As Hulland et al. (2018) point out, surveys remain a relevant method for acquiring new knowledge and insights in social science research. Therefore, this study utilizes survey data to examine the relationships between classic archetypes of organizational culture and digital company culture.

Regarding the sampling procedures, this study employed a non-probability purposive sampling method based on a deductive methodology (Hulland et al., 2018). Non-probabilistic samples, often referred to as targeted or convenience samples, involve a screening process guided by the research's specific characteristics rather than a statistical generalization criterion. Many quantitative and qualitative studies make use of such samples. Given our intention to examine the position and behavior of organizations already undergoing digital transformation, we opted for a convenience sample in this case. Consequently, our target population consisted of managers and staff members working in Spanish firms with over 50 employees and a reasonable level of digitalization. The decision to include not only managers but also other employees is based on the understanding that culture is not something that can be imposed top-down but is rather co-created by all members of the organization. According to the Spanish National Statistics Institute, the target population for this study comprised 2993 companies as of January 2022. Subsequently, in March 2022, we sent a link to an electronic questionnaire via email to 300 firms that met our selection criteria. Following the approach outlined by Baruch and Holtom (2008), cover letters were sent to provide an overview of the study's objectives, data collection methods, and confidentiality regulations. Ultimately, we collected 285 valid responses, resulting in a response rate of 16.6%, from a total of 50 companies.

In this study, data is collected through a gamified self-assessment test embedded in a website belonging to our research group. The data collection process is fully automatic and conducted online. Participants access the web application through a unique URL containing a token that identifies the organization under study in the database. This allows us to gather all the necessary data from the organization for subsequent analysis.

The web application presents several batches of questions, totaling eighteen in number. Each question offers five options related to corporate values, which the user must rank from highest to lowest based on their degree of identification with each value. Following the completion of the questions, an additional form is presented to the respondent to provide personal information. This information is highly valuable and includes details such as the department/area of the company in which they work, level of responsibility, location, gender, etc. These details enable the establishment of different levels of analysis in the subsequent stages of the study.

According to Kline (2005), the sample size used in this study, consisting of 285 individuals, would be considered small to medium. To verify the adequacy of the sample size, we conducted an a priori analysis using the Gpower 3.1 tool, as recommended by Faul et al. (2009). This analysis estimates the required sample size based on predetermined values for the desired significance level (α), statistical power (1-β), and population effect size. In our model, which includes four predictors, the Gpower test indicated that a minimum sample size of 74 individuals is needed to achieve a power of 0.95, with an alpha of 0.05 (see Fig. 3). Therefore, our final sample of 285 individuals surpasses the initial sample size requirements (Roldán & Sánchez-Franco, 2012).

Fig. 3.

G*Power plot.

(0.23MB).

Regarding the respondents’ profile, the sample consists of 285 individuals, of which 114 (40%) are male, 81 (28.4%) are female, and 90 (31.6%) preferred not to declare their gender. In terms of managerial level, the 285 respondents are classified as follows: 100 (35%) are first-line managers, 46 (16%) are middle managers, 42 (15%) are senior managers, and 97 (34%) do not have personnel under their supervision.

This section aims to elucidate the procedures employed to accomplish the goal of developing and validating a measurement instrument for the digital culture construct. In this study, an adaptation of the Organizational Culture Assessment Instrument (OCAI) was utilized, which has been previously employed and validated for assessing the classic cultural archetypes formulated by Cameron and Quinn (1999) and derived from the Competing Values Framework (Quinn & Rohrbaugh, 1981). These archetypes include Clan (also referred to as People) culture, Adhocracy (also known as Innovation) culture, Market (also termed Goals) culture, and Hierarchy (also denoted as Norms) culture. Additionally, the endogenous construct, Digital culture, was evaluated using a scale developed based on a comprehensive qualitative analysis conducted earlier, also drawing inspiration from the Competing Values Framework.

