Journal of Retailing and Consumer Services 41 (2018) 288–295 Contents lists available at ScienceDirect Journal of Retailing and Consumer Services journal homepage: www.elsevier.com/locate/jretconser Incentivized reviews: Promising the moon for a few stars a,⁎ a b Maria Petrescu , Kathleen O’Leary , Deborah Goldring , Selima Ben Mrad a b a T Nova Southeastern University, USA Stetson University, USA A R T I C L E I N F O A B S T R A C T Keywords: Incentivized reviews Exchange Theory Word-of-mouth Influencer marketing This paper studies the motivations behind incentivized consumer reviews generated via influencer marketing campaigns. Exchange theory is applied as a theoretical framework to analyze, in a qualitative and a quantitative study, the relationship between incentivized reviews and the satisfaction ratings assigned by consumers to a product. The main contributions of the study find that incentivized campaigns can contribute to a sustained increase in the number of reviews and have the potential to lead to higher purchase potential. Moreover, this study also uncovers that incentivized electronic word-of-mouth, in the form of consumer reviews, leads to increased consumer interest and desire to find out more about the product through search engines. Our findings also show that the scope of exchange theory can be broader, from an exchange between two parties to more complex relationships, between brands, influencers, and consumers, through an emerging, specialized word-ofmouth technique. 1. Introduction Incentives are a necessary component of an influencer marketing campaign because only a small fraction of social media influencers will write a positive or negative review without even the slightest incentive, which may include both money and product samples. On the influencer side, it is hard to self-identify, join, stay motivated and maintain a longterm commitment to brands, as incentives are regularly needed and real-life consumption is expected to be aligned with influencer communication. High-value influencers will only work with brands they perceive to be authentic and trustworthy. Influencers can find that their online fame also results in an ‘always on’ lifestyle, which can be physically and emotionally exhausting. Incentives are an effective method to engage with influencers and get them to recommend a product or service (Wirtz and Chew, 2002). Despite the efforts of businesses to legitimately engage in influencer marketing, the level of integrity and unethical conduct in fake reviews is still a problem. Amazon has recently filed a lawsuit against more than 1000 unidentified individuals who were allegedly selling fake reviews on Fiverr.com for products sold on Amazon (Soper, 2015). Businesses have started campaigns to incentivize buyers with a variety of discounts and promotions for posting positive recommendations. However, some companies still have disguised business-generated reviews as consumer recommendations in the anonymity of the Internet (Mayzlin, 2006, 2014). An entire industry has developed around influencer marketing, including managing both the brand side and the influencer side with sophisticated databases to manage campaigns and track influencer activity. These companies can support brands by finding the best Word-of-mouth marketing is a brand-initiated strategy of intentionally persuading consumer-to-consumer conversations (Kozinets, 2010). Social media enabled word-of-mouth marketing is a powerful mechanism for quickly disseminating positive message about brands. Some examples of WOM marketing tools include viral videos, contests, challenges, teaser advertisements, and influencer marketing. Its purpose is to encourage organic word-of-mouth, without intervening in the posting process and the content of consumer generated conversations. Influencer marketing, a relatively new word-of-mouth marketing technique, goes even further, by compensating connected social media participants who have the credibility, following, and motivation to drive positive word-of-mouth to a broader and salient segment of the market. Influencer marketing focuses on consumer-to-consumer campaigns in which the identified influencer receives incentives to post positive messages about a brand so that it permeates throughout their valuable network via electronic word-of-mouth (eWOM). In this case, consumer reviews are not organic. Reimer and Benkestein (2016) found that eWOM is positively impacted by consumer incentives; however, this also entails a possible risk for the perception of the company, especially when reviewers receive material rewards. Influencer marketing is not without challenges. On the brand side, it is difficult to identify, recruit, activate, and retain high-value influencers. Brands want influencers who have an authentic personal brand and are trustworthy. Influencer outreach is costly and time-consuming. ⁎ Corresponding author. E-mail addresses: mpetresc@nova.edu (M. Petrescu), koleary@nova.edu (K. O’Leary), goldring@stetson.edu (D. Goldring), sbenmrad@nova.edu (S. Ben Mrad). http://dx.doi.org/10.1016/j.jretconser.2017.04.005 Received 23 December 2016; Received in revised form 21 March 2017; Accepted 22 April 2017 Available online 02 May 2017 0969-6989/ © 2017 Elsevier Ltd. All rights reserved. Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. campaigns on consumers’ search for information, as well as purchase intent. The study also provides more questions for future research in detecting the differences between fake product reviews, incentivized product reviews, and organic/non-incentivized product reviews. The paper is organized as follows. First, the literature related to incentivized online reviews is summarized. An exploratory, qualitative study is performed on reviews from Amazon.com, to examine the differences between reviews from consumers who purchased the product and those who received incentives. Then, a conceptual framework is proposed to enable our hypothesized relationships, and a multimethods study is conducted to test the hypotheses. Finally, the results are presented with managerial implications. influencers and providing small incentives in exchange for posting on social media or product review sites. BzzAgent (www.bzzagent.com) is one of the largest influencer marketing agencies. This type of campaigns through third-party firms are seeded marketing campaigns (SMCs) (Chae et al., 2016). They consist of potential influencers responding to surveys, being matched with products and then directed to post their comments on different social media sites and online retailing websites. Smiley360 (smiley. socialmedialink.com) offers products and coupons in exchange for posting on personal blogs or social media. Influenster (influenster. com) requires that influencers have a blog, YouTube channel, or Twitter account to receive product samples. While posting fake consumer reviews is hard to control, this situation has resulted in manufacturers, retailers and third-party companies developing technology platforms for systematically managing customer reviews (Zhou and Duan, 2015). Research on the effectiveness of these marketing campaigns is scarce. There are few studies (Anderson and Simester, 2014; Chae et al., 2016; Zhou and Duan, 2015) that