Beyond Diffusion: A Systematic Literature Review of Innovation Scaling

: Innovation is essential for our ability to overcome global issues such as climate change, natural resource depletion, and inequality. A central aspect of innovation is the scaling process. While an abundance of studies on innovation scaling exist in many different disciplines, there is a lack of shared understanding of what scaling means and how it can be successfully achieved. This systematic literature review addresses both these issues by reviewing 147 articles on “innovation scaling” making several contributions to research on innovations and innovation scaling. First, in outlining the ontological differences between “diffusion” and “scaling”, clear conceptual boundaries are established, which provide clarity and support cross-disciplinary consilience. Second, based on the analysis of articles, eleven common modal contextual factors that inﬂuence the outcomes of innovation scaling across contexts and disciplines are presented. Third, an initial theoretical framework of the innovation scaling process is developed, outlining four theoretical propositions. As a fourth contribution, the article establishes a research agenda for the future development of innovation scaling research across many research domains.


Introduction
Innovation constitutes a cross-cutting and multi-disciplinary research field [1][2][3] spanning sustainability (e.g., sustainable agriculture, green growth, the circular economy, or sustainable urban development), smart cities, management, international development, health, and social welfare literatures.Cross-disciplinary academic interest in the topic stems from the fact that innovation occurs across all aspects of society and when done successfully, may deliver new solutions that address pressing issues and/or increase the efficiency and effectiveness of policies, products, and services [4][5][6].
The process of innovation is different from the process of invention [6] in that it describes the (collective) development and implementation of new ideas [4,5,[7][8][9][10][11].Innovation is generally understood as a series of steps: Following the initial inception of novelty, ideas are then iteratively tested, refined, and socio-politically legitimated, before they are ultimately implemented [4,5].Actors' motivations, as well as contextual drivers and hindrances, are central in determining the ultimate outcome of the innovation process [7].Once an innovation has been implemented, the processes of diffusion and scaling-which both refer to the longitudinal spreading of innovations-take central importance in determining the innovations' wider societal impact [4,7,[12][13][14][15].
Notably, in the current academic discourse, innovation scaling has received relatively less attention than the other aspects of the innovation process (e.g., inception of novelty or its diffusion).One field where scaling innovation is particularly popular, however, is in sustainability studies, where scaling and innovation are viewed as key components for sustainable management, development, and growth [16][17][18][19][20].Though popular, a clear and widely agreed upon definition of scaling remains absent, and so too is a comprehensive theoretical framework for understanding the entire scaling process.
For many, the concept of "scaling" revolves around the idea that there are empirical phenomena related to the spreading of innovations that are poorly captured by the population-level perspective of diffusion, which tends to highlight a kind of passive permeation of innovations [15,[21][22][23][24][25].Empirical phenomena considered as instances of "innovation scaling", on the other hand, tend to be characterized by proactive and strategic actors purposefully moving innovations into new contexts and degrees of quality [26][27][28].However, when scholars discuss scaling in this way, they tend to do so using relatively shallow or idiosyncratic conceptualizations [29].Some scholars implicitly or explicitly understand scaling to be a process of vertical movement ("scaling up"), while others understand it as one of horizontal movement ("scaling out/wide") or of qualitative change ("scaling deep") [30].Furthermore, there lacks a common definition of what it means to "scale" innovations.Rather, the term "scaling" is applied and used differently across different contexts (e.g., private sector compared to public sector) and across areas of research (e.g., social welfare compared to development, management, or public administration research).
This lack of conceptual clarity is problematic.It obstructs the development and exchange of scientific knowledge about the process of innovation scaling.To help effectively address pressing global issues through innovation, scholars must develop a systematic understanding of how high impact innovations may be proactively scaled.To do so, one not only has to integrate the research and knowledge that already exists across research disciplines, one also has to develop a conceptual common ground and shared research agenda that facilitates further advancement and knowledge sharing on the topic of innovation scaling.
In pursuit of these goals, this article presents the results of a systematic review of the innovation scaling literature.The review utilized the PRISMA approach (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) [31,32] and covered 147 research articles.The analysis of these articles was guided by four primary research questions: 1.
What does the landscape of research investigating innovation scaling look like? 2.
How can innovation scaling be defined?3.
What are the key drivers and barriers of innovation scaling? 4.
How can extant research be integrated into an initial overarching framework of innovation scaling?
Fitting the multi-disciplinary relevance of the innovation scaling process, this review covers articles from numerous research fields, including sustainability, development, agriculture, management, and health studies.The innovation scaling research that has been conducted within any one these disciplines has tended to remain largely isolated from the others.While there are fundamental contextual differences between these disciplines, integrating their perspectives into a common framework can help to provide a holistic lens upon the cross-cutting innovation scaling phenomenon.Moreover, where cross-disciplinary commonalities do exist, it is possible to identify essential aspects and building blocks that any conceptualization of innovation scaling must pay respect to and integrate them into an initial theoretical framework.
The results of this systematic literature review facilitate several important theoretical and practical contributions.First, this article presents a cross-disciplinary definition for conceptualizing "innovation scaling" that builds upon extant research while also clearly distinguishing "scaling" from the related concept of "diffusion".In doing so, the article offers new clarity that should aid in facilitating further systematic research and cross disciplinary knowledge exchange relating to innovation scaling.Second, this article reports on the eleven "modal contextual factors" that acted as drivers and/or barriers to the innovation scaling process across the 147 covered research articles.Third, based on our proposed definition and modal factors, the article develops an initial theoretical framework for understanding and studying the innovation scaling process.Fourth, this article presents a comprehensive agenda facilitating further systematic research into this important phenomenon.

