Dynamic capabilities and data-driven culture for digital transformation: evidence from agri-food SMEs Open Access

The agri-food industry faces mounting environmental pressures that are reshaping the socio-economic landscape, making the transition toward digital technologies increasingly urgent (Abbate et al., 2023). Many have recognized that digitalization can heighten efficiency, competitiveness, resilience, and sustainability in an industry historically viewed as less technology-intensive (Bičkauskė et al., 2020). Indeed, recent evidence suggests that digital tools substantially improve food sustainability and strengthen agri-food systems (Abbate et al., 2023; Ciasullo et al., 2022; Galanakis et al., 2021; Marvin et al., 2022; Masi et al., 2021). These gains stem from integrating new-age technologies such as artificial intelligence, big data analytics, and the Internet of things, which enable more prudent use of resources, reduce waste, and strengthen supply chain performance (Javaid et al., 2022). The expanding role of big data in this industry has been especially notable, as it equips firms with advanced capabilities to detect and preempt key production risks, such as irrigation shortfalls or pathogen attacks (Ciasullo et al., 2024; Cricelli et al., 2024; Kazancoglu et al., 2024).

Despite these advantages, introducing digital innovations often necessitates overturning established management practices, which, often times, demand a strategic reconfiguration of core assets (Ciasullo et al., 2025). Large enterprises may respond more effectively due to the equivalent (large) magnitude of resources they have at hand (Annosi et al., 2025), but small and medium-sized enterprises (SMEs) often face shortfalls in qualified personnel, inadequate digital expertise, and organizational resistance (Ates and Acur, 2022; Battistoni et al., 2023; Bičkauskė et al., 2020). These concerns are pressing given that SMEs constitute an essential driver of economic growth, employment, and social integration (Boumediene et al., 2022; Rao et al., 2023), and they play a dominant role in agricultural production (European Commission, 2018).

Extant research has frequently emphasized the barriers that hinder digital transformation (DT) (Benavides-Espinosa et al., 2024; Darra et al., 2023; Dwivedi et al., 2023; Malodia et al., 2023). Yet, limited work has shed light on how dynamic capabilities, paired with a supportive organizational culture, can expedite DT. This gap is problematic for four reasons. First, overlooking these drivers may leave agri-food SMEs exposed to intensifying global competition, causing stalled growth and missed opportunities (Ciasullo and Lim, 2022). Second, deeper insights on capabilities and cultures can guide agri-food SME managers toward prudent resource allocation and stronger operational effectiveness (Ismail, 2024). Third, these elements are indispensable for the resilience of supply chains that feed growing populations, given that DT can coordinate and streamline the delivery of agri-food supplies (Ciasullo et al., 2025). Fourth, the agri-food industry faces urgent pressures from climate change, geopolitical volatility, and public health crises, all of which demand a swift shift to flexible digital infrastructures (Belhadi et al., 2024; Kazancoglu et al., 2024). In this vein, this study addresses this gap by analyzing how dynamic capabilities and data-driven cultures facilitate a more effective adoption of digital solutions in the agri-food industry. Thus, the guiding research question (RQ) is:

Theoretically, this study employs dynamic capabilities as a theoretical lens to explore how processes, knowledge, and skills shape organizational culture and drive DT. Methodologically, this study employs structural equation modeling (SEM), based on 300 survey responses drawn from Italian agri-food SMEs, to test its hypotheses. Practically, this study sheds light on how agri-food SME managers can strategically leverage dynamic capabilities, especially big data analytics capabilities (BDAC) and digital capabilities (DC), combined with a DDC, to enhance their readiness for DT. The next section elaborates on the theoretical background and hypotheses. The section thereafter details the quantitative methodology while the subsequent sections deliver the results and discuss the theoretical and managerial implications. The last section concludes with limitations and avenues for future research.

The agri-food industry has historically lagged in technology adoption, a pattern often attributed to its structural complexity and stakeholders’ resistance to change (Farace and Tarabella, 2024). This issue is compounded by the prevailing perception of the industry as low-tech (Bucci et al., 2019; Finco et al., 2018), which further contributes to its slower pace of innovation compared to other industries. Yet, ongoing advances in digital solutions signal considerable benefits for SMEs, including opportunities to optimize production, mitigate risks, and improve decision-making flexibility (Cricelli et al., 2024; Fachrunnisa et al., 2020; Frau et al., 2022).

