BUSINESS (NO PLAGARISM A+ WORK, ON TIME)

EBSCO-FullText-10_10_2025.pdf

Home >Business & Finance homework help >Management homework help >BUSINESS (NO PLAGARISM A+ WORK, ON TIME)

Information Systems Research Vol. 24, No. 3, September 2013, pp. 596–612 ISSN 1047-7047 (print) � ISSN 1526-5536 (online) http://dx.doi.org/10.1287/isre.1120.0454

Promotional Marketing or Word-of-Mouth? Evidence from Online Restaurant Reviews

Xianghua Lu School of Management, Fudan University, Shanghai, China, 200433, [email protected]

Sulin Ba School of Business, University of Connecticut, Storrs, Connecticut 06269, [email protected]

Lihua Huang School of Management, Fudan University, Shanghai, China, 200433, [email protected]

Yue Feng School of Business and Management, Hong Kong University of Science and Technology, Hong Kong,

[email protected]

The value of promotional marketing and word-of-mouth (WOM) is well recognized, but few studies have compared the effects of these two types of information in online settings. This research examines the effect

of marketing efforts and online WOM on product sales by measuring the effects of online coupons, sponsored keyword search, and online reviews. It aims to understand the relationship between firms’ promotional mar- keting and WOM in the context of a third party review platform. Using a three-year panel data set from one of the biggest restaurant review websites in China, the study finds that both online promotional marketing and reviews have a significant impact on product sales, which suggests promotional marketing on third party review platforms is still an effective marketing tool. This research further explores the interaction effects between WOM and promotional marketing when these two types of information coexist. The results demonstrate a substitute relationship between the WOM volume and coupon offerings, but a complementary relationship between WOM volume and keyword advertising.

Key words : promotional marketing; online review; keyword sponsored search; online coupon; word of mouth; product price; product sales

History : Chris Dellarocas, Senior Editor; Ming Fan, Associate Editor. This paper was received March 7, 2011, and was with the authors for 10 months for 2 revisions. Published online in Articles in Advance January 28, 2013.

Introduction Economic and marketing studies have long shown that word-of-mouth (WOM) plays an important role in shaping consumer attitudes and behaviours (e.g., Buttle 1998, Katz et al. 1955). The Internet has enabled the creation of many online platforms where users make comments on products and/or services, such as epinions.com, cnet.com, and view- points.com, to name just a few. These platforms can effectively aggregate positive, negative, and neu- tral opinions into readily accessible online WOM, which has become a critical source of information that is reshaping the behaviour of online consumers by reducing product information asymmetry between firms and consumers (Gu et al. 2012). Indeed, the pub- lic has become increasingly reliant on online WOM information to make various decisions, ranging from what movie to watch (Dellarocas et al. 2007) to what stocks to invest in (Guernsey 2000) or which book to

read (Chevalier and Mayzlin 2006). Naturally, there has been an abundance of research focused on online WOM. These studies demonstrate that positive online WOM has a positive impact on product sales (Chen and Xie 2005, Chevalier and Mayzlin 2006, Duan et al. 2008b, Godes and Mayzlin 2004, Gu et al. 2012).

The proliferation of online review sites and the huge traffic they draw make them an attractive channel for online marketing promotions. Because third party online review platforms gather together a large num- ber of consumers who are interested in various prod- ucts or services (e.g., restaurants), advertisers believe they can reach specific groups of consumers via dif- ferent promotional marketing strategies on these plat- forms (Chen and Xie 2008). Consequently, users of these review platforms are often exposed to both the WOM information from peers and promotional marketing from vendors in the form of sponsored search results or online coupons. For example, when

596

a consumer is reading the reviews of one product, he may likely find a promotion coupon for the product from a vendor on the same Web page. Keyword spon- sored search (or paid search), as another form of pro- motional marketing from vendors, has not only been popular on search engine sites such as Google.com and Yahoo.com but also has been increasingly adopted by third party review platforms such as yelp.com.

Given the frequent coexistence of WOM and pro- motional marketing on these review platforms, many questions arise: which one is more dominant in influ- encing users’ decision making? In a world of wide- spread availability and rapid dissemination of online WOM, should firms still actively promote themselves through marketing mechanisms such as coupons to drive sales, or should they try to offer strategic incentives to stimulate word-of-mouth? Several online WOM studies have suggested that companies should actively try to create WOM communications (Chen and Xie 2008, Godes and Mayzlin 2004, Liu 2006, Rosen 2000), given that the primary purpose of users of these review platforms is to seek real and diverse product comments instead of other information cues such as coupons (Beuscart and Mellet 2008). But to what extent should firms invest financial resources to stimulate additional WOM? How does WOM com- pare to promotional marketing efforts such as key- word sponsored search? Are promotional marketing tools overshadowed by online WOM on these review platforms, or would the effect of online promotional marketing be enhanced in the presence of online WOM information?

The goal of this study is to investigate the com- parative influence of promotional marketing, such as sponsored search and online coupons, and WOM information on third party hosted online review plat- forms. We not only examine the direct impact of each on product sales but also explicitly assess the possible interaction effects between WOM and pro- motional marketing when both are offered to con- sumers simultaneously. The findings indicate both WOM and promotional marketing impact sales, even though the original purpose of users coming to a review site might only be to seek WOM information. In addition, there is a substitute relationship between the volume of WOM information and coupon offer- ings, contrary to our hypothesized complementary relationship. These findings are important for ven- dors making advertising decisions and also contribute to the knowledge body of WOM research by com- paring WOM with other forms of online marketing strategies. Moreover, the interaction effect between online WOM and promotional marketing on con- sumer decision making has never been formally

studied. We believe this research fills the gap by more precisely identifying the impact of WOM and pro- motional marketing and their interaction on product sales when WOM and marketing promotion coexist. The findings will be helpful to vendors in adjusting their online marketing strategies in the presence of third party review information.

The research makes several important contributions to the literature. First, although keyword sponsored search has been a popular research topic in the last few years, most prior research has taken an ana- lytical approach by examining different theoretical designs and their implications. Our research is one of the first that empirically investigates the effec- tiveness of keyword sponsored search on firm rev- enue in an online review context. Second, although online WOM has been extensively studied, no prior research has assessed the impact of consumer gener- ated content and vendor marketing strategies simul- taneously. By doing so, our research sheds light on the relative importance of these two types of informa- tion for consumer purchase decision making. Third, by explicitly investigating the possible interaction effect of WOM and promotional marketing tools on sales, our research findings provide guidance to ven- dors on how they should deploy their marketing tools in the context of consumer generated content. Finally, our study focuses on high-involvement prod- ucts, i.e., restaurant food and services. Our findings are largely consistent with previous WOM studies on low-involvement products, such as books and DVDs, which suggests the importance of online WOM in a service oriented setting.

The rest of the paper is organized as follows. We first review the literature on online WOM and online promotional marketing as well as their inter- action effect in other online contexts. Based on the literature review, a set of hypotheses regarding the effect of online WOM, promotional marketing, and their interactions are proposed. We then explain the research methodology and present our empirical results. Finally, we conclude the paper with future research directions.

Literature Review and Hypotheses Development Online Word-of-Mouth Word-of-mouth (WOM) is traditionally defined as “oral, person-to-person communication between a receiver and a communicator whom the receiver per- ceives as non-commercial, regarding a brand, a prod- uct, a service or a provider” (Arndt 1967). In recent years, online WOM has gained tremendous popu- larity because more and more consumers offer their

opinions on online review sites. Compared with inter- personal WOM, online WOM goes beyond the tradi- tional form of oral communication and is far more voluminous in quantity and content richness. Both positive and negative information is aggregated on online review websites from multiple sources, as opposed to a single piece of information that is either positive or negative in traditional valence (Chatterjee 2001). Moreover, conventional interpersonal WOM works only within limited social contact boundaries, the influence diminishing quickly over time and distance (Bhatnagar and Ghose 2004, Ellison and Fudenberg 1995). On the contrary, online WOM spreads with low cost and bidirectionality and allows individuals to reach many other people in a one-to- many process (Hennig-Thurau et al. 2004). Thus, the scale and scope of online WOM influence are consid- erably expanded.