Innovation Diffusion and Innovation Scaling
"Diffusion" has been identified as one of the core mechanisms in social theories seeking to explain macro-level societal changes [23,24].Diffusion itself is usually defined as the longitudinal spreading of (new) "ideas, structures, and practices" [24] (p.335) in social systems [33,34].Social scientists have researched the diffusion of innovations for more than seven decades (see, e.g., [35,36] for early accounts of diffusion dynamics) using various lenses.Two larger schools of thought may be identified and differentiated: "rational accounts" of diffusion, and "social accounts" of diffusion [25,33].
Rational accounts of diffusion emerged from economic research and make sense of the spreading of innovations by focusing on the benefits that innovations bring vis-àvis existing ideas, structures, and practices [25].Some scholars use an evolutionary lens, according to which individual and organizational actors who fail to adopt efficient practices get weeded out [37,38].Other scholars may adopt a more actor-centric lens, investigating the rational evaluation and decision-making by potential innovation adopters [39][40][41].Rational accounts have repeatedly found that the diffusion of innovations can be connected to their (perceived) cost-effectiveness [2,23].Accordingly, a central aspect of diffusion processes within this school of thought are "information cascades" [42][43][44][45] that transmit innovations and their associated utility expectations from one actor to another.
Social accounts of diffusion emerged from sociological research.In opposition to rational accounts, social accounts emphasize the "pressure toward social conformity" [25] (p.70) that grows as more and more actors in a field adopt an innovation.Drawing on institutional theory [46][47][48][49], social accounts of diffusion highlight how group pressures and bandwagon effects may replace rational concerns as an innovation becomes increasingly adopted in a field.These dynamics may even lead actors to adopt new ideas, structures, or practices that are inefficient for them [21,23] simply as a means of maintaining legitimacy [50].
Both rational and social accounts of innovation diffusion tend to assume a populationlevel perspective that emphasizes interorganizational conditions (information cascades or pressures for social conformity) as the central mechanisms explaining diffusion [25].In both, innovation diffusion is viewed as a natural phenomenon akin to gravity: It permeates through populations of individual and organizational actors as a passive force that may prompt human activity but is hardly driven by it.Accordingly, much diffusion research has been focused on measuring the speeds and rates of this force [2,15,35,36,[51][52][53].This has led to the well-established insight that, "when plotted over time, the cumulative number of adopters of an innovation approximates an S-shaped curve" [54] (p.544).
One limitation of these perspectives on the spreading of innovations is that they relatively discount the proactive agency and strategic decision-making of individual and organizational actors.Human agency fits into rational and social accounts of innovation diffusion as a reaction-i.e., an adoption decision prompted by information cascades and conformity pressures (see also [55])-but seldom as a proactive driving force.While [38]'s seminal review of diffusion research noted that "organized dissemination" of innovations falls under the roof of diffusion research, this research has, in practice, paid little attention to such a proactive and purposeful spreading of innovations by actors.Rather, such "evangelism" for specific innovations by individual actors has been characterized as "the reverse of diffusion" [24,56].
This article argues that in order to address pressing global issues through innovation, scholars must develop a systematic understanding not only of how beneficial innovations may passively diffuse through populations, but also of how strategic actors may purposefully and proactively disseminate them.This phenomenon of proactive and strategic dissemination of innovations is increasingly of interest to scholars from across academic disciplines, where it is variously referred to using terms such as "scaling" and "upscaling".This article uses the term "innovation scaling" to refer to this process.While a cohesive definition and common conceptualization of innovation scaling has not yet manifested, existing definitions gravitate around the understanding that strategic and purposeful actors may indeed be a driving force behind an innovation's movement into new contexts and degrees of quality [26,27,30].For example, [30] understands scaling to be a deliberate, guided process.According to [26], scaling "is often considered to refer to a series of processes to introduce innovations with demonstrated effectiveness through a program delivery structure with the aim of improving coverage and equitable access to the innovation(s)" [26] (p.2).
Accordingly, this article views innovation scaling as complementary to diffusion (see Table 1).Research into innovation scaling does benefit from the conceptualizations of field-level rationality and institutional factors developed in innovation diffusion research.At the same time, by investigating not what happens in a field (the focus of diffusion research), but, rather, what actors do, the "agentic account" offered by innovation scaling offers a complementary perspective that may contribute to scholars' overall understanding of how new ideas, structures, and practices spread.This benefit accrues in addition to the distinct practical urgency that exists to develop a systematic understanding of how innovations may be proactively and strategically scaled to address pressing global issues.Addressing both the theoretical and practical opportunities presented by a systematic review of innovation scaling research, this article now presents the methodological approach and findings of the systematic literature review.

Methodological Approach
This article follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodological approach [31,32].This approach presents a systematic way of reviewing literature due to its clear step-by-step instructions that can be transparently communicated and replicated.It is widely accepted as a standard for conducting systematic reviews due to its methodological clarity [31].Figure 1 presents an overview of our application of the PRISMA approach to the innovation scaling literature.The process began with the definition of eligibility criteria that would determine the inclusion of articles in our systematic review.The PRISMA process followed for this literature review.Source: Authors' own elaboration, based on PRISMA guidelines (see [32]).