Technologies such as artificial intelligence, blockchain, radio frequency identification, and the Internet of things have emerged as critical enablers in this context. Artificial intelligence can analyze soil quality, predict climate patterns, and forecast crop yields while reducing vulnerabilities associated with pathogen attacks (Pathan et al., 2020; Sharma et al., 2021). Blockchain strengthens traceability across the supply chain, thereby improving food safety and security (Hassoun et al., 2024; Kayikci et al., 2022; Zhang et al., 2022). Radio frequency identification offers real-time monitoring, which supports traceability and lowers operational expenses (Gautam et al., 2017; Kumari et al., 2015; Magalhães et al., 2019). The Internet of things allows automated processes through real-time sensor data, cutting the need for extensive manual intervention in irrigation, farm surveillance, and harvest planning (Cane and Parra, 2020; FAO, 2019; Kamilaris et al., 2016).

Yet, introducing these innovations without cultivating organizational capabilities, especially those essential for managing and interpreting complex data, can yield minimal returns (Ciasullo et al., 2025). While DT demands a comprehensive shift in resources, competencies, and processes (Warner and Wäger, 2019), many agri-food SMEs lack big data analytics and DC, along with a culture that values data-driven decision-making. This shortfall constitutes both a risk and a potential source of advantage, as establishing these capabilities enables firms to effectively adapt to current uncertainties and capitalize on emerging market opportunities.

The dynamic capabilities perspective elucidates how firms can reconfigure resources to operate effectively amid turbulence, a situation that is increasingly common in the agri-food industry (Schneider et al., 2024). As Teece et al. (1997) argue, dynamic capabilities capture a firm’s ability to sense and seize evolving opportunities, ensuring the protection, recombination, and adaptation of tangible and intangible resources (Teece, 2007). Eisenhardt and Martin (2000) reinforce this viewpoint, noting that dynamic capabilities are necessary for adapting and combining key assets in response to shifting market conditions. Evolving from the resource-based view (Penrose, 1959), dynamic capabilities enable firms to exploit internal resources more fully, particularly as market environments become increasingly volatile (Yeow et al., 2018). These capabilities shape how firms orchestrate expertise and knowledge, enhancing the adoption and use of big data technologies (Endres et al., 2020), and by extension, securing a more enduring competitive position.

DT requires deep structural changes across a firm and thus demands specific capabilities and cultures to fully harness digital technologies (Ciasullo and Lim, 2022). The agri-food industry increasingly regards big data and related technologies as strategically important (Bronson and Knezevic, 2016), which highlights the need to examine how dynamic capabilities—specifically BDAC and DC—shape a DDC and propel DT. In essence, BDAC represent a firm’s capacity to manage and interpret data for complex decision-making in a data-driven framework (Akter et al., 2016) while DC reflect a firm’s proficiency in deploying digital technologies, relying on accumulated experience and technical expertise (Arkhipova and Bozzoli, 2018; Belhadi et al., 2024). Noteworthily, recent work indicates that these capabilities underpin a DDC, understood as an orientation where data insights guide managerial choices rather than intuition (Gupta and George, 2016).

The relationship between BDAC and data-driven cultures has been examined through various lenses, often focusing on non-direct effects (Al-Khatib, 2022; Karaboga et al., 2023; Liu et al., 2022; Yu et al., 2021). For instance, Karaboga et al. (2023) found that data-driven cultures partially mediate the link between BDAC and firm performance in information technology settings. Liu et al. (2022) and Yu et al. (2021) documented the moderating roles of data-driven cultures in enhancing supply chain integration and supply chain finance, respectively. Few studies, however, have addressed the direct influence of BDAC on data-driven cultures. Karaboga et al. (2023) even reversed the direction, conceptualizing data-driven cultures as a precursor to BDAC, aligning with Gupta and George (2016). Almazmomi et al. (2022) and Duan et al. (2020) took a different view by showing how business analytics capabilities can fortify a firm’s data-driven mindset and improve the absorption and use of information. This article takes the position that BDAC serve as a catalyst for establishing data-driven cultures, especially in agri-food SMEs, where systematic collection and analysis of large datasets are critical for optimized decision-making. Without adequate analytical capabilities, firms may struggle to embed data insights into daily operations or reinforce a shared commitment to data-driven decision-making. Hence, the following hypothesis is proposed:

DCs have frequently been studied in relation to digital culture and DT (Baiyere et al., 2025; Shin et al., 2023; Slavković et al., 2023). However, fewer investigations have focused on specific, measurable dimensions of DC, particularly those that underscore the importance of training, recruiting, and shared learning across an organization. Recent work by Proksch et al. (2024) conceptualizes DC to include structured training programs, recruitment that values digital expertise, consistent usage of digital services, the presence of a qualified workforce to support further digitalization, and open dialog around digital projects—including lessons from both successful and flawed initiatives. These elements reflect a firm’s concrete actions to embed digital tools into everyday operations, rather than just the adoption of isolated technologies. More importantly, developing DC in this manner can strengthen an organization’s inclination to rely on data-driven insights. Noteworthily, Zhen et al. (2021) found that heightened digital know-how often spurs cross-functional collaboration, which increases the acceptance of digital technologies, in which data analytics is a part of, in daily decision-making processes. In such a scenario, team members who receive ongoing training and work in an environment that values transparent learning from digital successes and failures are more likely to be inclined to base their judgments on data rather than instinct. This evolution of mindsets resonates with the idea of a DDC, where shared beliefs, values, and assumptions guide members to prioritize empirical evidence (Gupta and George, 2016). In the agri-food industry, this alignment can be especially impactful, given the growing reliance on technologies for crop monitoring, resource allocation, and supply chain coordination. Consequently, this study posits that DC serve as an enabler of data-driven cultures, leading to the following hypothesis:

A growing body of work also maintains that organizational culture is pivotal in driving DT (Çetin Gürkan and Çiftci, 2020; Ghafoori et al., 2024; Karimi and Walter, 2015; Liu et al., 2023; Papanagnou et al., 2022). These works suggest that transformation efforts flourish when managers and employees share beliefs and routines that endorse continuous learning, experimentation, and openness to technological change. One branch of this literature highlights DDC as especially relevant to the agri-food industry, where the thorough use of data on crop conditions, resource optimization, and supply chains can yield strategic and operational gains. For instance, Ghafoori et al. (2024) introduced the concept of data-driven DT, which Papanagnou et al. (2022) portrayed as a process of realigning resources in firms that place data at the forefront of decision-making. Yet, most of these studies often consider the influence of culture at large or frame culture as a moderator, which, in turn, overlooks the direct role of DDC in initiating and reinforcing DT. This article argues that a strong data orientation spurs firms to embed analytical and DC more systematically, transforming them from optional enhancements into essential supports for transformative changes in firm operations (Xie et al., 2022). Thus, the following hypothesis is proposed, and Figure 1 presents the conceptual framework:

Data were gathered over four months – from July to November 2024 – using an online survey of agri-food SMEs in Italy. The Italian market was chosen due to its agri-food industry’s considerable contribution to the national economy, accounting for 2.3% of gross domestic product – higher than the European Union average of 1.9%, in addition to Italy ranking third among European Union members in value added generated by the agri-food industry, following Germany and France (Unioncamere, 2023).

A simple random sampling method was used to select firms from the AIDA (Analisi Informatizzata Delle Aziende Italiane in Italian) database (created and distributed by Bureau van Dijk – https://aida-r1.bvdinfo.com), which provides balance sheets, master data, and product information for active Italian firms. Simple random sampling is widely recognized for reducing selection bias and improving the generalizability of results (Lim, 2025; Saunders et al., 2009). Each chosen firm was contacted by email and provided with details about the research goals and instructions for completing the survey, ensuring that respondents represented a diverse range of companies (Bryman, 2016). The survey targeted founders, directors, or managers as they hold key decision-making roles. This approach enhances the validity of the dataset, given that respondents were selected randomly rather than through subjective criteria. The profile of participants is reported in Table 1.

