Drivers and consequences of word of mouth communication from the senders’ and receivers’ perspectives: the evidence from the Croatian adult population

Abstract This article addresses the issue of predicting and stimulating consumers’ word of mouth communication (WOM). It contributes to the WOM literature by examining and comparing the magnitude of effects of a set of antecedents and consequences of WOM information spread and WOM information seek in one model. The data collected from consumer survey in Croatia were analysed using structural equation modelling (SEM). The results reveal that consumer innovativeness, followed by price sensitivity and attitudes towards advertising are the most important drivers of both WOM variables, whereas the effects are stronger for WOM information seek than WOM information spread. The results also indicate that WOM information seek has a four times stronger effect on buying intention than WOM information spread, which can be utilised by retailers. Accordingly, retailers should focus on WOM information seek and try to stimulate it by introducing new products in the first place, but also with frequent sales, and appealing advertising messages.


Introduction
As competition in the retail market is fierce, the growth limited and expenses high, firms seek more effective ways of generating demand. In this sense, word of mouth (WOM) communication was suggested to be a more efficient communication channel, as it requires less investments than traditional advertising (Ansary & Hashim, 2018;Nielsen, 2012;Warren, 2018). WOM refers to consumer-to-consumer communication in which individuals share opinions, news, and market information with others in the forms of forwarding information (WOM information spread) and searching for information (WOM information seek) (Alsulaiman et al., 2015;Sweeney et al., 2008). The quest for more efficient types of promotion is evident nowadays, as traditional advertising is expensive and consumers tend to trust recommendations from friends and families rather than advertising.
WOM has been investigated for a long time in marketing and diffusion of innovation literature, either as a consequence of consumer behaviour or in a form of a stand-alone construct. WOM has been recognised as a communication source and a process of personal influence that impacts consumers' attitudes, decision-making and purchases (e.g., Ansary & Hashim, 2018;Mazzarol et al., 2007;Sweeney, 2018;Sweeney et al., 2008). The main challenge is, however, the difficulty of predicting WOM and there is no consensus on drivers and outcomes of WOM. The knowledge is fragmented and sometimes inconsistent, whereas some researchers argue that WOM literature lacks sound theoretical framework too (Alsulaiman et al., 2015).
Our study addresses the issue of predicting and stimulating consumers' WOM and examines the antecedents and consequences of WOM in Croatia. As antecedents we examine several individual variables (i.e., consumer innovativeness, attitudes towards advertising, price sensitivity and shopping enjoyment), and relevant demographic variables, while buying intention was taken as the outcome variable. The article contributes to WOM theory by examining and comparing the magnitude of effects of antecedents of WOM that can be influenced by retailers to some extent by employing the retail mix (e.g., new products, pricing, advertising and store atmosphere). There is no study that explored these variables in one model.
The comparison of drivers and consequences from the perspective of WOM information spread and WOM information seek in one model is a further contribution of our article. Most of WOM research focused either on a sender or a receiver of information, while very few studies have examined both perspectives in one model (e.g., Alsulaiman et al., 2015;Mowen et al., 2007;Sweeney et al., 2008). These studies were carried out in developed countries (e.g., the U.S., Korea and New Zealand), and suggest that there are differences in the drivers and effects of WOM.
A further contribution of our article is the Croatian setting and the sample of general population. Although Mowen et al. (2007) found that the drivers of WOM between the U.S. and Korean samples were quite similar, this study contained the student population only, and thus this framework needs to be further tested in cross-cultural contexts and on the sample of the adult population. As people vary in their reactions and WOM is not homogenous in its impacts (Sweeney et al., 2008), we might get some differences in WOM drivers and the effects in Croatia.
Our study contributes to the field of economics in the area of stimulating purchasing decisions and creating demand for goods and services by using WOM. It aims to provide marketing professionals with a better understanding of factors likely to influence senders and receivers of WOM messages, and WOM effects. As WOM is a part of communication mix and a complement to advertising efforts (Feng & Papatla, 2011), the results of our study might provide some recommendations on how to make marketing communication more effective by using WOM.
This article is organised as follows. In the second section, the theoretical background is presented, followed by the development of our hypotheses. The research method is described in section three. Research results are given in section four, followed by the discussion in section five. Section six concludes.