Eligibility Criteria
Four criteria were used to determine whether an article was eligible to be included in our systematic review: Topic: To be eligible for inclusion in the review, an article had to focus on the "scaling" (rather than "diffusion" or "adoption") of a specific innovation or multiple innovations.
Language: To increase the accessibility of the review's findings, only publications written in the English language were eligible.
Publication Year: Only articles published from the year 2000 onwards were included.Publication Type: To be eligible for inclusion, an article had to be published in a peer-reviewed journal and be a full journal article.This means editors' notes, research commentaries and book reviews were excluded from the review.
Due to the limited quantity of extant research on innovation scaling, and the interdisciplinary nature of this research field, this study included empirical as well as theoretical articles in the review and did not use the field of study as an eligibility criterion.

Literature Search Strategy
To find the initial sample of articles, a search in the SCOPUS database was conducted, searching articles' titles, abstracts, and keywords using the following query: the word "Scaling", "Scaling Up", "Scaling-Up", "Upscaling", or "Replication" had to be combined with words indicating the theme of "innovation" (see Table 2).

Eligibility Criteria
Four criteria were used to determine whether an article was eligible to be included in our systematic review: Topic: To be eligible for inclusion in the review, an article had to focus on the "scaling" (rather than "diffusion" or "adoption") of a specific innovation or multiple innovations.
Language: To increase the accessibility of the review's findings, only publications written in the English language were eligible.
Publication Year: Only articles published from the year 2000 onwards were included.Publication Type: To be eligible for inclusion, an article had to be published in a peer-reviewed journal and be a full journal article.This means editors' notes, research commentaries and book reviews were excluded from the review.
Due to the limited quantity of extant research on innovation scaling, and the interdisciplinary nature of this research field, this study included empirical as well as theoretical articles in the review and did not use the field of study as an eligibility criterion.

Literature Search Strategy
To find the initial sample of articles, a search in the SCOPUS database was conducted, searching articles' titles, abstracts, and keywords using the following query: the word "Scaling", "Scaling Up", "Scaling-Up", "Upscaling", or "Replication" had to be combined with words indicating the theme of "innovation" (see Table 2).
This initial search yielded 3985 articles.To improve the accuracy of the query, we restricted our review to articles from this sample that had one of the following exact keywords: "Scaling", "Scaling Up", "Upscaling", "Scaling-up", "Scale-up", or "Innovation" and applied the inclusion criteria noted above.This resulted in 628 remaining articles.
After this, each article's title, keywords, and abstracts were manually reviewed for relevance.Most articles that were removed at this stage stemmed from the natural sciences.This led to a final sample of 143 eligible articles.An expert reviewed this list of 143 articles and was then asked to suggest any additional articles that may have been falsely excluded.Four additional articles were suggested, thus leading to a final sample of 147 articles.The final list of articles can be found in the Supplementary Materials.

Coding Process
The 147 articles were coded by multiple coders using a blend of inductive and deductive coding processes.This means that the research began with a predetermined coding scheme informed by the authors' existing understanding of innovation scaling processes (deduction) but used also open codes to capture aspects that were not included in extant typologies and frameworks of innovation scaling (induction).During the process of coding, the coding scheme was regularly and iteratively revised as new open codes emerged.This was done until saturation was reached.In order to ensure intercoder reliability during the initial coding, the same subset of ten articles were coded by all coders at the beginning of the coding process.These results were then compared, and, in case of disagreements, coders discussed and cooperatively refined the deductive codes.
The coding scheme was designed to capture different aspects of the research articles.The codes captured basic information about the articles, such as region of the world, focus, research methodologies.They also captured the exact definition of scaling used (articles that did not provide a clear definition were coded as "no clear definition"), as well as actions and behaviors of strategic actors and contextual factors that were depicted as driving or hindering the scaling process.A factor deemed to be a driver or barrier was coded only once per article.
The coding scheme for drivers and barriers was initially based on the typology of innovation antecedents presented by [57] (p.155) but further refined and extended through the inductive part of our coding process.Relevant factors not captured by the initial categories were coded with an "open" code, regularly reviewed, discussed, and grouped into new categories.This process ensured that the analysis of the texts was informed by extant knowledge of the innovation process while at the same time being able to surface and systematize relevant aspects typically neglected in innovation scaling typologies and frameworks.Overall, it allowed us to conceptualize eleven modal contextual factors influencing the success of scaling projects as a driver or barrier depending on the situation.
For exploring the definitions provided by the articles, the definitions were coded and analyzed based on [30]'s framework differentiating between "scaling out" (the expansion of innovations to a larger group of actors), "scaling up" (the enactment of political and legal change), and "scaling deep" (enacting deep cultural and institutional change).This framework, which the authors became aware in the early phases of the literature review, represents one of the first systematic attempts to present a nuanced and multi-dimensional understanding of scaling.Due to its high relevance and wide breadth, it served as a starting point for the coding of the definitions of scaling.
The coding scheme can be found in the Supplementary Materials.

Findings
In this section, we present the results of our systematic review as they correspond to the research questions.

