Social media use in international marketing: Impact on brand and firm performance

In recent years, digital marketing has played an increasingly important role in firms’ international marketing and branding. The growing penetration of social media (SM) all over the world contributes to its more frequent use in marketing strategies targeted at both local and foreign customers. These activities are referred to as “social media marketing” (SMM) [Okazaki and Taylor, 2013; Tsimonis and Dimitriadis, 2014].

Scholars indicate several benefits of SMM from the company perspective, such as increased brand awareness, improved brand image, increased consumer brand engagement [Felix et al., 2016], and lower costs of promotion [Qualman, 2010]. As pointed out by Okazaki and Taylor [2013], the application of SM in international marketing, especially in promotion addressed to a foreign audience, is determined by its key features favoring interactions across geographic and cultural boundaries, such as the relative ease of creating networks and transferring brand image with SM, coupled with remarkable flexibility of use. However, studies so far on international digital marketing have focused primarily on the localization of transnational corporations and their brands’ Websites [e.g., Čermák and Smutny, 2018], while less attention has been paid to the application of SMM. Although, in recent years, a number of papers on SMM have been published, only a few have been devoted to the use of SM in international marketing, branding, and factors influencing it [Berthon et al., 2012; Okazaki and Taylor, 2013; Tsimonis and Dymitriadis, 2014]. Therefore, knowledge on the usage of SM in international marketing is still incomplete.

Moreover, referring to the concept of dynamic capabilities, many authors emphasize the role of dynamic marketing and branding capabilities, including brand orientation and customer orientation [Morgan et al., 2009; Brodie et al., 2016]. Brand-oriented firms tend to have organizational processes revolving around the creation, development, and protection of their brands in continuous interactions with target customers to achieve sustainable competitive advantages [Urde, 1999, pp. 117–118]. Past studies confirm that strong brand orientation influences international marketing strategies [Wong and Merrilees, 2007] and translates into improved BP [e.g., Lee et al., 2017], in both home and foreign markets [Ling-yee and Ogunmokun, 2001; Morgan et al., 2006]. Another focal marketing capability, namely, customer orientation, is a set of organizational behaviors and beliefs that prioritize customers’ interests and the creation of greater value for them [Rindfleisch and Moorman, 2003]. Firms with a strong customer orientation are characterized as customer-close, customer-flexible, and customer-accessible [Jeong and Hong, 2007]. Strengthening customer orientation requires the creation and use of information technology (IT) infrastructure, including such applications of SMM that support relationships with customers. Orientation toward customers is considered to be an important factor influencing firm performance [Kirca et al., 2005].

While the importance of brand and customer orientations is widely recognized in the literature, there is no thorough empirical investigation into the effects of both orientations as dynamic capabilities on a firm's engagement in international marketing and branding through SM. Similarly, not much empirical research has been published on the relationship between the application of SM in international marketing/branding and firm performance/BP in foreign markets. Addressing this research gap, the aim of our study is to determine associations along the entire chain of effects, from brand and customer orientations, through the scope of the firm's engagement in international SM brand marketing, to firm and BP in foreign markets. To enhance the validity and reliability of the findings, our conceptual model accounts for the moderating roles of content localization in SM, control over SM activities in foreign markets, involvement of local partners, and the company's capital structure by its country of origin (domestic or foreign).

Data for our study were sourced from a quantitative survey with the computer-assisted telephone interviewing (CATI) method on a sample of 245 randomly selected Polish medium and large firms that own consumer brands sold on foreign markets and use SM in international branding. Statistical analysis for testing the research hypotheses involved partial least squares structural equation modeling (PLS-SEM).

This paper is organized as follows. First, we summarize relevant literature sources and develop hypotheses. Then, we describe our research design and method. Research findings come next, to be followed by the Discussion section. The paper ends with an overview of limitations and recommendations for further studies.

This study contributes to the international marketing literature by simultaneous examination of the influence of the dynamic capabilities of a firm (such as brand orientation and customer orientation) on SM international marketing engagement and on brand and business performance in foreign markets. It also offers novel insights for brand/marketing managers into the influence of various dimensions of SMM engagement in a foreign market on brand and firm performance in this market.

