starting and ending years, and their "strength" reflecting burst intensity.Additional information on the algorithm used to identify such surges is provided by Kleinberg [182] and Chen [183].In this analysis, besides the sampled articles, we include those resulting from cocitation analysis.We limit this analysis to articles published between 2005 and 2022.
The outcome of this analysis is consistent with the cornerstones and trends discussed.In particular, we assist in categorizing the concept of ecosystem into different subtopics, such as service ecosystem, innovation ecosystem, entrepreneurial ecosystem, and platform ecosystem.Moreover, we note the rise of terms, such as machine learning and big data, with around 40 papers on these topics published on platform domains since 2020.This is in line with the analysis of the bursts of selected publications between 2005 and 2022, as shown in Table VI below.
Regarding research methods, we identify three main approaches.Studies are either based on observation and case studies, data collected through the platforms, and-in recent years-experimentation.In the earlier phase of platform research, scholars have derived novel insights from studying the history and moves of (very) successful players, formulating fundamental concepts that are still widely employed, such as "winner takes all" and "critical mass."Certainly, this approach comes with several shortcomings.First of all, there is the risk of selection bias because studies often rely on early players, such as Apple and Microsoft (and their ecosystems).However, they have been crucial in setting the vocabulary and the early direction This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.

TABLE V SELECTED KEYWORDS WITH THEIR CITATION BURSTS BETWEEN 2005 AND 2022
in the field.Another type of study, beginning around the early 2010s, employs systematic data collection from the field-for example, by observing prices in marketplaces or comparing the characteristics of different players.Moreover, in these studies, scholars focus on more diversified platforms-including new giants, such as Amazon and Uber-with more than 30 and 40 studies published since 2018 based on data collected from each.Finally, in the late 2010s, platform scholars started to use experimental methods (roughly 80 publications since 2020), in a few cases with the cooperation of the platform owners (not lab experiments).Moreover, scholars have investigated the long tail of less well-known platforms, including those based in Asia.Finally, another interesting aspect concerns institutions-and, therefore, authors' networks-in which research into platform business is performed.Table VII provides an overview of the total number of citations per institution, divided by the time span of published articles.It is interesting to observe a shift from the U.S. university hegemony to a more diversified situation in which European universities play an important role.
Such a tendency is better framed by considering the country of origin of the publications, again presented in two different time spans.While more than one-third of citations of publications dated from 2009 to 2017 originated from work coming from U.S. universities, that figure declined to roughly one-fourth considering the time span of 2018 to 2022, with the increasing importance of work performed by universities in countries, such as the United Kingdom, China, and Italy, as shown in Table VIII.Nevertheless, the platform domain remains a field primarily concentrated in the USA.
Despite such (declining) polarization, the platform domain is one of relatively high international collaboration-that is, coauthorship between scholars belonging to institutions located in different countries.This is confirmed by the analysis of the coauthorship network, graphically represented in Fig. 5 below covering the 2018-2022 time span.

V. AVENUES FOR FUTURE RESEARCH
The study of the fundamental literature (cornerstones) and of the latest research trends allows us to develop a research agenda for future studies on platforms and ecosystem domains.Reflecting on cornerstones, research trends, and the literature studied, we have identified three main macroareas for future article: 1) enlarge the scope of analysis; 2) explore the consequences of platformization; and 3) diversify the research methodologies.Table IX summarizes the main themes in each area.