Landscape of Innovation Scaling Research
The most popular journals included in the analysis were Sustainability, Agricultural Systems, International Journal of Agricultural Sustainability, Journal of Cleaner Production, Environmental Innovation and Societal Transitions, Health Research Policy and Systems, Globalization and Health, and BMC Health Services Research.Twenty-three journals had two articles each included in the review, and eighty journals had only one article each included in the review.The empirical phenomena addressed by the articles, accordingly, were diverse, though they primarily related to health, agriculture, and development.Combined, these areas made-up 78 percent of the included studies.Education, social welfare, business, and public administration were represented as well, but less so.
This result shows that interest in innovation scaling is widespread and multi-disciplinary.However, even though the topic of innovation scaling is cross-cutting, there has been relatively less innovation scaling research in business and public administration journals.This is particularly interesting considering that these journals have a rich history of general innovation scholarship.A possible explanation is that innovation scaling is still a nascent field research.This becomes clear particularly when looking at the yearly number of publications (Figure 2).In this millennium, the first innovation scaling research was published in 2004, but it took until 2010 for a growth trend to emerge.The vast majority of innovation scaling research was published in the last four years.

Landscape of Innovation Scaling Research
The most popular journals included in the analysis were Sustainability, Agricultural Systems, International Journal of Agricultural Sustainability, Journal of Cleaner Production, Environmental Innovation and Societal Transitions, Health Research Policy and Systems, Globalization and Health, and BMC Health Services Research.Twenty-three journals had two articles each included in the review, and eighty journals had only one article each included in the review.The empirical phenomena addressed by the articles, accordingly, were diverse, though they primarily related to health, agriculture, and development.Combined, these areas made-up 78 percent of the included studies.Education, social welfare, business, and public administration were represented as well, but less so.
This result shows that interest in innovation scaling is widespread and multi-disciplinary.However, even though the topic of innovation scaling is cross-cutting, there has been relatively less innovation scaling research in business and public administration journals.This is particularly interesting considering that these journals have a rich history of general innovation scholarship.A possible explanation is that innovation scaling is still a nascent field research.This becomes clear particularly when looking at the yearly number of publications (Figure 2).In this millennium, the first innovation scaling research was published in 2004, but it took until 2010 for a growth trend to emerge.The vast majority of innovation scaling research was published in the last four years.Methodologically, articles were overwhelmingly qualitative, with a majority of articles utilizing either single or multiple case studies.For the purposes of this review, these case studies provide a large body of empirical evidence that facilitated our identification of the eleven underlying modal contextual factors driving and/or hindering innovation scaling.An overview of both the field of study and the methodology is provided below as Table 3. Methodologically, articles were overwhelmingly qualitative, with a majority of articles utilizing either single or multiple case studies.For the purposes of this review, these case studies provide a large body of empirical evidence that facilitated our identification of the eleven underlying modal contextual factors driving and/or hindering innovation scaling.An overview of both the field of study and the methodology is provided below as Table 3.
Looking to the key sectors involved in innovation scaling, the largest share of articles presented cases in which a mix of different sectors was involved.When only a single sector was involved, the most common one was the public sector, followed by the private sector, and non-profit sector.As there is a large body of literature that focuses on scaling and social enterprises, they were given a separate code.Overall, the majority of articles present research investigating the importance of collaboration and multi-stakeholder involvement for the innovation scaling process.
Geographically, most research focused on Africa, followed by Europe, a Global perspective, Southeast Asia, North America, Asia, Oceania, South America, and then the Middle East.In some cases, research with multiple case studies had cases from different geographical areas and/or the innovation itself was focused globally, see, e.g., [58].
An overview of the key sectors and the geographical regions is shown below in Table 4.South America 7

Middle East 1
Notes: Number of Articles (N) = 147.For sector analysis, only one code was applied per article.However, 19 articles could not be classified as a distinct sector.These are not included in this table.For the regional analysis, in the event of multiple case studies from different countries, double codes were used.Thus, in the regional analysis 294 regions were coded within the 147 articles.

Definitions of Innovation Scaling
Regarding the conceptualizations and definitions of innovation scaling, somewhat unsurprisingly-and in line with previous research conducted by [57] on innovationapproximately 43 percent of articles did not provide a clear definition of innovation scaling.
Examining the consistency of the definitions, most definitions (61 articles) explicitly or implicitly referred to the concept of scaling out.A much smaller proportion referred to scaling up (19 articles).Scaling deep was referred to the least (10 articles).
Interestingly, 33 articles appeared to use the terminology for their definition incorrectly (e.g., referring to "scaling up", when what they were describing would more accurately have been labelled "scaling out" [30]).This further illustrates the conceptual ambiguity and lack of clarity that has characterized the cross-disciplinary innovation scaling literature.Beyond the themes of expansion (i.e., scaling out), and systems change (i.e., scaling up), we identified two additional themes in the definitions offered.The first is the consistent reference to active and deliberate processes, which supports the above argument that innovation scaling is distinct phenomenon from innovation diffusion.The second is the focus on impact: When characterizing innovation scaling, many articles emphasize that innovation scaling happens for the specific purpose of enhancing the impact of innovations.