Past literature has shown that online WOM indeed impacts product sales (Chevalier and Mayzlin 2006, Duan et al. 2008a, Godes and Mayzlin 2004, Gu et al. 2012, Liu 2006). In previous research, valence, vol- ume, and dispersion of online reviews are widely recognized as the three key measurements of WOM (Dellarocas et al. 2007). For example, Liu (2006) and Duan et al. (2008b) explored the impact of Yahoo! Movies reviews on box office sales. They found that the volume of movie reviews had a significant influ- ence on box office revenue. Similar conclusions have been reached by Godes and Mayzlin (2004) and Duan et al. (2008a); that is, a daily volume of online dis- cussions is highly correlated with new TV show ratings and movie box office sales. Apart from vol- ume, Dellarocas et al. (2007) studied the impact of movie reviews and found that the valence of online movie ratings is also a predictor of future movie revenues. Their explanation is that positive opinions will encourage other customers to adopt a product whereas negative opinions will discourage them.

Dispersion is defined as the extent to which product-related conversations are taking place across a broad range of communities (Godes and Mayzlin 2004). The higher the dispersion is, the wider the impact of WOM will be. According to Granovetter (1973), opinions spread quickly within communities but slowly across them. Ideas and opinions that exhibit strong dispersion across communities are thus likely to have substantial staying power. Godes and Mayzlin (2004) have looked at the dispersion of Usenet conversations about television shows and find that the dispersion of conversations among differ- ent newsgroups is significantly correlated with the Nielsen (viewership) ratings of these shows.

Because the context of this study is one online review site only (one of the most popular restaurant

review sites in China), dispersion of WOM across online communities cannot be measured here. There- fore, our research focuses on the other two key measures: valence and volume. In addition, prior literature has concluded that negative reviews and WOM are more powerful than positive ones (Dellarocas 2003); we therefore also include the per- centage of negative reviews in our study as a supple- ment to measure the impact of online WOM.

WOM Valence. WOM valence denotes reviewers’ evaluation score of a specific product. In theory, the higher the valence of a product, the more positive the consumers’ attitudes are toward the product, which leads to an increased adoption of the product.

Results from previous studies on the effect of valence have not been consistent. The experimen- tal study by Senecal and Nantel (2004) demonstrates that the purchase frequency of customers who have been exposed to positive review information is twice that of customers who have not. Chevalier and Mayzlin (2006) studied book review records from Amazon.com and found that an increase in valence of book reviews can lead to a sales increase. Clemens et al. (2006) conducted a survey of online reviews on beer manufacturers from 2001 to 2003 and found that the valence of reviews made by consumers is posi- tively related to beer sales because products with high valence are likely to be purchased again.

However, also using Amazon.com book review data, Chen et al. (2004) show that WOM valence is not correlated with sales. Similarly, Liu (2006) points out that whereas the volume of online conversations has explanatory power, their valence does not, sur- prisingly. Duan et al. (2008a) reached a similar conclu- sion in their study of online movie reviews. Hu et al. (2009) examined a set of Amazon.com book review data and concluded that the uneven distribution of the WOM data might be the reason why previous research yields inconsistent results.

Despite the inconsistent outcomes of previous research on valence, we follow the theory that pos- itive valence could induce more product sales and hypothesize a positive effect of valence.

Hypothesis 1A (H1A). The valence of online WOM has a positive impact on product sales.

WOM Volume. WOM volume refers to the quan- tity of reviews or comments on a specific product or service. The theory behind measuring volume is that when more customers discuss a product, there is a higher the chance that other customers will become aware of it (Dellarocas et al. 2007, Forman et al. 2008). Moreover, the herding theory indicates high volume of reviews of a product may generate more conver- sations about the product (Godes and Mayzlin 2004) and increase the curiosity and informedness of users

on the product, which in turn leads to its higher sales.

Volume has been widely included in measuring WOM in previous studies. Liu (2006) includes vol- ume as an independent variable when he examines the relationship between weekly movie reviews on Yahoo.com and box office revenue. His findings sug- gest that the volume of WOM accounts for most of aggregated and weekly revenues. Duan et al. (2008b) and Dellarocas et al. (2007) also examine the impact of movie reviews on box office sales and reach the same conclusion. Therefore, we hypothesize a positive effect of WOM volume on product sales:

Hypothesis 1B (H1B). The volume of online WOM has a positive impact on product sales.

Negative Review. Previous research has indicated that the effect of negative WOM is quite different from that of positive WOM. Chatterjee (2001) con- cludes that negative reviews on the Internet have a negative influence on both consumers’ trust and intention to purchase. In an experimental study of online auction markets, Ba and Pavlou (2002) demon- strate that negative ratings have a greater opposing effect than do positive ratings when a buyer forms his level of trust in a seller based on feedback information (WOM) from other buyers. Bajari and Hortacsu (2003) show, using eBay auction data, that negative WOM affects the probability of buyer entry into the auction. Milnik and Alm (2002) similarly show that negative WOM decreases selling prices on eBay. Aside from WOM studies using eBay feedback data, Chevalier and Mayzlin (2006) find, when studying online book reviews, that an incremental negative review is more powerful in decreasing book sales than an incremen- tal positive review in increasing sales.

Given the stronger influence of negative WOM compared to the positive, as demonstrated by previ- ous studies, we add the percentage of negative review as a supplementary measure of the value of word- of-mouth. We use percentage of negative reviews, instead of the absolute number of negative reviews, to avoid possible biases incurred by using valence and volume only because negative and positive volumes may have offsetting relationships with future sales that may cancel each other out in empirical estimates (Godes and Mayzlin 2004).

Hypothesis 1C (H1C). The percentage of negative online WOM has a negative impact on product sales.

Online Promotional Marketing With the rapid growth of e-commerce in the last few years, it is not surprising that there has been resilient growth in online advertising. According to the market research firm eMarketer, U.S. online adver- tising spending will more than double as a percentage

share of total media advertising spending, from 6% in 2006 to more than 15% in 2012 (eMarketer 2008). The fast growth of online advertising compared to traditional promotion methods is largely because of the increased ability of users to interact with firms in the online world, which enables a shift from “mass” advertising to more “targeted” advertising. Ansari and Mela (2003) show that targeted advertis- ing improves matches and therefore makes both con- sumers and advertisers better off.

There are various formats of online advertising/ promotion, such as display ads, paid search, email, classifieds, and sponsorships. These ads not only appear on major portal sites such as MSN or Yahoo .com but are also increasingly prevalent on third party review sites. In our paper, we will focus on two pro- motional marketing formats: coupons and keyword sponsored search. The main reason is that many third party review platforms discourage the practice of interruptive marketing (e.g., banner ads) because too much of this type of advertisement makes users feel the review platform is not acting in the best interest of users and, consequently, may lower user trust in the platform (Dreze and Hussherr 2003, Li and Hitt 2008). Coupons and keyword sponsored search, on the other hand, are based on consumers’ own queries and are hence considered far less intrusive than are online banner ads or pop-up ads (Chen et al. 2009) and are more “targeted” (Yao and Mela 2009).

Online Coupons. The coupon has long been a widely used promotional tool in competition between firms. Many studies have shown that coupons do induce consumers to switch brands and, in some cases, to accelerate purchases (Blattberg and Neslin 1990). Sometimes, a consumer who possesses a coupon may be induced to purchase the product sim- ply because of the possession of the coupon (Sen and Johnson 2000).

Online coupons, because of their greater ability to target, easiness to search for, and simplicity of use, have much lower distribution and redemption costs than do traditional coupons. Online coupons, therefore, have become a popular promotional tool (Chiou-Wei and Inman 2008). According to a sur- vey conducted by Simmons/Experian Research and Coupons, Inc., the number of American adults using online coupons rose by 39%, to 36 million, between 2005 and 2008. Meanwhile, newspaper coupon users declined from 96 million in 2005 to 92 million in 2007.1

Online coupon distribution grew faster than any other medium—up more than 80%—according to NCH Marketing (NCH 2009), although Internet coupons

1 http://www.mediapost.com/publications/?fa=Articles.showArticle &art_aid=87606.

represent less than 1% of all coupons printed. Like- wise, redemption volume of online coupons increased nearly 130% to 4.8% of all consumer packaged goods coupons redeemed in 2008 (Martin 2009). These trends indicate that marketers are increasingly using the Internet to reach consumers with coupons and that consumers, likewise, are responding at an increasing pace to Internet coupons.

