Do entrepreneurial and digital orientations impact e-marketing adoption? The COVID-19 pandemic context¹

In recent years, a progressive digitalization and digital transformation (DT) of firms has been observed, which is related not only to the development of digital technologies and their application in business but also to their marketing [Eller et al., 2020; Škare and Soriano, 2021]. Digitalization/DT denotes the process of organizational transformation through the adoption of digital technologies, manifested in organizations as digital artifacts, platforms, infrastructures, and digital business models [Guo et al., 2020], and refers to all changes that digital technologies can bring in a company’s business model, including processes, organizational structures, and products [Hess et al., 2016]. It is one of the key issues facing companies in the contemporary global economy that has changed the way businesses and consumers interact, and transforms the way modern companies do business [Yadav and Pavlou, 2014; Kraus et al., 2019, p. 520].

Growing interest in DT had been particularly relevant during the COVID-19 pandemic as Internet technologies have “come to the forefront of people’s lives” [Barnes, 2020, p. 1] and digitalization can significantly help companies in adapting to new realities and overcoming difficulties of the COVID-19 pandemic [Faraj et al., 2021]. The COVID-19 pandemic has changed the realities of how companies operate [Barbieri et al., 2020] as it is linked with various limitations of doing business (lockdown, injunctions related to sanitary requirements during a pandemic, restrictions on the operation of brick-and-mortar stores, etc.) [Fletcher and Griffiths, 2020; KPMG, 2020]. In the context of the pandemic, authors highlight the acceleration of digitalization [Almeida et al., 2020; Chashmehzangi, 2021] that refers inter alia to the adoption of digital marketing (e-marketing) tools. During the pandemic, the share of e-commerce, as well as the use of Big Data by companies from various industries, increased significantly [OECD, 2020]. Moreover, as indicated by Morgan Stanley [2022], “despite overall slowing economic activity, key trends suggest e-commerce still has room to grow.” Within the pandemic, companies have intensified not only online distribution, including e-commerce, m-commerce, and social commerce [Sonobe et al., 2021], but also online communication with customers [Thierry et al., 2020]. Many companies are abandoning stationary shops or choose to have a few stationary shops, complementary to online sales [Nanda et al., 2021]. The COVID-19 pandemic has contributed also to changes in consumer behavior [Hoekstra and Leeflang, 2020], and as, e.g., Zwanka and Buff [2021] suggest, these shifts in behavior may be long-term.

Referring to the resource-based view [Barney, 1991] and concept of dynamic capabilities, as well as strategic orientations (SOs) [Schweiger et al., 2019], researchers emphasize that SOs such as entrepreneurial orientation (EO) [Hakala, 2011; Meekaewkunchorn et al., 2021] and digital orientation (DO) [Kindermann et al., 2020] affect a firm’s strategies and tactics.

There are some publications on firm strategies during a crisis, including economic ones [Köksal and Özgül, 2007; Pereira et al., 2021], but it is impossible to translate them into the realities of the crisis caused by COVID-19 pandemic [Barbieri et al., 2020]. Although some researchers highlight the relationship between SOs and activities in response to the COVID-19 crisis (e.g., adoption of e-marketing tools), there are still only few publications addressing e-marketing adoption during the COVID-19 pandemic [Sonobe et al., 2021], and especially on the relationships between DO and EO, and e-marketing adoption during the pandemic crisis.

Considering the research gaps, the objective of this study is to verify the influence of such dimensions of the DO as digital learning orientation (DLO) and firm digital strategic vision orientation (DSO), as well as EO on e-marketing implementation (EMI), e-promotion, and e-distribution adoption during the COVID-19 pandemic crisis.

A survey of 241 Polish companies representing various industries was carried out, and to analyze the data and verify research hypotheses, the Partial Least Squares-Structural Equation Modeling (PLS-SEM) method was applied.

The present paper starts with a critical review of the literature about the importance of DO and EO as firm capabilities affecting firm’s strategies, including e-marketing adoption. Next, we present the conceptual model, hypotheses, research method, and findings. We conclude with a concise discussion of the results, limitations, and suggestions for further research.

The present study contributes to the extending knowledge on EO and DO as factors influencing e-marketing adoption [Dann and Dann, 2009; Sriayudha et al., 2018], taking into account the context of the COVID-19 pandemic.

Figure 1 presents the conceptual model reflecting relationships between the studied variables and hypotheses.

Figure 1.