According to some authors [e.g., Wong and Merrilees, 2007], relations between international marketing strategies, also those involving SM, and brand/firm performance can be considered through the lens of the resource-based view (RBV) and the concept of dynamic capabilities [Helfat and Peteraf, 2009]. According to RBV, a firm's strategy is a way of acquiring, developing, and deploying resources with the goal of achieving business success. As such, a company should develop new capabilities that enable building and strengthening of its competitive advantages [Eisenhardt and Martin, 2000] and improving its performance [Barua et al., 2004]. By comparison, the dynamic capabilities concept maintains that a firm achieves a competitive advantage and superior results on account of its dynamic capabilities, involving capacities to integrate, reconfigure, renew, and restore resources and operational competencies [e.g., Teece, 2007]. Brand orientation (with a focus on the ability to create and shape a brand) and consumer orientation (centered on developing relationships with customers), viewed as dynamic capabilities, can be positively related with brand/firm performance by fostering the capacity to use SM in marketing effectively. Thus, the conceptual context of both the RBV and dynamic capabilities helps explain the link between the company's brand and customer orientations with SM international marketing, as well as brand and firm performance in foreign markets [Morgan et al., 2004].

The conceptual model of the relations between variables illustrated in Figure 1 corresponds to the three research hypotheses and accounts for control variables to avoid the risk of spurious correlations due to the endogeneity of independent variables. The problem of endogeneity occurs when one or more independent variables are correlated with the error term, typically due to omitted independent variables [Wooldridge, 2020, pp. 496–498]. As Wooldridge explains [2020, p. 496], the lack of important independent variables in a regression model leads to biased and inconsistent estimates of regression weights, frequently of too large a magnitude, leading to overestimation of the effects of the retained predictors. Because the control variables included in the model describe various contextual factors that are not the focus of this study, we treated them as a means to improve the statistical properties of the model, without articulating their expected impacts on the model as separate hypotheses. The two arrows leading out of the rectangle with control variables represent the fact that their investigated role in the regression analysis will not only be as main effects but also as moderators of the relationships between the focal variables, which are embedded in the research hypotheses.

Figure 1

To verify the research hypotheses, we organized a CATI survey on a sample of 245 Polish medium and large enterprises, randomly selected from the general population of companies meeting the following criteria: (1) employment size of at least 50 employees; (2) headquartered in Poland; (3) owning at least one consumer brand offered on foreign markets; and (4) applying SM in brand marketing targeted at foreign customers. Data were collected in February and March 2019 by a hired professional research agency. The respondents were senior managers for SM marketing on the Polish and foreign markets. The final sample was a cross section of both relatively little-known and well-recognized companies; the latter group included such notable international names as CD PROJECT RED (one of the world's largest computer games developers), Inglot (counted among the world's most prominent manufacturers of make-up products), LPP (the owner of the Reserved fashion brand), CCC (one of the biggest shoe manufacturers in Europe with its own network of traditional and Internet stores), Wedel (the oldest chocolate brand in Poland), and Wyborowa SA (its vodka brand Wyborowa is one of the most popular in the world). The questionnaire used in our survey consisted of closed-ended questions, with Likert-scale items and dichotomous scales, complemented with a few open-ended questions. To identify eligible respondents, two filter questions were used, establishing whether the contacted firms sold consumer goods on foreign markets under own brand(s) and whether they used SM marketing tools in branding on foreign markets. The variables included in the conceptual model and hypotheses were measured, similar to earlier research by other authors, as reflective constructs using five-point Likert scales, with the end points labeled as “strongly disagree’ (1) and “strongly agree” (5). The content of each statement, together with the relevant literature sources, are presented in Table 1.