Theoretical background and hypotheses development
Several factors were shown to affect WOM, including demographics (Kundu & Rajan, 2017), individual features, perception and motivations, involvement and mood. Satisfaction, perceived value, trust, commitment, consumers' knowledge and experiences with a product, service or firm were also shown to predict WOM (Kundu & Rajan, 2017;Le et al., 2018;Konuk, 2019). Perceived risk has often been cited as the factor driving consumers' desire to seek information before purchase (Alsulaiman et al., 2015), especially when products are difficult to evaluate and uncertainty is high (Mazzarol et al., 2007;Thorbjørnsen et al., 2015). Several studies indicate that WOM might drive consumer purchases Sweeney et al., 2008). One stream of literature examines the role of WOM in the process of the adoption of new products and technologies in which early adopters have an important opinion leadership role by providing WOM in their networks (Sweeney et al., 2008).
WOM has been analysed mostly either from the perspective of information sender or receiver, while very few studies have examined both perspectives in one model, indicating some differences in the effects, depending on setting, the sample and a set of variables. Alsulaiman et al. (2015) showed that in New Zealand, utilitarian value significantly and positively drives WOM seek, but not WOM send information, whereas hedonic value had twice the effect on WOM send than on WOM seek information. Perceived risk has a stronger effect on WOM seek information, while the correlation between both WOM variables was moderately strong and positive. The study of Mowen et al. (2007) conducted on the student sample in the U.S. and Korea shows that among situational variables fashion innovativeness, shopping enjoyment and value consciousness all positively affected both WOM variables, while with respect to elemental and compound traits there are significant differences in their effects on WOM variables. These results can be interpreted, however, only from young consumers' perspective. The study of Sweeney et al. (2008) based on focus group further indicates that WOM drivers and effects depend on the nature of the sender and receiver relationship, the strength of the message and its delivery, and various personal and situational factors.
Our article fills the gap in the literature by examining the impact of relevant individual and demographic variables on WOM from the perspective of receiver and sender of information. The model also explores the impact of WOM on buying intention (see Figure 1). We assume that in the process of the exchange of the value between the seller and buyer, the buyer has control over WOM activities, whereas the seller might try to influence WOM by, e.g., introducing new products, changes in prices, advertising messages and store atmosphere. The more sellers understand the interaction between WOM behaviour and retail mix, the more they can influence this type of behaviour.

Demographic factors
Although some authors suggest that there are no gender differences in WOM (e.g., Sweeney et al., 2014), other researchers suggest that women are more willing to say positive words to others and recommend products and services to their friends then men (Loureiro & Ribeiro, 2014). They communicate more often (Kempf & Palan, 2006), and are more likely to express concern for others and exchange market information (Shen et al., 2014). On the other hand, men are more likely to dominate public discussions to build positive social image and present themselves as the experts and discuss the conflicts with a company through WOM communication in order to boost their status (Shen et al., 2014). As the receivers of information, women tend to be more receptive to WOM and more often seek advice from others (Kempf & Palan, 2006). Women have also higher risk perceptions and are more risk-averse than men (Abubakar et al., 2017). Thus: H1a: Women are more likely to engage in WOM information spread than men.
H1b: Women are more likely to engage in WOM information seek than men.
Several studies show that age has significant impact on WOM (Balaji et al., 2017), although some studies suggest the opposite (Martin & Lueg, 2013;Sweeney et al., 2014). Consumer needs, attitudes and patterns of behaviour vary with age, which impacts WOM activity differently. As shown in East et al. (2014), older consumers (aged 65 and more) experience social isolation, have fewer social interactions, concentrate on people they know well and process information at a slower speed which affects their decision-making process and has negative impacts both on spreading and seeking information. Younger consumers are, however, more likely to engage in WOM as opinion leaders and WOM information senders (e.g., Schiffman et al., 2012). According to  Wallace et al. (2009), 15-to 24-year-olds expect more personal and interactive experiences and tend to communicate more frequently on the Internet. Hence: H2a: Younger consumers are more likely to engage in WOM information spread than older consumers.
H2b: Younger consumers are more likely to engage in WOM information seek than older consumers.
Although there are studies suggesting that income is not related to WOM (Martin & Lueg, 2013;Sweeney et al., 2014), some empirical evidence shows that higherincome persons are more likely to engage in WOM as opinion leaders who spread information (e.g., Schiffman et al., 2012). They are also more likely to use the Internet, adopt new technologies earlier than the others, and get emotional enjoyment from novelty (Chen & Law, 2016;Kang et al., 2014). Enjoyment can stimulate them to spread information to others. From a receiver point of view, people with lower incomes tend to use WOM more often to reduce the risk before the purchase (Engelbertink & van Hullebusch, 2013). Hence: H3a: Higher-income consumers are more likely to engage in WOM information spread than lower-income consumers.
H3b: Higher-income consumers are less likely to engage in WOM information seek than lower-income consumers.
The relationship between education and WOM has also been less investigated. There is the evidence that people with higher education are more likely to browse the Internet and share their experiences (Chen & Law, 2016). They have more knowledge, are early adopters of new technology (Engelbertink & van Hullebusch, 2013), and are more likely to positively value usefulness, enjoyment and social influence (Kang et al., 2014). Kang et al. (2014) confirmed empirically that the link between perceived enjoyment and WOM was stronger for consumers with high school education and higher, who are more likely to spread positive WOM (Balaji et al., 2017). On the other hand, less-educated consumers do not show the proneness to process complex information (Kang et al., 2014) and exhibit a higher level of anxiety when purchasing new products. Therefore, they might be more interested to seek information in WOM communication before the purchase in order to reduce the risk of purchase. Hence: H4a: Higher educated consumers are more likely to engage in WOM information spread than less educated individuals.
H4b: Higher educated consumers are less likely to engage in WOM information seek than less educated individuals.

