Linking social media marketing to restaurant performance – the moderating role of advertising expenditure

Abstract

Purpose

Built upon customer engagement marketing theory and uses and gratification theory, this study examines the link between individual social media marketing (SMM) performance indicators and restaurant sales performance at the firm level. Moreover, the study investigates the moderating effect of advertising expenditure on this proposed relationship.

Design/methodology/approach

Random effect regression models were developed in Stata to examine the associations between SMM performance indicators, advertising expenditure, and restaurant firm revenue. Twelve years of SMM data from brands' Facebook pages were collected with a web scraper built in Python. Natural language processing was used to analyze the sentiment of user-generated content (UGC).

Findings

The results suggest that restaurant annual sales revenue increases as the volume of brand posts, “like”s, “share”s and positive comments on restaurants' Facebook pages increase. However, the total number of comments and the number of negative comments show non-significant associations with revenue. Firm advertising expenditure negatively moderates the relationships between sales revenue and the number of “like”s, “share”s, total comments and positive comments.

Practical implications

Restaurants benefit from making frequent posts on SNSs. Promotions that motivate online users to “like”, share, and comment on brand posts should be implemented. Firms with limited advertising budgets are encouraged to actively create buzz on SNSs due to evidenced stronger effects of UGC on sales performance than large advertisers.

Originality/value

This research bridges the gap by studying the effects of individual SMM performance indicators on restaurant financial outcomes. The findings support the effectiveness of SMM; and, for the first time, demonstrate that SMM could generate a more profound impact for firms with low advertising budgets.

Citation

Han, W., Ozdemir, O. and Agarwal, S. (2024), "Linking social media marketing to restaurant performance – the moderating role of advertising expenditure", Journal of Hospitality and Tourism Insights, Vol. 7 No. 4, pp. 1852-1870. https://doi.org/10.1108/JHTI-03-2023-0217

1. Introduction

Social networking sites (SNSs) such as Facebook and Twitter are popular marketing tools for businesses to enhance brand awareness and reputation (Baber and Baber, 2022). SNSs accelerate the spread of word-of-mouth (Alalwan et al., 2017), and the strategy to achieve consumer engagement through SNSs is known as social media marketing (SMM) (Choi et al., 2016). While restaurant operators use SNSs to promote food aesthetics and ambiance (Gambetti and Han, 2022), restaurant-goers rely on SNSs to search for information and share their dining experiences (Kim and Jang, 2018). As such, restauranteurs commonly adopt SMM for its effective customer reach and lower cost than mass media advertising (Kim et al., 2015).

Existing research findings support the positive effects of social media on consumer purchasing patterns and hotel performance (Ampountolas et al., 2019). Hotel research studying third-party review websites and SNSs finds that user-generated content (UGC) increases hotel value and financial performance (Aluri et al., 2016; Viglia et al., 2016). Despite the revealed importance in the hotel sector, SMM has been rarely studied for its effects on restaurant financial outcomes until recent years. Among the limited research, Wang et al. (2021) studied the financial impact of TripAdvisor reviews for restaurants in Iowa and found that review volume and rating scores influenced the annual profits of small and mid-sized restaurants. Interestingly, Luca (2016) found that the increase in star rating boosted the revenue of independent restaurants but had a non-significant impact on chain restaurants. The rationale is that consumers tend to rely on UGC when seeking information about an unfamiliar restaurant (i.e. food quality, price level, service quality), while depending more on their past experiences when deciding whether to visit a familiar brand (Wang et al., 2021).

Compared to online review platforms such as Yelp and TripAdvisor, SNSs excel in facilitating two-way communications between consumers and the brand (Michopoulou and Moisa, 2019). On SNSs, users can comment on brand posts and share the posts with friends. Some brands also selectively reply to the comments to provide users with more interactive experiences. In addition, SNSs are great tools for building customer loyalty, which the restaurant business relies heavily on (Han et al., 2018). However, only a handful of studies have examined the effect of SNSs on restaurant financial performance. For example, Kim et al. (2015) found that the level of participation in SMM directly impacts restaurant firms' value. They examined a single restaurant social media index score, and their sample covered 11 quarterly data totaling 178 analysis units. As their data featured a relatively short span, and the influence of various SMM performance indicators was yet to reveal, Kim et al. (2015) called for an examination of more complex metrics on a more expanded period of data to enrich the understanding of the relationship between SMM and restaurant performance. More recently, Li et al. (2021a) measured SMM performance of one casual restaurant chain and identified the significant impact of audience SNS engagement on sales, guest counts, and purchase frequency. However, as cross-sectional data were used and their sample only covered one casual-dining brand in the U.S., more empirical evidence is needed to generalize their research findings.

Overall, existing literature suggests that more research is needed to examine the interplay between individual SMM performance indicators and restaurant financial outcomes. This paper contributes to this line of research by examining the associations between SMM performance and publicly traded U.S. restaurants' annual revenue. Web data-mining technique is used to collect primary data from the most used SNS – Facebook. We gauge SMM performance generated by firms and consumers. In addition to assessing the volume measures, including the number of “like”s, “share”s, and comments, we identify consumer sentiment through natural language processing and examine how positive and negative valence comments link to firms' sales performance. To advance the understanding of these associations, we examine the effect of advertising expenditure in moderating the proposed direct relationships. To our best knowledge, this is the first study that examines the link between SMM and restaurant firm-level sales using multi-dimensional performance indicators. In addition, our application of web data scraping and sentiment analysis contributes to the emerging methodological trend in studying online marketplaces (Han and Anderson, 2021). The findings of this study identify SMM performance indicators that restaurant firms should monitor closely, especially for firms with low advertising budgets.