The statistical approach used to investigate evidence supporting and refuting our hypotheses was SEM with the SmartPLS software using the PLS method. PLS-SEM combines principal components factor analysis with multiple regression to identify latent variables and investigate their mutual relationships. PLS-SEM is preferred to the other popular method of covariance matrix-based SEM (CB-SEM) when the indicator variables do not follow multivariate normal distributions and the latent variables include not only reflective constructs but also formative ones [Hair et al., 2011]. Bootstrapping calculation of the confidence intervals for the estimator allows for the use of data with any kind of distribution. One major limitation of PLS-SEM is the lack of widely accepted goodness-of-fit tests and metrics for the entire model, a broad selection of which is readily available in CB-SEM. As a way around this problem, PLS-SEM relies on a number of separate metrics for specific aspects of model quality. In this paper, we followed the model validation procedure proposed by Hair et al. [2014, 2018], which involves three areas of analysis: 1)

Reliability and convergent validity were evaluated with factor loadings for individual indicators and mean factor loadings for all indicators assigned to respective latent variables, which is known in the literature as the AVE metric. We also report the coefficients of composite reliability, which are an arguably better alternative for Cronbach's alphas, not yielding upwardly biased values for constructs with larger numbers of indicators. With regard to these measures of reliability and convergent validity, a model is deemed to be acceptable if (1) its indicators (i.e., measurable variables) do not have very low correlation coefficients with their latent constructs (it is often assumed that 0.3 is the minimal admissible value), (2) every latent construct explains at least 50% of the variance in its indicator variables, i.e., all AVEs should be >0.5, and (3) the composite reliability coefficients for each construct are >0.7.

Prior to verifying the hypotheses, it is essential to ensure the interpretability of the regression weights in the structural model. This is contingent on meeting several quality criteria by the measurement model with regard to reliability and validity. As described in the preceding section (Section 3), the procedure for investigating reliability and validity explores the correlational patterns between the latent variables and their indicators, as well as among the latent variables themselves. Table 1 shows the individual factor loadings equivalent to the correlations between constructs and their metrics, together with the synthetic measures of AVE and composite reliability.”

Factor loadings with sufficiently high values imply that no indicator variable should be removed from the measurement model on account of measuring something else than its assigned construct. The average amounts of variance extracted by the latent variables from their respective sets of indicators are all beyond 0.50, which indicates that at least 50% of the information held by the indicators was retained by the latent variables. This allows for interpreting the constructs based on the meaning of their indicators. Each coefficient of composite reliability is greater than its recommended threshold of 0.7, suggesting acceptable levels of reliability. Another important aspect of structural model quality is its discriminant validity, assessed here with the Fornell–Larker approach (Table 2).

It is clear from Table 2 that discriminant validity is not an issue since the average correlations between the constructs and their metrics (along the diagonal) are all greater than the correlations with other constructs (off the diagonal). Thus, there is no risk that the latent variables share too much of the same content, which would present both interpretational and statistical difficulties. It should be stressed that the analyses above pertain only to reflective latent constructs, where the presumed direction of causality leads from the construct to its indicators and the indicators themselves are supposed to be strongly correlated with each other. By contrast, formative constructs are composed of metrics which “form” and not “reflect” the construct, so the directionality here goes from the metrics to the construct. Moreover, formative metrics do not have to be mutually correlated, as they often represent distinct aspects of the latent variable. Therefore, quality assessment of formative constructs should not involve the above-described procedure based on shared correlational patterns. The current research involves only one formative construct, namely, BP, which is the index of possible positive outcomes of using SM in international marketing. It was designed as the mean of six probable effects of SM on brand image and sales, measured on a scale of 1–5, according to their observed intensity. Since all six effects could occur and develop largely independently of each other (at least in the short-to-medium term), the choice of the formative format seems to be justified. Quality assessment of formative constructs typically involves studying the so-called face validity, which does not entail any statistical procedures. Instead, face validity is determined by content analysis of each item (based on common sense and specialist knowledge – here, in management, marketing, and economics) to find out if they indeed belong together as different elements of the same latent variable. Based on the proposal by Wong and Merrilees [2008], we applied the following component items of the formative BP index measuring the possible positive outcomes of SMM in foreign markets (respondents were informed that all statements concern what could have happened with their key brand in the most important foreign market in the past year; scale structured according to a five-point Likert format, where 1 – complete disagreement, 5 – complete agreement with each statement):

The image of our key brand has improved.