Drivers and Barriers of Innovation Scaling
The large number of case studies present in the articles facilitated the empirically grounded identification of eleven modal contextual factors acting either as drivers of barriers to the success of innovation scaling.Four of these are external factors (legislative environment; adaptation to local context; collaborations, partnerships, networks; and stakeholder engagement), two are project factors (technical factors; and performance management information), and five are organizational factors (financial resources; human resources; leadership; risk aversion; time).Tables 5-7 present these factors along with statistics of how common they were across the reviewed articles, and examples for how each modal factor may either act as a driver or barrier of innovation scaling.

External Factors
Legislative environment.As a driver, this external factor captures governments who were open to regulatory changes, had clear regulatory frameworks, and/or a coordinated institutional environment.These aspects greatly facilitated the scaling up of innovations.At the same time, this factor could act as a barrier when there was red tape, contradictory policies and regulations, or a lack of a fitting policy arena for the innovation.
Adapting to local context.As a driver, this external factor refers to projects where there was an emphasis on proactively engaging with local conditions, rather than taking a "one size fits all" approach.When projects did not take local conditions into consideration, they ended up meeting resistance in terms of uptake as well as problems in implementing the projects altogether.
Collaboration, Partnerships, Networks.The necessity to collaborate and coordinate was another external factor that seemed to drive scaling up when it was actively done, as it facilitated knowledge sharing, better coordination among actors, and the dissemination of ideas.At the same time, failure to collaborate acted as a barrier to innovation scaling by leading to a lack of coordination.
Stakeholder Engagement.Most articles referring to stakeholder interests as a driver concluded that engaging directly with stakeholders (e.g., ensuring training of users, introducing new products or processes) was a key driver for scaling.On the flipside, failure to do so often led to a lack of buy in, use and/or lack of interest from local communities.

Project Factors
Technical factors.Technical factors were a driver when innovations were complementary to existing systems, or adapted to existing, accepted, standards.They were a barrier when the innovation was too advanced for the users or when there was a general lack of technical capacity.
Performance management information.Pilot projects, performance evaluations and the use of performance management were seen as key drivers in scaling.They were used to identify critical weakness of projects, and to develop more reflective learning as projects begin to scale.As barriers, a lack of information and performance information stalled or derailed some projects as the implementers ran the risk duplicating errors or approaches that may have already failed in the past.

Barriers
Lacking information about good practice, performance information needed to understand how the scaling up process was happening, how they could be improved, and built upon.Financial Resources.One of the most commonly identified factors influencing the success of innovation scaling were financial resources, both in terms of projects costs of implementation as well as the cost of usage for the end user.Having stable financial resources for the entire duration of the scaling phase, including accurate budgeting forecasting, was a critical driver.As a barrier, a lack of financial recourses became a strain to many innovation scaling projects.These included fragmented budgets or a financial unviability of products.
Human Resources.As a driver, the capacity to hire staff and recruit the "right" people for a particular team was identified to contribute to the success of innovation scaling projects.In some cases, this was very much intertwined with the availability of financial resources.As a barrier, factors such as high turnover and a lack of skills and training resulted in projects failing to launch, or projects failing the scale effectively because the skills were either spread too thin or were not adequate at all.
Leadership.Having political commitment for the projects scaling phase, either at the local or national level, acted as a driver of innovation scaling.Not only did this provide legitimacy for the projects, it also often facilitated other factors such as funding and converging regulatory frameworks.By the same token, a lack of political commitment, especially in more hostile environments, created serious problems related to accessibility and acceptability in the local communities.Internal to the projects themselves, clear leadership skills, including adaptive leadership skills and the ability to negotiate, appeared as critical drivers.As a barrier, a lack of leadership and management skills created leadership vacuums, leading to projects either not realizing their intended outcomes, or not actually being completed in the first place.

Barriers
Not enough time to commit to the project, or the projects themselves did not have enough time to develop and adapt before their funding ended.

