Electronic word of mouth and the hidden impact of incidental influencers

Marketers are commonly assessed on key metrics such as brand recognition and returns on marketing spend. Another important target is to get people spreading positive messages about the brand across social networks. This is commonly referred to as “Word of Mouth” (WOM).

WOM is defined as “the process of conveying information from person to person” and includes consumers sharing their opinions about products and services (Jansen, Zhang, Sobel, & Chowdury, 2009). The impact of WOM on consumers is widely acknowledged (You, Vadakkepatt, & Joshi, 2015) and it is a significant marketing tool (Sundaram, Mitra, & Webster, 1998). While WOM is often transmitted in person, the internet and particularly social media are alternative channels for such transmissions. On digital platforms, opinions and information can be easily sent to a wide network of people through actions such as liking, sharing, and commenting on social media posts. Contacts who are exposed to this information can, in turn, share it with their own network. This form of WOM is commonly referred to as electronic word of mouth (eWOM) and eWOM is now a significant mode of WOM transmission. Due to its importance as well as the accessibility of measurements, the scope of this study will focus predominantly on eWOM.

Due to its profound brand impact, many organizations place considerable emphasis on WOM management and this is often part of the marketing function (Rialti, Zollo, Kim, & Yoon, 2024). In an attempt to spread WOM while maintaining control over its positivity, marketing efforts have typically focused on ensuring customer satisfaction, managing dissatisfaction effectively and organizing brand campaigns that target both current and potential customers (whom we will collectively refer to as Market Participants or MPs for short) (Zhong & Zhong, 2024). Similar strategies have also been adopted for eWOM (Velnadar et al., 2024).

On electronic platforms, targeting campaigns at MPs is made easier with built-in segmentation features. It is a sound strategic approach by marketing to narrow its focus on MPs who either bring current businesses or may do so soon. However, from our systematic behavioral observations, a counterintuitive hypothesis emerges that a substantial amount of eWOM is also created by “Non-Market Participants” (NMPs), people who are neither current nor potential customers. They seemingly have no interest or usual market relations with the brand, yet create eWOM about it.

If our hypothesis is correct, there is a considerable weakness in current eWOM strategies. Companies are so narrow in their focus on the MP segment that they risk missing out on a large group of people, the NMPs, who spread eWOM about their brands as well. While NMPs may seem to be unimportant since they are neither current nor potential customers in the near future, this is actually a myopic perspective. When eWOM is spread by MPs, brand messages will likely be diffused among people of similar profiles and social circles. This is grounded in the principle of homophily (Lazarsfeld & Merton, 1954), where people who buy the same products or brands tend to have the same background (Ma, Krishnan, & Montgomery, 2014). Therefore, there is a limit to how far these eWOM can be dispersed. Godes and Mayzlin (2004) observed that spreading messages across diverse social networks is more effective for increasing eWOM than narrow, targeted communication. NMPs are, therefore, the heterophilous links that allow brand messages to cross disconnected social circles and help in the wide dispersion of eWOM. When eWOM extends into new communities, it can expand brand awareness to previously untapped potential customers (Rogers, 1962; Granovetter, 1973).

While there are significant theoretical and practical implications for NMPs' role in eWOM diffusion, there is very limited research on this subject. This is the theoretical gap that we are addressing in this study.

To study the suspected phenomenon from our systematic behavioral observations of NMPs creating eWOM, we have two research questions (RQs). In RQ1, we will first determine whether our hypothesis of NMPs creating eWOM can be backed by empirical evidence:

If answers for RQ1 show that NMPs are indeed generating eWOM, it is important for us to then find out their motivations for doing so. As such, RQ2 is as follows:

Word of mouth as a communication medium is likely to be as old as the time when humans started communicating. However, it was not until the 1950s that this subject was studied in the business field. Most literature traces the term “Word of Mouth” to a 1954 magazine article by William Whyte (Kimmel & Kitchen, 2014; Pyle, 2010). While walking around a neighborhood, Whyte noticed that air-conditioning (a new product at that time) was fitted by households in clusters. A similar trend was observed with television antennas. From this, Whyte concluded that new product adoption behaviors were being influenced through communications with immediate neighbors.