In the context of third party review platforms, although the main purpose of users is to search for peer reviews on certain products, the “targeted” online coupon offers on product review pages are much less likely to be perceived as annoying. Some- times it is even welcomed by users because of the material gains they could get by redeeming the coupon (Chiou-Wei and Inman 2008). For example, a dishwasher manufacturer might offer a targeted online coupon for a particular model of dishwasher when a consumer searches for review information on dishwashers. This special price, as well as the WOM information on the product, will both influence the consumer’s evaluation of the product and intention to purchase (Suri and Monroe 2003). The incentive char- acteristics of online coupons, coupled with their more targeted nature and their convenience of use, lead us to hypothesize the following:

Hypothesis 2A (H2A). Offering online coupons through review platforms has a positive impact on product sales.

Different coupons have different characteristics that influence coupon redemption. Among those charac- teristics, coupon value is particularly important to marketers (Yin and Dubinsky 2004). Previous research has found that coupons with a higher face value are associated with higher redemption rates (Bawa and Shoemaker 1987, Bawa and Srinivasan 1997). As the face value of the coupon increases, the redemp- tion value to coupon-prone consumers will increase, which subsequently leads to increased sales (Leone and Srinivasan 1996). A study by Chiang (1995) also indicates that coupon face value has a significant and positive coefficient when regressed on product demand. In the online environment, there is no reason to think coupon value will work differently from that of traditional print-based coupons. In fact, because of the ease of searching and comparing coupons online, it is possible that coupon value may play an even big- ger role in product sales. Therefore, we hypothesize the following:

Hypothesis 2B (H2B). Online coupon value has a pos- itive impact on product sales.

Keyword Sponsored Search. Keyword sponsored search, also known as “paid search,” “pay-for- placement,” or “keyword auction advertising,” is a

practice where advertisers bid for premium spots on the search results page of an Internet search engine (Yao and Mela 2009). The appearance of a brand in the premium spot can create positive impressions among users, increase online traffic, and ultimately boost sales. Particularly, using sponsored search mechanisms, firms’ ability to target their advertis- ing to potential consumers, as consumers actively seek out certain information (via keywords), is greatly enhanced (Ghose and Yang 2009). Because listings appear when a keyword is searched for, an advertiser can reach a more targeted audience on a much lower budget. Firms are reallocating significant proportions of their advertising budgets from traditional media to keyword sponsored search and competing to be listed at the top of search results generated in response to a user’s keyword query. This type of marketing promo- tion is believed to relate a firm’s advertising expendi- tures more closely to outcomes.

Some prior research has supported this point of view. Research by Animesh et al. (2010) indicates that firms’ relative advertising expenditure serves as a stronger signal of quality in online sponsored search markets, which in turn impacts customers’ purchase decisions. Ghose and Yang (2009) reach a similar con- clusion that click-through rates decrease with ad posi- tion and the rate of change diminishes as one goes down the search engine screen. Katona and Sarvary (2010) also demonstrate that for less popular advertis- ers the performance of sponsored search is better than organic search results. Tellis (2004) argues that even if keyword clicks do not lead to immediate conver- sions in the short run, the mere act of repetitive expo- sure to a stimulus can increase the familiarity with the brand name and lead to a preference, which in turn enhances the effectiveness of future advertising. We therefore posit the following:

Hypothesis 2C (H2C). Keyword sponsored search on review platforms has a positive impact on product sales.

Interaction Between Online WOM and Promotional Marketing Online word-of-mouth is changing people’s be- haviour in subtle but important ways. Researchers suggest that WOM serves as a low cost and, poten- tially, effective channel for acquiring and retaining customers, complementary to advertising and promo- tional marketing (Bayus 1985, Chevalier and Mayzlin 2006). Trusov et al. (2009) present a model that enables marketing actions to be credited for their WOM- generating abilities. Hogan et al. (2005) show that the lifetime value of a customer acquired through pro- motional marketing can increase several-fold when the effect of a customer’s WOM is included in the calculation.

When firms actively promote their products and services on third party review platforms through the use of keyword sponsored search and online coupons, these third party review platforms effectively offer dual information channels that flow from vendors to users and from prior consumers to users. How would the two types of information influence users’ purchase decisions when users are faced with both? Compared with online WOM, can a seller’s promo- tional marketing activities produce a stronger impact on consumers’ purchase decisions? More importantly, is there an interaction effect between a seller’s promo- tional marketing and online WOM information about the seller’s products/services?

Chen and Xie (2005) study how third party reviews impact a firm’s marketing strategies. They find that a third party review has two conceptually differ- ent effects on a firm’s advertising function. First, a third party product review generates a substitutive effect because it reduces consumers’ need for adver- tising information. However, a third party review may also generate a complementary effect because it can increase the effectiveness of a firm’s advertis- ing. Their results indicate that there is an interaction effect between third party reviews and the promotion strategies of a firm. Although the third party reviews they study are based on independent laboratory tests or expert evaluations such as those that appear in PC Magazine or Consumer Reports, not consumer- generated word-of-mouth information, both types of reviews provide opinions and extra information to potential customers.

Interestingly, consumer-created information is likely to be considered more credible than seller- created information because the credibility of information is often related to the perceived trustwor- thiness of the information source (Wilson and Sherrell 1993). Consumer-created information could poten- tially enable a seller to implement some marketing strategies that may not be credible otherwise. Chen and Xie (2008) have shown that seller-created product attribute information and buyer-created review infor- mation on a seller’s product website do interact if the review information is sufficiently informative. How- ever, to the best of our knowledge, no prior research has investigated the interaction effect between firms’ promotional marketing and user-generated WOM on online third party review platforms. Although Chintagunta et al. (2010) examine the robustness of the effect of advertising and valence of user ratings on box office performances, their research does not explore the possible interaction effect between advertising and WOM.

Interaction Between Online WOM and Coupons. A person’s coupon usage behaviour depends not only on his or her inherent coupon proneness or desire

to use coupons but also on the attractiveness of the coupons encountered (Bawa and Srinivasan 1997), which is mainly based on the consumer’s expecta- tions about the product/experience. WOM informa- tion could influence such expectations. Arndt (1967) shows in a field experiment that people who received favourable WOM exposure were more likely to use their coupon to buy a new food product. Shimp and Kavas (1984) use the theory of reasoned action to argue that consumers’ intentions to use coupons are determined not only by their attitudes but also by their perceptions of others’ opinions. That is, whether important others (e.g., spouse) think one should or should not expend the effort to use coupons will influence one’s coupon redemption. We believe the same reasoning applies in the online review context as well. Online review platforms provide an opportunity for users to go beyond their own close circles of friends and family to access a large amount of information from many previous con- sumers. Although such information (e.g., opinions) does not come from people with whom they have strong ties, sociologists have long emphasized the importance of “weak ties” in the information diffu- sion process (e.g., Montgomery 1992). Therefore, it is possible that consumers are more likely to respond to coupon promotions when they read rich and pos- itive WOM information offered by previous users on an online review platform, which, in turn, influences product sales.

Therefore, we hypothesize that online WOM has a complementary interaction effect with online coupons:

Hypothesis 3A (H3A). The effect of online coupons on product sales is stronger in the presence of higher WOM valence.

Hypothesis 3B (H3B). The effect of online coupons on product sales is stronger in the presence of larger WOM volume.

Hypothesis 3C (H3C). The effect of online coupons on product sales is stronger when negative WOM percentage is lower.

Interaction Between Online WOM and Keyword Sponsored Search. The main advantage of keyword sponsored search is to display targeted information in response to user-generated queries. Animesh et al. (2010) show that firms’ keyword sponsored search expenditures are usually perceived as a strong signal of quality that will enhance consumers’ expectations for the product/experience. As we have previously argued, keyword sponsored search should have a positive impact on product sales.

Research, however, has shown that sponsored search advertising also carries the risk of reducing the

signalling properties of advertising (Animesh et al. 2010). Low-quality firms may be able to mimic the advertising strategies of high-quality firms if there is no intervention of other mechanisms by the review platforms. Online WOM might serve as such an inter- vention mechanism. Again, using the theory of rea- soned action (Ajzen and Fishbein 1980), consumers’ purchase intentions depend on their attitudes as well as subjective norms. Therefore, when online WOM information is sufficient and positive, the quality sig- nal of keyword sponsored search is reinforced. Con- sequently, the effect of keyword sponsored search on product sales could be enhanced.