The authors initiated the process by creating a preliminary set of 26 items that encompassed various aspects of digital culture. This compilation was based on an extensive review of the literature on organizational culture and digitalization, encompassing 262 papers published between 1980 and 2022. To conduct the review, a keyword search was conducted in the Web of Science (WoS) repository, focusing on terms such as "digital culture" (Topic) OR "organizational culture" AND "digital*" (Topic) within the Business Economics research area.

Following the literature review, in-depth interviews were conducted with managers to gain further insights into the cultural factors and drivers of digitalization. These semi-structured interviews involved specialists from academia and industry who have contributed to the ongoing discourse on digitization at the business level. The experts were presented with the five constructs and their definitions and were asked to discuss how they believed these cultural archetypes manifest in an organization, specifically identifying the values, behaviors, or organizational structures associated with the five cultures shaping the model. The interviews were recorded, transcribed, and coded, and the findings were combined with the literature review to generate potential new items.

Using the gathered information, a six-item instrument for measuring digital culture was developed, utilizing a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). The initial questionnaire draft underwent a proofreading and refinement process by a panel of practitioners and academic experts to ensure contextual and content validity. Minor language adjustments were made based on their feedback. After conducting pilot testing, a comprehensive survey was launched, and the collected data underwent various statistical tests to assess item reliability, composite reliability, convergent validity, discriminant validity, absence of multicollinearity, and other relevant factors.

The initial responses indicated that all the items achieved loadings exceeding the critical threshold of 0.707, thus satisfying the requirement for item reliability. Furthermore, the internal consistency of the constructs under examination ranged from 0.88 (Cronbach's alpha) to 0.91 (composite reliability), surpassing the recommended minimum value of 0.7 as suggested by Hair et al. (2010). Moreover, the average variance extracted surpassed the critical level of 0.5, attaining a value of 0.62, thereby confirming convergent validity.

Lastly, a panel of knowledgeable scholars convened as a Focus group to review a preliminary draft of the instrument. This panel of experts gathered in Toledo, Spain, in February 2022 during the XXXI Iberian Conference on Scientific Management. Their feedback and insights contributed to the final version of the instrument, which comprises 36 items. All items were measured using five-point ordinal scales.

This study employs partial least squares structural equation modeling (PLS-SEM), a variance-based SEM approach (Roldán & Sánchez-Franco, 2012), to test the conceptual model and hypotheses. PLS-SEM is a quantitative method widely used in the social sciences, particularly in management, marketing, and economics (Hair et al., 2012). It is well-suited for exploring relationships between latent constructs such as attitudes, beliefs, and behaviors in complex systems (Hair et al., 2014). The choice of PLS-SEM is motivated by the composite construct nature of the variables in our study model (Benítez-Amado et al., 2017). Previous studies (Henseler et al., 2014; Rigdon, 2012; Rigdon et al., 2017) and empirical simulations (Becker et al., 2013; Sarstedt et al., 2016; Felipe et al., 2020) have supported and praised the use of PLS-SEM for models with composite constructs, highlighting its consistency (Rigdon, 2016) and lack of bias (Sarstedt et al., 2016). Furthermore, this study aims to assess the predictive power of key constructs in relation to digital culture, making PLS-SEM an appropriate method for analysis. The measurement and structural models were estimated using SmartPLS v3.2.9 software (Ringle et al., 2015).

In addition to PLS-SEM, this paper complements the analysis with two advanced techniques: Importance-Performance Map Analysis (IPMA) and PLS-Predict. The combination of these methods enables researchers to gain a deeper understanding of the complex relationships between variables, as well as the relative importance and performance of these variables within a specific context. This integrated approach facilitates a comprehensive analysis of the data, enhancing our understanding of the underlying relationships between variables and their impact on the outcome of interest. The combination of PLS-SEM and IPMA offers valuable insights into the relative importance and performance of variables, while PLS-Predict allows for predictions based on these relationships.

Table 2 presents descriptive statistics for the manifest variables analyzed in the model. It includes the mean, median, standard deviation, as well as the minimum and maximum values.