The questionnaire was created in Google Forms and included 23 questions divided into three sections. The introductory portion explained the survey’s objectives, listed the identities of the researchers, estimated the time needed for completion, and outlined data usage and privacy policies. The first section comprised demographic and organizational characteristics of respondents while the second section measured the constructs in the conceptual framework using a five-point Likert scale ranging from “completely disagree” to “completely agree.” Specifically, five questions measuring BDAC were adapted from Hao et al. (2019), five items reflecting DC followed Proksch et al. (2024), five items evaluating DDC drew upon Yu et al. (2021), and four items relating to DT were based on Xie et al. (2022). The complete set of measurement items is reported in Table 2.

Assurances of anonymity were provided to minimize social desirability bias (Lim, 2025), a phenomenon in which respondents might alter their answers to appear more aligned with social norms (Nederhof, 1985). A pretest with experts was conducted to ensure content validity (Lim, 2024) while a pilot study with a sample of 20 respondents from the target population was conducted to confirm clarity and comprehension of all questions before full-scale data collection (Lavrakas, 2008) and thus establish face validity (Lim, 2024). A total of 300 responses were secured, adhering by precedence of sample size for recent and similar studies (Lim, 2025) involving agri-food (Ciasullo et al., 2025), and meeting Bentler and Chou’s (1987) recommendation of maintaining a ratio of at least 5:1 between the number of responses and estimated parameters for SEM. This sample size also satisfies Nunnally’s (1967) guideline of at least ten responses per item for regression-based analyses like SEM.

A covariance-based structural equation modeling (CB-SEM) approach using SmartPLS 4 (Ringle et al., 2024) was adopted to evaluate the relationships among the constructs. CB-SEM is a method for testing theoretical models because it estimates correlations between independent and dependent variables while accounting for unobserved structures, wherein an inherent advantage is its assumption that constructs operate as common factors, which helps reduce measurement errors and refine latent variable estimates (Rigdon et al., 2017; Sarstedt et al., 2016).

A two-step approach was conducted (Anderson and Gerbing, 1988). First, a confirmatory factor analysis (CFA) was carried out to assess the reliability and convergent validity while heterotrait-monotrait (HTMT) ratio of correlations was used to verify the discriminant validity of the measurement model. Second, a path analysis was conducted to empirically evaluate the hypothesized relationships among the constructs in the structural model. Related descriptives, correlations, and model fit indices were also reported.

The sample comprises 300 professionals from the Italian agri-food industry, with just under half (49%) falling within the 46–65 age range, suggesting seasoned leadership across participating organizations. The distribution by gender stands at 53% female and 47% male, indicating a balanced representation. Nearly half of the respondents (49%) hold a postgraduate degree, reflecting the increasing knowledge demands of the agri-food industry. Founders account for 53% of the sample, underscoring the influence of entrepreneurial perspectives in shaping strategic decisions.

A CFA was conducted to assess the adequacy of the measurement model in terms of convergent validity and reliability. As shown in Table 2, all factor loadings exceed the recommended 0.704 benchmark while average variance extracted values surpass 0.50 threshold, reflecting acceptable convergent validity (Hair et al., 2019, 2020; Lim, 2025). Cronbach’s alpha and composite reliability also exceed 0.70, indicating adequate reliability. Table 3 confirms discriminant validity by showing that the square roots of average variance extracted for each latent variable are greater than the corresponding inter-construct correlations and HTMT ratios remain below 0.85 (Henseler et al., 2015; Lim, 2025). These results, therefore, confirm that the measurement model meets established criteria for validity and reliability.

The means and standard deviations also provide insights into the overall level of digital maturity among the sampled firms. For instance, items related to DT exhibit relatively modest mean scores (approximately 2.50–2.65) alongside standard deviations above 1.00, suggesting a lower but varied adoption of digital initiatives. In contrast, items measuring BDAC, DC, and DDC each feature moderate average scores (ranging from roughly 2.90 to 3.50) with similarly high standard deviations, indicating that some firms are progressing toward more data-oriented and digitally equipped operations, whereas others remain in the initial stages.