Consumer innovativeness
Consumer innovators are the first buyers of new products in the market; who tend to share information with others (Jordaan & Simpson, 2006). Previous studies suggest that consumer innovativeness drives both WOM spread and seek information in fashion retailing context (Mowen et al., 2007). This is because consumer innovators are opinion leaders and risk takers, and as such they can have an important role in spreading information (Jordaan & Simpson, 2006). They are also more likely to obtain information, are open to new ideas and have higher incomes (Dobre et al., 2009). Innovativeness is also related with the need for knowledge (Wood & Swait, 2002), which drives WOM seek. Thus: H5a. Consumer innovativeness is positively related to WOM spread information.
H5b. Consumer innovativeness is positively related to WOM seek information.

Shopping enjoyment
Shopping enjoyment refers to fun and joy consumers obtain from visiting stores. Excitement and shopping enjoyment were shown to positively influence both types of WOM (Mihi c & Kursan Milakovi c, 2017; Mowen et al., 2007), and emotional pleasure and arousal are strongly linked to willingness to share knowledge and engage in WOM spread (Alsulaiman et al., 2015;Loureiro & Ribeiro, 2014). People like to talk about their emotions, which triggers WOM spread. On the other hand, consumers oriented towards utilitarian values, who do not derive enjoyment from shopping, are more careful, objective, rational and more precise in processing information (Alsulaiman et al., 2015). These consumers might want to reduce the risk by seeking information. Thus: H6a. Shopping enjoyment is positively related to WOM information spread.
H6b. Shopping enjoyment is negatively related to WOM information seek.

Price sensitivity
Price sensitivity describes how consumers react to price levels (Irani & Hanzaee, 2011). Price sensitive consumers seek to obtain the best value for money and search for lower prices (Sprotles & Kendall, 1986). Low prices and bargains they found stimulate them to inform other people about this (Irani & Hanzaee, 2011;You et al., 2015). Price sensitive consumers might also seek information from the others, especially in situations where prices are higher, several alternatives exist and they do not have enough knowledge about the product and enough time to do a proper market research. There is empirical evidence that price sensitive consumers are more prone to engage in WOM activity (Kursan Milakovi c & Mihi c, 2016). Hence: H7a. Price sensitivity positively influences WOM information spread.
H7b. Price sensitivity positively influences WOM information seek.

Attitudes towards advertising
People can learn about new products through advertising and WOM communication (Lamey et al., 2018). Peluso et al. (2016) argue that advertising messages can be perceived as barriers to consumers' sense of control, and can increase the likelihood of WOM information spread. The relationship between advertising and WOM is, however, not so straightforward. As some authors suggest, advertising may not increase WOM, but can even decrease it, which depends on product type and advertising frequency (Feng & Papatla, 2011). Despite this, some authors found positive relationship between advertising and WOM (Keller & Fay, 2009). Advertising reminds people of why they like the product, and can encourage them to learn more about the brand, which triggers seeking information (Tho et al., 2016). Thus: H8a. Attitudes towards advertising positively influence WOM information spread.
H8b. Attitudes towards advertising positively influence WOM information seek.