In 1955, Elihu Katz and Paul Lazarsfeld theorized that communication happens in a two-step flow with opinion leaders receiving information from mass media and subsequently dispersing it through WOM (Katz & Lazarsfeld, 1955). Scholars from the 1960s to the 80s continued to study this two-step communication model (Deutschmann & Danielson, 1960; Rogers, 1962; Sheth, 1971) and discovered that while WOM had reinforcement functions, most people still received their firsthand news directly from the mass media.

Interest in WOM research continued to intensify in the following decades. As people became more self-centered, there was an increased focus on the individual consumer (O'Brien, 1971; Sheth, 1971). O'Brien (1971) tracked the cognitive journey of a consumer from awareness to actual purchase decision. He listed out the different stages, including “advertising,” “attitude,” “awareness,” “favorableness of WOM” and termed it the hierarchy of effect theory. Advertising did not exert a direct impact on final purchase decisions. Instead, the strongest influence originated from WOM.

With this early research, the influence of WOM on consumer behavior became increasingly apparent to marketers. Subsequent studies shifted focus beyond examining WOM's impact to exploring strategies for encouraging consumers to generate positive WOM for brands (Akrimi & Khemakhem, 2012). Market segmentation emerged as an important tool to channel these efforts, aiming both to stimulate WOM and to ensure that such messages reached the most relevant audience segments (Ring, Tkaczynski, & Dolnicar, 2016; Riley, 2016).

Taking on a different perspective, Brown and Reingen (1987) looked at how the strength of social ties and homophily affected the diffusion of WOM on both macro and micro levels. They found that both weak and strong social ties had unique roles to play in the flow of WOM. Weak ties were the bridges that allowed information to flow across social groups, while strong ties were more influential. This, again, highlights the important role that NMPs play as weak ties to a brand and, therefore, a conduit for brand messages to spread across social groups. This social network approach to WOM research took on more complexity with the advent of the internet, social media and consequently eWOM. Brown, Broderick, and Lee (2007) believed that “existing theory maybe inappropriate to describe online WOM and its influence on evaluation and purchase.”

Research on eWOM increased in recent times (Chu & Kim, 2011; Kim, Sung, & Kang, 2014; Mardhiyah, Dharmmesta, & Purwanto, 2013). Chu and Kim (2011) focused their study on social media as a burgeoning platform for consumer-to-consumer brand-related eWOM. While these networking sites are important eWOM platforms, relatively little is known about why and how eWOM takes place on these sites. eWOM can be classified into three kinds – opinion seeking, opinion giving and opinion passing. While opinion seeking and giving are well examined with offline WOM, the lines between these roles tend to blur in online spaces as people take on different roles on different occasions. Opinion passing on social media comes from acts of “sharing” and is more likely to happen online due to click-of-the-mouse convenience (Chu & Kim, 2011).

Verma and Yadav (2021) discussed another emerging area of eWOM research. This is the recommender system adopted by platforms such as Netflix and Amazon. Referred to as a location content-aware recommender system (LCARS), it offers recommendations to consumers based on where they are, what they are interested in and options that are highly rated by others (Qiu, Gao, Cheng, & Guo, 2016). Such systems are becoming common for online platforms with growing user acceptance. As e-Commerce platforms gain popularity, Beck, Koskie, and Locander (2023) also studied the process through which these platforms earn the trust of users. Contrary to common belief, native reviews are perceived as eWOM from weak ties and are less trusted compared to reviews gained through “live” online interactions.

From our literature review, we can see that both WOM and eWOM research have been silent about NMPs. Like management practice, MPs are also at the center of attention. As such, a gap exists in current eWOM literature on understanding NMPs. If NMPs are indeed creating eWOM, it is important to address this gap.