3
Risk Disposition.The concept of risk disposition was primarily identified as a barrier to scaling processes: Professional resistance of leaders, or uncertainly of outcomes, led to hamstringing a project by leading to a lack of action.On the flipside, being able to address these risks directly through increased stakeholder engagement and reframing was also noted as a driver of scaling processes.
Time.The concept of time as a driver refers primarily to the ability to accurately forecast not only each project stage, but also the ability for projects to be implemented in a "timely manner" or at the "optimal time".In the cases where time was perceived as a barrier, this was mostly related to ill-planned projects (e.g., too short project cycles), or projects that were tied to hard-to-achieve external funding deadlines.Some articles also noted as a barrier the underestimation of the time for implementation that may occur within project teams.
To examine the consistency of the modal contextual factors and their manifestation as either driver of barrier, the coded segments were compared across different sectors and regions of the world.Based on this analysis, the top three to four drivers and barriers are presented in Tables 8 and 9.The drivers and barriers are ordered by based on the number of codes assigned to all articles for each respective sector/region.It is important to keep in mind that as this is initial exploratory research, this article cannot make a statement on the statistical significance of these observations, but future research may wish to explore this in greater detail.
The results show that many drivers manifest in a similar fashion across sectors and regions.For example, regardless of sector, collaboration, networking building, and adaptation to local contexts were key drivers.At the same time, some clear differences do emerge in terms of sector.For example, financial resources did not appear to be a salient driver of scaling (3.6%) in the private sector, while it was one of the most common drivers for non-profit (15.6%) and social enterprises (13.8%).Additionally, stakeholder engagement and user engagement appear to be strong drivers of scaling in the private (10.7%) and public (12.5%)sectors, but they did not appear to be as salient in the non-profit (6.3%) and social enterprise literature (5.2%).
In terms of barriers, financial resources, human resources, and a lack of coordination appear to be clear barriers across sectors.However, lack of stakeholder engagement appeared as a more common barrier for the non-profit sector (15%) and social enterprises (10.3%) compared to the public sector (5.4%) or the private sector (0%), while technical barriers appeared to be more problematic in the public sector (10.8%) and projects that are cross sectoral (11%) compared to the non-profit sector (5%) or social enterprises (3.4%).Examining the modal contextual factors as drivers and barriers by the most commonly studied regions, collaboration, partnerships, networks were identified as particularly salient (ranging from 11 to 17%).Stakeholder engagement appeared to be more salient in cases from Asia, Southeast Asia, Africa, and Europe (ranging between 8.5% and 13.5%), compared to North America (4.3%).Adapting to local context was more often referenced as a driver of scaling in North America (15.2%),Europe and Asia (both 8.5 and 8.1%), compared to Southeast Asia (2.6%).A lack of collaboration and coordination appear to be larger barriers in Southeast Asia (18.9%) compared to other regions (ranging from 0 to 10%).Government regulations were particularly salient in North America (16.7%) compared to the other regions (which ranged from 5% to 10%), financial resources were most commonly cited as barrier in Southeast Asia (21.6%),Europe (17.3%) and Africa (12.5%) and human resources were relatively more salient in North America (25%) and Southeast Asia (18.9%), and Africa (15%).The themes presented based on frequency of codes, where the top 3 most coded theme is presented, some sectors may have more themes presented as these themes had the same number of coded segments.The category 'other' is not included in this table.* Refers to themes with the same number of codes.** Refers to themes with the same number of codes.*** N, refers to the number of papers coded in this region.

Discussion
Drawing on the results of this systematic literature review, it is possible to start to develop a strong foundation for the discussion and theorization of innovation scaling.This foundation will be elaborated in this section and become the basis of four concrete contributions of this research.Three are discussed in this section, and the fourth is presented in the conclusion.
The first contribution is related to the conceptual differentiation between the topics of innovation diffusion and innovation scaling.While scaling and diffusion have often been used as synonyms, in this article it is argued that there are clear theoretical differences between the two terms that must be considered.While the general social science phenomenon of "diffusion" can be understood as the longitudinal spreading of (new) ideas, structures, and practices in social systems, based on extant theory and the systematic literature review, it is possible to differentiate between innovation diffusion and innovation scaling.
Innovation diffusion views spreading as a passive process where innovations permeate naturally throughout a population or field.Innovation scaling views spreading as a proactive and deliberate process driven by individual and organizational actors.As the results presented above show, research into the latter cuts across disciplines, is nascent but rapidly growing, emphasizes the increasing of an innovations social impact, yet currently lacks theoretical clarity.Drawing on extant theory and the systematic review presented here, it is possible for this article to offer a clear baseline conceptualization of the innovation scaling phenomenon: Innovation scaling refers to the process of proactively, strategically, and/or deliberately spreading an innovation.Individual and organizational actors engage in innovation scaling in order to enhance the beneficial impact of a given innovation by replicating it.