Buying intention
Several articles suggest that WOM influences consumers' purchasing decision, feelings and attitudes before and after the purchase (Thorbjørnsen et al., 2015). When consumers are loyal to the store and are satisfied with shopping experience, they are more likely to revisit the store and purchase again (Liu & Lee, 2016), and engage in WOM spread. Listeners' desires to act on received WOM depend on their attitudes towards recommended product, service and retailer (Martin & Lueg, 2013), and those attitudes will be stronger in the case of positive WOM. If consumers are not familiar with the product, as opposed to the situation when a listener is familiar with the product, purchase intention decreases. WOM seeking reduces uncertainty and risk, while positive messages lead to enthusiasm, confidence and optimism (Sweeney et al., 2008). WOM also influences trust, which is positively related to buying intention (Mikalef et al., 2017). Thus: H9a. WOM information spread positively influences buying intention.
H9b. WOM information seek positively influences buying intention.

Research methodology
Empirical research was conducted on the representative sample of 1,000 respondents. Croatian Census of population was used as the basis for assessing the representativeness of the sample (Croatian Bureau of Statistics, 2011). The survey was carried out by market research agency using the computer-aided personal telephone interviewing (C.A.T.I.) technique. Sample representativeness was ensured by probability sampling applying the random stratified sample. For this purpose, two-stage proportional stratification according to the size of the county and estates within the particular county was used. The stratification was conducted in two stages due to the size of counties and settlements. Proportional stratified sample requires the estimation of groups (according to the selected population characteristics), whereby all units within each group are selected randomly. All groups retained the proportions that they hold within the population that is being researched. Random selection of the respondents was achieved using the 'last birthday' method within each household (focusing on the adult individuals/consumers). If the person was not available, the household was called later until the respondent was reached. Household telephone numbers were selected randomly.
Previous studies on both WOM perspectives were done in developed countries and have some limitations (such as student sample and focus groups), and thus more research is needed to examine WOM framework across countries. People vary in their reactions, WOM is not homogenous in its impacts (Sweeney et al., 2008) and some variations in WOM effects might be found in Croatia. Our survey addresses general offline retailing, which is still valid for WOM research. Although the majority of WOM is nowadays conducted via electronic channels (Abubakar et al., 2016), offline environment is still very important. Keller and Fay (2012) argue that 90% of WOM is done in an offline environment and offline channels dominate in volume and impact of conversations, thus stressing the importance for marketers in terms of acknowledging the necessity of stimulating WOM in the offline world (Keller & Fay, 2016). Authors also argue that offline conversations are more credible and more likely to lead to purchase intentions than online conversations, which justifies revisiting offline setting for WOM research.
The measurement instrument was a highly-structured questionnaire containing 36 statements measured by using Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The items taken from literature were translated from English to Croatian, and then back to English. All items used in this research are presented in Table 1.

Research results
Data was analysed using the variety of the statistical methods and tests. Data was checked for multivariate and univariate outliers, for missing values and was assessed for the univariate and multivariate normality of distribution, bivariate and multivariate collinearity, heteroscedascity, as necessary preconditions for later structural equation modelling (SEM) conduction, as suggested by Kline (2011). These tests led to the final sample of N ¼ 960, while other necessary values were adequate. Reliability, validity and unidimensionality of the used measurement scales were checked through exploratory and confirmatory factor analyses, as explained in the later corresponding parts.

Exploratory and confirmatory factor analyses
For the purpose of sample adequacy Kaiser-Meyer Olkin and Bartlett's test of sphericity was analysed. The results show that the sample is adequate for exploratory factor analysis (EFA) conduction (Table 2).
EFA was performed on 28 items, whereby all items loaded on the corresponding factors with factor loadings higher than 0.4. The principal component analysis and Varimax rotation were used, while the Eigenvalue used was higher than 1. Varimax method was used due its characteristics of maximising variance of factor loadings by making high loadings higher and low loadings lower for each factor (Tabachnik & Fidell, 2007). In addition, using the Varimax rotation, for treating the factors independently, it is possible to load the smaller number of variables highly on each factor, which can provide a better interpretation of factors (Field, 2009). The factor structure resulted in seven factors explaining 66.3% of variance. The measurement scales were tested for reliability (using Cronbach alpha coefficients) and convergent and discriminant validity, both in EFA and confirmatory factor analysis (CFA) Namely, EFA showed that all measurement scales have high and adequate Cronbach alpha coefficients. Furthermore, it showed that items had high factor loadings on the corresponding factors and low for other factors. Moreover, correlations among the factors were lower than 0.85 (Kline, 2011). These measurement scale characteristics were additionally tested in CFA For this purpose, the measurement model was created based on the principles of Kline (2011), whereby all items (manifest variables) measuring one/ corresponding latent variable were loaded on that common factor, factor loadings were high, measurement errors independent and latent factors were correlated but  Pollay and Mittal (1993) I like to look at ads/ advertising. (ATAD03) Mehta and Purvis (1995) In general, advertisements present a true picture of the product advertised. (ATAD04) Pollay and Mittal (1993) Buying not exhibiting too high absolute correlation values. The CFA indicates a good fit of model (Table 3), which also confirms that the measurement scales are unidimensional. The model is free of common method bias.
CFA showed that specified measurement model fits the data well, while factor loadings are significant and high. In addition, composite reliability (CR) and average variance extracted (AVE) values (Table 4) suggest the presence of reliability and convergent validity.
Not too high correlations among the latent factors (i.e., lower than 0.85) along with the roots of AVEs values (Table 5) suggest the presence of discriminant validity of used measurement scales.
EFA and CFA analyses suggest that all measurement scales exhibit the characteristics of reliability, internal consistency and validity, confirming that the data is reliable and usable.