Most research on eWOM has been conducted in the field of marketing. Therefore, these works are mostly anchored on marketing-related theories and constructs (Chu & Kim, 2011). While they help us to understand more about MPs and eWOM, they do not provide insights into NMPs whose motivations may be very different. As we are researching eWOM on digital media platforms for this study, we need a foundational theory that allows us to take on perspectives about media users' behavior. As such, we believe that the “Uses and Gratifications” (U & G) theory is well-suited to help us find answers for RQ2.

The origin of U & G theory can be traced back to the 1940s, when research in mass communication shifted from the influence of mass media on a seemingly passive audience to an active audience who put media to use that suits them and the gratifications they receive along the way (Baran & Davis, 2009). One of the earliest works in U & G was by Hertha Herzog, who studied gratifications received by housewives from listening to soap operas on the radio (Herzog, 1944). These include emotional release, identification with daily realities and dreams of a better life. U & G has subsequently been adopted to study many other communication mediums, including television, magazine and cinema.

Over the years, critics of U & G questioned its legitimacy as a theory (Klapper, 1963). Katz, Blumler, and Gurevitch (1973) countered that U & G “does rest on a body of assumptions… that have some degree of internal coherence…” and that this theoretical framework consists of five key tenets:

1. The audience is active and goal directed.

2. Linking need gratification with media choice sits with the audience.

3. There are many ways of fulfilling human needs and media is only one out of many.

4. Users are self-aware and able to report their interests and motives in their media usage.

5. There should not be any value judgments on what gratifications audiences seek, to meet specific needs through specific media and content.

In recent times, U & G has also been used to study internet usage (Zhang, Tang, & Leung, 2011; Eginli & Tas, 2018). With social media, content consumers can now also be producers as the two roles become interchangeable (Quan-Hasse & Young, 2010). With an unprecedented level of interaction between audience and medium, it can be argued that U & G, and its theory on active audience, is perfectly suited to study the internet as a medium. It is also through this process of interaction that eWOM is generated, positioning U & G as an appropriate foundational theory for our study.

Lee and Ma (2011) recognized the “appropriateness of the U & G approach in the context of the online environment and its potential explanatory ability in predicting individuals' news sharing behaviors.” They also noted that most U & G research on social media found similar gratifications such as entertainment, socialization, information and status seeking. In the following section on methodology, we have also adopted Lee and Ma (2011) as a basis to design our survey.

Our research consists of two studies. For both, we used the automotive market in Singapore for data collection. There are three key reasons why this segment is suitable for our subject matter.

1. “Vehicles” is one product category with the most active WOM (Allsop, Bassett, & Hoskins, 2007). Cars are expensive consumer durables and there are substantial risks in a wrong purchase (You et al., 2015). Therefore, buyers will tend to seek opinions, including through eWOM, before making any decisions.

2. Singapore's automotive market is unique in several ways. Besides public access to data from the Land Transport Authority (LTA), most automotive brands are represented by only one dealer with one showroom. As such, many of these dealerships operate like a brand owner, distributor and retailer rolled into one. This streamlined structure helps to reduce noise in our data that may be present in usual multi-dealership markets.

3. Most automotive brands in Singapore are active with social media marketing, especially on Facebook. In addition, they have Singapore-dedicated pages which are separate from their international versions. This, again, helps to reduce noise in our data.

In Study 1, we investigate whether our hypothesis on NMPs creating eWOM (RQ1) holds true by employing a deductive methodology. In contrast, we can measure eWOM for a brand on social media by the number of “likes,” “shares” and “comments,” measuring the number of MPs and subsequently deriving the number of NMPs poses a challenge. From LTA data, we know the number of current customers for every automotive brand. However, it is difficult to quantify the number of potential customers.