2. Literature review

2.1 Social media marketing performance metrics-audience engagement

Social media is a credible information source that consumers refer to when making a purchase decision, especially for younger generations (Alalwan et al., 2017). Consumers' experience with social media influences their penchant for buying and thus affects brands' financial performance (Ampountolas et al., 2019). SMM can be defined as a social and manageable process that aims to enhance customer engagement in the products offered by the business (Chan and Guillet, 2011). It enables two-way interactions and value co-creation from user-generated content. SMM strategies are rooted in customer engagement marketing theory, which describes the phenomenon where firms initiate and manage marketing activities to motivate and empower customers' voluntary contributions to the company's marketing functions (Harmeling et al., 2017). Engagement marketing emphasizes building business-to-customer rapport through interactions to elicit value co-creation from customers. SMM is an example of engagement marketing, as it is customer-centric. The objectives of SMM are achieved through firm-generated content (FGC) and user-generated content (UGC), which are produced by business operators and consumers, respectively (Li et al., 2021b). Customer engagement on social media can be understood through the lens of uses and gratification (U&G) theory, which argues that people use media for functional purposes and to meet their intrinsic needs (Katz, 1959). U&G theory considers the audience as active media users rather than passive information receivers (Ruggiero, 2000). Convenience, information, and self-expression are the gratification factors that impact SNS users' behavioral intentions (Choi et al., 2016). Users visit brand SNSs to gather insightful information to assist their decision-making. In addition, self-expression motivates users to engage in conversations on SNSs, which directly triggers UGC. The convenience of use is a fundamental feature of SNS that the audience desire (Srinivasan et al., 2002), and Facebook proved its ease of use by having 2 billion daily active users by 2022 (Meta, 2023). Facebook is the most popular SNS for restaurateurs, and it is also the most examined SNS in marketing research (Alalwan et al., 2017). A survey of U.S. respondents shows that the primary motivation for Facebook use is sharing information and connecting with real-life friends, while other SNS, such as Instagram, Snapchat, and Tik Tok, are more likely to be used for self-expression and making connections to strangers (Kwok et al., 2022). Therefore, Facebook excels at building connections and establishing trust compared to other platforms. Indeed, over 93% of restaurant managers in Needles and Thompson's (2013) report indicated using Facebook for SMM initiatives.

Marketing and hospitality scholars identified the positive effect of SMM on customer spending, brand sales, and firm value as it boosts consumer demand for the marketed brand or product (Hu et al., 2008). However, they measured SMM efforts using integrated index scores (Kim et al., 2015) or customer surveys (Kumar et al., 2016). There is scant evidence regarding how individual SMM performance indicators are associated with overall sales. In this regard, the need to deploy actual social media performance data arises in studying the effectiveness of SMM on brand revenue.

Marketing research has extensively studied SMM initiatives and developed metrics to gauge SMM performance. The two most common metrics are audience engagement and sentiment (Etlinger, 2011; Michopoulou and Moisa, 2019). Engagement can be defined as the extent to which audiences' attention is held onto the brand, and it is quantified with volume measures, including the number of followers, “like”s, “share”s, and comments on SNSs (Etlinger, 2011; Michopoulou and Moisa, 2019). Audience engagement is the direct outcome of FGC, as repeat exposure to marketing stimuli increases brand familiarity and enhances consumer engagement with the brand (Ananda et al., 2019). Firms' SMM performance is partly reflected by audience engagement and eventually impacts purchases (Murdough, 2009). Therefore, we propose the following hypotheses:

H1.

There is a positive association between firm sales performance and (a) the number of brand posts on restaurants' Facebook pages, (b) the number of “like”s, (c) “share”s, (d) comments under restaurants' Facebook posts.

2.2 Social media marketing performance metrics-consumer sentiment

Criticism arises when marketers only emphasize volume measures because they cannot fully capture the effectiveness of SMM in achieving marketing objectives (Lovett, 2011). A successful marketing strategy aims to increase brand exposure and brand favorability. Since consumer attitudes and feelings for the brand are critical in developing brand favorability, it is essential to incorporate consumer sentiment as an indicator of SMM performance (Etlinger, 2011). Although the number of comments reflects the popularity of brand posts (de Vries et al., 2012), a large volume of comments does not necessarily translate into favorability, as consumer sentiments are not congruently positive (Li et al., 2021b). Sentiment represents audience attitudes and judgments toward the brand, reflected by the valence of online comments (Etlinger, 2011).

Informational or entertaining FGCs stimulate SNS users to consume, spread, and produce brand-related content (Muntinga et al., 2011). Unlike engagement metrics such as “like” and “share”, the comments posted by the SNS users are a joint outcome of FGC, their experience with the brand, and the existing comments from others. Due to individual differences, consumers could hold different attitudes regarding the same information, and prior attitude serves as an anchor when individuals evaluate new information. Therefore, people with different experiences, thus holding varied attitudes towards the brand, could react distinctly towards the same FGC. Besides, the valence of UGC can also be affected by the existing comments. Positive (versus neutral) comments could trigger empathy and inspire users to post similar content. In contrast, negative (versus neutral) comments could encourage the posting of concurring or dissenting opinions (de Vries et al., 2012).