Structural equation modelling
SEM was covariance based and applied according to Kline's (2011), whereby it was made sure that the sample size corresponds to the ratio of sample units' number and number of the parameters in the model, and that the structural model is identified. For the latter assumption, each latent variable had its own scale/metric, each latent variable had minimally two manifest variables and the number of parameters was less than a number of covariance matrix units. The results show that the model fits the theory very well ( Table 6).
The parameters in the model were estimated using the maximum likelihood method (ML). If the standardised structural coefficients are significant and have corresponding direction the hypotheses are accepted. Standardised structural coefficients along with the status of hypotheses are visible in Table 7.
The results show that gender significantly effects both types of WOM, which supports hypotheses H1a and H1b. Age is significant only for WOM information seek, unlike WOM information spread, which supports hypothesis H2b and rejects H2a.  Income and education do not influence WOM spread and seek, which rejects hypotheses H3a, H3b, H4a and H4b. Consumer innovativeness positively impacts both types of WOM, exhibiting a stronger influence on WOM information seek, which supports the hypotheses H5a and H5b. Shopping enjoyment positively effects WOM spread, which is not the case with WOM seek. Therefore, hypothesis H6a is supported and H6b is rejected. Our  analysis further shows that price sensitivity and attitudes towards advertising have a stronger effect on WOM seek than WOM spread. Thus, hypotheses H7a, H7b, H8a and H8b are supported. Finally, both WOM information spread and seek positively effect buying intention, whereby a greater influence can be seen in terms of WOM seek. Hence, hypotheses H9a and H9b are accepted.