To overcome this challenge, we restrict our scope to a few unique car brands where, due to price positioning, the number of people who currently own or can afford to have these cars on their consideration lists is very limited. These are niche, luxury brands that include Aston Martin, Bentley, Ferrari, Lamborghini, Rolls-Royce and McLaren. The number of MPs for these brands is limited to a very exclusive group of people, making it easier for estimation. Since the MP-NMP classification is mutually exclusive and collectively exhaustive, if we can find out the number of eWOM creators for one of these brands and compare that with the estimated number of MPs for that brand, it is then a deductive step to take the difference (if positive) as the number of NMPs creating eWOM. This is illustrated with the following formula:

To conduct this exploration, we use Bentley as our brand of investigation as it is the most prominent in this luxury segment by virtue of market share. Before we estimate the number of MPs for Bentley, we provide our specific definition of an MP as follows:

1. A current owner of the brand or

2. Someone looking to buy a car of that brand in the next 12 months.

For Bentley, we know from LTA data that there are an average of 782 owners at any one time over the last 10 years (https://onemotoring.lta.gov.sg). To estimate the number of potential customers who may be looking to purchase a Bentley in the next 12 months, we first tabulated the number of Bentley cars sold per annum over the last 10 years (see Table 1).

From Table 1, we can see that Bentley's sales volume is relatively small but consistent, averaging at 72 units per annum. Therefore:

Estimated Total Number of Bentley MPs = Current Owners (782) + Estimated Potential.

Customers in the Next 12 Months (72) = 854.

Next, to estimate the number of people who create eWOM for Bentley, we accessed the brand's official Singapore Facebook page – their most active social media channel.

We looked at every post uploaded by the company for six alternate months, reducing the possibility of influence from external events on our data set. From every post, we manually noted the names of individuals who had created eWOM by “liking.” “commenting” or “sharing” that post. We then removed duplicate names to arrive at a list of unique individuals. For posts that had too many such engagements and were impractical for us to list out all the names, we removed duplicates based on proportions derived from posts for which we were able to do so.

The results, annualized into a 12-month period, are tabulated in Table 2 below.

We can now calculate the estimated number of NMPs creating eWOM for Bentley using the earlier mentioned formula:

As we can see, the number of NMPs creating eWOM for Bentley in a 12-month period stands at 2,946, significantly exceeding the number of MPs (854). This assumed that all MPs contributed to eWOM in this period. If any of them did not, the number of NMP contributors would be higher. Earlier, we defined a potential customer as someone looking to buy a car of that brand in the next 12 months. Even if we broaden the definition and use 24 or 36 months, the number of NMPs creating eWOM still exceeds MPs by nearly threefold (2,874 vs 926 for 24 months; 2,802 vs 998 for 36 months).

For Study 2, we adopt both semi-structured interviews and surveys in a convergent parallel mixed method design (Rech et al., 2024) to:

1. Reinforce Study 1 on whether NMPs are contributing to eWOM (RQ1) and

2. To uncover the motivations that lead these NMPs to do so (RQ2).

Semi-structured interview is often utilized in the field of social sciences (Judd, Newton, Newton, & Ewing, 2014) and their self-reporting nature is suitable for uncovering motivations as expounded by U & G's core tenet (Katz et al., 1973). This qualitative method is also useful to “better understand any phenomenon about which little is yet known” (Hoepfl, 1997; Strauss & Corbin, 1990), which is the case for our study on NMPs creating eWOM.

In our interviews, we asked participants the following questions:

1. Do you drive at least once a week?

2. If yes, which brand of car do you drive? Do you have plans to buy another car in the next 12 months? If yes, which brand?

3. If you do not drive at least once a week, do you have plans to buy a car in the next 12 months? If yes, which brand?

4. What were the car brands that you have engaged (“like,” “share,” “comment”) with on social media?

5. What were your motivations for doing so?

Questions 1 to 4 were designed to determine whether the interviewee is an MP or NMP. During our pre-study interviews, participants indicated uncertainty for any time frame beyond 12 months. Hence, 12 months were used in Q2 and Q3. Question 5 was an open-ended question that allowed the interviewees to freely share their motivations for creating eWOM.

There were two methods we used to recruit interview participants. First, we recruited using convenience, purposive sampling – people from our own contacts, whom we knew to be creators of eWOM for cars on Facebook. Our second recruitment method was similar to Choi, Folwer, Goh, and Yuan (2015). We accessed car Facebook pages originating from Singapore to randomly seek out individuals who had engaged with posts from these brands. In the process, we covered as many car brands as possible, ensuring a balanced mix of both entry-level and premium cars. These recruitment methodologies meant that our data came from people who created actual eWOM rather than simply possessing the intention to do so (Mardhiyah et al., 2013). This adds to the originality of our article.

A total of 30 participants were interviewed (Roscoe, 1975). Three of the interviews were conducted in person and the remaining over the phone. All responses were recorded verbatim and our initial findings are as follows:

1. Of the 30 interviewees, only 1 turned out to be an MP who owned a Korean car and created eWOM only for that brand. This further reinforces our findings from Study 1 and supports RQ1.

2. The 29 NMPs had all created eWOM for brands that they were not looking to buy from in the next 12 months.

3. Of the 29 NMPs, 12 (41%) predominantly engaged with entry segment brands and 17 (59%) with premium brands. This indicates that our research samples have a relatively balanced mix between the two segments and complement our samples from Study 1, which focuses on the luxury segment.

Next, we adopted thematic analysis by partially utilizing NVivo to systematically identify findings from the verbatim responses to Q5. Thematic analysis has been commonly done in qualitative research and can help to produce reliable findings (Nowell, Norris, White, & Moules, 2017; Braun & Clarke, 2006). We adopted Braun and Clarke (2006) six-step framework, making a distinction between top-down and bottom-up analysis. A top-down analysis takes guidance from the research questions, while a bottom-up approach is inductive from the data itself (Maguire & Delahunt, 2017; Braun & Clarke, 2006). With the top-down approach, we conducted several rounds of analysis on the full verbatim data. We listed out relevant quotes after removing those that were too repetitive or rhetorical. We subsequently went through every quote and searched for emerging themes with RQ2 motivations in mind. From this process, we identified six themes as follows:

1. Aspirational needs

2. Informational needs

3. Social needs

4. Entertainment needs

5. Financial gains

6. Verification

We next adopted Braun and Clarke (2006) bottom-up approach. This was done through two methods. First, we ran our verbatim responses through NVivo that helped us to identify keywords frequently used by our respondents. As shown in Figure 1 below, words such as “Share” (52 times), “Information” (34 times), and “Friends” (21 times) appeared with the highest frequency.

While we picked up “Information” and “Friends” (Social Needs) from our top-down analysis, “Share” was a new theme that might be important for us. This specific word was often used throughout our interviews (“Share interesting news,” “Share innovative stuff,” “Share for general knowledge,” “Share news to inform friends”). To gain a deeper insight, we ran another query with NVivo to find out words that were most frequently used in conjunction with “Share.” Results showed that words such as “Information” (48 references/24 joint occurrences) and “Friends” (30 references/15 joint occurrences) were among the top in the list. With high joint occurrences between “Share” and “Friends,” we posit that sharing is a manifestation of social needs and can be grouped together.

The second method we adopted as part of our bottom-up analysis was the use of two independent coders (IC). They had little to no knowledge of our research topic and questions, providing us with fresh perspectives that were not influenced by our RQs. We gave them the verbatim responses and asked them to list out themes that they could identify. The results are in Table 3 below.

From Table 3, we identified a new theme coming from IC 1, which might point to another motivation for NMPs. This was to gain authority in the subject matter, which in the context of our study referred to automobiles. We also calculated intercoder reliability scores using NVivo, comparing our identified themes with those from the ICs. We arrived at an agreement score of 85.46% and Cohen's Kappa coefficient of 0.69.

To reinforce our findings for RQ2, we also administered a survey, which was conducted after the qualitative interviews for the 29 NMPs. This provided us with quantitative empirical data to supplement the qualitative set. The survey was designed based on U & G literature that came closest to our context, namely Lee and Ma (2011) and Choi et al. (2015). Some refinements were done to suit our specific research needs. In total, there are six factors and four indicators for each factor (Table 4). The factors matched the themes that emerged from our interviews. Participants were asked to rate the indicators on a 5-point Likert scale, with 1 being “Strongly Disagree” and 5 being “Strongly Agree.”