Within the past ten years, companies have increasingly embedded social media into integrated marketing communication. SMM elicits multi-way communications among the brand, fans, and friends of the fans. Popular SNSs (i.e. Facebook and Twitter) have a function that enables a user's friends to see his or her “like”s and comments on brand pages. The value of UGC is profound when page followers or commenters extend brand-related conversations to their friends (Lipsman et al., 2012). Empirical evidence suggests that while favorable UGC benefits the brands, unfavorable UGC reduces customer visits and harms firm profitability (Han et al., 2018; Hu et al., 2008). Therefore, it is necessary to distinguish the valence of UGC when examining its impact. The current study measures consumer sentiment with the valence of UGC, which is gauged by the number of positive/negative comments posted by Facebook users (Murdough, 2009). Since positive/negative consumer sentiment leads to stronger/weaker purchase intentions, we propose the following hypotheses:

H2a.

There is a positive association between firms' sales performance and the number of positive comments on restaurants' Facebook pages.

H2b.

There is a negative association between firms' sales performance and the number of negative comments on restaurants' Facebook pages.

2.3 Moderating effect of advertising expenditure

Consumer-oriented factors such as reviewers' characteristics and reviews' quality have been found to moderate the relationship between UGC and sales performance (Ghose and Ipeirotis, 2011; Hu et al., 2008). However, research attention has rarely been paid to studying firm-oriented moderators. While SMM is increasingly implemented, advertising remains a key marketing activity for restaurants. Advertising is a sponsored (paid) communication tool commonly used to promote products and increase brand exposure. Advertising literature suggests that exposure is the initial step when consumers process brand-related information (Hovland, 1957). This notion is conceptualized with the mere exposure effect, which states that individuals develop a favorability for familiar stimuli compared to novel ones (Zajonc, 1968). Hospitality research has identified the positive impact of advertising investment on corporate financial performance. Advertising helps create a favorable image to boost brand reputation, which positively influences consumer dining intentions (Ryu et al., 2012) and restaurant sales performance (Kim and Kim, 2005). Prior research also evidences the moderating effects of advertising expenditure. For example, consumers' perceived product value has a stronger impact on their repurchase intentions for brands with lower advertising investment (Ou et al., 2017). Similarly, brands that spend less on advertising benefit more by engaging in corporate social responsibility (CSR) activities (van Doorn et al., 2017).

Advertising differs from SMM in that the former initiates one-way communication, and the latter emphasizes the interactions among media users (Chan and Guillet, 2011). In other words, the collaboration of UGC distinguishes SMM from traditional advertising. In practice, social media management costs are low and are reported as part of the digital marketing expense, separated from advertising expenditure for large-scale firms (Cawley, 2021). Given the empirical evidence regarding the effect of advertising on marketing effectiveness, the current study proposes that advertising expenditure moderates the association between SMM performance and firms' sales revenue. Accordingly, the following hypotheses are developed. In addition, the research framework is depicted in Figure 1.

H3.

Advertising expenditure moderates the association between firms' sales performance and (a) the number of posts, (b) “like”s, (c) “share”s, (d) comments, (e) positive comments, (f) negative comments on restaurants' Facebook page so that the association is stronger for firms with lower advertising expenditure.

3. Methodology

3.1 Sample and procedure

Due to the availability of financial data, we only include publicly-traded U.S. restaurant firms (SIC: 5812) in our sample. We cross-check to screen out firms that do not operate Facebook pages. The final sample includes 39 restaurant firms' data from 2009 to 2020. Because each firm's IPO date and the date of Facebook page creation are different, the sample periods for every firm are not completely identical. The annual financial data is derived from COMPUSTAT, and the SMM data is retrieved from each brand's official Facebook pages. We gather SMM data in two steps. Firstly, data crawling on Facebook pages is performed to collect SMM performance data generated between January 1, 2009, and December 31, 2020. A web crawler written in Python based on the Scrapy framework (Myers and McGuffee, 2015) is used to collect the total number of posts, “like”s, “share”s, and comments for each brand within each observation period. Secondly, we collect the most relevant comments under the posts, which are identified with the comment ranking algorithm implemented by Facebook. The ranking algorithm aims to show the most liked, replied to, and highest-quality comments at the top of the comment list (Dhaoui et al., 2017). Due to the vast number of comments accumulated under most posts, a random sample of 400 most relevant comments is extracted per firm each year (Kaur et al., 2019). The extracted comments are processed with sentiment analysis using Valence Aware Dictionary and Sentiment Reasoner (VADER) in Python to acquire the ratio of each sentiment (Hutto and Gilbert, 2014). Using this tool, we first assign each word in the sentence an individual polarity score on a scale from −4 (most negative) to +4 (most positive). We then combine the individual polarity scores to derive a composite score ranging from −1 (most negative) to 1 (most positive). Hence, comments with composite polarity scores between 0 and 1 are classified as positive; those lying between −1 and 0 are considered negative, and those with a 0 polarity score are considered neutral. The numbers of positive and negative comments are estimated based on the total number of comments and the ratio derived from sentiment analysis.