The five cultural typologies in the CFAM model have been estimated as Mode B (formative) composites. This modeling approach aligns with previous studies that also employed Mode B to estimate the cultural archetypes inherent in the OCAI Model (Sanz-Valle et al., 2011; Roldán et al., 2012; Naranjo-Valencia et al., 2017; Felipe et al., 2017). Consequently, these composites require a different evaluation compared to the constructs modeled in Mode A (reflective). Instead of assessing factor loadings and various measures of composite reliability, convergent validity, and discriminant validity, it is necessary to test for the absence of multicollinearity between the different manifest variables and examine the weights of these indicators in shaping their respective constructs. Concerning multicollinearity between items, Petter, Straub and Rai (2007) suggest that a variance inflation factor (VIF) higher than 3.3 indicates high multicollinearity. In our case, the maximum VIF value for indicators was 1.481, which is well below this critical threshold. Therefore, it can be concluded that the analyzed model does not exhibit any multicollinearity issues. Next, we assessed the magnitude and significance of the weights, as presented in Table 3. These weights provide insights into the contribution of each indicator to its respective composite (Chin, 1998), enabling the classification of indicators based on their contribution.

To evaluate the structural model, several aspects need to be considered. Firstly, the coefficients of variance explained, also known as the coefficient of determination (R2), for the endogenous variables must be assessed. Secondly, the magnitude of the path coefficients (β) and the strength of these relationships should be evaluated. The significance of these relationships can be analyzed using the Bootstrapping procedure, which involves generating 5000 random samples.

Evaluation of the coefficient of determination (R2)

The coefficient of determination (R2) is a measure of the predictive power of a model, indicating the proportion of variance in an endogenous construct explained by its predictor variables. Ranging from 0 to 1, the R2 score reflects the model's ability to predict the variable of interest. It is essential for R2 values to be sufficiently high to demonstrate at least some explanatory power (Urbach & Ahlemann, 2010). However, the appropriate levels of this indicator may vary across disciplines. Falk and Miller (1992) suggest that the explained variance of endogenous variables should be greater than or equal to 0.1, as lower values, despite being statistically significant, provide limited information and have minimal predictive ability. Chin (1998)) categorizes R2 values close to 0.6 as substantial, those close to 0.3 as moderate, and those below 0.2 as weak. Similarly, Hair et al. (2011) describe values above 0.75 as substantial, above 0.5 as moderate, and above 0.25 as weak. In our study, the endogenous construct Digital Culture has an R2 value of 0.509 (refer to Table 4 and Fig. 4), which can be classified as moderate to high. Therefore, we can conclude that the model demonstrates an adequate level of explanatory power based on these values.

Table 4.

Structural model results.

Relationship	Β	t	p-value	5%	95%
H1: People culture → Digital culture	0.215 Sig.	1.719	0.043	0.007	0.357
H2: Innovation culture → Digital culture	−0.275 Nsig.	0.944	0.173	−0.466	0.244
H3: Goals culture → Digital culture	−0.375 Sig.	5.469	0.000	−0.490	−0.270
H4: Norms culture → Digital culture	−0.261 Sig.	3.736	0.000	−0.341	−0.149
R2Digital culture = 0.509

Note: Sig. = statistically significant relationship; Nsig. = statistically non-significant relationship.

Fig. 4.

Display of structural model results.

(0.23MB).

As previously mentioned, this research is primarily focused on prediction. According to Shmueli and Koppius (2011), the predictive performance of a model refers to its ability to generate accurate predictions for new interpretable observations, whether they are temporal or cross-sectional. Therefore, this paper assesses the predictive ability of our model through cross-validation with retained samples (Evermann & Tate, 2016), with a specific emphasis on the key dependent construct, Digital Culture. To conduct this evaluation, we employed the PLS-predict algorithm (Shmueli et al., 2016), which is available in SmartPLS software version 3.2.9 (Ringle et al., 2015).