A. Enlarge the Scope of the Analysis
More types, industries, and geographies: Estimating the number of operating platforms is not an easy task.Market research estimates the number in the order of thousands worldwide for e-commerce marketplaces only.Yet, research to date has focused on a very limited sample of successful B2C or C2C platforms, omitting the long tail of other players and largely neglecting unsuccessful platforms.To develop normative theories on platform ecosystems, we propose to enlarge the sample of platforms in studies, including successful and failed cases.Another feature to consider is the geographical aspect.Indeed, the data show that more than 60% of revenues from e-commerce are generated in Asia alone [187].However, most publications and case studies deal with American-based marketplaces.That applies not only to marketplaces but also to other kinds of platforms, such as social media and intermediaries.Moreover, researchers focus on some of the most prominent platforms-for example, Twitter concerning social networks-and omit others.It is important to shift attention to certain new platform categories that have been developing over recent years and those that are currently emergent.Indeed, it would be beneficial to further explore a wider range of industries and fields.Examples include intermediaries for working and providing services, e-learning platforms, B2B niche marketplaces, aggregators, newly developed social networks, platform conglomerates, and freelancer services platforms.Moreover, scholars can further investigate the effect of platforms to foster economic growth in underdeveloped countries [188].
Furthermore, we observe that large incumbent firms have been adopting platform business models.For example, Volkswagen has already embraced the concept of modular product platforms,  and it has now taken the next step by placing its new modular car platform for its family of electric vehicles at the heart of its future business platform.The combination of a modular physical and digital car architecture allows the company to create and orchestrate a platform ecosystem with complementors offering complementing services to be used with its vehicles.For example, the technological option for bidirectional charging and temporary storage of energy is likely to create a wide variety of new services that will connect energy service providers to households in ways that were unthinkable in the past.The main issue concerning the study of such settings is the lack of publicly available data.Here, cooperation with such less prominent but equally interesting platforms might allow scholars to obtain access to fascinating data that will uncover mechanisms behind the functioning of platforms.This will facilitate a more complete understanding of the realm of the platform, reducing the risk of bias from too narrow a selection of cases.
Workforce conditions and welfare issues: A consequence of the surge in sharing economy platforms is the working opportunities generated by them.Examples include ride-sharing, food delivery, lodging, and professional services portals for freelancers.Often beginning as a means to integrate earnings for their users, they have evolved to become permanent or semi-permanent workforce conditions.Despite the rising interest of scholars in the field [124], [189], [190], [191], [192], many aspects still need to be clarified, including the impact of platformization on the job market.Linked to this research stream are the regulation of platforms and the welfare implications for society.While discussions on topics, such as taxation, welfare, privacy, and competition policy are intensively debated in the public arena, research on such matters-although present [193], [194], [195], [196], [197], [198], [199], [200]-still needs to be fully developed.
Impact of latest technologies: Certainly, a fundamentally important factor behind the evolution of the ecosystem is the This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.development of new technologies.These reduce barriers to entry and generate new opportunities for entrepreneurship because advanced technical capabilities are no longer strictly required.On the other hand, the evolution of technology offers new opportunities to well-established platform firms (Google and Meta perhaps being the most representative examples) to expand their operations, for example through artificial intelligence [201], [202].For both new entrants and established giants, blockchain solutions and the Internet of things are the two technologies where scholars call for greater in-depth research as they are likely to affect the evolution of existing platforms and the emergence of alternative platform architectures.For example, blockchain technology might lead to the emergence of "blockchain platforms" [203] with alternative structures for platform governance that shift from a traditional centralized governance mode to a decentralized one.Future articles could also address the contingency factors that determine the performance of different platform designs, and the impact of big data.

B. Explore the Consequences of Platformization
Platformization and traditionally offline players: The consequences of the "platformization" of the economy are still largely unclear, and several avenues for future research need to be explored.The winner-takes-all paradigm seems to be outdated [48], leading to the presence of multiple platforms in the same environment.These are competing for market share, and the mechanisms behind such competition between platforms still need to be fully explored.
Another element concerning the competitive landscape that requires exploration by researchers is the response of traditional offline players to the presence of online platforms, particularly concerning e-commerce.These firms can either integrate their services by adding some online elements to the value journey (for example, with "click-and-collect" options) or try to enter the platform business by positioning themselves as platform leaders [134], [204], [205].Furthermore, we suggest that traditional firms opening their closed architecture to independent complementors (Volkswagen being a clear example) would offer an interesting opportunity for future article.Finally, they can resist the introduction of platforms in their sector (such as Uber in multiple countries).
Development path of nascent platforms: As mentioned in the third research trend, the evolution of the ecosystem offers opportunities for entrepreneurship.Yet, many questions remain.One of the main interesting research opportunities lies in the development of nascent platforms [206].While researchers have analyzed the road to success of first-generation platforms that were often first entrants at the end of the 1990s and beginning of the 2000s, such strategies would need to be revised to apply to current nascent platforms.Here, experimental methods might prove useful in answering questions related to the drivers of platform takeoff [207].Incidentally, the increasing number of operating platforms facilitates the development of new typologies of business to allow users to navigate complex ecosystems.For example, in the e-commerce domain, platforms that facilitate sellers being simultaneously present in multiple marketplaces are receiving substantial finance from venture capitalists.
Coexistence of platforms and multihoming: Ecosystems are becoming increasingly complex and recent research has shown that multiple platforms can coexist in the same environment.This has consequences for the decision-making of users and firms other than the platform leader operating in the ecosystem.Studies on multihoming-both on the users' side and the complementors' side-despite increasing in past years [166], [167], [168], [208], [209], are still needed to uncover the full picture.The COVID-19 pandemic seems to have accelerated this process of platformization since it is argued that people have turned to online platforms for working and entertainment, but the consequences, in the long run, are still unclear.
Channel conflicts: Closely related to the previous topic is the subject of channel conflicts, that is, issues with business partners in the value chain.Channel conflicts have been studied by scholars for decades [210].Yet, their influence on established players converting to platform business models is largely unexplored because most studies have focused on the risk of cannibalization between offline and online sales channels.For example, incumbent B2B operating firms willing to open their ecosystems and become platform leaders must deal with challenges such as overcoming the commitment of businesses to their traditional suppliers who are part of the supply chain in which the new platform will operate and that are largely unexplored by researchers.