3.2 Measurement

The independent variables in this study include SMM performance indicators measured by audience engagement and consumer sentiment. We use the annual number of brand posts (FGC), “like”s (Like), “share”s (Share), and comments (Com) on Facebook pages to gauge audience engagement. In addition, the numbers of positive (Pos) and negative (Neg) comments left under firms' Facebook posts are used to represent consumer sentiment. The dependent variable is firm financial performance. Firm performance studies in the business domain adopt sales revenue as one of the fundamental indicators of financial performance (Eggert et al., 2014; Omondi-Ochieng, 2019; Paniagua et al., 2020). Sales volume is affected by consumer preferences and firms' marketing efforts (Paniagua et al., 2020). It measures the direct economic impact of UGC from the sellers' perspective (Ghose and Ipeirotis, 2011). Therefore, revenue is superior to other accounting measures, such as ROA and ROE, in examining the effects of SMM. Prior research studying SMM has used the profit of the restaurant group (Wang et al., 2021) and sales revenue (Hu et al., 2008; Kumar et al., 2013) to gauge firm financial performance. Hence, we adopt sales revenue (Rev) to scale a firm's financial performance because it quantifies financial effectiveness (Omondi-Ochieng, 2019). The dependent variable Rev_it is defined as the annual gross revenue earned by firm i at time t.

This study examines the moderation effect of advertising expenditure. Advertising expense (Ad) data are retrieved from COMPUSTAT and are log-transformed in the regression analysis to alleviate the potential violation of the normality assumption. The U.S. Securities and Exchange Commission (SEC) defines advertising expenses reported in SEC filings as costs on promotional activities that are spent to stimulate the consumption of goods or services, and the costs include those incurred from the use of commercial media, mailings, billboards, and brochures (US Securities and Exchange Commission, 2001). Correspondingly, the footnotes to the firms' consolidated financial statements in 10-k filings specified that advertising expense primarily consists of agent service fees, advertising content production fees, and sponsorship fees on commercial media such as television and radio (McDonald's Corporation, 2016; Denny's Corporation, 2021).

Based on existing literature, this study controls the firm size, financial leverage, firm age, franchising condition, restaurant type, internationalization, year effect, GDP percentage change from the preceding period (GDP), and consumer price index (CPI) in the analyses. The size of a firm could affect financial return because large companies tend to get better deals from the supplier and therefore keep their expenses lower (Kim et al., 2015). Firm size was found to be positively associated with restaurant firms' financial performance (Koh et al., 2009). This study scale firm size by the log of total assets. A firm's investment risk can be represented by its capital structure, which is the financial leverage calculated as the ratio of total liabilities to total assets. A high leverage ratio makes investors doubt the firm's capability to pay new debt and weakens the firm's ability to borrow capital (Choi and Lee, 2018). Therefore, this study controls firms' risk level for its potential impact on business expansion and revenue generation. Firms with a long history could benefit from previous operating experience in brand building, customer retention, and cost control (Madanoglu et al., 2011). The current study controls firm age calculated by subtracting the year a firm was established from the year being analyzed. Franchising lowers the risk of a service firm and enhances its competitive advantage (Hua and Dalbor, 2013). In the US, over half of fast restaurants are franchises (US Census Bureau, 2018). Franchising condition was found to influence restaurant firms' value, stock price, and risk-adjusted financial performance (Hua and Dalbor, 2013; Koh et al., 2009; Madanoglu et al., 2011). The current study incorporates the franchising effect with a dummy variable where 1 represents franchised firms and 0 otherwise. Different types of restaurants manage their operating expenses differently, and full-service restaurant firms were found to perform better than limited-service restaurant firms in the internationalization process (Rhou and Koh, 2014). The restaurant type incorporated in this study is indicated by a dummy variable where 1 for full-service restaurants and 0 for limited-service restaurants. Internationalization relates to whether a firm operates its business across national borders, and it impacts organizational expense, idiosyncratic risk, overall value, and financial performance (Ozdemir et al., 2020). Following Hua and Lee (2014), internationalization in this study is expressed as a dummy variable, which is coded 1 if the firm reports foreign pretax income in a given year and 0 otherwise. In addition, we include year dummies to account for time-variant effects that might affect all restaurant firms in the same manner (Wang et al., 2021).

During the sampling period, the U.S. economy experienced two major downturns due to the global financial crisis of 2007–2008 and the COVID-19 recession in 2020. These incidents severely impacted national employment, GDP, the stock market, and consumer expenditure in the hospitality sector (Chowdhury et al., 2021). Therefore, this study controls macroeconomic prosperity reflected by GDP variation and the inflation captured by CPI (Wang et al., 2021). GDP statistics are acquired from the Federal Reserve Bank of St. Louis, while CPI data are retrieved from the Bureau of Economic Analysis database.

3.3 Data analysis

Regression analyses are performed on the panel data that covers 12 years of annual observations over a cross-section of restaurant firms. Two sets of empirical models are used to test the effect of SMM performance on firm sales revenue [Equation (1)] and the moderating effect of advertising expenditure on the relationship between SMM and revenue [Equation (2)]. Three estimation methods, including fixed-effects regression, random-effects regression, and pooled ordinary least squares regression (OLS), are tested to determine the model that best fits the data structure. For the main effect and the moderation effect analysis, the Hausmann test statistic is not significant (main: chi-sq (16) = 23.60; Prob > chi-sq = 0.0987; moderation: chi-sq (17) = 12.29; Prob > chi-sq = 0.78), which is in favor of the random-effects regression. We also run the Breusch-Pagan Lagrangian multiplier test for random effects versus pooled OLS regressions. The significant test results (main: chibar2 = 345.85; Prob > chibar2 = 0.0000; moderation: chibar2 = 310.22; Prob > chibar2 = 0.0000) conclude that random effects are more appropriate for model estimation.