To verify the hypotheses, the computer-assisted web interview/computer-assisted telephone interview (CAWI/CATI) survey on 241 Polish companies with owners or managers responsible for digital marketing was conducted (Table 1). The survey was carried out in July–August 2021, preceded by a pilot study on 10 companies by a research company.

The structure of the surveyed companies by industry is diversified: more companies are manufacturers and the minority are large ones. Taking into account the criterion of being or not being a part of an international corporation – the vast majority is not a part of such a corporation.

To identify the general characteristics of the studied companies, IBM SPSS Statistics 26 was used. To analyze the data and verify the above-mentioned hypotheses, the authors applied PLS-SEM and SmartPLS 3, which are normally used for exploratory research and theory development [Hair et al., 2017] and are recommended when the variables measured do not have normal distribution [Hair et al., 2014, p. 15] – a common case encountered when using the Likert scale. Structural equation models for PLS-SEM are developed based on the observed variables, which are combined to produce latent variables (unobservable) [Witek-Hajduk, 2020, pp. 227–228]. To estimate the errors’ distribution and determine the statistical significance of the regression paths, PLS-SEM uses bootstrapping [Hair et al., 2014, pp. 133–134].

To assess the quality of the model (reliability, validity), the following coefficients were estimated: composite reliability (CR), convergent validity (CV), divergent validity, and average variance extracted (AVE) [Bido et al., 2014], as well as the discriminant validity – using the Fornell-Larcker [Wong, 2013, p. 22] criterion. A model is acceptable if: measurable variables do not have too low values of factor loadings (i.e. >0.3) describing their correlations with the hidden variable, the CR is at least equal to 0.7, and the AVE for each latent variable >0.5 [Bido et al., 2014; Sarstedt and Cheah, 2019]. As the PLS-SEM algorithm fits the model to best estimate the parameters maximizing the explained variance of the latent variables, relatively high R² values indicate that the construct is well explained by the variables pointing to it in the paths of the structural model [Hair et al., 2014, pp. 82–168].

Considering that the common method bias (CMB) may occur when applying PLS-SEM as a result of e.g., instructions provided within the questionnaire and may affect respondents’ answers, and thus – indicators may share common variation, even if the model meets given above standard assessment criteria [Kock, 2015, pp. 2–3], further assessment was undertaken. To do so, the SPSS software was used and Harman’s single factor test for the CMB was estimated. This indicated as to whether the variance was caused by substantive reasons, and not due to the usage of the same type of scale (Likert’s scale) for each respondent [Fuller et al., 2016]. Therefore, all the metrics were loaded on a single factor, and the variance extracted was 37.555% (below the assumed maximum equal to 50.0%), which means that there is a lack of bias resulting from applying the same measurement method.

Prior to the verification of the research hypotheses, the authors ensured that the obtained regression coefficients can be interpreted. Hence, the quality of the measurement model based on the correlations between the hidden variables and their indicators was assessed, i.e., an assessment of the reliability and the convergent and divergent validity. The correlations between the reflective constructs and their measurable variables (factor loadings) and synthetic measures for individual constructs (AVE and total reliability) are presented in Table 2.

According to Table 2, individual factor loadings’ values are >0.3 – thus no variables should be excluded. All CR are >0.7 – the model is internally consistent, and measurable variables are not the same indicators because they are <0.95 [Bido et al., 2014]. To assess the discriminant validity of the model, the Fornell-Lacker criterion was applied [Wong, 2013, p. 22] (Table 3).

According to Table 3, all the square roots of the AVE are higher than the correlations of analyzed constructs with other constructs, i.e., model satisfies the criterion – discriminant validity.

Figure 2 shows the structural model of the studied relationships with standardized regression weights representing relationships between constructs’ strength, with acceptable values (“-1” to “1”). The estimates of the variance’s explained value by the remaining variables are included inside the circles for endogenous variables, i.e., dependent ones.

Figure 2.

To estimate the predictive value of the structural model, for each endogenous (hidden) variable, R² were calculated [Hair et al., 2017]. The structural model allows for accurate prediction of: EMI – with 66.5% of the variance explained), e-marketing adoption during COVID-19 pandemic both in terms of EMAP – 64.3% and EMAD – 45.2%.

Next, the bootstrapping procedure was applied (Table 4) to present total effects (sum of direct and indirect effects). The significance of the total effects should determine the hypotheses verification [Hair et al., 2017].