The model fit indices also exhibit generally favorable results. The relative chi-square (χ²/df) is 3.660, which falls below the recommended cutoff of 5. The comparative fit index (CFI = 0.959), normed fit index (NFI = 0.945), and Tucker–Lewis index (TLI = 0.952) all exceed 0.90, indicating acceptable model fit. Although the root mean square error of approximation is 0.094, marginally higher than the 0.08 threshold, the standardized root mean square residual is 0.036, which remains below 0.08. Overall, these indices suggest that the estimated model achieves a good fit (Hair et al., 2019).

The hypothesized relationships in the structural model were evaluated through CB-SEM. Table 4 indicates that all coefficients are positive and statistically significant. BDAC show a strong effect on DDC (β = 0.860, p = 0.000 < 0.001, t = 9.974 > 2.576), confirming H1. DC also exert a positive and statistically significant impact on DDC (β = 0.251, p = 0.003 < 0.01, t = 3.021 > 2.576), supporting H2. Finally, DDC positively influences DT (β = 0.499, p = 0.000 < 0.001, t = 12.846 > 2.576), validating H3.

All three relationships hypothesized in the conceptual model were supported by the results in the structural model, indicating that big data analytics capabilities and digital capabilities each have a meaningful influence on data-driven culture, which, in turn, significantly contributes to digital transformation. Notably, the strong effect of big data analytics capabilities on data-driven culture suggests that advanced analytical tools and technical expertise not only improve operational efficiency but also shape an organization-wide mindset anchored in empirical reasoning. Meanwhile, digital capabilities appear crucial for embedding collaborative routines and open dialogue around data use, allowing managers and employees to integrate digital insights into everyday decisions. These results collectively underscore the pivotal role of dynamic capabilities and data-oriented practices in enabling agri-food SMEs to adapt their core processes and organizational structures in the face of evolving market demands and thus is consistent with the call to engage in context-sensitive theory development (Homer and Lim, 2024). The next section explores the theoretical implications of these findings and clarifies how they enrich the existing literature on digital transformation while the subsequent section thereafter outlines the managerial implications to guide practitioners in applying these insights.

This article adopts a dynamic capabilities perspective to investigate how big data analytics and digital capabilities and data-driven culture advance digital transformation in the agri-food industry. Several noteworthy findings emerge from the results.

First and foremost, the descriptive statistics suggest that many agri-food SMEs remain in the early stages of DT, displaying a relatively low rate of technology adoption despite its acknowledged importance (Bičkauskė et al., 2020; Bucci et al., 2019; Finco et al., 2018). The substantial standard deviation for DT indicates that the level of engagement with digital technologies varies considerably across firms, suggesting a mix of early adopters and late movers. This observation underscores the possibility that resource constraints, regulatory complexities, or strategic skepticism contribute to uneven progress in technology integration.

Furthermore, the strong link between BDAC and DDC (H1) extends theoretical insights into the agri-food industry, where prior research has typically emphasized big data’s potential for crop management (Abbate et al., 2023), food traceability (Misra et al., 2020), and supply chain optimization (Belaud et al., 2019). These studies, however, offered only scattered evidence on how these capabilities shape organizational culture. The results herein address this theoretical gap by establishing that having sufficient analytical expertise not only leads to better data utilization but also fosters a collective mindset oriented toward evidence-based decision-making. This insight implies that big data analytics is more than a set of technical tools, rather, it is a driver of cultural transformation that can shift managerial attitudes and employee behaviors toward more rigorous methods of identifying market opportunities and minimizing risks.

Moreover, DC also exert a significant impact on DDC (H2), reinforcing previous work on how digital knowledge supports overall organizational culture (Velyako and Musa, 2024; Zhen et al., 2021). Yet, this study contributes a more in-depth perspective by illustrating that DC can spark collaborative routines and transparency around data use, specifically in agri-food environments where technical proficiency often lags (Farace and Tarabella, 2024). Such capabilities appear essential for motivating employees to engage with “agri-data,” a term referring to both structured and unstructured data in agriculture (Kamilaris et al., 2017; Wolfert et al., 2017). When digital knowledge is consistently diffused, teams become more aware of how analytics can optimize resource allocation, mitigate production risks, and respond strategically to environmental changes. These findings align with European Commission objectives for a sustainable and data-driven agri-food sector (https://agridata.ec.europa.eu), as per the Farm to Fork Strategy and Common Agricultural Policy (Luyckx and Reins, 2022).