C. Diversify the Research Methodologies
Opportunities from mixed methods for theory development: In the literature on platform ecosystems, a great variety of strategic choices are presented but, to date, a holistic view of how these choices combine into recipes leading to platform success is missing.An opportunity for further article can be found in employing a holistic, configurational perspective on the theory development of digital ecosystems.For multifaceted digital phenomena, such as digital platform ecosystems, Park et al. [211] propose relying on qualitative comparative analysis (QCA).QCA can be used to study how causally relevant elements, including distinct attributes related to structural design and governance choices, combine into configurations that consistently outperform other configurations.In contrast to regression-based methods, this approach allows the researcher to study causal complexity in terms of conjunction and equifinality, for example [212].
This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.

TABLE X KEYWORDS AND COMPARISON WITH OTHER LITERATURE REVIEWS
Conjunctural complexity means that performance can only be explained by the joint presence (or absence) of certain design choices.Equifinality implies that distinct combinations of conditions might exist that are equally effective [213], [214].This approach can enrich current theories and provide an important step in creating a typology of ecosystem design and governance choices.
More empirical evidence: Regarding methodologies for platform strategy research, we call for more empirical studies in this area.Certainly, we can point to an increase in experiments on platforms over recent years [207], [215], [216], [217], [218], [219].However, these are limited in quantity and scope, given the possibilities to explore the mechanisms of platform adoption from the perspective of not only end users but also complementor firms.Here, the cooperation of scholars with platform firms would be beneficial in obtaining more diverse data and more widely generalizable results.

VI. CONCLUSION
We are confident that we have provided a clear and synthetic presentation of the literature on platform business, identifying both the cornerstones and the most recent trends based on the deployment of bibliometric methods.Of course, the scope of such a venture is broad, and we are aware of the presence of limitations.In particular, the relevant number of considered sources might have directed the focus to certain topics at the expense of others.Moreover, despite our efforts to be objective, the risk of bias due to the personal training of the authors is impossible to eliminate entirely.
We have identified the following five cornerstones: 1) the business ecosystem; 2) network effects and competition dynamics: modeling the multisided markets; 3) ecosystem governance, value creation, and value capture; 4) the service dominant logic and the rise of the sharing economy; and 5) platform architecture: open innovation in a modular context.All of these are well-established research areas in the domain of platform business.We expect to have provided a short yet solid summary of those areas, in particular by underlying their fundamental pieces.
The research trends have required a more challenging descriptive effort.The four identified trends are 1) platforms in the sharing economy; 2) new roads to value in the evolving ecosystem; 3) entrepreneurial opportunities in the ecosystem context; and 4) platform strategies and ecosystem orchestration.Unlike cornerstones, the content of research trends is much more variegated and, therefore, it is more difficult to label and summarize those trends.Undoubtedly, in future years, new research trends will continue to appear.Nevertheless, we believe that the summary we have provided is exhaustive as far as the main and most robust trends that have appeared in recent years are concerned.
Finally, we have suggested avenues for future platform research.Those are based on the following macroareas: 1) enlarge the scope of the analysis; 2) explore the consequences of platformization; and 3) diversify the research methodologies.
We believe there are three main outcomes from this literature review.First, platforms are here to stay.This is demonstrated by the dramatic increase of interest of scholars in the past decade as well as the presence of publications addressed to a broader public of managers and entrepreneurs.Second, the research streams are coherent.This is possible to determine thanks to the continuity between the identified cornerstones and research trends.Third, research opportunities are many.These can be derived from the analysis of the research trends and are highly diversified.Certainly, there is no lack of areas of inquiry.
We are convinced that this overview would prove useful for both established scholars as well as for other readers interested in the topic.

APPENDIX I KEYWORDS AND COMPARISON WITH OTHER LITERATURE REVIEWS
In our review, we combine all the keywords and terms used in previous literature reviews (Table X).About the review of Facin et al. [10], this is the case because we include the term "platform" in our query without other constraints.
This article has been accepted for inclusion in a future issue of this journal.Content is final as presented, with the exception of pagination.

Manuscript received 25
October 2021; revised 28 May 2022 and 31 October 2022; accepted 11 January 2023.This work was supported by Maykers.Review of this manuscript was arranged by Department Editor M. Dabic.(Corresponding author: Matteo Di Domenico.)

Fig. 3 .
Fig. 3. Representation of bibliographic coupling, adapted from van Oorschot et al. [13].Colored circles represent the knowledge base-namely, articles included in the sample-while arrows indicate citations.Bibliographic coupling only considers the articles in the sample, clustering them according to the shared bibliography.

TABLE VI SELECTED
PUBLICATIONS WITH THEIR CITATION BURSTS BETWEEN 2005 AND 2022

TABLE VIII TOP
10 COUNTRIES FOR THE NUMBER OF CITATIONS OF ARTICLES PUBLISHED IN THE 2009-2017 AND 2018-2022 TIMEFRAMES

TABLE IX SUMMARY
OF THE FUTURE RESEARCH AREAS