(1)LnRevit=α+ß1LnSMM Performanceit+ß2LnSizeit+ß3Financial Leverageit+ß4Ageit+ß5FranchisingDit+ß6Restaurant TypeDit+ß7InternationalizationDit+ß8YearDit+ß9LnGDPit+ß10LnCPIit+εit

(2)LnRevit=α+ß1LnSMM Performanceit+ß2LnAdvertising Expenditureit+ß3LnSMM Performanceit*LnAdvertising Expenditureit+ß4LnSizeit+ß5Financial Leverageit+ß6Ageit+ß7FranchisingDit+ß8Restaurant TypeDit+ß9InternationalizationDit+ß10YearDit+ß11LnGDPit+ß12LnCPIit+εit

Note: LnSMM Performance_it is substituted with LnFGC_it, LnLike_it, LnShare_it, LnCom_it, LnPos_it, LnNeg_it, in each empirical model, respectively.

4. Results

4.1 Descriptive statistics

Table 1 provides a summary of the variables included in the analyses. The sample covers 39 restaurant companies' annual data from 2009 to 2020. The sales revenue per firm (Rev) averages $1,123.19 million. Firm SMM input (FGC) ranges from one to 749 posts annually. Consumer engagement indicators (Like, Share, Com, Pos, Neg) range from zero to millions-level participation. Consumer engagement statistics show large standard deviations, indicating that firms gained differentiated visibility on Facebook due to heterogeneous SMM strategies. However, this is acceptable as prior literature supports the value of SMM performance evaluation research using data with large deviations among metrics (Xie et al., 2017; Wang et al., 2021). Nevertheless, the maximum values of size (total assets) and advertising expenditures broadly deviate from their minimum values, suggesting that the sample firms varied widely in scale and advertising budget. In addition, 19 out of 39 firms operate limited-service restaurants. In 62.78% of the firm-year observations, a restaurant firm is a franchisor; in 58.99% of the cases, it has an international presence.

4.2 Test of measurement models

Excluding the missing data on each variable, the main analyses cover 39 firms with 287 firm-year observations, and the moderation analyses cover 36 firms with 257 firm-year observations. Table 2 illustrates the correlations between each pair of variables. UGC elements (i.e. number of “like”s, “share”s, comments) positively correlate with firm sales revenue, which is consistent with our expectation except for the direction of the number of negative comments (ρ = 0.475, p < 0.05). The positive correlations between UGC and revenue indicate that sales revenue increases as consumer engagement increases. In addition, the correlation between FGC and revenue is not statistically significant. Firm size ((ρ = 0.934, p < 0.05), firm age (ρ = 0.401, p < 0.05), advertising expense (ρ = 0.828, p < 0.05), franchising (ρ = 0.185, p < 0.05), and internationalization (ρ = 0.153, p < 0.05) are positively correlated with revenue. These results imply that a restaurant's sales revenue increases as firm size, years of operation, and advertising expenditure increase. Furthermore, firms that allow franchising and operate globally tend to gain higher revenue.

Table 3 presents the estimation results of six empirical models under Equation (1). Consistent with our expectation, the results of Model 1 show that lnFGC has a positive coefficient of 0.043 (p < 0.05), which suggests that a 1% increase in the FGC in a given year leads to a 0.043% increase in a firm's sales revenue. Similarly, lnLike (β = 0.019, p < 0.05), lnShare (β = 0.026, p < 0.001), and lnPos (β = 0.023, p < 0.05) are positively associated with revenue. Despite a positive association between the total number of comments and revenue being evidenced in our sample, this relationship is not statistically significant (β = 0.014, p > 0.1). In addition, lnNeg is found to have a non-significant effect on revenue (β = 0.004, p > 0.1). Restaurant type (β vary from 0.352 to 0.380, p < 0.05), GDP (β vary from 0.298 to 0.433, p < 0.001) and CPI (β vary from −6.193 to −4.037, p < 0.001) significantly influence sales revenue. More precisely, the sales revenue is approximately 0.35–0.38 points higher for full-service restaurants than for limited-service restaurants. Macroeconomic conditions also determine firm revenue, with total revenue increasing as GDP grows and the inflation rate lowers. These findings imply that a restaurant firm's sales performance improves as the number of brand posts, consumer “like”s, “share”s, and positive comments increase. Moreover, the revenue for full-service restaurant firms is higher than those that provide limited service. Contrary to hypotheses H1d and H2b, the total number of comments and the number of negative comments show non-significant correlations with sales performance.

Table 4 presents the estimation results of six empirical models under Equation (2). The results of Model 2 show that the interaction term lnAd*lnLike has a coefficient of −0.011 (p < 0.05), suggesting that advertising expenditure negatively moderates the correlation between the number of “like”s and firm sales performance. The negative coefficients on the interaction term imply that as firms invest more in traditional advertising channels, the financial impact of “like”s on social networking sites becomes weaker. Conversely, firms with low advertising budgets benefit more from SMM. In a similar vein, advertising expenditure negatively moderates the effect of lnShare (β = −0.009, p < 0.05), lnCom (β = −0.011, p < 0.05), and lnPos (β = −0.015, p < 0.05) on sales revenue. However, the moderation effects of lnAd on the relationship between lnFGC and lnRev (β = −0.003, p > 0.1), as well as between lnNeg and lnRev (β = −0.004, p > 0.1), are non-significant.