Research in the information search area could also shed some light on the possible interaction effect between WOM and keyword sponsored search. In a study that investigated how people performed per- sonally motivated searches on the Web, researchers found that instead of jumping directly to their infor- mation target using keywords, their study partici- pants navigated to their target using their contextual knowledge as a guide, even when they knew exactly what they were looking for in advance (Teevan et al. 2004). On third party review platforms, WOM infor- mation forms a rich context for firms promoting their products and services. Therefore, it is reasonable to conjecture that users rely on the contextual informa- tion provided by WOM to help them decide whether they want to click on the sponsored search links. We hypothesize the following:

Hypothesis 4A (H4A). The effect of keyword spon- sored search on product sales is stronger in the presence of higher WOM valence.

Hypothesis 4B (H4B). The effect of keyword sponsored search on product sales is stronger in the presence of larger WOM volume.

Hypothesis 4C (H4C). The effect of keyword sponsored search on product sales is stronger when negative WOM percentage is lower.

Research Methodology Research Context Online WOM sites are typically of two types: retailer- hosted (such as Amazon book reviews) or third party- hosted (such as CNet or Epinions reviews). Some of the WOM reviews are about tangible products such as electronics, whereas others are more focused on ser- vices. According to Bansal and Voyer (2000), WOM plays a more important role when the product in question is more risky or uncertain and when con- sumer’s involvement with it is higher. Thus WOM has been found to be especially effective in decision making regarding services (Murray 1991). Zeithaml and Bitner (1996) point out that there is a higher

level of risk associated with the purchase of ser- vices, primarily because services are intangible, not standardized, and usually sold without guarantees or warrantees.

For our research, we have chosen a third party restaurant review site ABC.com2 as our research con- text. Restaurant service is a high involvement prod- uct on which consumers often spend a considerable amount of time searching for information to make the right decision (Gu et al. 2012). It provides an ideal context to study the effectiveness of WOM on third party platforms (Gu et al. 2012, Luca 2011, Mangold et al. 1999). ABC.com is one of the biggest restau- rant review platforms in China, with reviews for restaurants across 390 cities. Any registered member can post review information and leave comments on restaurants. Along with text reviews, members are also asked to report the average price per person of their patronage and rate the service quality, environ- ment, and taste of food on a scale of poor, ordinary, good, excellent, and outstanding. The scales of these three indicators are then automatically calculated into a valence value by ABC.com’s proprietary algorithm. Because the credibility of valence depends on the number of persons who leave reviews, ABC.com does not provide a valence value for any restaurant that has fewer than four reviews. ABC.com also reports the real time volume of reviews for each restaurant. Users who register with ABC.com are provided with a membership card that entitles them to certain dis- counts at restaurants that have formed an affilia- tion with ABC.com. All restaurant transactions are recorded when membership cards are used. Adver- tising is the main revenue source of ABC.com. Two typical advertising methods are sponsored search and online coupons. Sponsored search works quite simi- larly as search engine advertising. When users search for restaurants with a keyword, the restaurants that bought the keyword will be listed at the top of the search results followed by the organic results. For example, when users search for “Szechuan cuisine,” those restaurants that bought the keyword “Szechuan cuisine” would be displayed at the top of the search list, and the organic search results are listed after. In order to protect user experience, ABC.com limits the maximum number of sponsored search results to 20 and marks the displays as “sponsoring vendor.” The order of the sponsoring vendors, however, is ran- domized by ABC.com. ABC.com charges sponsoring restaurants according to the popularity of the key- words to simplify the keyword purchasing process. Restaurants can buy multiple assorted keywords at the same period to capture more potential relevant searches.

2 The review site that provided us data has requested anonymity.

The other main form of promotional marketing on ABC.com is online coupon offering. Coupons are available for users to print or download by cell phones (see the Appendix A for the screenshot of a keyword search result page and Appendix B for a coupon page). There are three ways by which users can find the coupons. First, from the main page of each restaurant, users can easily find the link at the top to view the detailed offer. There is also a “coupons” tab at the top of every Web page on ABC.com where users can search for coupons accord- ing to their needs. The display order of the coupon search results is random for each coupon search. Finally, coupon offering is also displayed as a small icon on the right side of each search result for those restaurants that are offering a coupon. In the third scenario, the effect of coupon promotion may interact with that of sponsored search because the coupons of top listed restaurants are more likely to be found by users. In Appendix A, both restaurants shown engaged in coupon promotion as well as keyword sponsored search.

The data ABC.com provided to us are all from restaurants in Shanghai where users can use ABC.com membership cards. This enables us to obtain detailed transaction data of these restaurant visits by mem- bers. The sales revenue and review information of these restaurants are organized as weekly panel data.

Table 1 Variable Description and Operationalization

Variables Description Operationalization

ln4Sales it 5 The log of revenue of restaurant i in week t The log value of the total revenue of restaurant i in week t generated by customers who use ABC.com membership cards

Valit Valence of review of restaurant i in week t The listed rating value of restaurant i in week t on ABC.com

ln4Vol it 5 The log of review volume for restaurant i in week t The log value of the cumulative number of reviews restaurant ireceived up to week t

Nrit Negative review percentage of restaurant i in week t We define a review as negative if its rating is lower than 2/3 of the average rating of all restaurants. We compute the percentage of these negative reviews as Nrit .

Ocit Whether restaurant i offered an online coupon or not in week t

Dummy variable: 1 means the restaurant offered an online coupon and 0 otherwise in week t .

Cvalueit The monetary value of the coupon offering of restaurant i in week t

On ABC.com, a coupon can be a fixed deductible amount (e.g., 30 RMB off), or a percentage discount on the total consumption (e.g., 15% off), or a voucher for a free dish or drink. We monetized all of these formats into coupon monetary values using the sales data of each transaction.

Kwit The number of keywords The number of keywords restaurant i bought in week t for sponsored search advertising

ln4Priceit 5 The log of average price per person reported by all reviewers

The log value of the average price per person reported by all reviewers for restaurant i in week t

Compit Number of competitors of restaurants i We input the address of each restaurant into Google Maps to compute the number of restaurants within a 1-kilometer radius of restaurant i.

Dishit The number of dishes recommended by reviewers The number of special dishes recommended by reviewers for restaurant i in week t

Holit If week t included any holiday Dummy variable: 1 means week t included a holiday and 0 otherwise.

Hcit Average historical consumption of all reviewers of restaurant i in week t

We extract all the reviewers of restaurant i in week t and compute their average eat-out consumption in all restaurants in our data set before week t .

In total, the data set has around 20,000 records involv- ing 420 restaurants from May 2005 to March 2008. Upon further examination of the data, we deleted the restaurants that had fewer than three records to ensure the efficiency of time series analysis. The final number of valid records is 18,130.

Empirical Model Specification Based on the research hypotheses presented above, we set up the following econometric model:

ln4Salesit5 = �0 +�1 Valit+�2 ln4Volit5+�3 Nrit+�4 Ocit

+�5 Cvalueit+�6 Kwit+�7 Ocit ∗Kwit

+�8 Valit ∗Ocit+�9 Valit ∗Kwit

+�10 ln4Volit5∗Ocit+�11 ln4Volit5∗Kwit

+�12 Nrit ∗Ocit+�13Nrit ∗Kwit

+�14 ln4Priceit5+�15 Holt+�16 Compit

+�17 Dishit+�i+�it0 (1)

The dependent variable is the weekly restaurant revenue captured through members’ use of ABC.com cards. The revenue is log-transformed, consistent with prior literature that examines the relation- ship between WOM and retail sales (Chevalier and Mayzlin 2006). The independent variables are defined

in Table 1, where Val, Vol, and Nr are WOM vari- ables and Oc, Cvalue, and Kw are promotional mar- keting variables. Vol is log-transformed to account for the diminishing marginal effect of the number of reviews (Chevalier and Mayzlin 2006). Both Oc and Cvalue are coupon related: Oc is a dummy variable to indicate whether a coupon is offered or not in a certain week; Cvalue, on the other hand, represents the monetary value of the coupon. To investigate the interaction effect between WOM and coupons, we use the coupon dummy variable for the interaction terms. The reason for doing so is that the detailed mone- tary value of a coupon can only be seen after one enters a coupon Web page, where there is no WOM information. The interaction effects between coupon value and WOM would be less likely to occur in such a setting.