To analyze the results, our attention should be directed towards the Q2 value. Positive Q2 values indicate that the prediction error of the PLS-SEM results is lower than the prediction error obtained by simply using mean values. Consequently, the model demonstrates adequate predictive ability. In our case, the model satisfies this criterion for the Digital Culture construct and all of its indicators, except for Tolerance to failure (Table 5).

The PLS results presented earlier are further complemented by the application of the "Importance-Performance Map Analysis" (IPMA) technique, also known as Priority Map Analysis (Ringle & Sarstedt, 2016). In essence, the IPMA technique compares the total effects, which signify the relevance of antecedent constructs in determining a specific construct (referred to as the target construct), with the mean values of latent variable scores, which indicate their performance (Ringle & Sarstedt, 2016). The objective is to identify the antecedents that hold relatively greater importance in influencing the target construct. This information becomes crucial for decision-makers as it highlights the major areas that require focus or improvement.

Therefore, the relative levels of importance/performance of different constructs for the target construct indicate their logical order of deployment. Managers can initiate the process by increasing the deployment levels of constructs located in the lower right quadrant. Constructs/indicators found in the upper right quadrant are also significant and exhibit higher performance/deployment levels. Monitoring them would be sufficient to maintain or enhance their performance. On the other hand, constructs/indicators in the upper left quadrant are of lesser importance but demonstrate performance/deployment levels above the mean. This situation may suggest inadequate resource allocation.

As illustrated in Fig. 5, the People culture emerges as the most crucial factor in determining the target construct of the assessment, namely Digital culture. In terms of importance and level of deployment, the People culture surpasses the average and occupies the upper right quadrant. Following closely in importance is the Innovation culture, while the Norms and Goals cultures are relatively less influential, albeit still significant. Notably, the Innovation culture finds itself in the lower-right quadrant, indicating a noteworthy managerial implication. Despite its above-average importance, this culture lags in terms of deployment. Consequently, organizations should prioritize the promotion of disruptive values within their organizational culture to foster the growth of digital culture.

Fig. 5.

IPMA matrix.

(0.17MB).

The main goal of this study has been to provide insights into the characteristics of a digital culture within organizations. This is evident through the prevalence of values such as customer adaptation, analytical approach, cooperation, tolerance to failure, learning orientation, and openness to change. These values promote the formation of cross-functional teams and encourage learning to generate optimized and agile responses. Moreover, it has been confirmed that the development of human capital is a genuine priority in companies with a strong digital culture.

Furthermore, we have validated the explanatory power of the classical approach consisting of the four cultural archetypes: People Culture, Innovation Culture, Goals Culture, and Norms Culture, originally proposed by Cameron and Quinn (1999) and based on the Competing Values Framework (Quinn & Rohrbaugh, 1981). This framework aids in understanding the formation of the new archetype, Digital culture. The findings are reinforced by the high level of the R2 coefficient and the positive Q2 values, indicating that the prediction error of the PLS-SEM results is lower compared to using mean values alone. Consequently, the model exhibits adequate predictive capability.

At the hypothesis level (Table 6), it is particularly noteworthy that both Goals culture and Norms culture have a significant, yet negative, impact on the emergence of Digital culture. This implies that certain values associated with Goals culture (Results orientation, Excellence, Competitiveness, Pragmatism, Ambition, Leadership) and Norms culture (Formality, Coordination, Stability, Responsibility, Perseverance, Efficiency) are in competition with those of Digital culture, impeding its proper development and hindering the digital transformation process.

In line with this, the IPMA analysis (Fig. 5) reveals that People culture and Innovation culture play the most crucial role in shaping digital culture. However, only People culture surpasses the average in terms of performance. Hence, companies aiming to effectively implement digitalization should emphasize values such as Closeness, Commitment, Generosity, Loyalty, Teamwork, and Trust, which are characteristic of People culture. Simultaneously, they should not overlook values like Autonomy, Creativity, Daring, Dynamism, Innovation, and Risk tolerance, which are representative of Innovation culture. Therefore, at the organizational level, companies dominated by Goals culture or Norms culture that aspire to undergo a digital transformation should recognize the potential challenges they may face. Consequently, they should prioritize learning, team development, and cultivate a culture that supports this profound process of change (Kane, 2019; Nicolás-Agustín et al., 2021).