Working from this understanding, it is possible to explore the drivers and barriers of innovation scaling that have been identified captured by extant research.This is the second contribution of this article.Analyzing the reported drivers and barriers, it was possible to identify 11 modal contextual factors influencing the success of innovation scaling projects as either drivers or barriers, and which could be further categorized into three primary groups (external factors, project factors, organizational factors).The fact that two of the three most commonly identified drivers appear to be inter-personal in nature highlights that innovation scaling is characterized by a stark inter-dependency of actors.These actors are not only the recipients of the innovations, but also employees, citizens, organizations, regulators, and governments alike.Therefore, innovation scaling must progress utilizing a systems perspective that considers not only the scalability of the innovation itself, but also its impact and relevance to the population in which it is to be scaled.
Based on these insights and extant research, it is possible to derive an initial overarching framework of the innovation scaling process.This is the third contribution of this article.As described above, one of the clearest and most popular existing frameworks for understanding innovation scaling is a typology developed by Moore et al. [30], who distinguish three types of scaling: Scaling up, scaling deep, and scaling wide.Moore et al. [30] argued that specific strategies and processes could be identified for each type of scaling depending on one's goals, but also acknowledged the cross-cutting nature of these strategies.The approach of distinguishing between strategies for different types of scaling can also be seen in previous models and frameworks [19], or in papers that are attempting to develop "success strategies" for different types of scaling [17].
Based on the above analysis of how innovation scaling may be conceptualized and how relates to the modal contextual factors, the authors of this paper would go a step further and argue that scaling deep, scaling wide, and scaling up are indeed important, but are best conceptualized not as distinct types of innovation scaling but rather as different facets of the same fundamental scaling process.
To successfully scale an innovation may at one stage require the legislative environment to change (typically related to "scaling up"), and at another stage the engagement with a wider range of stakeholders may become critical (typically related to "scaling wide").These appear to be less distinct types of scaling processes and rather different aspects of the same fundamental scaling process that become salient at different points in time or in different situations.Furthermore, it is also possible that when one engages in scaling wide, it leads to a change in the legislative environment (scaling up) (this holds true for other combinations as well), thus demonstrating the close linked nature of these processes.
It is also important to recognize that the relationship between these facets of innovation scaling and the modal contextual factors is bilateral.The presence of modal contextual factors affects the outcomes (ultimate success or failure) of an innovation scaling project and leads to the enactment of specific facets of the innovation scaling process (i.e., wide, up, out).At the same time, the enactment of these facets may in turn change the modal contextual factors, such that they may flip from being a barrier (e.g., unsupportive legislative environment) to a driver (e.g., supportive legislative environment).Within innovation scaling processes, modal contextual factors and facets of scaling therefore are likely to exhibit a co-dependent dynamic in which both change across time.
These insights facilitated the development of an initial overarching framework of innovation scaling processes, which is presented in Figure 3.The framework proposes three essential components of a scaling process: modal contextual factors, facets of innovation scaling, and eventual outcomes.Each component is co-dependent on the others, both influencing them and being influenced by them.
Explicating Figure 3, four initial propositions delineating the innovation scaling process can be proposed: Proposition 1-Innovation scaling is a multi-faceted process.Depending on the specific situation, the three facets of scaling out, scaling up and scaling deep may become enacted to differing degrees.It is possible that they become enacted to equal degrees, or that one or two facets overshadow the others.
Proposition 2-The multi-faceted innovation scaling process both affects, and is affected by, eleven modal contextual factors: financial resources, human resources, leadership, risk disposition, time, technical factors, performance management information, legislative environment, adaptation to local context, collaboration, and stakeholder engagement.Depending on the specific situation, these modal contextual factors may either act as drivers or barriers of the scaling process.
Proposition 3-The modal contextual factors and innovation scaling facets exist in a dynamic interplay.A specific arrangement of modal contextual factors may prompt the enactment of a certain arrangement of innovation scaling facets, which in turn may shift the modal contextual factors (e.g., changing one from acting as a barrier to acting as a driver).
Proposition 4-The dynamic interplay between modal contextual factors and innovation scaling facets determines the ultimate outcome of the innovation scaling process, i.e., whether it is a failure or success.As this framework is derived from a cross-disciplinary literature review, the pro sitions are likely to be applicable across many domains.At the same time, further resea is needed to explore the framework in more depth, after which our initial propositi may need to be revisited.The next section presents promising avenues for such resea and the limitations of the study.