Several control variables are included in the estima- tions. Specifically, the effect of average price per per- son of each restaurant is controlled for because price is highly relevant to revenue. In addition, given that the number of popular dishes a restaurant offers may have an effect on sales revenue, the number of dishes (Dishit5 recommended by members is included in the estimation. If a restaurant has more dishes recom- mended by customers, customers would more likely patronize. We also include competition in this model by introducing the number of restaurants within a one-kilometer radius of the focal restaurant because the degree of competition may influence a restau- rant’s marketing strategies (i.e., offering a coupon) and its sales revenue. In addition, a dummy vari- able indicating whether week t contained a holiday is included (Holt5. Finally, �i denotes the restaurant- specific fixed effects that capture the idiosyncratic characteristics associated with each restaurant, such as décor, location, or the type of cuisine. The fixed effects capture all unobserved differences across restaurants.

In this model, self-reported average price per per- son may be endogenous when it is included in the sales model (Villas-Boas and Winer 1999). Therefore, we use an instrumental variable to solve this problem. We believe that consumers’ historical eat-out con- sumption level (i.e., budget for eating out) will impact how much they pay when eating out. If they have an abundant budget, they are likely willing to pay more for their meal. However, a consumer’s overall eat-out budget is not likely to directly impact any par- ticular restaurant’s revenue because the budget will be spread among many different restaurants. There- fore, the historical eat-out consumption of review- ers (Hcit5 is an appropriate instrumental variable for price. Specifically, Hcit calculates the average histori- cal eat-out consumption of all reviewers of restaurant i before week t.

Table 2 Descriptive Statistics

Variables Mean Min Max Std.

Sales 4K RMB5 50135 1 125.6 70726 Valence 10726 0 4 00531 Volume 20404 1 21994 270046 Negative review percentage 00376 0 0.89 00256 Online coupon promotion 00123 0 1 00328 Coupon monetary value 3002 0 298 140433 Number of keywords 00107 0 11 00613 Average price per person 870048 0 11800 770467 Number of competitors 610657 0 130 390587 Number of recommended dishes 10753 0 15 10225 Holiday 00099 0 1 00299

Note. N = 181130.

Data Analysis and Results We present the descriptive statistics of the sample data in Table 2. The average weekly sales revenue captured through ABC.com’s membership cards is 5,135 RMB,3 with an average of 87 RMB per per- son (about US$14). Every week, a restaurant receives an average of five reviews from users. Among these reviews, 37.6% are considered negative. The average valence is 1.72, with 4 being the highest rating pos- sible. As for promotional marketing activities, 12.3% of the restaurants have coupon promotion each week on average, with the average monetary value of 30.2 RMB, about the value of a medium priced dish in China. Some restaurants are also actively engaged in keyword search promotions, buying 11 keywords at a maximum. However, from the whole sample, only 689 observations (3.8% of the sample) have keyword purchases, indicating keyword search promotion on ABC.com was still in an early stage during our sam- ple period.

Prior studies noted the endogeneity problem between the number of reviews and sales of books on Amazon.com because online WOM is not only a driver of retail sales but could also be an out- come of retail sales (Godes and Mayzlin 2004). A Durbin-Wu-Hausman test indicates that endogene- ity is not a concern here in our data set (�2 = 7032).

The two-stage least squares (2SLS) estimation results, run using SAS, are shown in Table 3. Model 1 and Model 2 report the direct effects of online WOM and promotional marketing on sales, respectively. The interaction effects of WOM and promotional market- ing activities are shown in Model 3. Our discussions will mainly focus on the results of Model 3.

Discussions Consistent with prior research on WOM (Liu 2006, Duan et al. 2008b, Dellarocas et al. 2007), we

3 Roughly US$815, based on a conversion rate of US$1 = 6030 RMB.

Table 3 Effects of Online Word-of-Mouth and Promotional Marketing on Restaurant Sales

Category Variables Model 1 Model 2 Model 3

WOM Valence 00033∗∗ 00036∗∗ 00030∗∗

Volume 00341∗∗∗ 00333∗∗∗ 00345∗∗∗

Negative review −00065∗∗ −00061 −00071∗∗

percentage

Promotional Coupon promotion 00239∗∗∗ 00275∗∗∗

marketing Coupon value −00001 −00001 Number of keywords 00054 00027 Coupon ∗Keyword −00072∗∗ −00092∗∗

Interaction Valence ∗Coupon 00038 between Valence ∗Keyword 00039 WOM and Volume ∗Coupon −00151∗∗

promotional Volume ∗Keyword 00096∗∗

marketing Negative review ∗ −00041 Coupon

Negative review ∗ −00043 Keyword

Control Average price 00042∗∗ 00044∗∗ 00044∗∗

variables Competition 00005 00006 00006 Number of 00031∗∗∗ 00030∗∗∗ 00030∗∗∗

recommended dishes Holiday −00040∗∗ −00039 −00038

Fixed effects of Included Included Included restaurants

R2 00686 00687 00688

F value 84066∗∗∗ 84031∗∗∗ 83018∗∗∗

Notes. The dependent variable is the log of weekly revenue ln4Sales5. Average price is instrumented with the historical consumption of reviewers.

∗∗p < 0005, ∗∗∗p < 0001.

found that more positive review information (higher valence) and a larger number of reviews (greater vol- ume) both have a positive impact on restaurant sales (H1A and H1B are supported). Negative reviews also turn out to play a detrimental effect on sales revenue, supporting H1C.

With regard to promotional marketing, the avail- ability of coupons (i.e., a coupon was offered) has a positive impact on restaurant sales (i.e., H2A is sup- ported). Furthermore, the coefficient of the coupon promotion dummy variable indicates that the restau- rants that offered online coupons on ABC.com had on average more than 26.7% (exp4002395− 1) higher rev- enue than those that did not. This finding is consis- tent with prior study results that coupons are a very effective way to attract users and increase revenue (Chiou-Wei and Inman 2008). It also indicates that promotional marketing tools such as coupons are not only effective on portal websites such as MSN.com and search engines such as Google (Chiou-Wei and Inman 2008, Ghose and Yang 2009) but also effective on third party review platforms.

Interestingly, the monetary value of coupons is found insignificant here (H2B not supported), which

suggests that coupon availability is more important than the actual value of the coupon in terms of attracting customers to patronize the restaurants. This result is consistent with research by Chiang (1995) that although coupon face value has a significant and positive influence on product demand, the probabil- ity of becoming a coupon user increases only 3% in absolute terms when the average coupon face value increases by 25%. Consumers appear to give less con- sideration to the face value of the coupon. Leone and Srinivasan (1996) argue that coupons serve two purposes: they have a redemption value to coupon- prone consumers and advertisement value to coupon- indifferent consumers. This may explain why coupon availability is significant, whereas coupon face value is not: the mere availability of an online coupon advertises the coupon-offering restaurant to the users and can therefore attract both coupon-prone and coupon-indifferent consumers, whereas the face value of a coupon only plays a role in the purchase deci- sions of coupon-prone consumers.

Keyword sponsored search, contrary to our Hypothesis H2C, does not appear to have a significant impact on revenue, despite the popularity of online sponsored search as a marketing tool. However, as indicated earlier, only 3.8% observations in our sam- ple have keyword purchasing. The insignificant result of keywords may be due to this data selection bias. In order to exclude this concern, we compare the before and after sales revenue for those restaurants that adopted the keyword search promotion strat- egy during our sample period. The result indicates that the after weekly revenue (mean = 71984 RMB) is significantly higher than the before revenue (mean = 51861 RMB) (t = −5077, p < 00000). There- fore, H2C actually is supported in our study when excluding the restaurants that never purchased any keywords.

Surprisingly, the interaction term between key- words advertising and coupon promotion is sig- nificant but negative, which indicates these two promotional marketing approaches are not mutually enhancing. This result is contradictory to the finding of Yang et al. (2012) that sponsored search with pro- motion information leads to more clicks. To exam- ine whether this result is due to the small number of restaurants with keyword advertising, we run a post hoc analysis using the smaller sample again (i.e., using only those restaurants that purchased keywords during our sample period); the interaction coefficient is still negative but no longer significant. A possi- ble reason for the negative interaction between key- word advertising and coupon promotion could be due to the inherently different nature of these two promotion methods. Coupons are appealing because

of their potential material rewards. However, key- word advertising changes an organic search result into a commercial one, which might lower the appeal of the coupons: when users see that a restaurant is promoting itself by both offering coupons and spon- soring keyword searches, they might feel that the restaurant is being too aggressive, and a “truly good” restaurant would not need to spend all this money and effort to promote itself. As a result, the infor- mation conveyed by the keyword sponsored search and coupon might be intentionally disregarded and even impact users’ impression of the restaurant in negative ways.

Given that both promotional marketing and con- sumer WOM share the same purpose of inform- ing future customers but they present information from different sources, we have hypothesized that the two effects have a complementary relationship. The results, however, demonstrate that most of the WOM and promotional marketing interaction terms are insignificant. The only exceptions involve WOM volume. Contrary to H3B, the negative interaction between volume and coupon promotion actually demonstrates a substitute relationship between WOM and promotional marketing. That is, WOM volume has a significant negative moderating effect on the relationship between coupon promotion and sales. This suggests that for a restaurant with a high WOM volume, coupon promotions are not as effective as for those restaurants with less WOM volume. The effect of coupon promotion in attracting customers is marginally decreasing with the increase of restau- rant WOM volume. Inversely, if a restaurant has a low WOM volume, offering a coupon would attract more business to the restaurant. This result, although not consistent with our theoretical hypothesis based on prior literature, seems to make intuitive sense. Consumers may choose restaurants with rich WOM information whether the restaurants offer coupons or not. The restaurants’ needs for promotional market- ing are reduced in the presence of a large number of online reviews. Rao and Monroe (1988) propose that price is less likely to have a significant effect on buyers’ perceptions of quality in the presence of other information or when buyers are familiar with the product. Therefore, when a restaurant already has a large amount of WOM information available to the consumers, offering coupons with price cut may not matter to the consumers because they have already learned about the quality of that restaurant. Con- sumers may be even willing to pay a higher price for the vendors with rich and high volume of WOM to reduce the risk of uncertainty about food qual- ity (Ba and Pavlou 2002). On the other hand, for restaurants with less WOM, marketing efforts often play a more important role in attracting customers.

Price sensitive consumers will still patronize a restau- rant simply because of the possession of a coupon (Sen and Johnson 2000). Therefore, promotional mar- keting is more important when there is little WOM information.

However, WOM volume works differently with keyword promotions. The significant positive inter- action between volume and keyword advertising is consistent with our hypothesis of a complementary relationship between the two (supporting H4B). Key- word advertising usually leads to higher visibility for the restaurants. Consumers are more likely to make a purchase decision when the effect of keyword advertising is reinforced by the higher popularity, evi- denced by a large WOM volume, of an advertising restaurant.

It is worth noting that the impact of WOM vol- ume is slightly bigger than that of coupon pro- motions: the Wald test indicates the coefficient of online coupons is less than that of review volumes after considering the standard errors (F = 4040, p < 00036). The distinct insight from this result is that the consumers care more about WOM than promotion on third party review platforms. This conclusion is important for both third party review platforms and vendors.

Robustness Check As pointed out by prior literature (Duan et al. 2008b, Sonnier et al. 2011), there is a positive feedback loop between WOM and product sales: WOM leads to more product sales, which in turn generate more WOM. Similarly, a restaurant’s marketing promotion activities may also be influenced by sales: depend- ing on its sales volume, a restaurant may decide to increase or decrease its coupon offering or keyword promotions. To account for the dynamic relationships between WOM and product sales, and marketing pro- motion and sales, we adopt the simultaneous equa- tions model as used by Duan et al. (2008b) to check the robustness of our results.

In addition to Equation (1), the following three equations are added to the system of equations:

ln4Volit5= �0 +�1 ln4Salesi4t−155+�2 ln4Voli4t−155

+�3 Holt +�i +�it1 (2)

Ocit = �0 +�1 ln4Salesi4t−155+�2 Vali4t−15

+�3 ln4Voli4t−155+�4 Holt +�i +�it1 (3)

KWit = �0 + �1 ln4Salesi4t−155+ �2 Vali4t−15

+ �3 ln4Voli4t−155+ �4 Holt +�i +�it0 (4)

We include the lagged sales and WOM volume in each equation to capture their possible impact on

promotional marketing efforts and WOM in the cur- rent period. In addition, we control for restaurant- specific fixed effects and the impact of holiday in each equation.

We run three-stage least squares (3SLS) to esti- mate the four simultaneous equations. The results are presented in Table 4. Compared with the results in Table 3, no significant difference is found between the static model and the dynamic model, which indi- cates the robustness of our estimation results after incorporating the dynamic relationships. Moreover, the coefficient of sales is positive and significant for WOM volume and promotional marketing, suggest- ing a strong dynamic evolution between sales with the volume of WOM and self-promotion in the latter stage. An interesting finding is that the coefficients of WOM volume in Equation (3) and (4) are negative

Table 4 3SLS Estimation Results

Category Variables Coefficient

Equation 1: ln(Sales) as DV WOM Valence 00026

Volume 00332∗∗∗

Negative review −00068∗∗

percentage

Promotional Coupon promotion 00264∗∗∗

marketing Coupon value −00001 Number of keywords 00023 Coupon ∗Keyword −00085∗∗

Interaction Valence ∗Coupon 00049 between Valence ∗Keyword 00041 WOM and Volume ∗Coupon −00143∗∗

promotional Volume ∗Keyword 00091∗∗

marketing Negative review ∗Coupon −00059 Negative review ∗Keyword −00015

Control variables Average price 00043∗∗

Competition 00004 Number of 00028∗∗∗

recommended dishes Holiday −00034

R2 00691 F value 82089∗∗∗

Equation 2: Volume4t5 as DV Independent ln4sales54t−15 00322∗∗∗

variables Volume4t−15 00162∗∗∗

Holiday4t5 00007

Equation 3: Coupon promotion4t5 as DV Independent ln4sales54t−15 10456∗∗∗

variables Volume4t−15 −00396∗∗∗

Valence4t−15 −00082∗∗∗

Holiday4t5 00039

Equation 4: Number of keywords4t5 as DV Independent ln4sales54t−15 00390∗∗∗

variables Volume4t−15 −00086∗∗∗

Valence4t−15 −00015 Holiday4t5 −00008

and significant, which suggests restaurants tend to reduce their marketing efforts if they are relatively popular on ABC.com. This result also supports the substitute relationship between WOM volume and coupon promotion, consistent with our finding in the main model.

Conclusion and Future Research The objective of this study is to analyze the effects of online WOM and promotional marketing on third party review platforms. With users’ increasing dependence on online WOM for purchase decisions (Dellarocas 2003, Forman et al. 2008, Li and Hitt 2008), third party review platforms are generating huge traf- fic that in turn attracts promotional marketing activ- ities of vendors. However, WOM and promotional marketing present two totally different types of infor- mation. Their coexistence on the same third party review platform naturally raises the questions of which one has more impact on users’ purchase deci- sions and whether they are each other’s complement or substitute. The findings of this research provide several important insights from both the theoretical perspective and practical perspective.

Theoretical Implications In this study, we look at the impact of WOM and promotional marketing on users’ purchase decisions in the context of third party review platforms. Com- pared with prior literature that only focuses on the effects of online reviews (Duan et al. 2008b, Li and Hitt 2008, Liu 2006), this study contributes to the body of literature on WOM and promotional marketing in several ways.

First, the findings add to extant research conducted on the commercial value of online WOM in the con- text of restaurant products/services in China. Prior literature (Mangold et al. 1999) indicates that because of the intangible nature of restaurant services and high involvement nature of food, restaurants provide an ideal context to study the effectiveness of WOM. However, most relevant prior studies are conducted in the context of movies, books, and electronic prod- ucts in the United States (Chevalier and Mayzlin 2006, Dellarocas et al. 2007, Duan et al. 2008b, Forman et al. 2008, Godes and Mayzlin 2004, Li and Hitt 2008, Liu 2006). Our findings, which are largely consistent with previous studies, demonstrate the importance of online WOM in a service setting. This result confirms the findings of Gu et al. (2012) that WOM on third party review sites is particularly important for high involvement products.