This paper contributes significantly to the literature on organizational culture and digital transformation. Firstly, it enhances our understanding and analysis of the values associated with digital culture, thereby expanding the conventional perception of organizational culture to include the digital mindset alongside traditional cultural archetypes. Recognizing the digital component of culture is crucial for businesses that heavily rely on digital technology for their operations and stakeholder relations. To this end, the Cultural Fit Assessment Method (CFAM) (Leal-Rodríguez et al., 2023) presents a methodological approach that is both innovative and robust, as it builds upon elements of the Competing Values Framework (Quinn & Rohrbaugh, 1981) -a widely accepted and validated framework in the management of corporate culture. Hence, a significant contribution of this study is the adaptation and extension of the OCAI scale (Cameron & Quinn, 1999) to measure digital culture based on its fundamental underlying values.

Secondly, we contribute to the framework of functionalist theory by integrating it with structuralism, thereby providing a more comprehensive explanation of social processes. Functionalism focuses on how social units fulfill specific roles and contribute to the stability and maintenance of society. On the other hand, structuralism emphasizes the underlying patterns and linkages within social structures. A functionalist approach to assessing a social institution, such as an organization, would examine how it meets the needs of various stakeholders. In contrast, a structuralist approach would analyze the organizational relationships, power dynamics, and cultural significance associated with these interactions. By merging these perspectives, we can achieve a more comprehensive understanding of the organization as a social institution and how its functions are influenced by the underlying patterns and connections within it. Consequently, the combination of functionalism and structuralism enables a more nuanced and thorough comprehension of cultural change (Dyer, 1985; Wilkins & Dyer, 1988) by providing a theoretical framework that aids in better understanding the internal dynamics of cultural change, specifically the role of dominant cultural archetypes in shaping, developing, and transitioning to other archetypes, as observed in the case of digital culture.

Thirdly, we advance a research agenda that focuses on measuring culture based on archetypes (Richter et al., 2016; Grover et al., 2022) rather than the traditional approach based on discrete variables or values. This endeavor aims to contribute to a deeper understanding of digital culture. Archetypes represent universal patterns of behavior and experience that embody the collective unconscious of a culture. By measuring culture based on archetypes, researchers can gain profound insights into the underlying cultural patterns that shape an organization. In contrast, classical approaches that rely on discrete variables only provide a superficial understanding of cultural differences. In the context of the emergence of digital culture, we have a unique opportunity to develop a new cultural archetype that reflects the shared patterns of values among individuals in the digital age. This new archetype would offer a more comprehensive understanding of how digital technologies influence cultural norms, beliefs, and the ways in which people engage with the world. By incorporating digital culture into a research agenda based on archetypes, we can enhance our ability to compare and contrast different cultures, encompassing both traditional and digital realms. This approach yields valuable insights into how cultural values are expressed and perpetuated across diverse contexts, as well as how these values are shaped by technological advancements.

In summary, this paper makes a valuable contribution to the field of research on digital culture by defining the organizational values necessary for its implementation. It sheds light on the question of what type of organizational culture is required to embrace digitalization. This study represents a crucial initial step towards comprehending the impact of culture on successful digitalization. By identifying the cultural values that are crucial for digital transformation, we establish an ideal target digital culture. Consequently, we lay the foundation for future research on the role of organizational culture in digital transition. Moreover, this paper has broader implications for organizational culture research in general.