The study predicts the margins for levels of SMM inputs and advertising expenditure from low to high (5%, 25%, 50%, 75%, and 95% percentiles). Figure 2 shows the predictive margins across individual SMM performance indicators and advertising expense levels. As evidenced by the slope of the line that represents firms in the lowest percentile of advertising expenses (5%), the financial impact of SMM is much more significant for firms with a lower level of advertising expense. As advertising expenses increase, the slope of the lines flattens, corresponding to a lower marginal effect of SMM on predicted revenue. At high advertising expense levels (75% and 95% percentile advertising expenditure line), the lines are almost flat, implying a minimal effect of SMM on revenue. In other words, an enhanced SMM performance barely increases the revenue for firms that invest aggressively in advertising. Overall, these findings support that SMM boosts restaurants' sales performance, and its effect across several dimensions is stronger for firms with less advertising budgets.

4.3 Test of hypotheses

Overall, the model analysis results support positive correlations between sales performance and the number of brand posts on Facebook (H1a), the number of “like”s (H1b), “share”s (H1c), and the number of positive comments (H2a). Meanwhile, the correlations between sales revenue and the total number of comments (H1d) and the number of negative comments (H2b) are non-significant. In addition, advertising expenditure moderates the relationship between revenue and the number of “like”s (H3b), “share”s (H3c), the total number of comments (H3d), and the number of positive comments (H3e), but not for the number of brand post (H3a) and the number of negative comments (H3f).

5. Discussion

5.1 Conclusion

This study contributes to the SMM literature by underlining the effectiveness of SMM in boosting the financial performance of publicly traded U.S. restaurant firms. It proposes the direct enhancement effect of online content engagement (H1) and consumer sentiment (H2) on firm annual sales revenue. The results suggest that revenue increases as firms increase their social media presence through brand posts, consumers' “like”s, and “share”s. While a larger number of positive comments is associated with higher revenue, the effects of the total number of comments and the number of negative comments on revenue were non-significant. In addition, this study explores the intervening role of advertising expenditure in the proposed links between SMM and revenue (H3). It finds that advertising expenditure negatively moderates the relationships. The effect of SMM is stronger for firms with lower advertising expenditures. The findings of this study provide valuable insights for researchers and industry practitioners, which we discuss below under theoretical contributions and managerial implications.

5.2 Theoretical implications

With web data-crawling and sentiment analysis, this study contributes to the methodological evolution of studying SNSs for hospitality research. Data collected from surveys or quasi-experiments often fail to capture the complexity of consumers' behavior (Han and Anderson, 2021). This study applied web scraping as encouraged by recent hospitality scholarly work (Han and Anderson, 2021). The findings support the proposition that SMM helps improve restaurant firms' financial performance (Kim and Kim, 2005; Li et al., 2021a; Wang et al., 2021). These findings expand the uses and gratification theory by showing that sales performance is affected by UGC. Acquiring information about the restaurant through SNSs satisfies consumers' cognitive and emotional needs, which are developed intrinsically to assist purchase decisions. These findings align with hospitality research contending that brand posts and users' comments on hotel and restaurant SNSs affect consumer visiting intentions by influencing their service quality expectations (Ho et al., 2022; Simonetti and Bigne, 2022). The results also confirm the customer engagement marketing theory by revealing that UGC creates value for businesses, and firms implementing SMM observe the positive effects of value co-creation on sales.

The positive association between FGC and revenue shown in this study is consistent with Kim et al.’s (2015) finding regarding the impact of SMM on firms' value. In addition, the positive associations between sales revenue and “like”s, “share”s, and positive comments are consistent with Li et al.’s (2021a) findings that consumer engagements on SMM platforms enhance restaurants' sales performance. Interestingly, this study finds a non-significant effect of negative comments on restaurant revenue, implying that negative comments on SNSs may not impede firms from accomplishing their marketing goals. This finding contradicts Book et al. (2016) and Nazlan et al. (2018), stating that negative UGC weakens consumer purchase intentions for hospitality products. However, it corresponds to Gao et al. (2020) and Phillips et al. (2017), suggesting that negative UGC about service quality and food and beverage quality do not significantly impact hotel sales performance, especially for highly reputable brands. These shreds of evidence imply that the research scope might have played a critical role in determining if negative UGC is impactful. Specifically, a significant impact of negative UGC is found when the response variable being examined is consumer intentions, as negative contents reinforce the unfavorable impression of the brand in consumers' minds. However, consumer perceptions do not necessarily generate the same level of impact on actual sales because other factors, such as promotional campaigns and branding efforts, also determine sales performance. This study further identifies the moderation effect of advertising expense on the relationship between SMM performance and revenue, supporting van Doorn et al.’s (2017) argument that specific types of marketing communication can be more effective for firms with low advertising expenditure. This novel finding contributes to SMM literature by revealing a moderator that impacts SMM effectiveness.