To conclude, the measurement model demonstrates adequate levels of validity and reliability, thus enabling further analysis of regression weights in the structural model and testing of the research hypotheses.

Figure 2 shows the estimated structural model with standardized regression weights (ranging from −1 to +1), representing the strength of the associations between the constructs. Circles for endogenous (dependent) variables contain estimates of the amounts of variance explained by all connected independent variables.

Figure 2

The centerpiece of the structural model is BP, which is the main endogenous construct with most of the regression lines connecting to it. By explaining 74% of the variance, the model demonstrates a high capacity for predicting BP.

Return on assets (ROA) is another endogenous variable and serves as the measure of overall firm performance. It is the only variable that was not measured via the survey method but was obtained from an external, third-party-validated database called EMIS. The link between BP and ROA is weak, but significant. This is not unexpected, since BP is quite narrow in scope, considering only one aspect of operational performance – though the most important – i.e., brand. In comparison, ROA encompasses the whole of organizational performance, including debt, unexpected events, and taxes, while accounting for the efficiency of using assets. Moreover, the full impacts of changes in BP would be registered by ROA with some delay, possibly of >1 year. However, the existence of a positive, statistically significant relationship provides evidence for the nomological validity of our BP index. Table 3 displays the outcomes of the significance tests for all regression weights in the model. Shades of gray are used to improve readability, with light gray indicating weaker effects, significant at the 0.10 level, and dark gray highlighting stronger patterns significant at the 0.05 level.

The above test results demonstrate that, of the four SM engagement areas, only two are significantly and directly correlated with BP at the 0.05 level: brand equity creation and sales support. However, this is not the full picture of the correlational patterns in the model because, in addition to the main effects of SM engagement, their interactions with other variables should also be investigated. In order to not increase multicollinearity unnecessarily, which would raise standard errors, both the diagram and the table include only interactions significant at the ≤0.1 level. The interactions were obtained by multiplying pairs of pertinent variables and indicate how the regression weight of one variable changes depending on the level of the other variable. In other words, a significant interaction reveals that regression weights for the interacting variables are not constant (and significant) across all members and subgroups of the study population. Considering the significant interactions in the model, one could make the following interpretations: 1)

The effect of SM engagement in market research could turn out to be significant for those companies that demonstrate strong customer orientation (beta = 0.089) and do not involve partners (e.g., influencers, bloggers, and other SM personalities) in their foreign SMM activities (beta = −0.152). The negative consequence of partner involvement for benefits from marketing research might be due to the limited access to complete data documenting their contacts with customers. This underscores the importance of a firm's direct relationships with its customers, with as limited participation of intermediaries as possible.

Overall, our findings validate the notion that the use of SM is advantageous for both the brand itself and the firm at large. Based on the regression weights for direct effects, some applications of SM seem to be more effective than others (brand equity creation and sales support), but even when the direct effect is insignificant, SM use could still be beneficial through how interactions from other variables boost its positive outcomes. SM engagement in information dissemination and market research shows such a pattern of relationships with BP, without meaningful effect on its own, but potentially beneficial with particular configurations of other variables describing a company's strategy on foreign markets (i.e., high customer orientation, low local partner involvement, and high content localization).

As such, the first research hypothesis (H1) about a positive relationship of SM engagement in international brand marketing with BP and firm performance can be considered supported.

In the conceptual model for the study, it was assumed that engagement in SM and its outcomes will be driven by two major aspects of a firm's organizational culture: brand orientation and customer orientation. This idea was formally expressed as hypotheses H2 and H3.