offer an explanation on how consumers respond to these techniques or if they might perceive any of them as disingenuous. Not a lot is known about how product reviews on a manufacturer's website influence those on a retailer's portal or if fake reviewers contribute to sales and positive word-of-mouth, although reviews, in general, have a positive impact on consumers. Most of the research has looked at the characteristics of ideal seeds (Hinz et al., 2011), the relationship between loyalty and consumer seeds (Godes and Mayzlin, 2004), the spillover effect of seeded WOM on marketing campaigns (Chae et al., 2016), and the types of brands that will generate more WOM. However, to our knowledge, there have not been studies looking at the consequences of word-of-mouth seeds on consumers. Therefore, this research examines the consequences of consumer reviews generated by influencer marketing campaigns as a result of receiving a product for free. We draw on exchange theory as the theoretical framework (Gatignon and Robertson, 1986; Ryu and Feick, 2007). Our study is the first attempt using exchange theory as the main framework for understanding the motivation behind being involved in incentivized reviews. Exchange theory has been used for referral and reward programs (Ryu and Feick, 2007), but not in the case of online incentivized reviews. We analyze influencers’ motivations behind involvement in incentivized reviews using both a qualitative and a quantitative study. Our objective in the qualitative study is to differentiate between verified reviews and incentivized reviews regarding the positive versus negative sentiments that consumers show, as well as regarding the primary characteristics of the product, brand, and purchasing process that are important, including price and quality, in the two types of reviews. These differences are then further explored in a quantitative analysis in which we analyze the relationship between incentivized reviews and the satisfaction ratings assigned by reviewers to the respective product. Also, we focus on the receiving end of the communication model and analyze the reactions of consumers who are the recipients of the message. Considering the word-of-mouth literature, this paper looks at the relationship between incentivized review campaigns and their effect on the quantity and the sentiment of reviews. It also explores the relationship between incentivized review campaigns and consumers’ further search for information about the product, as well as their purchase intentions. Overall, this study characterizes several aspects of incentivized consumer-generated reviews, an area of considerable importance to sellers of products and services online and to consumers who rely on the authenticity of these reviews. We also analyze the differences in the number of reviews and the satisfaction ratings that happen even after the campaign, as well as their impact on purchases. From a theoretical standpoint, this paper fills a gap in the literature by providing an understanding of incentivized reviews motivations. From a managerial perspective, this article can show the effect of seeded marketing 2. Incentivized online reviews A wide range of information is available to consumers during the buying decision process, including electronic word-of-mouth and consumer-generated content. Because of this array of information, consumers have gained considerable power, by communicating their satisfaction or disappointment with a product or purchase experience in the online environment, usually by clicking from one to five stars in addition to writing about their product experiences. Even though there are many tools through which consumers can communicate, including social media posts, blogs, and forum discussions, the most widely used are online customer reviews (Casaló et al., 2015; Kostyra et al., 2016; Moon and Kamakura, 2016). There are different platforms where customers can post comments and reviews (positive and negative), including the manufacturer's website, the online retailer's website (Walmart, Amazon, and others), as well as on review aggregators, whose primary purpose is to host reviews, such as Yelp and TripAdvisor (Anderson and Simester, 2014; Chevalier and Mayzlin, 2006; Godes and Mayzlin, 2004; Munzel, 2015). Research has even found that customers show more interest in usergenerated product information on the Internet than toward information provided by businesses, while online reviews are the second most trusted source of product information, after family and friends (Salehan and Kim, 2016). Online customer reviews have been defined as “peergenerated product evaluations posted on a company or third–party websites” (Mudambi and Schuff, 2010). Customer reviews generate more sales (Mudambi and Shuff, 2010; Chen et al., 2008; Clemons et al., 2006), affect consumers’ trust (Pavlou and Gefen, 2014) and create more word-of-mouth spillover effect (Chae et al., 2016; Kostyra et al., 2016; Moon and Kamakura, 2016). Some reviews are posted spontaneously by non-seeded consumers, and some others are posted by seeded users, which are usually incentivized by manufacturers, retailers or third-party companies offering free samples or products free of charge in exchange for consumer reviews and electronic word-of-mouth in social media. In return for these seeded WOM reviews, consumers can have access to free trials or sampling of electronic products, such as software programs, either with limited functionality or for a limited period, and free book previews on Amazon and Google (Zhou and Duan, 2015). Influencers who participate in these programs are usually motivated to be as active as possible to be included in other future incentivized opportunities. For example, each consumer with an account at BzzAgent receives points for reviews posted and messages shared on social media, and the standing and ranking of the account changes according to this activity. The better the ranking of the account, the more incentive options the particular consumer will receive. In summary, incentivized reviews are based on an exchange between the manufacturer, distributor or a third-party company and the influencer. Influencers are motivated to post reviews when they receive a sample, a discount coupon or another material incentive in return for their posts. Therefore, we define incentivized reviews or seeded reviews as online product and service reviews posted on e-tailers or review websites as a result of an incentive received by the reviewer. 