5.3 Practical implications

In an increasingly technology-driven business environment, SMM becomes an essential strategy as a cost-efficient approach to increase public exposure and engage consumers. Social media changed its users from passive consumers of information to active participants. Social media users are encouraged to take active roles in co-creating their experiences, such as creating and sharing information. Like other businesses that establish and maintain strong customer bonds, restaurants should invest in SMM to increase customer reach and encourage value co-creation. The current research highlights the financial benefits that publicly traded U.S. restaurant firms received from such efforts. Therefore, restaurant firms should strive to create more high-quality content on SNSs to facilitate audience participation in brand-related discussions. Promotional tactics such as ‘share this post, get a free drink’ should be encouraged to motivate consumers to like, share, and comment on brand posts. Proactively managing negative UGC has been a common practice for large hospitality firms (Xie et al., 2017). However, the findings of this study imply that restaurant firms should not be overly concerned about negative UGC on SNSs, as we did not find evidence supporting the impact of Facebook's negative comments on sales revenue. In addition, the findings on the moderation effect of advertising expenditure are particularly insightful for practitioners. While large-scale companies rely more on traditional advertising to enhance sales performance, SNSs are great platforms to initiate marketing communication for firms with limited advertising budgets. Restaurants with insufficient public exposure should develop creative SMM tactics to generate buzz and trigger UGC. Holding events, posting viral funny videos, and collaborating with influencers are some examples that maximize “share”s and comments on SNSs.

5.4 Limitations and future research

Like other research, this study has several limitations. First, the findings apply to US restaurant firms and must be applied to different contexts (e.g. other industries, other countries) with caution. Second, the observation period was chosen based on the availability of Facebook data and did not trace back to firms' SMM activities conducted before 2009. In addition, due to constantly updated two-way communication, firm contribution (FGC) and audience contribution (UGC) regarding the same Facebook post may not coincide. Third, our sample covers only publicly-traded restaurant firms. Thus, delicate attention should be paid to derive implications for privately-held restaurant firms. Private companies may practice different SMM strategies and be strategically more aggressive as they are not subject to close public scrutiny.

There are several directions that future research could work toward. First, scholars could expand the research scope to other hospitality sectors such as hotels, cruises, or theme parks. As the mental budgets for the service provided by these sectors can be substantially higher than that of restaurants, consumers may be more cautious about negative comments due to loss aversion tendencies. Second, future research could examine other financial performance indicators such as bottom-line accounting ratios (e.g. net income, earnings per share) and business efficiency measures (e.g. return on assets, equity, and sales). Lastly, mediators and other potential moderators should be explored to explain how SMM impacts sales and what factors might change the direction or magnitude of the effects of SMM.

Figures

Figure 1

Research framework

Figure 2

Moderation effect of advertising expenditure on the relationship between SMM performance and sales revenue

Table 1

Descriptive statistics

Variables	N	Mean	SD	Min	Max
Rev ($ Millions)	314	1123.19	2822.43	1.48	26508.6
FGC	295	203.19	142.33	1	749
Like	295	145025.30	356776.2	0	4,001,503
Share	295	15481.96	33,947.11	0	258,118
Com	295	18261.22	37032.73	0	348,471
Pos	295	1568.98	4090.25	0	53314.31
Neg	295	234.10	1188.06	0	18195.19
Ad Exp ($ Millions)	278	29.83	69.25	0.03	506.76
Size (Total Asset in $ Millions)	316	1017.81	2652.19	4.25	29374.5
Financial Leverage	309	0.72	0.43	0.04	4.35
Firm Age	317	37.50	17.13	0	73
Franchising	317	0.63	0.48	0	1
Restaurant Type	317	0.54	0.50	0	1
Internationalization	317	0.59	0.49	0	1
GDP	317	3.46	1.85	−2.2	5.4
CPI	317	1.68	0.74	−0.36	3.16

Source(s): Author's own creation

Table 2

Correlation matrix

		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1	lnRev	1.000
2	lnSize	0.934*	1.000
3	Financial Leverage	0.071	0.048	1.000
4	Firm Age	0.401*	0.425*	0.128*	1.000
5	lnCPI	0.008	- 0.030	0.052	0.003	1.000
6	lnGDP	−0.046	−0.130*	−0.131*	−0.093	0.180*	1.000
7	lnFGC	−0.082	−0.076	−0.052	0.040	−0.138*	0.018	1.000
8	lnLike	0.443*	0.453*	0.237*	0.230*	−0.117*	−0.081	0.510*	1.000
9	lnShare	0.400*	0.433*	0.269*	0.242*	−0.139*	−0.136*	0.443*	0.890*	1.000
10	lnCom	0.515*	0.529*	0.307*	0.280*	0.019	−0.122*	0.429*	0.924*	0.830*	1.000
11	lnPos	0.475*	0.473*	0.183*	0.210*	−0.026	−0.024	0.459*	0.920*	0.789*	0.937*	1.000
12	lnNeg	0.417*	0.424*	0.207*	0.254*	−0.028	−0.046	0.330*	0.655*	0.513*	0.698*	0.656*	1.000
13	lnAd	0.828*	0.809*	0.236*	0.324*	0.026	−0.039	0.055	0.519*	0.478*	0.575*	0.516*	0.496*	1.000
14	Restaurant Type	−0.070	−0.167*	−0.307*	0.012	0.000	0.056	−0.114*	−0.298*	−0.280*	−0.307*	−0.226*	−0.289*	−0.256*	1.000
15	Franchising	0.185*	0.191*	0.226*	0.251*	−0.041	−0.022	0.070	0.367*	0.319*	0.375*	0.329*	0.400*	0.420*	−0.497*	1.000
16	Internationalization	0.153*	0.171*	0.186*	−0.293*	0.006	−0.042	−0.023	0.315*	0.266*	0.286*	0.290*	0.191*	0.212*	−0.230*	0.306*	1.000