The research findings reveal a lack of relationships between brand orientation and any form of SMM engagement. There is also no evidence for either direct or indirect impact of brand orientation on BP, controlling for the other latent variables in the structural model. This lack of relationship could be due to a high level of multicollinearity in the model, but this is not the case since the only other predictor of SMM engagement – customer orientation – was only mildly correlated with brand orientation (r = 0.545 in Table 2). As will be explained later, there were differences between Polish-owned companies (PCs) and companies with foreign capital (FCs) regarding the role of BP in the model, but this did not change the outcomes at the pooled level (when the whole sample of companies was accounted for). Thus, H2, anticipating a positive link between brand orientation, SM engagement, BP, and firm performance seems to be false.

It is quite different with the role of customer orientation in the model. Regression weights point to the pattern whereby the stronger the customer orientation, the stronger is a firm's engagement in SM, with the sole exception of sales support. To assess the second part of H3, which anticipates a positive link between customer orientation and performance metrics for the brand and the firm, it is necessary to not only look at direct regression links but also consider longer, indirect regression paths going through SM engagement. In the following list, all variables considered as determinants of BP in our model are tabulated according to the absolute strength of the total effect on BP in descending order.

The above list shows the metrics of total associations between pairs of variables in the model, which combine direct and indirect effects along all regression paths. From this arguably more complete viewpoint, the total effect of customer orientation on BP is significant and positive, with the beta metric at 0.185. However, the link between BP and firm performance is so weak that any indirect regression path leading to firm performance considered in this study is insignificant. In conclusion, our data provide partial support to H3, except SM engagement in sales support, and with the additional caveat that it applies to BP but not to firm performance.

An interesting complement of the analysis so far is a comparison between companies with only Polish capital (n = 181) and those controlled fully or partially by foreign owners (n = 64). Table 4 shows the regression weights for these two groups of firms.

The general conclusion from Table 4 is that FCs use SMM in international markets more effectively than their Polish-owned counterparts (PCs). It is highlighted by the significantly higher regression coefficients for the FCs for the relationship between SMM engagement in brand equity creation and BP, with all other applications of SMM showing similar associations with BP in both groups of firms. Among other meaningful differences, the FCs were less sensitive than PCs to the involvement of foreign partners in SMM. For the FCs, brand orientation had significant positive correlations with two aspects of SMM engagement (information dissemination and market research), which was not true for the PCs. FCs and PCs displayed different patterns of relationships between customer orientation and BP, with the former having a stronger direct link, while the latter group was characterized by mediated associations through SMM engagement. The differences between FCs and PCs revealed by this comparative analysis could be explained by FCs’ better knowledge of international markets, especially those co-owned by foreign firms from the same industry, and higher marketing sophistication, possibly due to their links with multinational corporations. Operating within the capital group of a multinational gives them access to rich resources of know-how, but also it might require following certain corporate procedures and implementing elements of the general marketing strategy. This seems to be reflected in the fact that FCs had systematically higher levels of brand orientation (z-statistic for the Mann–Whitney U-test was −2.192, p = 0.028), which was the only significant difference between the two groups of firms for all latent variables in the model. A lower average brand orientation might be the reason why PCs do not show positive correlations of this variable with SMM engagement and BP. This could happen if the positive effects become detectable only after a certain critical threshold has been crossed, which apparently was not the case for the Polish firms. The positive correlation between brand orientation and the two aspects of SMM engagement in FCs is in agreement with the literature implying the existence of such a relationship. The lack of a similar relationship in PCs is turning this effect insignificant at the pooled sample level, as there were more than twice as many PCs than FCs in our study.

The primary aim of our research was to investigate how engagement of a company's main brand in SMM in a key foreign market is related to its BP and firm performance. The outcomes suggest that SMM engagement is generally advantageous, but not all areas of engagement are equally important. This conclusion is in line with the research by Felix et al. [2016], Ainin et al. [2015], and Alarcón-del-Amo and Rialp [2015, 2016] confirming a positive impact of branding in SM on BP and firm performance.