289 Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. reactions to deceptive attempts. For example, Amazon has recently filed a lawsuit against more than 1000 unidentified individuals selling fake reviews for products sold on amazon.com (Soper, 2015). Moreover, Amazon has also started deleting incentivized reviews that seem to have higher ratings than the average review (Anderson and Simester, 2014; Perez, 2016). Many companies who have used incentives to create biased user reviews have encountered the risk of reducing their usefulness, especially in the case of branded, well-known products (Mayzlin, 2006, 2014). Even the use of manipulated reviews from third-party companies, who provide influencer marketing and online reputation management services represents a risk of deceiving consumers (Munzel, 2015; Zhou and Duan, 2015). Studies have shown that external user reviews, including reviews from retailer websites, can be more influential on sales than internal comments from a manufacturer's web page. Research also noted that the volume of consumer reviews on a retailer's website increases consumer awareness of the brand and contributes to the bottom line (Petrescu, 2011; Zhou and Duan, 2015). The question that arises is how can we distinguish between the incentivized review and the non-incentivized one. Understanding the differences between these two types of reviews is paramount, and an exploratory study is conducted for this purpose. Research has found that around 10% of goods are subject to online review manipulation (Hu et al., 2012). Moreover, an analysis of Amazon by ReviewMeta has also shown that incentivized reviews tend to have a higher rating than non-incentivized ones, which pushed Amazon to begin deleting them, even retroactively (Perez, 2016). Other studies found that deceptive reviews are a growing problem overall, but with different growth rates across communities. The growth rates are influenced by signaling costs associated with deception for each review community, including posting requirements (Ott et al., 2012). Researchers also noted that the prevalence of useful comments depends on the platform used, media type and the degree of polarization among commenters (Momeni et al., 2013). For example, reviews posted via mobile devices are more timely, shorter, and more negative than web reviews. At the same time, positive reviews are on average shorter than negative reviews (Piccoli and Ott, 2014). Even though researchers have focused on electronic word-of-mouth and consumer reviews, they did not consider all incentivized reviews aspects. They have not studied extensively the impact of online customer reviews on the consumption decision process or the consequences of online reputation marketing as well as corporate meddling in the reviewing process, by posting fake reviews or paying for consumer reviews (Anderson and Simester, 2014; Kostyra et al., 2016; Reimer and Benkestein, 2016). For example, Dellarocas (2006) found that firms are locked in a competitive ‘rat race’ and are forced to spend resources on review posting activities, in the hopes of minimizing perceptions of brand damage. However, the author also notes that the social cost of this online manipulation can be reduced by a filtering technology that allows the elimination of biased or dishonest reviews. Moon and Kamakura (2016) also analyzed ways to deal with reviewer bias and writing style to obtain consumer insights from reviews, such as identifying common words and writing styles that these reviewers would use. At the same time, there are also deterrents that uncover and penalize fake reviews, including fines for businesses that publish fake reviews. In 2013, Samsung was fined $340,300 for hiring people to write fake positive reviews about their products and posting negative ones on their competitors (Munzel, 2015). For example, during the month of May 2016, Yelp issued 59 new Consumer Alerts, which are notices that are placed on a business page, indicating that they attempted to pay for better reviews (Streitfeld, 2016). A website, reviewskeptic.com, was developed by Cornell University researchers to identify fake hotel reviews with approximately 90% accuracy. Some businesses are even trying to remove their name from review websites or take action against the trend of fake reviews, by encouraging their customers to post funny, unfavorable reviews for them (Streitfeld, 2016). Finally, in May 2015, the FTC issued new guidelines for both brands and influencers entitled The FTC's Endorsement Guides: What People Are Asking, which provides detailed information on solicitation and disclosure that provides greater transparency to the consumer on incentivized or compensated influencer marketing campaigns. The FTC suggests tagging incentivized reviews with words like sponsored, promotion, or ad with or without a hashtag (#). Research has shown that online review websites are among the Internet platforms that are most affected by deceptive communication efforts (Munzel, 2015). Mayzlin (2006) analyzed firms who manipulate the anonymity of online reviews that promoted their products. The study provides a model for a unique equilibrium where the promotional chat remains credible despite the use of deceptive messages. Studies also show that using deceptive messages is costly to the firm, which means that it is not optimal to produce high volumes of these messages (Mayzlin, 2006). Articles have also noted that using deceptive practices is much easier in the online environment than offline since deception detection in electronic media is usually harder than in face-to-face circumstances (Anderson and Simester, 2014). Nevertheless, this topic requires more attention and research, due to new technologies and new business initiatives, innovations, as well as 3. Exploratory analysis In a qualitative study, we analyzed incentivized consumer reviews and their positive or negative tone, versus reviews posted by customers who bought the product and posted their reviews organically and spontaneously. We explored similarities and differences among reviews to identify any possible trends that might be further helpful in the quantitative analysis. This type of analysis is especially important, considering the difficulties faced by major e-tailers, such as Amazon, which started deleting incentivized reviews. Consumers are starting to become aware of review manipulations and are paying more attention to the text of the review and noticing any positive/negative sentiment (Bambauer-Sachse and Mangold, 2013). Therefore, text mining methods to analyze consumer text data regarding associated sentiment and word usage are considered a promising approach to take advantage of the potential of this information (Agnihotri and Bhattacharya, 2016; Rese et al., 2014). We collected consumer reviews from Amazon for a facial hair trimmer from a major brand. Amazon includes both qualitative reviews, in the form of comments, and summary statistics, in the form of fivestar ratings (Kostyra et al., 2016). The incentivized review campaign ran through BzzAgent. Influencers received the product for free and were asked to post comments on social media websites, as well as reviews on Amazon. They were instructed to write in their reviews that they received the product for free in a “program run” by the major brand company. The incentivized reviews were posted on Amazon during a 30-day period. For our qualitative analysis comparison, we used a total of 1558 incentivized reviews and 305 verified reviews posted from actual buyers from Amazon. We performed a content analysis using NVIVO, version 11, and the first technique used was a simple, yet efficient standard term frequency measure (Hu et al., 2012). We created a word cloud illustrated in Fig. 1, based on verified and incentivized reviews. Regarding the most frequent words, it can be observed in Fig. 1 and 2 that the discussions are centered around the same issues, such as the battery, blades, the attachment tools