Our study also contributes to the promotional mar- keting literature. Online coupon promotion, as a nonintrusive and “targeted” marketing approach, is

found to be quite effective in terms of generating pur- chases and increasing sales in our study. Few prior studies have examined how consumer reviews and advertising can be used together to achieve better marketing results. Chen and Xie (2008) take the first step in this direction by examining consumer reviews and advertising together in the context of a seller- hosted review platform. Our study extends this line of research into the context of third party review platforms, where vendors cannot control the WOM content and dispersion process, and reaches the con- clusion that well designed promotional marketing can still play an important role to influence users’ pur- chase decisions, although this effect is slightly less than the WOM effects. A post hoc comparison of the average weekly revenues between the restaurants that have promotional marketing activities (coupons or keyword sponsored search) (average weekly sales =

7161407 RMB) and those that do not (average weekly sales = 4177402 RMB) indicates a significant difference (t = 16058, p < 00000). These results highlight that even in the context of WOM where consumers seem to rely on information from other consumers for deci- sion making, promotional marketing can still be an effective marketing tool.

Online WOM and vendors’ promotional marketing activities both provide information to potential con- sumers. The relationship between these two types of information, however, has not been examined before, although their coexistence is becoming ever more prevalent. Do they complement each other or serve as each other’s substitute? To the best of our knowl- edge, our research is the first one to empirically exam- ine this relationship. Although most of the interaction terms between WOM and promotional marketing are not significant, we did find that the volume of WOM and coupon offering exhibit a substitute relationship. That is, the firms’ marketing effort through coupon offering is less effective when there is a substantial amount of WOM information. On the other hand, if consumers cannot find enough WOM information, they are more likely influenced by firms’ coupon promotions.4 We also find that keyword advertising has a positive marginal effect when there is a large WOM. The visibility brought by keyword advertis- ing complements the popularity of high WOM vol- ume in influencing consumers’ decision making. This is an interesting theoretical insight highlighting that although the sources of the information are com- pletely different, with supposedly different levels of credibility, they nevertheless share the same function of informing and drawing customers.

4 We thank an anonymous reviewer for this insight.

Practical Implications We believe that our work has a number of impor- tant implications for vendors who are interested in pursuing marketing strategies on third party review platforms. First, as our research and previous research have shown, WOM does play a role in consumers’ decision making process. Therefore, vendors should try to strategically stimulate users to generate more positive WOM information.

However, even on third party review platforms where users actively seek WOM information from fellow consumers, promotional marketing by ven- dors is still an effective marketing tool, especially online coupon offerings. In fact, many of today’s savvy netizens often search for online coupons before they decide where to shop. Vendors need to real- ize that with ever increasing competition from other vendors, and the abundant amount of WOM infor- mation available, getting consumers’ attention is crit- ical. Online coupons are an effective way to place one in front of the consumers. Failing to leverage online marketing tools, on the other hand, may lose important revenue generating opportunities. In addi- tion, our research results demonstrate that allowing promotional marketing activities by vendors does not hurt the credibility of these review platforms. Online coupon offers not only help vendors but also are an important revenue source for the review platforms. They should actively develop noninterruptive and nonintrusive marketing approaches to help vendors find their target customers. By doing so, both vendors and review platform providers can benefit.

In addition, our research has shown that coupon offering and keyword sponsored search do not com- plement each other. That is, these two types of mar- keting tools do not enhance each other’s effect. On the contrary, sponsored search seems to diminish the effect of online coupons. We conjecture that the opposing effect of the two may be due to the overly explicit “push” to attract customers. Users might feel that a good restaurant with high quality food does not need to engage in so many promotional activities. Therefore, vendors need to carefully consider which marketing tool is most suitable for them and focus on that specific tool instead of doing everything all at once.

Moreover, the mixed interaction effect between WOM volume and the two different promotional mar- keting approaches suggests that for popular restau- rants that have garnered lots of WOM on a review platform, it is not necessary to offer too many online coupons because the marginal effect of online coupons decreases, but to be ranked at a premium position through keyword advertising does help increase sales. This mixed result is actually quite sur- prising to us because we have theorized that online

coupons would enhance more patronage that in turn would increase WOM; a higher WOM volume would motivate users to seek online coupons to use at the restaurant. But the substitute effect between coupons and WOM, as demonstrated by our data, suggests to vendors that although coupons are generally effec- tive in attracting customers, the context of the coupon offering should also be taken into account when the coupons are offered on a third party review platform.

Future Research Directions The study can be strengthened in several ways in future research. First, this study focuses on nonintru- sive promotional marketing approaches. The effect of intrusive advertising tools such as banner ads or pop- up ads is not examined. Previous studies have indi- cated that the use of banner advertisements can lead to positive outcomes (Chatterjee 2005) . However, in the context of third party review platforms, where users tend to trust peer reviews more than vendor information in banner advertisement, the effects of banner advertisement need to be further investigated.

Second, our study analyzes at an aggregate level the effects of WOM and promotional marketing. It does not model individual level perceptions and response to these two different information cues. We believe that measuring how users process and react to different information cues is important to a more in-depth understanding of the effectiveness of different marketing tools. Future research may be con- ducted along this line to complement our study.

Third, the effects of online coupons and keyword sponsored search are far from being fully explored in this research. Online coupons play a complicated

Appendix A

Restaurant name

Self-reported price WOM valence

WOM volume

“Sponsoring vendor” indicator

Coupon offering indicator

role in consumer decision making because they not only raise consumer awareness of products but also reduce purchase prices. Other marketing efforts and WOM, however, are information-based only. Further comparison studies between coupons and other mar- keting tools may need to control for the price sensi- tivity of individual consumers as well as to separate the redemption value of coupons from their advertis- ing value. In addition, only a very small percentage of restaurants in our sample experimented with the key- word search strategy. Although keyword sponsored search is a popular marketing tool on search engine sites, how this tool actually works in a third party review context warrants further investigation. When more firms adopt this strategy on third party review platforms, how it works in combination with coupon offerings and how it interacts with WOM information remain interesting research questions.

Finally, our data set only includes the transactions of ABC.com members, which accounts for only a small portion (on average less than 15%) of a restau- rant’s total revenue. The impacts of WOM and pro- motional marketing on nonmember users are not considered. Therefore, the effects of online WOM and promotional marketing may be underestimated in our study. Practitioners should take this into account when applying our research results to their online WOM and marketing efforts.

Acknowledgments The authors would like to thank the Senior Editor, the Asso- ciate Editor, and the three anonymous reviewers for their thoughtful reviews and constructive suggestions during the review process. The authors also acknowledge the sup- port of the National Natural Science Foundation of China [Project 71229101, 71128002, 70828003].

Appendix B

References Ajzen I, Fishbein M (1980) Understanding Attitudes and Predicting

Social Behavior (Prentice-Hall, Englewood Cliffs, NJ).

Animesh A, Ramachandran V, Viswanathan S (2010) Quality uncer- tainty and the performance of online sponsored search mar- kets: An empirical investigation. Inform. Systems Res. 21(1): 190–201.

Ansari A, Mela CF (2003) E-customization. J. Marketing Res. 40(2): 131–145.

Arndt J (1967) Role of product-related conversations in the diffu- sion of a new product. J. Marketing Res. 4(3):291–295.

Ba S, Pavlou P (2002) Evidence of the effect of trust building technology in electronic markets: Price premiums and buyer behavior. MIS Quart. 26(3):243–268.

Bajari P, Hortacsu A (2003) Winner’s curse, reserve prices and endogenous entry: Empirical insights from eBay auctions. RAND J. Econom. 34(2):329–355.

Bansal HS, Voyer PA (2000) Word-of-mouth processes within a services purchase decision context. J. Service Res. 3(2): 166–177.

Bawa K, Shoemaker RW (1987) The effects of a direct mail coupon on brand choice behavior. J. Marketing 24(4):370–376.

Bawa K, Srinivasan SS (1997) Coupon attractiveness and coupon proneness: A frame work for modeling coupon redemption. J. Marketing Res. 34(4):517–526.

Bayus BL (1985) Word of mouth: The indirect effects of marketing efforts. J. Advertising Res. 25(3):31–39.

Beuscart JS, Mellet K (2008) Business models of the Web2.0: Adver- tising or the tale of two stories. Comm. Strategies (November): 165–182.

Bhatnagar A, Ghose S (2004) Online information search termination patterns across product categories and consumer demograph- ics. J. Retailing 80(3):221–228.

Blattberg RC, Neslin SA (1990) Sales Promotion Concepts, Methods, and Strategies (Prentice-Hall, Englewood Cliffs, NJ).

Buttle FA (1998) Word of mouth: Understanding and managing referral marketing. J. Strategic Marketing 6(3):241–254.