The first managerial implication of this study relates to fostering a digital culture within firms. Specifically, HR departments should establish systems to identify their corporate values and utilize them as catalysts for digital change. It is crucial to recognize and strengthen poorly established digital culture values through cultural transformation techniques that facilitate an efficient digitalization process. Therefore, conducting an effective organizational culture diagnosis based on archetypes becomes a vital procedure in supporting the adoption of new technologies. To ensure the success of digitization, it is essential to perform an initial cultural diagnosis and engage in ongoing monitoring, as corporate culture is dynamic. More specifically, we provide practitioners with a target culture upon which they are advised to build their cultural change activities by identifying values that are essential for achieving or reinforcing a digital culture. The Cultural Fit Assessment Method (CFAM), based on the competing values framework (Quinn & Rohrbaurgh, 1983), can be employed by practitioners as a tool for managing culture during the digital transformation process. This tool enables practitioners to examine their current culture and identify areas that require cultural change as part of the digital transformation process.

A second managerial implication that arises from our results concerns the importance of fostering an enthusiastic, positive, and open mindset toward upcoming technological challenges and digital disruption. It is crucial for organizations to address their members' resistance to digitalization by adopting transparent practices aimed at motivating and engaging employees while reducing knowledge gaps regarding the transformation process. Although most transformation processes are primarily top-down, the willingness and commitment of employees to understand and embrace digitalization significantly impact its performance. Therefore, organizations must learn how to motivate both themselves and their staff. Employees need to recognize that their company's survival hinges on undergoing digitalization. Consequently, they should commit to enhancing their skills and supporting less digitally competent colleagues whenever possible. In this regard, investing in intergenerational mutual mentoring could be beneficial, where senior members are paired with younger ones to learn about the digital shift from them. The goal is to bridge the generational gap in the digital environment by facilitating the exchange of values, ideas, expectations, and abilities between older and younger individuals.

Thirdly, managers need to recognize the following points: (1) the formation of a digital culture archetype within organizations is a complex phenomenon that comprises multiple dimensions or sets of interacting values; (2) the detection and measurement of cultural archetypes provide a qualitatively superior methodology for cultural analysis compared to the use of measurement scales based on individual cultural dimensions, such as values; (3) within the same organization, multiple cultural archetypes may coexist simultaneously or to varying degrees of intensity; and (4) while a dominant archetype may exist in an organization due to strategic orientation, isomorphic pressures from other organizations, or external and internal environmental influences, cultures also evolve through conflicts and interactions between different cultural archetypes. These interactions can have both positive and synergistic effects or negative effects, acting as barriers to the development of other archetypes, as seen in the case of digital culture. Today's organizations are characterized by cultural complexity, where multiple archetypes and cultural values coexist and complement each other, but can also contradict each other. For instance, many organizations emphasize gender equality, yet discrimination against women at certain levels of their structure remains prevalent. This cultural complexity, with potential conflicts between archetypes, values, and microcultures, must be measured and analyzed when implementing any cultural intervention program.

The final implication of this study entails recognizing that merely harnessing new technologies and realizing the advantages of big data, as well as understanding how to utilize AI in data analysis, is inadequate. Consequently, the diagnosis and transformation of corporate culture should be perceived and fostered as an essential prerequisite to effectively navigate the digital revolution. By embracing both functionalist and structuralist paradigms, organizations acknowledge that cultural complexity, with its network of dynamic interactions and diverse social manifestations, is indispensable for organizational functioning. The abundance and diversity of cultural archetypes serve as valuable assets for organizations to navigate novel situations and challenges, including digitalization.

However, this work is subject to several limitations. Firstly, the sample size remains modest, focusing exclusively on businesses within a specific geographic area (Spain). Therefore, future endeavors aim to expand the scope of the study to include other countries and economic sectors, enhancing the robustness and generalizability of the findings and implications. Secondly, the study solely relies on one method to gather the perceptions of the surveyed individuals. Exploring alternative methodologies, such as fsQCA, could unveil additional insights that contribute to a deeper understanding of the interactions between digitalization and performance (Ribeiro-Navarrete et al., 2021). Future research endeavors also aim to assess the impact of different cultural archetypes, including the proposed new digital culture archetype, on various financial and organizational performance metrics. This aligns with the trajectory of recent empirical studies on this topic, such as those conducted by Khan et al. (2022) and Naveed et al. (2022).