and the price. Incentivized reviews are also focused on the fact that influencers received the product for free in exchange for a review. Based on our qualitative analysis, we found that price has a higher importance for consumers who bought the product than the ones who received the sample for free. The product's quality and its potential to last for years are also more important for actual buyers, as seen in the word map in Table 1. Actual product customers discuss price, money, 290 Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. to only 27% for incentivized reviews. The same difference is encountered regarding very positive sentiment, with 30% for verified reviews versus 40% for incentivized reviews. The top negative words were identified in reviews as being: bad, duped, expensive, disappointed, disappointing, broke, poor, waste, weak and terrible. These adjectives were more commonly encountered in the text for the reviews posted by buyers (0.37% of text) than the reviews posted by influencers who received incentives (0.14%). In summary, the qualitative analysis shows that influencers who participate in the incentivized campaign are more likely to focus on the positive aspects of the products and to use positive words in their reviews, and are recommending the product reviewed more than the non-incentivized ones. Overall, this exploratory study shows differences that will be further explored on a theoretical basis. 4. Conceptual framework Fig. 1. Verified Reviews. The relationship between the influencer and the company in the incentivized process is based on the exchange between both. As found in our qualitative study, influencers tend to recommend the product when they receive an incentive. Also, positive words are being used in their posts to express their positive feelings towards the product. Therefore, we use exchange theory framework to examine how consumers respond to incentivized review programs (Gatignon and Robertson, 1986; Ryu and Feick, 2007). This theory states that the transmission of information is based on a cost/benefit analysis by the communicator or influencer, following the same rules of utility maximization in economic decisions (Gatignon and Robertson, 1986). In other words, reviewers will engage in an incentive program by weighing the rewards and the costs of such a post. The potential rewards for the communicator from this exchange of information come from decision support and justification, social status and power. While influencers engage in such a relationship for an economic gain, they also share their experience to reduce cognitive dissonance, help others in their choices, and be perceived well by others (Ryu and Feick, 2007). However, the greatest risk in the incentivized program is mainly economic since the influencer does not have a direct relationship with the consumer. iInfluencers might spend time and energy posting their reviews, responding to others reviewers' comments (Ryu and Feick, 2007) and even dealing with dissatisfaction in case the receiver of the information is unhappy. However, the economic cost that the influencer will encounter is much greater than the social cost. In case the review is found to be inaccurate, the influencer might be perceived as biased, and therefore he might encounter a low internal rating by the company. Therefore, the influencer might face a risk in which his relationship with him/her and the receiver might suffer, but also an economic one since he won’t be able to access free or highly discounted products (Ryu and Feick, 2007). This framework was used by previous research to analyze the impact of rewarded referral programs. Studies found that rewards increase consumers’ referral likelihood and referrals without any extrinsic rewards may create feelings of inequity (Ryu and Feick, 2007). In the context of eWOM, Reimer and Benkestein (2016) concluded that monetary and material incentives increase the likelihood of online recommendations. An increase in reviews can also be obtained through altruistic incentives, such as donations to a social Fig. 2. Incentivized Reviews. and how expensive the product is. Moreover, buyers also talk about their customer service experiences when contacting company representatives regarding the problems they encountered when using the product. However, service experience, durability, and quality seem not to be topics of interest for the participants in the incentivized campaign (Table 2). Nevertheless, it is worth noticing that recommending the product appears to be more important for individuals who received the shaver for free in the review campaign. Our qualitative research did not focus only on the most important attributes for each user of the product, but also on the feelings expressed by the two types of reviewers. Therefore, sentiment analysis is used to identify the positive and negative language in the text, using a classical text mining method. The method consists of employing a technique of machine learning and natural language processing performed by NVIVO. The negative versus positive sentiments were automatically generated based on sentiment lexicons of negative and positive words (Hu et al., 2012). Thus, we found that 34% of the text analyzed for verified reviews exhibited negative sentiments, compared Table 1 Word map verified reviews. like works battery love good well easy price bought call blade time blades recommend manager nice money 291 quality batteries best attachment expensive cheap shower expect Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. Table 2 Word map incentivized reviews. free like good nice battery time works shower love price attachment smooth trimmers size work quality recommend easy well best blade blades handle attachments razors influencers are significantly motivated to post reviews that are considered of higher quality since they do not want to be publicly accountable for comments such as “did not find it useful” or "report this comment." Even though they are not explicitly told to post positive reviews, social desirability bias might intervene in this type of situation and influencers will tend to post favorable reviews. In fact, based on the exploratory study previously performed, influencers who receive a gift to write a review tend to use a more positive tone and a more favorable formulation of the review. The qualitative analysis of Amazon reviews showed that incentivized influencers tended to recommend the product more often and had a more positive sentiment than buyers of the product. Therefore, we believe that influencers on average are more likely to post positive reviews due to the incentives they receive, especially if their activity can lead to more free products and rewards. While influencers seem to be still bound by the social risks of posting bad reviews, the potential benefits they get outweigh the potential hazards. Thus, we posit that incentivized reviews will tend to be more positive than regular reviews. project and the incentive to help other consumers and the company itself. Considering exchange theory in the context of incentivized user reviews, the exchange is even more complicated than in the case of word-of-mouth, because it includes communication between multiple parties. The influencer