Conclusions
The concept of innovation scaling is of high interest to a wide number of discipli This article sought to address the conceptual ambiguity that exists in the field using a cr disciplinary systematic literature review.As a result of this research, four critical con butions are made to the scaling innovations literature.The first contribution outlined p As this framework is derived from a cross-disciplinary literature review, the propositions are likely to be applicable across many domains.At the same time, further research is needed to explore the framework in more depth, after which our initial propositions may need to be revisited.The next section presents promising avenues for such research and the limitations of the study.

Conclusions
The concept of innovation scaling is of high interest to a wide number of disciplines.This article sought to address the conceptual ambiguity that exists in the field using a cross disciplinary systematic literature review.As a result of this research, four critical contributions are made to the scaling innovations literature.The first contribution outlined presents the conceptual differences between the concepts of innovation diffusion and innovation scaling.The second contribution identified eleven modal contextual factors that influence the outcomes of innovation scaling.These factors were consistent across many innovation projects, suggesting an element of universality across research domains and contexts.The third contribution presented a framework of innovation scaling accompanied by four theoretical propositions that emerged therefrom.The framework developed consists of three main components: the modal contextual factors, the facets of innovation scaling, and the outcomes.The co-dependency of these three components highlights the complexity of the scaling process.This is by no means a finished framework-but the beginnings of a larger research agenda in understand how innovations scale of which we invite scholars to participate in.
In making these contributions, this systematic literature review facilitates a burgeoning agenda for future research-this is the fourth core contribution of this article.
The theoretically and empirically grounded framework of innovation scaling points to the multifaceted relationship between action and modal contextual factors in the scaling process.Where much extant theory of innovation scaling had treated contextual factors primarily as antecedents of the scaling process influencing outcomes (see, e.g., de Vries [57]), the analysis presented above suggests that contextual factors exist in a dynamic relationship with innovation scaling efforts.Overall, the findings and framework call to attention Gidden's [59]'s seminal structuration theory explicating the dynamic relationship between agency (the actors doing the innovation scaling as a strategic project) and structure (contextual factors).As such, a first line of future research is related to the theoretical understanding that contextual factors such as the legislative environment are not antecedents but changeable modal factors.As such, new questions come into view.For example, how do innovations that are scaled in different contexts, change modal contextual factors from hindering their deliberate spreading to supporting it?How can strategic actors engaged in innovation scaling projects make such changes to the modal contextual factors?Which facets of innovation scaling are enacted as part of these processes?Answering research questions like these will require in-depth qualitative studies utilizing process analysis methods [60] to explicate how the scaling unfolded.
A second promising avenue for research regards the variability of innovations during the scaling process.Extant theoretical frameworks of innovation scaling treat the innovations themselves as unchanging.The assumption is that, as strategic actors spread innovations into new contexts, the innovations maintain their original form and function.However, looking to the results of this literature review-specifically the prevalence of the external factor "adaptation to local context"-this assumption appears flawed.Indeed, within the literature on innovation diffusion (which, as discussed above, also investigates the spreading of innovations) it has become well established that the adoption of innovations by new actors frequently also requires the adaptation of those innovations [25,[61][62][63][64]. Innovation diffusion research, due to its population-level perspective, has focused on how adopters may adapt innovations when information cascades and/or legitimacy pressures bring those innovations to their attention.Mirroring these investigations, further empirical and theoretical research is needed to systematically investigate and explicate how the actors attempting to scale innovations may deliberately adapt them to make them more palatable for the targeted audience-or in contrast, study the resistance to adaption of an innovation.As this article demonstrates, some research is already paying explicit attention to this factor.Accordingly, this article has sought to acknowledge the potentially central role of adaption in the proposed framework of innovation scaling.But further research is needed to develop a more fined grained understanding of adaption within innovation scaling by systematically importing more theoretical insights from innovation diffusion research on this topic and engaging in empirical research that probes the process of deliberate adaptation in-depth.
Despite the benefits and rigor of the PRISMA approach, several limitations should be acknowledged.First, this research focused primarily on articles that referred directly to the concept of "scaling".This means that it is possible that articles outlining innovations that scaled but were not described as such were not captured by this review.Related to this, in the process of deciding upon the filters for this systemic review, and to reduce the sample to a manageable number, keywords were used.As keywords are not standardized, some relevant articles may not have been captured.While a thorough review of keywords from well-known studies examining innovation scaling was conducted, it is still possible that some articles were missed.This limitation was somewhat minimized by the manual review of the final list of included articles by an expert in the field.Third, there is always a risk of coding error.This was minimized though various coding strategies developed between the coders.In addition, due to the transparency of the PRISMA process as well as making the coding scheme available in the appendix, any reader can assess the internal validity of the results presented.Fourth, as with all systematic reviews there is a threat of survivorship bias.Articles included in this study almost all focus on successful innovations that have been scaled up or are in the process of scaling up.Finally, by taking a multi-disciplinary approach, we run the risk of oversimplifying the scaling processes across different domains.While the review did show many similarities between different disciplines and areas of innovation, any form of theoretical framework must always be critically reviewed and applied in a case-by-case fashion.
Despite these limitations, several implications emerge from this research.For anyone who is interested in the topic of scaling, this paper provides the most comprehensive overview currently available in the academic literature.Thus, for future research focusing on scaling or innovation scaling, this paper will be a key starting point in guiding this research.Second, this paper develops a strong foundation for future cross disciplinary research on the topic of innovation scaling.This is accomplished by setting a common vocabulary and establishing clear definitions of scaling and diffusion.In doing so, this paper offers a strong initial attempt at both enabling future researchers to establish consilience between one another and encouraging cross-disciplinary learning.Third, by identifying commonly occurring contextual modal factors that affect the scaling process, it is possible to begin to develop an understanding of how each of these is likely to affect the scaling process itself.Even more interesting, however, is that the identification of these factors should enable the development of new strategies and approaches to innovation scaling.Finally, the implications of our theoretical model would suggest that scaling needs to be understood as a continually changing process, both in terms of the interdependency regarding the type of innovation scaling (up, out, or deep), but also the contextual factors that influence this possibility in the first place.
Beyond the theoretical and academic contributions, practitioners, too may find this article relevant through its clarification of scaling and the identification of eleven driver and barriers of scaling (identified as modal contextual factors) that influence the success of scaling projects.The conceptualization of scaling as a multi-faceted process also offers practitioners and scholars alike a new way to look at and understand how to scale a given innovation.Situating this within the developed framework offers a new tool to understand either why a given innovation did scale, or why the outcome of scaling did not go as planned.Overall, this article provides a strong foundation for future theoretical exploration and development of the concept of innovation scaling across disciplines.

Figure 2 .
Figure 2. Included publications by year of publishing.Source: Authors' own elaboration.

Figure 2 .
Figure 2. Included publications by year of publishing.Source: Authors' own elaboration.

14
modal contextual factors are presented based on frequency of codes, where the top 3 most coded modal contextual factors are presented.Some sectors may have more modal contextual factors presented because these factors had an equal number of coded segments.* Refers to modal contextual factors with the same number of codes.** Refers to modal contextual factors with the same number of codes.

Figure 3 .
Figure 3. Scaling Innovations Theoretical Framework.Source: own elaboration based on own analysis.

Figure 3 .
Figure 3. Scaling Innovations Theoretical Framework.Source: own elaboration based on own analysis.

Table 1 .
The relationship between innovation diffusion and innovation scaling research.Source: Authors' own elaboration.

Table 3 .
Empirical Areas and Methods in Innovation Scaling Research.Source: Authors' own elaboration.
Notes: Total number of articles is 147.Articles were double-coded if they focused on multiple fields of study or combined multiple methods.

Table 4 .
Key Sectors and Regions Studied Scaling up Research.Source: Authors' own elaboration.

Table 8 .
Modal Contextual Factors as Drivers and Barriers by Sector.Source: Authors' own elaboration.

Table 9 .
Modal Contextual Factors as Drivers and Barriers by Region.Source: Authors' own elaboration. Notes: