Business value of SME digitalisation: when does it pay off more?

ABSTRACT Small and medium enterprise (SME) digitalisation involves the reinforcement, modification, and renewal of business models with the help of digital technologies. It is widely considered imperative for SMEs to stay relevant in the digital age. Yet, little is known about the conditions under which SME digitalisation improves the performance of SMEs in the IS literature. Guided by the SME literature, we postulate that the business value of SME digitalisation – based on its impact on improving financial firm performance – is dependent on two factors that are particularly relevant to SMEs due to their smallness and flexibility: radical orientation and organisational rigidity. Using data from multiple waves of surveys in 2019–2020 and archival financial data from Dutch SMEs, we demonstrate that the positive impact of SME digitalisation on performance improvement strongly depends on SME characteristics. SMEs who are oriented towards radical change and are more rigid are disadvantaged and attain lower returns on SME digitalisation.


Introduction
Small and medium enterprises (SMEs) increasingly face challenges regarding the use of digital technologies (Cenamor et al., 2019;Soluk et al., 2021).Digital technologies such as big data analytics, artificial intelligence, cloud computing, and blockchain have become ubiquitous and accessible to SMEs, allowing them to optimise existing processes, enhance their value proposition as well as improve customer experience (Ainin et al., 2015;Bouncken et al., 2021;Matt et al., 2015).The resulting digitalisation of business models (hereafter: SME digitalisation) is of vital importance to SMEs to stay relevant in the digital age (Benitez, Chen, et al., 2018;Fischer et al., 2020;Kane et al., 2015;Sebastian et al., 2017).Thereby, SME digitalisation concerns the degree of digital change and ranges from the incremental use of digital technologies to reinforce existing business models (Tumbas et al., 2018) to the development of entirely new business models, also known as business model renewal (Volberda et al., 2018) that uses digital technologies to deeply and thoroughly alter firms' value creation (Verhoef et al., 2021;Vial, 2019).
While much research about digitalisation and digital transformation has been performed in the context of large incumbents (see Firk et al., 2021;Wang & Wang, 2020;Wessel et al., 2021), little is known about its performance implications, i.e., how firms reap financial benefits from SME digitalisation (Eller et al., 2020;Li et al., 2018).With SMEs representing more than 90% of the firms in the European Union as well as 50% to 70% of the workforce (Müller et al., 2018), it is highly important to shed light on the business value implications of SME digitalisation.Especially to better understand inhibitors and enablers of the relationship between SME digitalisation and performance improvement.
This study responds to recent calls to investigate the enablers and inhibitors of SME digitalisation (Chan et al., 2019;Hanelt et al., 2021), including organisational culture (identity, leadership) and organisational barriers (inertia, resistance) (Chan et al., 2019;Proksch et al., 2021;Vial, 2019;Volberda et al., 2018).In doing so, this paper integrates specific factors from the SME context to explain the diverging performance outcomes of SME digitalisation, namely: radical orientation-that is, an SME's cultural dimension that reflects the firm employees' orientation towards having ideas that differ substantially from existing practices and alternatives (Bala & Venkatesh, 2013;Luo et al., 2012;Proksch et al., 2021;Vial, 2019), and organisational rigidity-that is, an SME's organisational barrier that reflects the firm's inability to adapt and innovate (Chan et al., 2019;Fan et al., 2020;Gilbert, 2005;Lucas & Goh, 2009;Polites & Karahanna, 2012;Vial, 2019).Using data from multiple waves of surveys in 2019-2020 and archival financial data from Dutch SMEs, we test whether a radical approach hampers the business value of digitalisation by impeding organisational learning as well as how rigidity may limit the required flexibility to overcome path dependence and adjust routines for SMEs in particular (Chan et al., 2019;Karhade & Dong, 2021b).
This study contributes to IS literature in three notable ways.First, we extend knowledge on the degree to which SMEs derive financial value from using digital technologies (Baiyere et al., 2020;Soluk & Kammerlander, 2021;Volberda et al., 2018).We specifically integrate knowledge on the contextspecific inhibitors for SME digitalisation (Hong et al., 2014) and find that, against the promoted strategic imperative to increase digitalisation efforts, the business value of SME digitalisation is hindered by an SME's radical orientation and organisational rigidity (Chan et al., 2019;Proksch et al., 2021;Soluk et al., 2021).We also contribute new insights about the role of organisational factors that influence the business value of IT for SME firms (Benitez, Castillo, et al., 2018;Benitez, Llorens, et al., 2018;Trang et al., 2021).Second, from an empirical point of view, we expand existing methods and evidence in the literature on the business value of SME digitalisation by combining data from a multi-year survey and financial data to increase the validity of the findings.Due to the unique characteristics of SMEs, most studies resorted to qualitative methods and thus lacked quantitative insights into performance.Third and more broadly, our work also contributes to the literature on digital transformation.We question existing contentions that treat digital transformation as a binary outcome of transformed vs. not transformed (Vial, 2019;Wessel et al., 2021).Instead, we use variance logic to theorise about the degree of change.Against the common contention that digital transformation requires radically new ideas (Baiyere et al., 2020;Vial, 2019;Vom Brocke et al., 2021;Wessel et al., 2021), our study finds that such a radical orientation can limit the potential business value of SME digitalisation.Furthermore, incumbents may struggle with high rigidity from their established organisations to realise the potential business value of SME digitalisation.By incorporating these two important inhibiting factors, we warn against the transformation imperative of "the more the better" (Baiyere et al., 2020;Vial, 2019;Vom Brocke et al., 2021;Wessel et al., 2021).We provide initial findings that assumptions made for larger firms may not always hold for smaller firms and inform future theory development for digital transformation about the key role of organisational characteristics.

SME digitalisation
There is an increasing interest in the IS field in the implications of digital phenomena like IT-enabled change, data-driven business models, digitalisation, and digital transformation, 1 as exemplified by the multitude of recently published works (e.g., Hanelt et al., 2021;Verhoef et al., 2021;Vial, 2019).Such digital phenomena (e.g., Chanias et al., 2019;Gupta & Bose, 2022;Li et al., 2018) often overlap and require researchers to clearly define their digital construct (Verhoef et al., 2021).In this study, we focus on SME digitalisation 2 which focuses on business model reinforcement, modification, and renewal with the help of digital technologies by SMEs (Baiyere et al., 2020;Soluk & Kammerlander, 2021;Volberda et al., 2018).Current contentions in the digital transformation literature are almost exclusively based on the investigation of larger incumbent companies (Hanelt et al., 2021;Vial, 2019;Wessel et al., 2021) and often take a limited binary view (transformed vs. not transformed) which limits our understanding of SME digitalisation.First, popularised by its high potential, many scholars and practitioners assume that digital transformation is imperative for all firms (Baiyere et al., 2020;Vial, 2019;Vom Brocke et al., 2021;Wessel et al., 2021), thereby promoting a "the more, the better" mentality that recommends firms to increase their digitalisation efforts and adopt a variety of technologies to improve their business performance.However, as explained by Eller et al. (2020, p. 120): "the overwhelming majority of resource-constrained SMEs are not equipped for this level of complexity".The lack of specialised IT personnel (Bouncken & Barwinski, 2020;Wee & Chua, 2013), the integration of large quantities of digital technologies (Eller et al., 2020) as well as the complexity of realising deep changes to organisational identity which are required for digital transformation (Wessel et al., 2021), are overwhelming for SMEs.Second, while digital transformation forms a process that unfolds over time (e.g., Verhoef et al., 2021;Vial, 2019), several studies take on a limited perspective by assuming that firms are either digitally transformed or not (Gao et al., 2022;Libert et al., 2016).Thereby ignoring the potential gains from undergoing changes by altering "some" elements of the underlying valuecreating structures as often occurs in the SME context.Hence, the current theory on digital transformation falls short of accounting for the context of smaller firms and regarding the measurement of SME digitalisation.Given these issues, it is important to integrate current research on business model digitalisation in the SME literature (Chan et al., 2019;Li et al., 2018;Proksch et al., 2021;Soluk & Kammerlander, 2021;Soluk et al., 2021).
Based on current definitions of digital transformation (Hanelt et al., 2021;Verhoef et al., 2021;Vial, 2019;Wessel et al., 2021), we define SME digitalisation as the degree of change to the valuecreating structures of an SME rooted in the utilisation of digital technologies.SMEs can realise high levels of SME digitalisation when they use digital technology to strongly alter multiple value-creating structures underlying their business model (Vial, 2019), including changes to business operations, relationship management, and strategic partnerships.For SMEs, managing digital technologies to reinforce and upgrade existing or develop new business models has been considered a strategic imperative for future success (Benitez, Llorens, et al., 2018;Cannas, 2021;Soluk et al., 2021).SMEs differ significantly vis-à-vis larger organisations, as their smallness and flexibility may speed up the uptake of new technologies improving the rate at which they can innovate (Bouncken & Barwinski, 2020).At the extreme end of SME digitalisation, SMEs use digital technologies to fundamentally alter and renew their current business model by creating new value-creating structures (Vial, 2019;Wessel et al., 2021) that ultimately lead to new ways of capturing value; hence, extremely highly levels of SME digitalisation likely change the business logic of the firm and how it earns its money (e.g., Zott et al., 2011).At lower levels of digitalisation, SMEs may use new digital technologies to replicate successful or further improve existing parts of their business model by, for instance, realising efficiency gains or gaining more control over business operations without altering the business logic of the firm.
The potential business value of SME digitalisation derives from the degree to which SMEs can leverage and enhance their value-creating structures through the use of digital technologies by altering the firm's value propositions (e.g., shifting from products to services by using data), value networks (e.g., becoming a platform), and digital channels (e.g., enhancing distribution, sales or logistics through digital technology).Ultimately, SME digitalisation alters the value creation and delivery to customers and the conversion of payments received into profits (Teece, 2010).While recent studies have identified the antecedents of SME digitalisation (Agarwal et al., 2010;Chanias et al., 2019;Gupta & Bose, 2022;Li et al., 2018;Matarazzo et al., 2021;Proksch et al., 2021), much less is known about its performance implications, and in particular, what factors explain why some firms are successful whereas others are not (see Table 1 for an overview).A higher degree of SME digitalisation does not uniformly guarantee performance gains for all SMEs as rigidity (Beliaeva et al., 2019;Chan et al., 2019;Vial, 2019) and radical orientation (Bala & Venkatesh, 2013;Proksch et al., 2021;Soluk et al., 2021;Vial, 2019) may act as contextspecific inhibitors moderating the effect of SME digitalisation.Next, we explain how the two factors may hamper (or foster) the degree to which SME digitalisation results in performance improvement.

Enablers and inhibitors to benefit from SME digitalisation
We draw on two contingency factors that reflect the organisation's culture and organisational barriers to implementing digitalisation and digital transformation (Vial, 2019): radical orientation (Bala & Venkatesh, 2013;Luo et al., 2012;Osiyevskyy & Dewald, 2015;Proksch et al., 2021;Soluk et al., 2021) and organisational rigidity (Bala & Venkatesh, 2013;Chan et al., 2019).These two contingency factors are key to understanding the business value of digitalisation in the SME context, as they explain the conditions under which SME digitalisation converts into performance improvement.
SME digitalisation may require a company culture that is open and receptive to substantially new ideas because higher forms of SME digitalisation introduce greater change and require that employees think in novel ways that differ from existing practices (Vial, 2019).In an SME context, radical orientation plays an important role in determining whether SMEs benefit from digitalisation, as there needs to be a willingness to pursue the risky endeavour of SME digitalisation.The smaller size of SMEs and lack of slack resources make higher levels of SME digitalisation particularly risky.Consequently, SMEs with a higher radical orientation may feel a greater urgency to transform fast to update (without much deliberation) their business models in response to the advent of fast-changing digital technologies and increased competition from digital entrants (Soluk et al., 2021).Yet, the culture towards introducing substantially different ideas to the business may skip essential organisational learning that is crucial to SME's successful handling of digital technologies (Bala & Venkatesh, 2013;Matarazzo et al., 2021;Proksch et al., 2021), such that SMEs with a strong radical orientation may miss out on the advantages offered by more incremental, agile approaches that facilitate organisational learning.In their rush to the market, these SMEs may implement SME digitalisation too quickly and with inferior quality, which ultimately harms their performance.Moreover, given the limited resources of SMEs, a strong radical orientation may easily require significant resources and shift attention away from their traditional business, which may reduce their performance in their existing business.
However, their smaller size may limit the development into new areas and promote rigidity by hindering the widespread adoption and use of digital technologies (Cenamor et al., 2019).Despite their nimbleness, overcoming rigidity remains a key factor for SMEs to benefit from the new opportunities presented by digital technologies (Chan et al., 2019;Eller et al., 2020).A lack of organisational rigidity helps to create business value from digitalisation (e.g., Benitez, Chen, et al., 2018;Benitez, Llorens, et al., 2018) because it facilitates the effective transfer of ideas into digital prototypes and innovations.

Hypotheses development
Figure 1 shows our research model that proposes that SME digitalisation is positively associated with performance improvement (H1) and that this positive relationship is attenuated by radical orientation (H2) and organisational rigidity (H3).Below, we explain the associated hypotheses.

SME digitalisation and performance improvement
Digital technologies enable changes in value-creating structures that can alter or redefine the business model in various ways (Vial, 2019).First, firms can change the value propositions with the help of digital technologies, as shown by Netflix, which switched from renting out physical videos to streaming and leveraging big data to provide personalised recommendations to customers (e.g., Dremel et al., 2020;Gomez-Uribe & Hunt, 2015).Second, firms can change their value networks in response to digital technologies, resulting in the use of digital platforms between different types of stakeholders to co-create value (e.g., De Reuver et al., 2018).Third, firms can change their use of digital channels by introducing new social media, e-commerce, and m-commerce channels to better interact with and serve customers (e.g., Suseno et al., 2018).In all of these instances, digital technologies enable changes in value-creating structures underlying business models.
SME digitalisation allows firms to not only create value but also appropriate greater financial returns.Digital technologies can help firms translate changes in value-creating structures into increased revenues and subsequent profits.Digital business models are more profitable because they require less labour, are less dependent on physical infrastructure, have greater scalability, and lower distribution costs (Constandinides et al., 2018;De Reuver et al., 2018;Verhoef et al., 2021).SME digitalisation may also stimulate revenues and profits by entering new markets and serving new customer segments (Wang et al., 2018).Empirically, we expect to observe that higher levels of SME digitalisation, on average, are positively associated with performance improvement.Hence, we propose the following hypothesis: H1: SME digitalisation has a positive relationship with performance improvement.

The moderating role of radical orientation
While we predict that SME digitalisation, on average, leads to performance improvement, we also predict that this impact varies according to the firm's radical orientation.An SME's radical orientation is an important cultural inhibitor (Proksch et al., 2021;Vial, 2019).We define radical orientation as the degree to which a firm's employees are oriented towards substantive change by having ideas that differ substantially from existing practices and alternatives (Bala & Venkatesh, 2013;Luo et al., 2012).Research on digital transformation (Osiyevskyy & Dewald, 2015;Proksch et al., 2021;Soluk et al., 2021) suggests that the urgency created by digital disruption promotes firms to develop radical approaches to change, but also that such radical orientations may inhibit organisational learning.
Organisational learning is critical for SMEs to adopt new technologies (Matarazzo et al., 2021).Due to perceived urgency and the need for new digital technologies, SMEs may be pressured into taking rapid action (Soluk et al., 2021).Thereby, the dynamism promoted by digital technologies combined with urgency for change and SME culture (Osiyevskyy & Dewald, 2015;Proksch et al., 2021) may push firms towards more substantive and hastily approaches of change.However, tensions resulting from changes to the value-creating structures (Vial, 2019) may create dysfunctional dynamics for SME firms that lower firm performance.To effectively address these tensions during transformation, recent studies point towards the necessity of a stepby-step approach to change characterised by experimentation and incremental development of new knowledge through organisational learning from success or failure in every step (e.g., Chanias et al., 2019;Verhoef et al., 2021;Visnjic et al., 2022).Hence, a firm's radical orientation towards change may play a key role in creating tensions stemming from digitalisation, which can induce high "dynamic adjustment costs" as hindrances to successful business model innovation (Karhade & Dong, 2021a).
To digitalise the business model in a valueincreasing manner, organisational learning from related experiences with digital technologies is critical (Chanias et al., 2019;Estrada & Dong, 2020).To reduce technological and market uncertainty, firms need to rely on iterative experiments (Chanias et al., 2019;Karhade & Dong, 2021a).These experiments can take different forms, such as the formation of separate units outside the existing organisation (Sia et al., 2016), but also rely on newly formed teams operating within organisational structures and that connect formerly unconnected departments and business units (Dremel et al., 2017).Such experiments minimise risk and guarantee the accumulation of new (incremental) knowledge that can be used to further improve the firm's business model (Yeow et al., 2018).
SMEs with a strong radical orientation have employees who think in substantially different ways that deviate from existing practices and that challenge the status quo (Bala & Venkatesh, 2013;Proksch et al., 2021).Such a culture does not align well with an iterative, step-by-step procedure for making incremental changes but rather fosters a "waterfall" approach that combines heavy upfront project design with limited changes and performance feedback during and between project stages (Dong, 2021;Mahadevan et al., 2015).A radical approach to change tends to interfere with existing business practices and cause ambiguity and confusion to employees that may subsequently harm performance (Bala & Venkatesh, 2013;Loebbecke & Picot, 2015;Sebastian et al., 2017;Wang et al., 2018).SMEs with a strong radical innovation orientation are less fit to learn from assessing market needs (e.g., based on learning from product launches) and adjust their business model accordingly to attain performance gains (Christensen et al., 2016).Given their limited experimentation and organisational learning, we hypothesise that SMEs with a strong radical orientation approach benefit less from increased digitalisation efforts because they are less effective in aligning and integrating digital technologies when updating their business models via desirable means of value creation and appropriation.Hence, we expect that the positive relationship between SME digitalisation and performance improvement will be weaker for firms with a stronger radical orientation.

The moderating role of organisational rigidity
Organisational rigidity refers to the degree to which a firm is unable to implement innovations (Bala & Venkatesh, 2013;Chan et al., 2019;Lucas & Goh, 2009;Vial, 2019) and reflects a firm's inertia to change resulting from path dependence.For incumbent firms that rely on long-established organisational routines and processes, rigidity can make it extremely hard to explore new ways of value creation and appropriation (Gilbert, 2005).Employees may resist digital technologies and decide not to engage in changes or alter their ways of working (Lapointe & Rivard, 2005).We argue that firms with greater rigidity against changes in innovation activity may be disadvantaged at higher levels of SME digitalisation that require profound changes to the value-creating mechanisms underlying SMEs' business models.
High levels of SME digitalisation involve deep changes to the value-creating structures underlying business models (Verhoef et al., 2021;Vial, 2019), in which organisational learning plays an integral part as it facilitates the use of new knowledge and innovative ideas (Abraham & Junglas, 2011;McKeown & Philip, 2003).As the transformation of organisational routines is fundamental to digital business models (Chan et al., 2019;Karhade & Dong, 2021b), firms must remain open to new ideas and innovation (e.g., Lokuge et al., 2019;Quinton et al., 2018).Subsequently, organisational rigidity represents the barriers to change that inhibit SME digitalisation due to the difficulty of accommodating and translating new ideas into concrete innovation (Chan et al., 2019).It leads to an absence of openness that subsequently leads to firms being stuck in established routines rather than developing new ones (e.g., Abraham & Junglas, 2011;Chan et al., 2019;Lokuge et al., 2019).
Rigid SMEs are less likely to question their beliefs and procedures related to business models (Quinton et al., 2018) and are less likely to learn from and incorporate new knowledge and update their routines (Alavi et al., 2005;Karimi & Walter, 2015).Conversely, less rigid SMEs are more likely to challenge the established business models, enabling the introduction of more effective and efficient practices (Day & Schoemaker, 2006;Quinton et al., 2018).Organisational rigidity can inhibit SME digitalisation because it harms the adoption of new digital technologies, inhibits knowledge sharing, and makes subsequent business model innovation difficult (Teece, 2018).It undermines the firm's ability to adapt to changes and implement new routines needed for innovation.In firms characterised by high organisational rigidity, employees tend to be less willing to put effort into the new digital venture and rather stick with the old business model (Karimi & Walter, 2015).Thus, organisational rigidity enhances path dependency and makes business model innovation more difficult and costly, thereby lowering the performance impact of SME digitalisation (Gilbert, 2005).Hence, we hypothesise that the relationship between SME digitalisation and performance improvement will be weaker for firms with greater organisational rigidity: H3: Organisational rigidity weakens the positive relationship between SME digitalisation and performance improvement.

Data collection
To test our hypotheses, we relied on two waves of survey data collected in March 2019 and March 2020.The dataset of 2019 was used to test our main hypotheses, and the dataset of 2020 helped to verify our findings.We had access to and could participate in a larger annual survey project that studies the business and innovation activities of SMEs in the Netherlands.The first survey sampled about 5000 Dutch firms with a firm size below 500 employees and had an average response rate of 9%, covering a variety of industries ranging from manufacturing, construction, IT, and retail sectors to public/nonprofit organisations.The lion's share of respondents (90.7%) was either an owner (26.1%), an executive manager (51.5%), or both (13.1%). 3 While survey research has inherent methodological limitations, our current setup provides two practical advantages.First, the decision power tends to be rather concentrated in SMEs, which makes it relatively easy for key respondents, who play a major role in defining the firm's digitalisation efforts, to track the degree of digitalisation and adequately assess its performance effects.In contrast to large multinational corporations, where digitalisation activities might be heavily dispersed among multiple units and across geographical locations, the digitalisation efforts of SMEs tend to be more concentrated and localised, which makes it easier to capture possible digitalisation changes using survey-based approaches.
Second, by being part of a larger annual survey project, we reduced the possible self-selection bias because we did not address the purpose of SME digitalisation upfront but asked SMEs about their business and innovation activities.To check for the potential non-response bias, we compared the early and late respondents (median split) in the 4-week period of data collection, but no statistically significant differences were found in firm size (t = −0.201,p = 0.841) and firm age (t = −1.516,p = 0.131).We also did not find any significant mean differences between early and late respondents for our main construct of SME digitalisation (t = 1.305, p = 0.192), radical orientation (t = 0.5484, p = 0.584), rigidity (t = −0.657,p = 0.512), and performance improvement (t = −0.916,p = 0.360).
The first wave of survey in 2019 that we used for our main analysis yielded a total of 462 responses.After list-wise deletion, we obtained a net sample of 175 observations.The sampled SMEs had, on average, 37 employees and were, on average, 3 years old.In the second wave of survey in 2020, we again collected data for the dependent variable (i.e., performance improvement) and obtained a total of 594 responses.By matching the two waves of survey based on firm identifiers, we obtained 101 firms participating in both surveys that were used for robustness checks.
To assess the sample's representativeness, we analysed our sample's industry distribution using selfreported sector codes and the official industry classification as used by the Dutch Central Bureau of Statistics (CBS).To make the industry classification consistent across these two independent sources, we grouped the firms into four sectors, including 1) manufacturing, 2) retail and wholesale, 3) services and 4) public/nonprofit.Among the initial sample, only 271 SMEs reported about their industry.Tables 2-3 show that our sample corresponds to the actual Dutch SME population in terms of age.Our sample contains fewer selfemployed firms (firms consisting of 1 person) than the population.Hence, our sample corresponds more strongly to the economic impact of SMEs.Table 4 shows that our sample largely matches the Dutch SME population in terms of industry, yet the retail and wholesale are somewhat under-represented.Given our focus on assessing the strength of associations instead of attaining population estimates, we consider the sample fitting for the purpose of our study.

Independent variable
Given the lack of validated scales for SME digitalisation, we followed recommended procedures for scale development and validation (MacKenzie et al., 2011).
In the first phase, we generated a pool of measurement items (Rossiter, 2002) based on a literature review and interviews with business practitioners and experts.We conducted semi-structured interviews with managers or owners of 11 SME firms.To identify key informants, we interviewed either the owner (for smaller firms) or the person responsible for digital technologies (for medium-sized firms).Our interview was guided by the main question of what digitalisation meant for their business.After conducting the interviews and scouting the literature (e.g., Parida et al., 2019;Verhoef et al., 2021;Vial, 2019), we developed an initial list of items covering the elements of creating novel offering configurations enabled by digital technology, understanding customer needs concerning digital solutions, creating value through collaboration, adjusting operational processes, and revised roles and responsibilities in the value system.In the second phase, we invited academics who are knowledgeable about digitalisation and digital transformation to improve the wording and consistency of our items (interrater reliability of 0.72).In the third phase, we conducted a pilot test for SME digitalisation to assess the reliability and face validity of our items.The values for composite reliability and AVE were satisfactory and demonstrated predictive validity as it was positively related to firm performance.In the final phase, we selected six items to measure SME digitalisation (see Table 5).All items were measured on a 5-point scale.We took the mean of these six items to calculate a composite score where a higher score indicates a higher degree of SME digitalisation. 4

Dependent variable
In line with Wang et al. (2012), we measured performance improvement based on two items asking about the additional revenue streams and efficiency gains realised with the help of digital technologies.These items measure both innovative (Svahn et al., 2017;  Vial, 2019) and financial (Karimi & Walter, 2015) dimensions of performance.By exclusively measuring the performance improvement with the help of digital technologies, we can alleviate the performance impact of omitted variables that are not relevant to digitalisation.For both items, we used a 5-point scale.We decided to use a perceptual measure of performance because it is oftentimes very challenging to collect objective and secondary data for SME firms.Therefore, a subjective performance measure is an appropriate measure for the research context. 5We also treat performance improvement as a composite construct because an improvement in revenues may not necessarily correspond with realising greater operational benefits (Jarvis et al., 2003).We calculate a composite score based on the means of the two items.Higher scores correspond to greater performance improvement.

Moderators
We measured radical orientation based on three items to measure the firm's employee's orientation towards having ideas that differ substantially from existing practices.The items of Bala and Venkatesh (2013), which refer to radical innovation, were adjusted to fit the new context.The items were measured on a 5-point scale, in which higher scores indicate a stronger radical orientation towards change.We measured organisational rigidity based on two items regarding the difficulty of using innovative ideas and translating them into concrete solutions.The questions were also adapted from Bala and Venkatesh (2013).The items were measured on a 4-point scale, in which higher scores correspond to greater organisational rigidity against change.We again, calculate a composite score for both variables.

Control variables
We controlled several factors that may confound with SME digitalisation and affect performance improvement.First, we controlled for a firm's digital capabilities as it may directly influence SME digitalisation.Digital capabilities have been widely studied in the literature on the business value of IT, and we controlled for it by using six items measuring a firm's capabilities to 1) utilise digital technologies, 2) analyse data and use insights gained, 3) find the best digital technologies, 4) integrate digital technologies, 5) change business processes, and 6) manage data ethically, again we use a composite score to measure the variable.Second, we used the number of employees to control for firm size (e.g., Rai et al., 2006), and took the natural logarithm to reduce the skewness of this variable.Third, we controlled firm age by the number of years of existence (e.g., Karimi & Walter, 2015), and again took the natural logarithm to normalise this variable.Fourth, the financial performance of a firm may drive its investment in and use of digital technologies (Anand et al., 2020;Dong et al., 2021).As there is a lack of archival performance data for Dutch SMEs, we controlled for SMEs' selfreported profit margin.Finally, we controlled industryfixed effects by including three dummies.

Assessment of measurement properties
We used partial least squares structural equation modelling (PLS-SEM) to assess the measurement properties of our five latent constructs (the independent and dependent variables, moderators, and a control variable digital capabilities) based on the final sample (N = 175).Following Benitez et al. (2020), we first evaluate the overall fit of the saturated model to assess the validity of the measurement and the composite models.Our SRMR value of 0.058 supports adequate model fit as its value is below the .08threshold.Furthermore, the discrepancy measures d ULS and d G are below the 95% quantile of their reference distribution (HI 95 ), finding empirical support for our incorporated latent independent and dependent variables in the model.
To assess the validity of the constructs, we find that all our constructs demonstrate sufficient convergent validity (see Table 5), as all item loadings are high and statistically significant, and latent variables have an average variance extracted (AVE) above 0.5.Despite the high expected correlations between the latent construct of SME digitalisation with digital capabilities (r = 0.694) and between digital capabilities and digital performance (r = 0.716), we found evidence for sufficient discriminant validity as none of the correlations of our predictors (±2 standard errors) include unity (Bagozzi & Yi, 1988).The Heterotrait-Monotrait (HTMT) test revealed no apparent discriminant validity issues among our independent variables (highest ratio between SME digitalisation and digital capabilities of 0.782).Yet, this test showed that our independent variable of digital capabilities and the dependent variable of digital performance has a value of 0.928, which is higher than the threshold of 0.9, which can pint to discriminant problems (Henseler et al., 2015).To reduce collinearity and reduce overlap between the constructs, we orthogonalised SME digitalisation and digital capabilities, which is a linear transformation widely used to reduce high correlation without biasing hypothesis testing (e.g., Sine et al., 2006).After orthogonalisation, we found that the square root of AVE exceeds the correlations with all other constructs, suggesting sufficient discriminant validity (Fornell & Larcker, 1981).Finally, we used composite scores for further analysis.

Assessment of common method bias
We assess common method bias for bivariate relationships 6 in multiple ways.First, to safeguard against common method bias, we kept the questionnaire relatively short to avoid that respondents shift from response accuracy to response speed (Podsakoff et al., 2003) and used mixed scales (e.g., 5-point, 4-point, and binary, etc.) to limit acquiescence effects (Lindell & Whitney, 2001).Table 6 suggests that common method bias is not substantial, as we observe insignificant correlations (Podsakoff et al., 2003).
Second, we formally assess the potential common method bias by Harman's single factor test (Malhotra & Kim, 2006).For Harman's single-factor test, using a principal component factor analysis, we found that the first factor explains 39% of the variance, which is below the common threshold of 50%.Furthermore, using covariance-based structural equation modelling (CB-SEM), we found that the model demonstrates poor fit (χ 2 /df = 6.212;GFI = 0.645; CFI = 0.641; RMSEA = 0.159).
Third, we followed Lindell and Whitney (2001) using the smallest (r = 0.016) and second smallest (r = 0.024) positive correlations as the proxies for common method variance (CMV).We found that all partial correlations remained statistically significant.Furthermore, using CB-SEM, we constructed a measurement model including a common latent factor (Podsakoff et al., 2003).Following the work of Liang et al. (2007), we found that none of the 19 estimated paths from the common latent factor to each indicator was statistically significant, while all 19 estimated paths to their intended construct remained statistically significant.The five latent constructs (59.15%) explained much more variance than the variance explained by the common latent factor (2.60%).In a subsequent step, we also assessed a structural model to assess whether the effect of SME digitalisation on performance improvement holds while adjusting for CMV.The structural model showed that the effect of SME digitalisation reduced slightly but remained statistically significant (β = 0.418, p < 0.001 vs. β = 0.312, p < 0.05).
Finally, we collected additional data from independent sources to cross-validate our self-reported survey data.For firm age and size, we were able to collect archival data for a small sample (around 10%) based on filings from the Dutch Chamber of Commerce.We found statistically significant and positive correlations between the two independent sources for firm age (r = 0.675, p < 0.001) and firm size (r = 0.820, p < 0.001), respectively, corroborating the accuracy of our data in the lack of common method bias.Taken together, all the evidence jointly suggests that common method bias is not a serious issue for our data.

Hypothesis testing
We tested our hypotheses based on ordinary least squares (OLS) regression analysis due to the small sample.To avoid multicollinearity, we mean-centred SME digitalisation, radical orientation, and organisational rigidity and then calculated their interaction terms.We checked the variance inflation factor (VIF) and found that multicollinearity is not a major concern (mean VIF = 1.560, maximum VIF = 2.530).
Table 7 reports the OLS regression results for hypothesis testing.As suggested by H1, we found a significant and positive relationship between SME digitalisation and performance improvement (β = 0.686, p < 0.001, see Model 2) in support of H1.To test the moderating effects as proposed in H2 and H3, we estimated a full model with moderators and their interaction terms to predict SME digitalisation (see Model 3).SME digitalisation remained to have a positive relationship with performance improvement (β = 1.422, p < 0.001).We found a statistically significant and negative interaction effect between SME digitalisation and radical orientation on performance improvement (β = −0.142,p < 0.01), in support of H2.The interaction term between SME digitalisation and organisational rigidity is statistically significant and negative (β = −0.137,p < 0.05), in support of H3.
To facilitate the interpretation of the interaction effects, we plot them and perform a simply slope test (see Figures 2-3).We found that the positive relationship between SME digitalisation on performance improvement is less steep at a higher level of radical orientation (high radical orientation: β = 1.280, p < 0.001; low radical orientation: β = 1.564, p < 0.001).We also found that the positive relationship between SME digitalisation and performance improvement is less steep at a higher level of organisational rigidity (high organisational rigidity: β = 1.285, p < 0.001; low organisational rigidity: β = 1.559, p < 0.001).Thus, the interaction plots show patterns in support of H2 and H3.

Alternative dependent variable (archival data)
In addition to the survey-based measure, we collected archival data from the Dutch Chamber of Commerce.
As SMEs in the Netherlands (less than 50 employees and less than €12 million in revenue per year) are not required to publish financial information (e.g., income statement or cash flow), we faced difficulty in collecting data on profitability (only 13 firms provided an income statement).Nonetheless, we obtained one solvency ratio from the balance sheet, namely, "equity/ total assets" in the following year 2020 for a smaller sample of 106 firms.The equity-to-asset ratio is commonly utilised to predict firm failure (Beaver, 1966), in which higher values represent that a firm has a greater share of assets relative to its liabilities.We consider this a suitable intermediate performance indicator to assess our research model.The results reported in Table 8, confirmed a positive relationship between SME digitalisation and our alternative dependent variable (β = 1.454, p < 0.01), as well as the negative moderating effects of radical orientation (β = −0.299,p < 0.05), and of organisational rigidity (β = −0.290,p < 0.05).Therefore,  our findings also hold for using an alternative dependent variable from archival data.

Alternative dependent variable (survey data)
As a second robustness check, we used an alternative measure for performance improvement to examine the robustness of our findings.Following recent IS literature, we examined the performance impact of SME digitalisation in terms of innovation performance (Karhade & Dong, 2021a;Trantopoulos et al., 2017) instead of financial performance.We measured innovation performance improvement alternatively based on three questions: (1) improvement in processes of goods and services, (2) improvement in logistics, and (3) improvement in distribution.The results reported in Appendix A, Table A1, confirmed the positive relationship between SME digitalisation and our alternative measure of performance improvement (β = 0.143, p < 0.001), and the negative moderating effects of radical orientation (β = −0.047,p < 0.05), and of organisational rigidity (β = −0.046,p = 0.090) that is marginally significant.Therefore, our findings also hold when using an alternative measure for performance improvement.

Time lag
To overcome the cross-sectional nature of single surveys, and in addition to the archival data, we have collected.We used data from the subsequent wave of the survey in 2020 and assessed the lagged effect of SME digitalisation, using the same performance improvement measured in 2020.The time lag also  helped to rule out reverse causality.Due to the relatively small matched sample with only 101 firms across two waves of the survey, and after removing missing values, we were left with a rather small sample of 51 observations.While we did find a positive and significant effect for the direct relationship for performance improvement in 2020 (β = 0.657, p < 0.01) as well as innovation performance in 2020 (β = 0.127, p < 0.05), we did not find support for the moderation effect.The small sample size may explain the inability to detect the more complex moderation relationship.
The results are reported in Appendix A, Table A2.

Alternative model specification
We assessed the robustness of our findings using an alternative specification of our model.As research highlighted the possibilities of nonlinear relationships due to curvilinearity (Haans et al., 2016;Kohtamäki et al., 2020) or synergies between related concepts such as SME digitalisation and digital capabilities (Nambisan et al., 2019;Witschel et al., 2019), we tested for these possibilities but neither find evidence for nonlinearity (the quadratic term of SME digitalisation is not statistically significant: β = −0.035,p = 0.461) nor for the interaction effect between SME digitalisation and capabilities (the interaction term is not statistically significant: β = −0.186,p = 0.731).Furthermore, to ensure that the insertion of digital capabilities, which is highly correlated with SME digitalisation, did not distort our findings, we tested a model without digital capabilities.The results reported in Appendix A, Table A3, were qualitatively the same: the effect of SME digitalisation on performance improvement (β = 0.645, p < 0.001) remained statistically significant and positive, and negative interaction effects are found for radical orientation (β = −0.183,p < 0.01) and organisational rigidity (β = −0.142,p = 0.055).Hence, our findings were not distorted by the inclusion or exclusion of digital capabilities in the model.

Endogeneity
In the presence of endogeneity (caused by simultaneity), OLS can produce biased and inconsistent parameter estimates.SMEs with better performance may have more financial resources available for digitalisation, which could question the causality between SME digitalisation and performance improvement.Omitted variables may also simultaneously affect SME digitalisation and performance improvement, leading to a spurious correlation.To further examine the endogeneity concern, we used the two-stage Heckman model.Following prior studies (e.g., Bharadwaj et al., 2007;Dong et al., 2017), we took the median of SME digitalisation in our sample and created a dummy variable indicating 1 if the SME's digitalisation score is greater than the median, and 0 otherwise.In the first stage, we estimated a probit model regressing this new dummy variable on an exclusive restriction as a digital specialist, a binary variable indicating if the SME had a digital specialist who is responsible for digital strategy related to data, technology, and digitalisation (0 = no, 1 = yes), and our moderators and control variables.Prior literature suggests that digital specialist plays a key role in leading digitalisation (Eden et al., 2019;Hess et al., 2016).The probit model demonstrated a good fit (see Table A4, Appendix A).The inverse Mills ratio (IMR), which represents the propensity of SME digitalisation being endogenously determined, was not statistically significant (β = −0.351,p = 0.384), which suggested the absence of endogeneity.After controlling for the IMR, the main and moderating effects remained consistent with our OLS results.

Theoretical implications and contributions
First, this study contributes to existing research on the business value of SME digitalisation by extending our knowledge of the degree to which SMEs derive value from using digital technologies for reinforcing, modifying, and renewing their business model (Baiyere et al., 2020;Soluk & Kammerlander, 2021;Volberda et al., 2018).In response to calls, we integrate knowledge of the context-specific inhibitors for SME digitalisation (Chan et al., 2019;Hong et al., 2014;Proksch et al., 2021).While our study provides important insights into the, on average, positive business value of digitalisation for SMEs, we find that heterogeneity exists in the performance outcomes and that the business value of SME digitalisation can be severely hindered by an SME's radical orientation and organisational rigidity.By demonstrating the moderating effect of the contingency factors, we contribute to our understanding of the organisational factors that influence the business value of IT for SME firms (Benitez, Castillo, et al., 2018;Benitez, Llorens, et al., 2018;Trang et al., 2021).
While more pervasive forms of SME digitalisation like digital transformation are connected to radical changes in the fabric and identity of the organisation (Wessel et al., 2021), we find that having an organisation culture that is supportive of such radical ideas (Bala & Venkatesh, 2013;Proksch et al., 2021;Vial, 2019) limits the potential business value of SME digitalisation due to lower levels of organisational learning.Our study also has implications for how digital transformation should be managed.The inhibiting role of radical orientation may seem counterintuitive, given that substantive changes are required for the organisation.Yet, the finding makes sense given the importance of organisational learning and the emergent nature of digital transformation in which firms learn by running small and incremental experiments that help to test the market assumptions that firms have (McGrath, 2010;Orlikowski, 1996).The dynamic nature of digitalisation, paired with the perceived urgency to change (Soluk et al., 2021), may pressure SMEs with high radical orientation to rush digitalisation.This radical orientation may contribute to undesirable outcomes of SME digitalisation because it foregoes step-by-step experimentation and learning (Karhade & Dong, 2021a).
Furthermore, we find that firms that are rigid in implementing ideas into innovation also benefit less from increased digitalisation efforts.The complexity of changes to the value-creating structures introduced by digital technologies can often be overwhelming (Eller et al., 2020).Due to their size, lack of resources and competencies, and the concomitant survival risks of implementing digital transformation, SMEs may suffer from inertia that hampers the successful deployment of SME digitalisation (Cenamor et al., 2019;Chan et al., 2019).Our findings corroborate that as we find that SMEs face difficulties in overcoming established routines in favour of new ones, which in turn decreases the business value of digitalisation.Moreover, we emphasise the findings of earlier studies that stress the importance of organisational flexibility, an important factor in creating business value from digitalisation (e.g., Benitez, Chen, et al., 2018;Benitez, Llorens, et al., 2018).
Second, from an empirical point of view, we expand existing methods and evidence in the literature on the business value of SME digitalisation by combining data from a multi-year survey and financial data to increase the generalisability of the findings.While prior research focused either on larger SMEs (see Table 1) and excluded firms under 50 employees (e.g., Soluk et al., 2021), or new ventures (Proksch et al., 2021), less is known about the digitalisation efforts within incumbent micro-and small-sized enterprises.With these smaller SMEs representing a significant share of the overall SME firms (e.g., 90% of SMEs within the Netherlands, according to the Dutch central Bureau of Statistics), our study contributes important insights into the role of digitalisation for smaller incumbent SMEs.Moreover, due to the unique characteristics of (smaller) SMEs, most studies resorted to qualitative methods and thus lacked quantitative insights into performance (see Table 1).Our combination of multi-year survey and objective financial data may help to overcome some of the inherent weaknesses of cross-sectional survey studies and structurally assess the contingencies for a so-far neglected set of SMEs.
Third, our study also contributes to the literature on digital transformation.While current inquiries into digital transformation focus on larger legacy firms, our study shows that the findings of these studies may not hold in an SME context and that additional research is wanted to understand the business value of SME digitalisation.Furthermore, while existing studies treat digital transformation as a binary outcome, our study uses a variance logic that allows for a more finegrained understanding of changing the valuecreating structures to some degree (i.e., SME digitalisation).By incorporating two important contingency factors, we show that SME digitalisation does not pay off for all SMEs and warn against the transformation imperative of "the more the better" (Baiyere et al., 2020;Vial, 2019;Vom Brocke et al., 2021;Wessel et al., 2021).We infer that organisational learningderived from test-and-learn iterations with incremental improvements -plays a crucial factor for SMEs in explaining how those firms develop competences to handle digital technologies (Matarazzo et al., 2021;Proksch et al., 2021;Soluk et al., 2021).Thereby, we shed light on the importance of organisational factors.In sum, our study provides attention to current theoretical blind spots in the broader digital transformation discourse and helps to increase our understanding of when SMEs can derive business value from their digitalisation efforts.

Managerial implications
Our study offers SME managers important guidance when being confronted with the need to digitalise and transform their ventures.We show evidence that SME digitalisation is beneficial for SME firms and while risky, it is well worth the effort.Thus, SME firms should also aim to transform digitally, as it increases their business performance.In an increasingly digital world, abstaining from investments in digital technologies and the transformation of business processes is not a wise strategy for SMEs.This challenge is becoming even more important as large enterprises invest heavily in digital transformation, and create hypercompetitive environments (Troise et al., 2022).To respond to this challenge, SME managers should look beyond the binary classification of digitally transformed versus not transformed.Digital change is about the degree of change and thereby covers a wider range of possibilities.SME managers can gradually shift from being less digital to more digital, with investing in specific technologies also depending on their starting situation.While such a task may seem daunting to SMEs, not every SME has to become the next Amazon, Uber, or Netflix.Our study shows evidence that SMEs can improve their performance by moving through different degrees of SME digitalisation.
Our study also stresses the importance of organisational characteristics and how they function as important contingencies for the business value of SME digitalisation.Managers should, for instance, be aware of the necessity of organisational learning before reaping the benefits of digitalisation.As radical orientation inhibits the performance impact of digitalisation, managers are recommended to treat digitalisation, not as an overnight revolution in which they sprint to their strategic goals.SMEs with a strong radical orientation towards change should be wary of their lower ability to realise performance improvement from their digitalisation efforts.A recommended approach would be to develop an organisational culture focused on experimentation.Through the use of small-scale experiments, and installing mechanisms to learn from each step instead of making drastic changes, SMEs can learn over time from the implementation of these changes.Moreover, the financial and organisational investments in such smaller changes reduce the risk of digital transformation initiatives (McGrath, 2010).
Our study also reminds managers about the crucial inhibiting role of rigidity to change.Organisations with high rigidity face greater difficulties to implement creative ideas, which hinder the business value of digitalisation.Rigid organisations should be cautious and invest in agility to improve their returns on digitalisation when they simultaneously resolve inertia and resistance to change.SMEs can become more agile by investing in strategic planning (building responsiveness to external developments) and in technology and expertise (building infrastructure and attracting digital-savvy staff) (Troise et al., 2022).In so-called VUCA (volatility, uncertainty, complexity, and ambiguity) environments that are characterised by strong dynamics (i.e., technological leaps, the COVID-19 pandemic), this is even more important.
In sum, our study helps managers to develop a business case for engaging in digitalisation!However, it also warns of a strong radical orientation and an organisational rigid organisation when improving digitalisation.

Limitations and future research
Despite the merits, this study also has certain limitations and provides avenues for future research.First, this study uses mainly self-reported survey data from a common source.Although we tried to minimise biases in data collection and improve reliability by using multiple waves of surveys and collecting archival data, some methodological concerns remain.While we do not find that response bias, common method bias, or endogeneity harmed our results, we cannot fully rule out these concerns.Future studies are recommended to collect longitudinal data from multiple sources to overcome issues inherent to our research design.Ideally, we would have had access to secondary data for more profit-or growth-driven performance metrics as well to enhance our empirical measurements.We believe that our design has merits because it allows for fine-grained insights into the nature of digitalisation of smaller SMEs on a sample where secondary data is generally hard to come by.
Second, although digitalisation is purposefully conceptualised and measured in a broad manner to be applicable across industries, our approach may ignore distinct industry contexts and their implications for SME digitalisation.While we controlled for industryfixed effects in our study, we have not specifically focused on whether SME digitalisation and performance should be measured differently and/or has different effects across industries.Thus, future research may refine our approach and further explore SME digitalisation's contingencies by examining the robustness of our findings for specific industries.For instance, in line with this argument, future research can hypothesise and test specific moderators that play different roles across industry contexts.
Third, our sample is limited by its scope on Dutch SMEs.Future research could expand the generalisability to other types of firms (different types of SMEs or larger corporations), and that are active in other countries where digitalisation may have different strategic importance and performance outcomes.Lastly, our study integrates knowledge from both the digitalisation of SME literature and the digital transformation literature.We especially urge future research to better consider the role of firm characteristics when building theory about digital transformation.

Conclusion
By taking an SME perspective, this study conceptualises and operationalises the construct of SME digitalisation.Based on data from multiple waves of surveys and archival data, we shed light on how SME digitalisation impacts firm performance and provide timely insights into the business value of digitalisation.We identify two contingency factors that explain why digitalisation initiatives within SMEs may not always meet strategic goals to improve performance.We hope that our research incentivises other scholars to further explore the idiosyncrasies of digitalisation that help to better understand the business value of digitalisation.

Notes
1.In defining digital transformation, we follow the literature (Hanelt et al., 2021;Verhoef et al., 2021;Vial, 2019;Wessel et al., 2021) and define it as the change caused by the introduction of digital technologies to the value-creating structures and organisational identity of the organisation.2. In this study, we investigate digitalisation at the organisational level (e.g., Baiyere et al., 2020;Morton et al., 2020;Singh et al., 2020;Vial, 2019), while profound digital changes may also occur at the individual level (e.g., Baptista et al., 2020), ecosystem level (e.g., Tan et al., 2020), industry level (e.g., Karimi & Walter, 2015;Lanamäki et al., 2020), or societal level (e.g., Dengler & Matthes, 2018).3. We assessed whether systematic differences exist in the perceptions of SME digitalisation across respondent groups based on functional roles, but independent sample t-tests between the four groups (owners, executive managers, both, and others) show no statistically significant differences (p > 0.05) across the different types of respondents.4. We treat all our constructs as formative/composite because our items determine the latent variable instead of the other way around.For example, for SME digitalisation, altering a component of the business model does not necessarily covary with changing another digital component (Jarvis et al., 2003).We believe that all relevant aspects of SME digitalisation were covered by our items, and that it is unlikely that we misconstrue SME digitalisation construct through the omission of important aspects, because there is a strong overlap between our business model components and those used in other studies (e.g., Parida et al., 2019).For other constructs, we apply a similar reasoning and treat them as formative/composite. 5.In a robustness check, we augmented our survey data with archival financial data for equity-to-asset ratio, which generated consistent results.6.While common method bias can potentially inflate correlations for bivariate relationships, it attenuates the significance of interaction terms and cannot artificially create statistically significant interaction effects (Siemsen et al., 2010).Finding statistically significant interaction effects, "despite the influence of CMV in the data set should be taken as strong evidence that an interaction effect exists." (p.470).

Figure 2 .
Figure 2. Performance impact of SME digitalisation at varying levels of radical orientation.

Figure 3 .
Figure 3. Performance impact of SME digitalisation at varying levels of organisational rigidity.

Table 3 .
Firm size distribution.

Table 6 .
Descriptive statistics and correlations.< 0.001.SME digitalisation and digital capabilities are orthogonalised.The square roots of AVE are on the diagonal based on PLS-SEM.Correlations in bold are the proxies for CMV.

Table 7 .
OLS regression results for hypothesis testing.

Table 8 .
Results for alternative measure of performance improvement (archival data).