will communicate with both consumers who read the review, and with the manufacturer, retailer or influencer marketing company who manages the incentivized campaign. Influencers who receive a particular incentive to engage in word-ofmouth about a product will base their decisions and actions on the perceived costs and benefits of the exchange between them and the influencer marketing company (Gatignon and Robertson, 1986; Ryu and Feick, 2007). Potential costs included in the transaction involve a low internal rating received from the influencer marketing company managing the campaign, if the review is not considered acceptable, which might put the reviewer on hold regarding future campaigns and free products. Potential benefits, however, will consist in an economic gain with free or discounted products. Incentivized reviews managed by third-party companies are usually based on offering free samples or products free of charge in exchange for consumer reviews and electronic word-of-mouth in social media, such as from BzzAgent. The digital format also allows for free trials or sampling of electronic products, such as software programs, either with limited functionality or for a limited period, on CNET download and free book previews on Amazon and Google (Zhou and Duan, 2015). In this case, considering the Exchange Theory framework and the cost/benefit ratio, we hypothesize that there will be a significant increase in the number of reviews from a third-party free sampling action, even though consumers are not paid to post reviews and access to free products is not conditioned by posting of reviews. H2. There is a positive relationship between the incentives provided for reviews and the average overall positive ratings assigned to the reviewed product. At the same time, researchers and practitioners are interested not only in how influencers respond when being asked to write incentivized reviews, but also on whether their posts will contribute to the creation of more organic word-of-mouth and, more importantly, sales (Duan et al., 2008a; Verlegh et al., 2013; Zhou and Duan, 2015). Positive word-of-mouth reviews with positive sentiment are more read and perceived to be helpful (Salehan and Kim, 2016). Previous research has also demonstrated the positive relationship between online consumer reviews and sales (Chevalier and Mayzlin, 2006; Kostyra et al., 2016). One of the risks that incentivized reviews might present is deception, or, as the persuasion knowledge model shows, a possible change in the effectiveness of the review if the reader feels a persuasion attempt (Godes and Mayzlin, 2004). The persuasion knowledge model is focused on how people use their knowledge of persuasion motives and tactics to interpret, evaluate and respond to influence attempts from marketers and others. This can be especially applied in the type of incentivized reviews where influencers state that they received the product for free to write a review about it, an occurrence very common on different online platforms. In these cases, if consumers perceive an ulterior motive of the influencer for the review or recommendation, it might even have negative consequences, depending on many factors, such as the profile of the company (Bambauer-Sachse and Mangold, 2013; Godes and Mayzlin, 2004; Mayzlin, 2006; Reimer and Benkenstein, 2016; Verlegh, 2013). Moreover, research on online product reviews found that users employ different information cues, such as disclosed information about the reviewer, the consensus between a review and the average ratings of the product (Munzel, 2016). In the case of incentivized reviews, a consensus among reviewers due to higher and positive ratings could have an impact on trustworthiness and further purchases. The sentiment of reviews, the quantity, and H1. There is a positive relationship between offering incentives and the number of reviews posted online. While influencers are usually driven by the economic gain of the exchange relationship between them and other consumers, they are also driven by social obligations and do not want to be accountable for their untruthful recommendation (Gatignon and Robertson, 1986). Reviewers on some platforms, such as Amazon, are easier to identify and hold accountable, as it not as easy to open an Amazon account every day. For other platforms, it is more facile to use throw-away accounts just for particular review purposes, with a lower social risk. From this point of view, consumers who post online reviews are held within ethical and truthful boundaries by their social obligations, but also by the possibility of being held accountable publicly with mechanisms such as “report this comment,” “did not find useful,” down-voting and other tools. Also, based on the exchange theory model, consumers use word-of-mouth for their desire to help and reduce any post-purchase dissonance (Ryu and Feick, 2007). That is why influencers are usually driven by posting reviews as complete and honest as possible (Gatignon and Robertson, 1986; Ryu and Feick, 2007). Influencers who participate in incentivized campaigns have a profile with the company managing the incentivized campaign and are evaluated based on the number and quality of their reviews. Then, 292 Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. consumers rate the satisfaction with the product from one to five “stars,” with five being the highest level of satisfaction. The number of verified reviews was used as a proxy variable for purchase potential, as verified reviewers are only persons who already purchased the product. We also measured daily consumer search for information related to the product by using the Google Insights index for the exact name of the product, reflecting the searches performed for that keyword in the Google search engine. To test the effect of the intervention and to see if the incentivized review campaign will lead to an increase in organic reviews even after the campaign, we started our analysis with an event study, a time-series analysis in the form of an autoregressive integrated moving average model (ARIMA) (Box et al., 2008; Kling et al., 2009). ARIMA is a type of model that accounts for autocorrelation and divides the pattern of a time series into three components: the autoregressive component, which describes how observations are related to each other because of being close together in time, the differencing component, which is used to make a time series stationary, and the moving average component, which describes outside “shocks” or interventions to the system. After coding the period when the incentivized reviews were posted as the intervention period, the ARIMA analysis was performed in SPSS, for the variable representing the number of reviews. The results of the analysis are presented in Table 3. As the results show, the intervention is significant (p < 0.001) for the number of reviews, which underlines the fact that the incentivized campaign had a significant effect (p < 0.001) on the number of reviews posted about the respective product during the period when the campaign was run. We continue the quantitative analysis with a MANOVA analysis in which we test the differences as a function of the period (pre, during and post-campaign), in the number of incentivized reviews and the number of verified reviews, the satisfaction stars assigned to the reviewed product and the Google search index representing consumers’ external search for information. We analyzed 74 days before the campaign, 39 days during, and 57 days post-campaign. The results of the MANOVA procedure are presented in Table 4. The four core indicators (Pillai's Trace, Wilks' Lambda, Hotelling's Trace and Roy's Largest Root) showed significance for the overall model analyzed, with p < 0.001. Regarding the variables of interest, the number of satisfaction stars awarded to the product was the only variable not found significant in the model, which will be discussed in the results section. We also performed a multiple comparisons test to assess the differences in the values of the variables analyzed as a function of the three time periods examined. This test is shown in Table 5. These results will be discussed in the following section. quality of information in reviews are also determinants that affect consumer purchase intentions (Agnihotri and Bhattacharya, 2016; Floh et al., 2013; Plotkina and Munzel, 2016). However, research has found that online word-of-mouth can be used successfully as a marketing tool because consumers have the power to regulate it through comments and votes. Research has underlined an active feedback mechanism between electronic word-of-mouth and sales because word-of-mouth leads to sales, which then leads to more word-of-mouth (Dellarocas, 2006; Duan et al., 2008a, 2008b; Godes and Mayzlin, 2004). In this case, word-of-mouth is seen not only as an exogenous variable but also as an endogenous one. For example, research considers the effect of online user reviews as an indicator of the intensity of word-of-mouth, which plays a significant role in box office revenues (Duan et al., 2008b). In this case, an incentivized campaign will also tend to have an effect of attenuation on possible organic negative reviews, decreasing the impact of negative reviews on sales. Therefore, despite the potential risks, and since more word-ofmouth can generate more sales, we hypothesize that there is a positive relationship between running a review campaign and further purchases. H3. There is a positive relationship between incentivized review campaigns and purchase potential. Nevertheless, the number of reviews available online has a significant effect on sales and can also lead to increased awareness (Duan et al., 2008b; Petrescu, 2011). Awareness is a critical outcome because even if consumers are not convinced of the review or recommendation they read, the Internet offers them a wide range of other sources of information to verify the objectivity of what they read. Exchange theory posits that the degree to which information from one source is compatible with the receiver's information as will enhance the acceptance of the information by the receiver (Gatignon and Robertson, 1986). Since online reviews, especially incentivized, can be unclear regarding their source characteristics, including credibility, which is an essential element in exchange theory, consumers might want to search for further information about the product before making a purchase decision. Therefore, we estimate that consumer information search for the product subjected to an incentivized review campaign will increase so that the receiver verifies whether the source is a credible one. H4. There is a positive relationship between incentivized review campaigns and consumer information search. After performing a qualitative analysis in the previous section and considering the theoretical support provided by literature, this study will now focus on a quantitative analysis of the relationship between offering rewards in exchange for reviews and the outcome of these campaigns. 6. Results and discussion The results of the ARIMA procedure show that the intervention of the incentivized reviews campaign has created an effect that lasted within the post-campaign period when the reviews posted on Amazon regarding the analyzed product came primarily from the influencers who purchased the product. The ARIMA and MANOVA procedures both provide support for Hypothesis 1 and show that the campaign had a positive effect in creating word-of-mouth and buzz around the product. The multiple comparisons analysis indicates that the total number of reviews is the highest during the campaign, followed by the post- 5. Methodology and analysis We continue the exploratory study performed with data collected from Amazon reviews with a quantitative analysis. We manually downloaded and analyzed the reviews posted for a shaving and trimming product from a major brand on Amazon from January 1st to September 30th. During this time, the incentivized review campaign managed by a third-party company was run between May and June, and it was reflected between May 21st – June 28th on Amazon's website. During the nine months analyzed, 1558 reviews posted were incentivized, and 305 were from verified shoppers. According to Amazon's online information, when a product review is marked Amazon Verified Purchase, it means that the customer who wrote the review purchased the item at Amazon.com. We quantified the number of reviews and the number of verified reviews per day and the average satisfaction “stars” per day, where Table 3 ARIMA Results. Non-Seasonal Lags Regression Coefficients AR1 Period Estimates Std. Error t Approx. Sig −0.222 44.488 0.076 7.458 −2.934 5.965 0.004 0.001 Melard's algorithm was used for estimation. 293 Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. other individuals or avoiding criticism and negative comments from other consumers (Gatignon and Robertson, 1986). Nevertheless, while the satisfaction rating might not be significantly different from one period to another, as the qualitative study showed, the text of the reviews posted by incentivized influencers had a positive sentiment in a higher proportion than that of actual buyers. The non-significant difference in review “stars” can also be due to consumers’ importance placed on the text of the reviews, especially considering that consumers are starting to be aware of manipulated reviews and need more information than just a “star” rating (Bambauer-Sachse and Mangold, 2013; Petrescu, 2011). Moreover, research has also found that consumers can discover the manipulation of numeric ratings, which might be another reason why numeric rating alone is no longer a significant indicator of potential sales (Hu et al., 2012). Looking over the remaining results of the MANOVA procedure and the multiple comparisons, we also find support for Hypothesis 4. The results show that consumers’ search for information on Google differs as a function of the period pre, during and post-campaign. Consumer interest in the product and external search for information is significantly higher in the third period, during the post-campaign months. This might be influenced by word-of-mouth from other sources, including social media posts that were encouraged during the same incentivized campaign from BzzAgent. Table 4 MANOVA results. Dependent Variable Mean Square F Sig. Nr. reviews Satisfaction Nr. verified reviews - Purchase potential Information search 22835.98 1.768 4.254 4329.394 209.037 1.901 8.53 17.838 0.001 0.153 0.001 0.001 campaign months. The campaign also influenced an increasing trend in sales, leading to a steady increase in the number of reviews even after the campaign has ended. This applies in the context of verified reviews too, used as a proxy for purchase potential, since the results of the MANOVA analysis provides support for Hypothesis 3. This shows a positive relationship between running an incentivized review campaign and possible purchases, once consumers have a significant amount of positive reviews to read about the product. The multiple comparisons analysis underlines the fact that the number of verified purchases is significantly higher during the postcampaign period. The results confirm previous literature noting that electronic word-of-mouth has a significant effect on product purchases, if we consider that there is a very clear correlation between product purchase potential and the number of verified reviews (Chevalier and Mayzlin, 2006; Dellarocas, 2006; Duan et al., 2008b; Godes and Mayzlin, 2004). The multiple comparisons test also confirms previous studies that found that the number of reviews can positively influence consumers’ opinion about a seller, as a surrogate for the seller's experience and time on market (Petrescu, 2011). An interesting observation that can make the subject of future studies is related to the fact that the number of verified reviews is the lowest during the campaign, lower even than during the pre-campaign months. Looking over previous literature, including the exchange theory (Gatignon and Robertson, 1986) and research related to free sampling (Zhou and Duan, 2015), there could be two main explanations for this occurrence: product buyers were less motivated to post a review due to negative influences from the campaign or the hundreds of free products offered through this campaign cannibalized potential sales. The fact that incentivized reviewers had to disclose that they received the product for free might have influenced this outcome. This can bring further support for some studies which found that the potential interest of the review writer might decrease its credibility, create some skepticism for review readers and reduce trust (Bambauer-Sachse and Mangold, 2013; Reimer and Benkenstein, 2016). However, when analyzing the MANOVA results, the satisfaction ratings assigned to the product in the form of “stars” from one to five do not differ as a function of the period, despite some signs that there are potential linguistic and sentiment differences found in the qualitative analysis. This might be explained by the exchange theory model of interpersonal communication and the reviewers’ process of weighing costs versus benefits (Gatignon and Robertson, 1986). Some of the factors that could have determined whether incentivized influencers provided low ratings might have been social, including impressing 7. Conclusions This study analyzes the evolution and outcomes of incentivized review campaigns, in a context where more companies are starting to use revolutionary marketing techniques to increase the number of product reviews on internal and external websites. The main contributions of the study are both theoretical and managerial. From a theoretical standpoint, this paper is the first one using exchange theory to understand the relationship between an influencer and other consumers when using seeded reviews. Exchange theory has been used in word of mouth reward programs (Ryu and Feick, 2007), but not in incentivized reviews. Our work shows that the scope of exchange theory can be broader, from an exchange between two parties to more complex relationships between influencers, consumers, and influencer marketing agencies, and it can also be applied to specialized online word-of-mouth situations. Our findings show that incentivized campaigns, even managed by third-party companies, result in influencers posting positive reviews because of the benefit they are getting from posting the reviews and the potential for further incentives and campaigns. Engagement in the incentivized reviews depends on a cost/benefit analysis similar to reward programs. In an incentivized review, an influencer is motivated by an economic gain (the sample he/she is getting in return). As a result, the influencer can contribute to a lasting increase in the number of reviews with a more positive sentiment and with the potential to lead to higher sales. Regarding the practical implications of these results, managers should know that dealing with business-generated word-of-mouth or consumer-generated reviews based on rewards is a process that may have the appearance of deception. There are different interpersonal Table 5 Multiple Comparisons. Mean Variable Nr. reviews (per day) Satisfaction (number of stars) Nr. verified reviews (per day) – Purchase Potential Information search (Google search index) Pre-campaign (0) 1.324 4.1655 1.176 49.649 Mean differences Campaign (1) 40.410 4.5295 0.872 57.795 * Significant at 0.001 level. 294 Post-campaign (2) 1.561 4.3547 1.474 66.018 (0–1) −39.086 −0.3639 0.304 −8.146* (1–2) * (0–2) * 38.849 0.1748 −0.602* −8.223* −0.237 −0.1892 −0.298 −16.369* Journal of Retailing and Consumer Services 41 (2018) 288–295 M. Petrescu et al. Duan, W., Gu, B., Whinston, A.B., 2008b. The dynamics of online word-of-mouth and product sales—An empirical investigation of the movie industry. J. Retail. 84 (2), 233–242. Floh, A., Koller, M., Zauner, A., 2013. Taking a deeper look at online reviews: the asymmetric effect of valence intensity on shopping behavior. J. Mark. Manag. 29 (5–6), 646–670. Gatignon, H., Robertson, T.S., 1986. An exchange theory model of interpersonal communication. Adv. Consum. Res. 13, 534–538. Godes, D.B., Mayzlin, D., 2004. Using online conversation to study word-of-mouth communication. Mark. Sci. 23 (4), 545–560. Hinz, O., Skiera, B., Barrot, C., Becker, J.U., 2011. Seeding Strategies for Viral Marketing: an Empirical Comparison. J. Mark. 75, 55–71. Hu, N., Bose, I., Koh, N.S., Liu, L., 2012. Manipulation of online reviews: an analysis of ratings, readability, and sentiments. Decis. Support Syst. 52, 674–684. Kling, G., Ghobadian, A., O'Regan, N., 2009. Organic growth and shareholder value: a case study of the insurance industry. Int. J. Res. Mark. 26 (4), 276–283. Kostyra, D.S., Reiner, J., Natter, M., Klapper, D., 2016. Decomposing the effects of online customer reviews on brand, price, and product attributes. Int. J. Res. Mark. 33 (1), 11–26. Kozinets, R.V., de Valck, K., Wojnicki, A.C., Wilner, S.J.S., 2010. Networked narratives: understanding word-of-mouth marketing in online communities. J. Mark. 74 (2), 71–89. Mayzlin, D., 2006. Promotional chat on the Internet. Mark. Sci. 25 (2), 155–163. Mayzlin, D., 2014. Promotional reviews: an empirical investigation of online review manipulation. Am. Econ. Rev. 104 (8), 2421–2455. Momeni, E., Cardie, C., Ott, M., 2013. Properties, Prediction, and Prevalence of Useful User-generated Comments for Descriptive Annotation of Social Media Objects. Association for the Advancement of Artificial Intelligence (〈www.aaai.org〉). Moon, S., Kamakura, W.A., 2016. A picture is worth a thousand words: Translating product