Note(s): *p < 0.05

Source(s): Author's own creation

Table 3

Estimation results for main analyses

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
lnFGC	0.043** (2.68)
lnLike		0.019** (2.02)
lnShare			0.026*** (2.61)
lnCom				0.014 (1.34)
lnPos					0.023** (2.26)
lnNeg						0.004 (0.45)
lnSize	0.795*** (23.88)	0.794*** (23.72)	0.790*** (23.66)	0.793*** (23.49)	0.789*** (23.46)	0.799*** (23.74)
Financial Leverage	0.222*** (3.81)	0.211*** (3.62)	0.204*** (3.52)	0.202*** (3.45)	0.210*** (3.62)	0.204*** (3.48)
Firm Age	0.007* (1.69)	0.006 (1.44)	0.007 (1.49)	0.007 (1.51)	0.007 (1.55)	0.007 (1.64)
Franchising	0.072 (0.65)	0.063 (0.57)	0.079 (0.72)	0.060 (0.54)	0.064 (0.58)	0.045 (0.40)
Restaurant Type	0.366** (2.25)	0.373** (2.29)	0.380** (2.34)	0.366** (2.25)	0.367** (2.26)	0.352** (2.17)
Internationalization	0.127 (1.30)	0.131 (1.33)	0.138 (1.41)	0.141 (1.43)	0.137 (1.40)	0.141 (1.42)
lnCPI	−4.845*** (−3.66)	−4.834*** (−3.57)	−6.470*** (−4.06)	−4.403*** (−3.32)	−4.654*** (−3.53)	−4.037*** (−3.10)
lnGDP	0.342*** (4.83)	0.342*** (4.71)	0.433*** (5.01)	0.318*** (4.47)	0.329*** (4.67)	0.298*** (4.29)
Intercept	4.626*** (4.18)	4.692*** (4.13)	6.193*** (4.56)	4.364*** (3.90)	4.59*** (4.11)	4.094*** (3.71)
YearDummy	Yes	Yes	Yes	Yes	Yes	Yes
N	287	287	287	287	287	287
R2	0.888	0.891	0.890	0.890	0.890	0.888

Note(s): t statistics in parentheses; *p < 0.1; **p < 0.05, ***p < 0.01

Equation (1), lnRev as Dependent Variable

Source: Author's own creation

Table 4

Estimation results for moderation analyses

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
lnFGC	0.031 (1.55)
lnLike		0.034** (3.16)
lnShare			0.042*** (3.98)
lnCom				0.032** (2.61)
lnPos					0.045*** (3.65)
lnNeg						0.010 (0.88)
lnAd	0.207*** (3.64)	0.308*** (6.11)	0.265*** (7.89)	0.296*** (6.15)	0.286*** (7.33)	0.210*** (7.33)
lnAd*SMM Performance Variable	−0.003 (1.55)	−0.011** (−2.55)	−0.009** (−3.14)	−0.011** (−2.42)	−0.015** (−3.03)	−0.004 (−1.08)
lnSize	0.665*** (17.98)	0.654*** (17.96)	0.657*** (18.36)	0.650*** (17.55)	0.646*** (17.75)	0.667*** (18.04)
Financial Leverage	0.136** (2.27)	0.139** (2.37)	0.140** (2.43)	0.124** (2.11)	0.131** (2.25)	0.123** (2.06)
Firm Age	0.001 (0.23)	0.001 (0.20)	0.001 (0.21)	0.001 (0.24)	0.001 (0.22)	0.001 (0.25)
Franchising	−0.020 (−0.19)	−0.021 (−0.20)	−0.007 (−0.07)	−0.029 (−0.28)	−0.023 (−0.22)	−0.022 (−0.20)
Restaurant Type	0.352** (2.04)	0.365** (2.12)	0.370** (2.27)	0.345** (2.00)	0.347** (1.98)	0.352** (2.04)
Internationalization	0.041 (0.43)	0.030 (0.32)	0.029 (0.31)	0.026 (0.27)	0.030 (0.32)	0.043 (0.43)
lnCPI	−2.322* (−1.47)	−2.643** (−2.01)	−4.318** (−2.80)	−2.208* (−1.71)	−2.435* (−1.90)	−1.781*** (−1.34)
lnGDP	0.186** (2.58)	0.202** (2.83)	0.297** (3.52)	0.180** (2.56)	0.190** (2.74)	0.153** (2.21)
Intercept	3.277** (3.04)	3.461** (3.23)	5.005** (3.86)	3.187** (3.01)	3.414** (3.26)	2.849** (2.71)
YearDummy	Yes	Yes	Yes	Yes	Yes	Yes
N	257	257	257	257	257	257
R2	0.887	0.897	0.896	0.896	0.898	0.889

Note(s): t statistics in parentheses; *p < 0.1; **p < 0.05, ***p < 0

Equation (2), lnRev as Dependent Variable

Source(s): Author's own creation