What comes to the fore, as the most important driver of BP, is a firm's SMM engagement in brand equity creation, including shaping a favorable brand image [Kim and Ko, 2010; Felix et al., 2016]. It is also beneficial to use SM in supporting the sales process, which involves not only the transaction itself, but also after-sales services, including complaint handling [Bhanot, 2012]. The other two SMM functionalities – market research and information dissemination – can be beneficial under certain conditions, as indicated by statistically significant moderation effects. This conclusion is compatible with the results of Nguyen et al.'s [2015] work validating the influence of knowledge acquisition from SM on brand innovation, which in turn may help improve BP and firm performance. Similar outcomes about the role of SM and information dissemination were reported by Bagozzi and Dholakia [2006] and Trainor [2012]. Our research suggests that it could be particularly dangerous to rely too much on so-called Internet personalities (also known as influencers), whose excessive involvement in international markets can severely degrade the positive impacts obtained from SMM engagement in creating brand equity and, in addition, reduce the value of data and knowledge flowing from the company's SM. This might be related to a certain fickleness of these popular figures and their tendency for self-promotion even at the expense of the brands that they have committed to support. Zaborek and Mazur [2018] describe an illustrative case involving a Master Chef Competition participant to promote brands of a certain Polish food manufacturer. A series of YouTube videos sponsored by the firm, even though popular with Internet users, were promotionally underwhelming due to the lack of proper exposure of the brand names, which went largely unnoticed by the viewers. It is worth noting that this negative effect from involving independent partners to support SMM in international markets is much weaker for FCs, with these also attaining greater benefits from using SMM in creating brand image. Considering that firms with international ownership ties should display a greater familiarity with foreign markets than solely PCs, their choice of local partners should also be more suitable to the sensitivities of the targeted consumer segments. This could reflect a more general issue of inadequate cultural adjustment of SM content to the specificity of foreign markets’ national culture [Okazaki and Rivas, 2002], content type (rational, emotional, and transactional content) [Shahbaznezhad et al., 2021], and the language of brand messages [Pezzuti et al., 2021]. This conclusion is reinforced in our study by a negative effect of too much centralization of SMM in markets that are foreign to the brand owner. Excessive centralization appears to reduce the effectiveness of SMM, but this is less pronounced among foreign-owned enterprises, where the centralization effect is marginally positive. Such a pattern is not unexpected and reflects earlier research by Araujo and Neijens [2012], as well as Felix et al. [2017], who found that foreign-owned enterprises, which are usually subsidiaries of global brand owners, may prefer global-reach SM and SMM standardization that corresponds with the centralized way these activities are organized.

The above findings enable us to propose recommendations for managers who would wish to use SM brand marketing abroad more successfully. First of all, SM seems to be an effective tool for shaping brand image and supporting sales, but it is important to have an intimate knowledge of local market conditions. When faced with the alternatives of either relying on its own local branches or getting local partners involved, the better option seems to be upgrading the company's local presence and giving it more autonomy as regards SM activities. Overall, independent partners can disturb the effective use of SM, so the participation of intermediaries and subcontractors should be limited. It is also important to cultivate the right organizational culture among employees – in particular, promoting a strong focus on customers and their needs, as well as encouraging frequent contacts with them from the widest possible groups of employees. This should contribute to creating more suitable conditions for value cocreation with consumers, not being only limited to the task of customization but also involving them in developing new offerings, which was found to improve business performance at both operational and financial levels [Zaborek and Mazur, 2019].

The main limitation of this study stems from the data collection method used, which involved telephone interviews with business managers. This approach can be a source of misrepresentations and inaccuracies related to imperfect memory of respondents, their lack of sufficient knowledge about the organization, or their confabulations to give answers that are considered socially acceptable. To reduce the associated measurement errors, we made every effort to carefully design the questionnaire based on validated measurement scales taken from reliable scientific publications. The questionnaire was thoroughly pretested, and the entire data collection process was outsourced to a professional research agency with extensive experience in similar projects. It is also worth mentioning that our conclusions concerning FCs are less reliable than those regarding businesses with only Polish ownership. This is due to the relatively small size of this subgroup, consisting of only 64 units. Considering how important differences between firms with different ownership turned out to be, future research on the same topic should ensure oversampling of FCs in order to conduct comparative analyses with more scholarly rigor.