Last but not least, DDC emerges as a pivotal catalyst for DT (H3), an outcome that both confirms the influence of cultural factors on organizational change (Çetin Gürkan and Çiftci, 2020; Liu et al., 2023) and extends the literature by focusing on a specific type of culture grounded in data-centered practices. This insight is of particular relevance to agri-food enterprises, which often face resource constraints and require a clear rationale for digital investments. The positive link between DDC and DT implies that strategic shifts in culture—toward continuous learning and reliance on empirical evidence—are instrumental in reconfiguring processes, managerial structures, and decision-making protocols. When taken collectively, these findings not only advance the theoretical understanding of how dynamic capabilities function in industries with varied technological maturity but also illuminate DDC as the key connector between technological investments and the realization of transformative change.

Several actionable insights emerge for agri-food practitioners aspiring to develop or accelerate DT.

First and foremost, the findings emphasize the need to strengthen data-related capabilities—big data analytics and digital knowledge—within organizations. In this regard, financial outlays in collaborative learning platforms, training, and staff upskilling may prove vital for bridging skill gaps and ensuring that complex analytics tools are both understood and effectively utilized. The logic is that firms with well-trained personnel stand a greater chance of discovering production inefficiencies and market trends, leading to more adaptable and forward-thinking operations.

Next, agro-food managers should view DDC as both a strategic goal and an enabler of technology adoption. In particular, rather than limiting investment to hardware or software upgrades, decision-makers may consider complementary initiatives that reshape attitudes toward and routines involving data. For instance, firms could introduce brown bag sessions or forums to discuss analytics-based success stories and failures, thereby creating an open learning environment where insights from agri-data become integral to day-to-day tasks. Such practices, in turn, may reinforce employees’ trust in empirical evidence and embed data into processes like inventory management, product innovation, and risk assessment.

Last but not least, a DDC can help align firm-level changes with sustainability goals set forth by international organizations like the European Commission and the United Nations. Technologies that enable advanced crop forecasting, real-time monitoring of water usage, and targeted fertilizer application are examples that not only hold the potential to improve efficiency and strengthen competitiveness but also meet emerging regulatory expectations as well as increasing consumer demands for environmentally responsible production. In this regard, agro-food managers who adopt a data-oriented mindset may discover additional benefits in areas of brand reputation, compliance, as well as social responsibility. Therefore, agri-food firms that integrate these elements stand to navigate rapidly evolving market conditions with greater agility, and, in turn, sustain their competitive advantage in an industry under pressure to reconcile productivity with sustainability.

This study investigated how the configuration of dynamic capabilities – i.e. BDAC and DC – and a DDC can encourage the adoption of digital technologies among Italian SMEs within the agri-food sector. The findings highlight that both types of capabilities positively influence DDC, which, in turn, promotes DT. In practice, managers who prioritize investments in analytics resources, collaborative routines around data, as well as tech-oriented staff development are more likely to shift their organizations toward a mindset guided by empirical insights, thereby increasing their readiness to integrate digital tools into core operations. This line of reasoning underscores that developing a data-oriented culture is not an optional add-on but rather a strategic requirement to harness the full advantages of digital innovation in an industry with escalating demands for productivity and sustainability.

Nevertheless, this study has limitations that open avenues for future inquiries. First, the sample consists solely of agri-food firms based in Italy, raising questions about the wider applicability of the findings and suggesting that extended research across other industries or regions would further validate the results. Second, data collection relied on self-reported surveys. Future investigations could integrate multiple data sources, including case studies, interviews, or longitudinal observations, to capture richer insights and temporal changes in how capabilities and culture evolve. Such mixed-method designs would allow for deeper exploration of contextual factors – such as market volatility and policy shifts – that can shape how agri-food enterprises pursue DT.