reviews into a product positioning map. International Journal of Research in Marketing, http://dx.doi.org/http://dx.doi.org/10.1016/j.ijresmar.2016.05.007. Mudambi, S.M., Schuff, D., 2010. What makes a helpful online review? A study of customer Reviews on Amazon.com. MIS Q. 34 (1), 185–200. Munzel, A., 2015. Malicious practice of fake reviews: experimental insight into the potential of contextual indicators in assisting consumers to detect deceptive opinion spam. Rech. Et. Appl. En. Mark. 30 (4), 24–50. Munzel, A., 2016. Assisting consumers in detecting fake reviews: the role of identity information disclosure and consensus. J. Retail. Consum. Serv. 32, 96–108. Ott, M., Cardie, C., Hancock, J., 2012. Estimating the Prevalence of Deception in Online Review Communities. WWW 2012 – Session: Fraud and Bias in User Ratings, April 16–20, 2012, Lyon, France, 201-210. Pavlou, P.A., Gefen, D., 2014. Building effective online Marketplaces with institutionbased Trust. Inf. Syst. Res. 15 (1), 37–59. Perez, Sarah, 2016. Amazon makes good on its promise to delete incentivized reviews. Tech Crunch. Retrieved from 〈https://techcrunch.com/2016/11/23/amazon-madegood-on-its-promise-to-delete-incentivized-reviews/〉. Petrescu, M., 2011. Online price dispersion – more than imperfect information. J. Product. Brand Manag. 20 (7), 541–548. Piccoli, G., Ott, M., 2014. Impact of mobility and timing on user-generated content. MIS Q. Exec. 13 (3), 147–157. Plotkina, D., Munzel, A., 2016. Delight the experts, but never dissatisfy your customers! A multi-category study on the effects of online review source on intention to buy a new product. J. Retail. Consum. Serv. 29, 1–11. Reimer, T., Benkenstein, M., 2016. Altruistic eWOM marketing: more than an alternative to monetary incentives. J. Retail. Consum. Serv. 31, 323–333. Rese, A., Schreiber, S., Baier, D., 2014. Technology acceptance modeling of augmented reality at the point of sale: can surveys be replaced by an analysis of online reviews? J. Retail. Consum. Serv. 21, 869–876. Ryu, G., Feick, L., 2007. A penny for your thoughts: referral reward programs and referral likelihood. J. Mark. 71, 84–94. Salehan, M., Kim, D.J., 2016. Predicting the performance of online consumer reviews: a sentiment mining approach to big data analytics. Decis. Support Syst. 81 (1), 30–40. Soper, S., 2015. Amazon targets web freelancers peddling fake customer reviews. Bloomberg. October 16th. Retrieved from 〈http://www.bloomberg.com/news/ articles/2015-10-16/amazon-targets-web-freelancers-peddling-fake-customerreviews〉. Streitfeld, D., 2016. Online reviews? Researchers give them a low rating. The New York Times, June 7, 2016. Retrieved from 〈http://nyti.ms/1WDOSbn〉. Verlegh, P.W.J., Ryu, G., Tuk, M.A., Feick, L., 2013. Receiver responses to rewarded referrals: the motive inferences framework. J. Acad. Mark. Sci. 41, 669–682. Wirtz, J., Chew, P., 2002. The effects of incentives, deal proneness, satisfaction and tie strength on word‐of‐mouth behavior. Int. J. Serv. Ind. Manag. 13 (2), 41–162. Zhou, W., Duan, W., 2015. An empirical study of how third-party websites influence the feedback mechanism between online word-of-mouth and retail sales. Decis. Support Syst. 76, 14–23. communication issues that contribute to consumers’ decision to review a product, including reviewers’ social motivations to seek power and recognition or to help others. This study noted many aspects that need to be considered regarding these consumer decisions. Regarding how consumers process an incentivized review, deception is also an important factor. In our example, potential deception was mitigated by asking reviewers to announce that they received the product for free; however, this has the potential to negatively influence consumers’ intentions to pay for the product. Regarding the effectiveness of such a review, even if consumers might not trust a review, they may have a higher interest in the product and search for more information. Nevertheless, as the results of this study show, an incentivized review campaign, even led by a third-party influencer marketing company, has significant positive benefits on consumers’ word-of-mouth, interest, and sales. Future studies can further develop this model and use primary data to assess influencers’ review behavior when participating in an incentivized campaign, as well as focus on ways to identify this potentially deceptive word-of-mouth, by analyzing the content of reviews and the different types of technologies already used in practice. Data from other websites and incentivized campaigns, with different product types, can make these results more generalizable can also help. Even though Amazon is screening fake reviews, inflated reviews seem to be also another problem that additional research needs to focus on. Also, future studies can analyze consumer response to the reviews posted as a result of this campaign, differences in attitude formation, as well as their interest and ability to make distinctions between regular and incentivized reviews. In addition, future studies should focus on understanding whether the trustworthiness of reviews submitted is directly influenced by higher review ratings provided by incentivized reviewers. Indeed, trustworthiness merits further examination. Research can study at what point the "value" of the product to be reviewed motivates/demotivates prospective users to participate in the review process, or determines them to "adjust" their commitment to providing an in‐depth versus a superficial review. Future research can also focus on the type of incentives being used and their impact on the quality and quantity of the reviews. References Agnihotri, A., Bhattacharya, S., 2016. Online Review Helpfulness: role of Qualitative Factors. Psychol. Mark. 33 (11), 1006–1017. Anderson, E.T., Simester, D.I., 2014. Reviews without a purchase: low ratings, loyal customers, and deception. J. Mark. Res. 51 (June), 249–269. Bambauer-Sachse, S., Mangold, S., 2013. Do consumers still believe what is said in online product reviews? A persuasion knowledge approach. J. Retail. Consum. Serv. 20, 373–381. Box, G.E.P., Jenkins, G.M., Reinsel, G.C., 2008. Time Series Analysis: Forecasting and Control, 4th edition. John Wiley and Sons, Inc., New York. Casaló, L.V., Flavián, C., Guinalíu, M., Ekinci, Y., 2015. Avoiding the dark side of positive online consumer reviews: enhancing reviews' usefulness for high risk-averse travelers. J. Bus. Res. 68, 1829–1835. Chae, I., Stephen, A.T., Bart, Y., Yao, D., 2016. Spillover effects in seeded word-of-mouth marketing campaigns. Mark. Sci. 36 (1), 89–104. http://dx.doi.org/10.2139/ssrn. 2676047. https://ssrn.com/abstract=2676047. Chevalier, J.A., Mayzlin, D., 2006. The effect of word of mouth on sales: online book reviews. J. Marketing Res. 43. pp. 345–354. Clemons, E., Gao, G., Hitt, L., 2006. When Online Reviews Meet Hyperdifferentiation: a Study of the Craft Beer Industry. J. Manag. Inf. Syst. 23 (2), 149–171. Dellarocas, C., 2006. Strategic manipulation of Internet opinion forums: implications for consumers and firms. Manag. Sci. 52 (20), 1577–1593. Duan, W., Gu, B., Whinston, A.B., 2008a. Do online reviews matter? — An empirical investigation of panel data. Decis. Support Syst. 45, 1007–1016. 295