Discussion
Our findings confirm the importance of consumer innovativeness, price sensitivity, attitudes towards advertising and shopping enjoyment for the WOM generation, while the effects of demographic variables appeared to be insignificant (income and education) or less important (e.g., gender and age). Overall, WOM information seek and spread have more or less similar antecedents, but the magnitude of the effect is different, which is in line with some past research (Mowen et al., 2007).
In line with the study of Mowen et al. (2007), consumer innovativeness is the most important driver and has positive effect on both types of WOM, indicating that a higher level of consumer innovativeness leads to a higher tendency to engage in WOM activity. Our results show that consumer innovativeness has a bit stronger effect on WOM information seek than WOM information spread, while the study of Mowen et al. (2007) showed the opposite. This can be explained by the fact that consumer innovators are open to new ideas, experiences and new knowledge (Jordaan & Simpson, 2006;Wood & Swait, 2002), and as such tend to seek additional information before the purchase, especially in situations when there is a higher risk related to new products. Croatian consumers have lower income than American or Korean ones and as such they will try to offset the risk more than consumers in developed countries. As expected, there is a positive effect of consumer innovativeness on WOM spread, because consumer innovators are more likely to spread information due to their enthusiasm in sharing their experiences with the others.
Price sensitivity is the second most important driver that is positively related to both types of WOM, indicating that price sensitive shoppers have a stronger tendency to engage in WOM. This effect is a bit stronger for WOM seek than WOM spread, which can be explained by the fact that price sensitive consumers want to reduce the risk by acquiring additional information about the product before the purchase. When these consumers find good sales and bargains, they tend to share this information with the others, which triggers WOM information spread. This is in line with past research (e.g., Irani & Hanzaee, 2011;You et al., 2015). In Croatia, middle income country, price sensitivity plays an important role in generating WOM, as compared to developed countries.
Our findings also show that the more consumers are involved with advertising, the more it is likely that they will engage in WOM activity. This is in line with past research (Keller & Fay, 2009). The effect is stronger for WOM seek than WOM information spread, as advertising tends to encourage consumers to learn more about the brand, which might stimulate them to seek for additional information before purchase. When campaign triggers an emotion, consumers are likely to share their experiences with the others.
In our study, shopping enjoyment is positively related only to WOM information spread, as compared to Mowen et al. (2007) whereby this variable had the same impact on both types of WOM. Our study is in line with the study of Alsulaiman et al. (2015), indicating that hedonism drives WOM information spread two times as much as WOM information seek. If consumers like to shop, it is also more likely that they will share their excitement with others.
Among demographic variables the drivers of WOM were only gender and age (only for WOM information seek). As compared to men, females are more involved in WOM than men, and their engagement is stronger for WOM information spread than seek. Females are more emotional and like to help the others with sharing information (Shen et al., 2014). They are also more risk-averse and like to seek advice from others (Abubakar et al., 2017).
Furthermore, our results show that younger consumers are more likely to engage in WOM information seek. Descriptive analysis shows that younger and older consumers are innovative and price sensitive to almost the same extent, which can be seen as a potential explanation for non-significant influence of age on WOM spread.
Our results also show that income and education are not predictors of WOM. When it comes to income, descriptive insights show that consumers are price sensitive and enjoy shopping to the same extent, i.e., regardless of the income range. In addition, both less-and more-educated consumers are equally innovative and hold positive attitudes towards advertising, which might explain the non-significant effects of income and education on WOM information spread and seek.
As for the outcomes, the results of our study show that both types of WOM influence buying intention significantly and positively. The effect of WOM information seek on buying intention is four times stronger than the impact of WOM information spread. This is because consumers tend to ask the others for additional information about the products, brands, and sales before the purchase to reduce the risk of purchase and to find a better bargain. The more consumers are engaged in WOM information seek, the more likely they will be motivated to buy the product they do not have. When WOM spread is concerned, it is possible that consumer already possesses this product which decreases buying intention. It is also possible that a consumer does not have this product, but he or she is very knowledgeable and might plan to buy the product, which also increases buying intention.

Conclusions
Our article proposes a model that examines antecedents and consequences of WOM information spread and seek. Although Mowen et al. (2007) indicate that both WOM variables might have similar antecedents, our findings indicate that some drivers are indeed the same, such as gender, consumer innovativeness, price sensitivity and attitudes towards advertising. However, the magnitude of effects is different for all drivers when we compare these two WOM variables. The impact of WOM information seek on buying intention is four time larger than WOM information spread, suggesting the importance of WOM information seek in Croatia. Our results also show that retail-based factors have stronger effect on WOM than demographic variables, among which only gender and age generate effects.
Our findings have several managerial implications. Namely, marketers need to look for smarter ways to engage customers in WOM communication. The most important thing is to be unique in terms of product offerings and advertising appeals. Taking into consideration high impact of consumer innovativeness on both types of WOM, marketers should attempt to influence and create awareness among innovators who might act as opinion leaders and send information to others through the variety of activities, such as an introduction of new and unique customised offerings combined with creative advertising and direct communication in stores. Here some benefits that would signal the exclusivity might be offered, e.g., something only for the first 30 customers or limited editions. Marketers can also provide better information to their shoppers about new products and brands through interesting, unusual advertising appeals and presentations in stores using, e.g., storefronts, window and in-store displays that will arouse additional interest among information seekers and information senders. This can be combined with appealing store atmosphere relying on colours, music, lightning, designs, staff, and layout to influence shoppers' feelings and create shopping enjoyment, which additionally contributes to WOM spread. Taking into consideration the impact of price sensitivity, it might be beneficial for marketers to additionally emphasise in advertising appeals and store promotions the rational arguments and savings, along with the frequently offering discounts and sales. Finally, marketers can organise different events designed to surprise the audience and create buzz. This study has several limitations. As the outcome of WOM we only examined buying intention, which might not necessarily lead to actual purchases. Furthermore, this research was conducted at one point of time, whereas the longitudinal approach might provide different insights due to potential consumer preference shifts.
Future research might replicate our study in online environment, which could provide additional insights into the differences between offline and online WOM communication. In addition, future research might explore cross-cultural differences in the context of WOM information spread and seek. The existing model could be also tied to specific product or service categories, while the comparison across various industries might shed some additional light on WOM generation.

Disclosure statement
No potential conflict of interest was reported by the authors.