According to the presented tests of significance of regression weights for total effects:

DLO has weak, negative relationship with EMI [H1a – not supported];

Considering the tests of significance of regression weights in relation to the indirect effects presented in Table 4, there is no statistically significant relationship.

With reference to the determinants of e-marketing adoption by the Polish companies during the COVID-19 pandemic, various significant relations exist. DLO, DSO, EOI, EOP and EOR having positive relationships with EMAP and EMAD during COVID-19 pandemic were identified. Moreover, DLO, DSO and EOR have relationships with EMI, with positive relationships for DSO and EOR, and negative ones in the case of DLO.

Based on the empirical study on a sample of 241 Polish companies representing fashion, cosmetics, consumer electronics, and household appliances industries, the authors confirmed that DLO, DSO, EOI, and EOP have had statistically significant and positive impacts on the EMAP and EMAD during the COVID-19 pandemic. This aspect has also been supported in the previous research. From the results of the studies of Sriayudha et al. [2018], Wang [2020], and Quinton et al. [2018], those determinants positively influence companies’ digitalization, including e-marketing adoption. The identified positive impacts of DLO on DSO and EMA are also supported in the light of recent studies, e.g. by Kindermann et al. [2020], according to whom DO increases a company’s digitalization. Furthermore, these authors add that DO has important impacts on a company’s strategies and tactics, being simultaneously one of the firm’s SO domains. Considering that applying various digital technologies allows companies to position themselves ahead of competitors [Niemand et al., 2017], especially in the realities of increasing competition [Martincevic, 2021], this seems to be necessary not only from the perspective of a company’s growth, but also for their survival in general.

According to this study, EOR’s relations with both e-promotion and e-distribution adoption during the COVID-19 pandemic are negative. It means that the companies with higher tendencies toward risk-taking are simultaneously less likely to adopt e-promotion and e-distribution, although, as results from the literature show, firms with a higher propensity to take risks are also more likely to undertake different online solutions [Mathews et al., 2018]. The surprising conclusions with regard to the studied Polish companies may be due to that these medium-sized and large firms had already adopted e-promotion and e-distribution prior to the COVID-19 pandemic, and therefore had no need to implement such solutions during the ongoing pandemic and do not consider it from the prism of risk-taking. It also may explain the identified lack of a statistically significant relationship between EMI and EMAD, and EMAP – as EMI refers to the e-marketing implementation before the pandemic. Furthermore, many companies and brands representing diversified industries even before the pandemic used to have their own e-commerce (e.g., Adidas, Bosch, Lego, and Samsung) or sell via Internet intermediaries, such as multisided-digital platforms (e.g., Amazon) or online multi-brand retailers (e.g., Zalando). As a result, they may not have reported any significant changes with reference to the above as an outcome of the COVID-19 pandemic.

This study contributes to the existing body of knowledge on e-marketing and SOs by simultaneous examination of both DO and EO influence on EMI and e-marketing adoption in the context of the COVID-19 pandemic. Furthermore, it can be insightful for the decision-makers inter alia with reference to indicating the required DO and EO supporting boosting the e-marketing adoption (considering e-promotion and e-distribution) within a company for its future market success [Tsiotsou and Vlachpoulou, 2011].

One may indicate that applying the PLS-SEM method for the data analysis in this research is a limitation of this study. As some authors [e.g. Kock, 2015] note, with it, the CMB is likely to occur, even for the models that satisfy the standard assessment criteria for the confirmation factor analysis. Such biases can occur as a result of the instructions given in a survey that may have impact on respondents’ answers, and thus – final indicators may share a common variation. Nevertheless, other authors, including Schwartz et al. [2017] highlight that it is rather a kind of so-called “urban legend” as the issue of correlations between various variables measured when using the same methods (usually the case of the self-reported surveys) is simply overstated [Spector, 2006].

One of the main limitations of the survey is that the research sample was restricted to the Polish companies, however, representing different industries. It would be worth conducting a study on a sample of companies representing other countries, as well as various industries to compare the results. It would also be worthwhile to include in the further studies not only medium-sized and large companies, but also small ones, and take into account other moderators than company size and industry, e.g., year of foundation.

Moreover, only DO and EO as determinants of the e-marketing adoption were included, while the influence of, e.g., managers’ previous experience in using modern technologies cannot be excluded [Gutierrez et al., 2015]. Additionally, the impact of DO and EO on only e-promotion and e-distribution, excluding other e-marketing tools, was examined. Thus, further studies referring to the remaining marketing-mix tools, i.e., product and price, may be postulated for further investigation.