Chatterjee P (2001) Online reviews: Do customers use them? Adv. Consumer Res. 28:129–133.

Chatterjee P (2005) Changing banner ad executions on the Web: Impact on clickthroughs and communication outcomes. Adv. Consumer Res. 32:51–57.

Chen J, Liu D, Whinston AB (2009) Auctioning keywords in online search. J. Marketing 73:125–141.

Chen P-Y, Wu S-y, Yoon J (2004) The impact of online recommen- dations and consumer feedback on sales. Paper presented at the Internat. Conf. Inform. Systems, Washington, DC.

Chen Y, Xie J (2005) Third-party product review and firm marketing strategy. Marketing Sci. 24(2):218–240.

Chen Y, Xie J (2008) Online consumer review: Word-of-mouth as a new element of marketing communication mix. Management Sci. 54(3):477–491.

Chevalier JA, Mayzlin D (2006) The effect of word of mouth on sales: Online book reviews. J. Marketing Res. 43(9):345–354.

Chiang J (1995) Competing coupon promotions and category sales. Marketing Sci. 14(1):105–122.

Chintagunta PK, Gopinath S, Venkataraman S (2010) The effects of online user reviews on movie box office performance: Account- ing for sequential rollout and aggregation across local markets. Marketing Sci. 29(5):944–957.

Chiou-Wei S-Z, Inman JJ (2008) Do shoppers like electronic coupons? A panel data analysis. J. Retailing 84(3):297–307.

Clemens E, Gao G, Hitt L (2006) When online reviews meet hyper- differentiation: A study of the craft beer industry. J. Manage- ment Inform. Systems 23(2):149–171.

Dellarocas C (2003) The digitization of word of mouth: Promise and challenges of online feedback mechanisms. Management Sci. 49(10):1407–1424.

Dellarocas C, Zhang X, Awad NF (2007) Exploring the value of online product reviews in forecasting sales: The case of motion pictures. J. Interactive Marketing 21(4):23–45.

Dreze X, Hussherr F (2003) Internet advertising: Is anybody watch- ing? J. Interactive Marketing 17(4):8–23.

Duan W, Gu B, Whinston AB (2008a) Do online reviews matter?— An empirical investigation of panel data. Decision Support Sys- tems 45(4):1007–1016.

Duan W, Gu B, Whinston AB (2008b) The dynamics of online word- of-mouth and product sales—An empirical investigation of the movie industry. J. Retailing 84(2):233–242.

Ellison G, Fudenberg D (1995) Word-of-mouth communication and social learning. Quart. J. Econom. 110(1):93–125.

eMarketer (2008) US online advertising: Resilient in a rough economy. http://alacrastore.com/storecontent/markintel/ EMARKETER-50188978.

Forman C, Ghose A, Wiesendfeld B (2008) Examining the rela- tionship between reviews and sales: The role of reviewer identity disclosure in electronic markets. Inform. Systems Res. 19(3):291–313.

Ghose A, Yang S (2009) An empirical analysis of search engine advertising: Sponsored search and cross-selling in electronic markets. Management Sci. 55(10):1605–1622.

Godes D, Mayzlin D (2004) Using online conversations to study word-of-mouth communication. Marketing Sci. 23(4): 545–560.

Granovetter M (1973) The strength of weak ties. Amer. J. Sociol. 78(6):1360–1380.

Gu B, Park J, Konana P (2012) The impact of external word-of- mouth sources on retailer sales of high-involvement products. Inform. Systems Res. 23(1):182–196.

Guernsey L (2000) Everybody’s an expert on everything. The New York Times.

Hennig-Thurau T, Gwinner KP, Walsh G, Gremler DD (2004) Elec- tronic word-of-mouth via consumer-opinion platforms: What motivates consumers to articulate themselves on the Internet? J. Interactive Marketing 18(1):38–52.

Hogan JE, Lemon KN, Libai B (2005) Quantifying the ripple: Word-of-mouth and advertising effectiveness. J. Advertising Res. 44(3):271–280.

Hu N, Zhang J, Pavlou PA (2009) Overcoming the J -shaped distri- bution of product reviews. Comm. ACM 52(20):144–147.

Katona Z, Sarvary M (2010) The race for sponsored links: Bid- ding patterns for search advertising. Marketing Sci. 29(2): 199–215.

Katz E, Lazarsfeld PF, Roper E (1955) Personal Influence: The Part Played by People in the Flow of Mass Communications (Free Press, New York).

Leone RP, Srinivasan SS (1996) Coupon face value: Its impact on coupon redemptions, brand sales, and brand profitability. J. Retailing 72(3):273–289.

Li X, Hitt LM (2008) Self selection and information role of online product reviews. Inform. Systems Res. 19(4):456–474.

Liu Y (2006) Word-of-mouth for movies: Its dynamics and impact on box office receipts. J. Marketing 70(3):74–89.

Luca M (2011) Reviews, reputation, and revenue: The case of Yelp.com. Working Paper 12-016, Harvard Business School, Cambridge, MA.

Mangold WG, Miller F, Brockway G (1999) Word-of-mouth com- munication in the service marketplace. J. Services Marketing 13(1):73–89.

Martin TW (2009) Coupons are hot. Clipping is not. The Wall Street Journal (February 25), http://www.wsj.com/article/ SB123551425475363603.html.

Milnik MI, Alm J (2002) Does a seller’s reputation matter? Evidence from eBay auctions. J. Indust. Econom. 50(3):337–349.

Montgomery JD (1992) Job search and network composition: Impli- cations of the strength-of-weak-ties hypothesis. Amer. Sociol. Rev. 57(5):586–596.

Murray K (1991) A test of service marketing theory: Consumer information acquisition activities. J. Marketing 55(1):10–25.

NCH (2009) Surge in online activity and overall coupon usage, http://www.reuters.com/article/2009/03/11/idUS191459+11 -Mar-2009+PRN20090311.

Rao AR, Monroe KB (1988) The moderating effect of prior knowl- edge on cue utilization in product evaluations. J. Consumer Res. 15(2):253–264.

Rosen E (2000) The Anatomy of Buzz: How to Create Word-of-Mouth Marketing (Doubleday, New York).

Sen S, Johnson EJ (2000) Mere-possession effects without possession in consumer choice. J. Consumer Res. 24:515–534.

Senecal S, Nantel J (2004) The influence of online product rec- ommendations on consumers’ online choices. J. Retailing 80(2):159–169.

Shimp TA, Kavas A (1984) The theory of reasoned action applied to coupon usage. J. Consumer Res. 11(3):795–809.

Sonnier GP, McAlister L, Rutz OJ (2011) A dynamic model of the effect of online communications on firm sales. Marketing Sci. 30(4):702–716.

Suri R, Monroe KB (2003) The effects of time constraints on con- sumers’ judgments of prices and products. J. Consumer Res. 30(1):92–104.

Teevan J, Alvarado C, Ackerman M, Karger D (2004) The perfect search engine is not enough: A study of orienteering behav- ior in directed search. Paper presented at the Proc. SIGCHI Conf. Human Factors Comput. Systems, New York.

Tellis G (2004) Effective Advertising (Sage, Thousand Oaks, CA).

Trusov M, Bucklin RE, Pauwels KH (2009) Effects of word-of-mouth versus traditional marketing: Findings from an Internet social networking site. J. Marketing 73(5):90–102.

Villas-Boas M, Winer R (1999) Endogeneity in brand choice models. Management Sci. 45(10):1324–1338.

Wilson EJ, Sherrell DL (1993) Sources effect in communication and persuasion research: A meta analysis of effect size. J. Acad. Mar- keting Sci. 21(2):101–112.

Yang S, Lu S, Lu X (2012) Modeling competition among paid search advertisers. Working paper, Marshall Business School, Univer- sity of Southern California, Los Angeles.

Yao S, Mela CF (2009) Sponsored search auctions: Research oppor- tunities in marketing. Foundations Trends Marketing 3(2):75–126.

Yin W, Dubinsky AJ (2004) Framing effects of coupon face value on coupon redemption: A literature review with propositions. J. Marketing Management 20(7/8):877–896.

Zeithaml V, Bitner MJ (1996) Services Marketing (McGraw-Hill, New York).

Copyright 2013, by INFORMS, all rights reserved. Copyright of Information Systems Research is the property of INFORMS: Institute for Operations Research and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.