Results from the survey show that “Information” has the highest average rating at 4.3, followed by “Aspiration” at 4.1 and “Social Interaction” at 3.7. Cronbach's Alpha for all six factors came in close or above 0.7 with “Information” being the only exception.

We will now discuss each of our seven themes (six from top-down analysis, one from bottom-up analysis) in detail.

Many of our NMPs (79%) had no plans to buy a car in the next 12 months. However, a handful of them continued to keep track and share news of cars that they aspired to own, even if they were out of reach (“Aspiring to change car”; “It's aspirational”).

This is another reason why NMPs can be an important group of people that marketers inadvertently miss out with narrow segment targeting. Current NMPs for a brand can become MPs in the future, especially if that brand continues to be an aspiration. In the short to medium term, they may be NMPs due to issues of affordability or they may simply be a part of the late majority (Rogers, 1962). To reinforce our qualitative findings, “Aspiration” was rated highly in our survey as well with a score of 4.1. Despite significant changes from analog to digital media, this gratification of aspirational needs has remained consistent since the early days of U & G research (Herzog, 1944).

64% of our verbatims show that informational needs motivated NMPs to create eWOM. They engaged with social media posts to gather information and knowledge (“Follow and engage for information”). Some also shared Facebook posts when they felt that the information might be useful for their network of friends (“The information is useful for my friends”; “Share for general knowledge”). These information covers a variety of topics ranging from pricing (“Share Mercedes prices to show that it is attractively priced”), product features (“Share information to find out specifications about other cars”) to new and upcoming products (“News on new product”; “Like to share on pre-launch models”).

There were several instances where we could identify informational needs in conjunction with the gratification of another need, all captured in a single quote. One of the recurring examples was where the interviewees indicated that they created eWOM as NMPs to share information with their friends. This might show that NMPs were gratifying their social needs simultaneously as well (“Share information on cars that friends drive”; “Has lots of car crazy friends”). There were also some NMPs who shared car news so that he or she could keep abreast with new information that would help in social interactions (“Share information for own self-keeping and discussion with friends”; “Keeping touch with news so as not to be left out”).

“Information” as a gratification also features prominently in U & G literature. Traditional media provides information for daily living (Katz et al., 1973) while modern digital media has been described as the optimal channel where individuals can both seek and create information (Ruggiero, 2000).

40% of our verbatims pointed to a gratification of social needs. We discussed earlier that many of the respondents implicitly cited social needs in conjunction with their informational needs when generating eWOM. These people shared social media posts when there was information that they deemed useful for friends (“Share interesting news”; “Share innovative stuff”). NMPs appear to be actively exchanging news and information with their digital network, similar to interpersonal conversations that we have with our family and friends.

However, from our verbatim data, we observed that not all sharing to gratify social needs were altruistic in nature. Some did so to expand their own network, some looked for exclusive social gatherings to attend and some simply wanted to fit in with certain peer groups (“Use sharing to network”; “Like to find out about private events and attend”; “Believe others behave the same way as well”). Two of the participants mentioned that they shared car news on social media to manage the perception that their friends had of them (“Let friends know that I can afford to spend”; “Shows that I am knowledgeable”). While this is related to social needs as an overarching construct, it also shows the desire to enhance social status as a deeper motivation. “Social needs” as a gratification is also supported by our survey with a relatively high rating of 3.7 and a Cronbach Alpha of 0.79.

Similar to aspirational and informational needs, gratification of social desires is frequently raised in U & G research as a motivation of media use (Ruggiero, 2000).

Several interviewees disclosed that they created eWOM as NMPs due to a general interest in automobiles. For some, the interest can be intensive (“Like cars and have passion for cars”; “Interested in continental cars for its performance and quality”) while others appeared to be less emotionally engaged. Besides cars, these people might also share other news as long as it was interesting (“Share interesting news”; “Share anything interesting”). For both groups, creating eWOM helped to gratify their need for entertainment and passing time.

Social media is well-suited to entertain users due to its highly interactive nature. This finding further strengthens U & G's perspective of active media users seeking entertainment, among other gratifications (Ruggiero, 2000; Ha & James, 1998).

Seven of the respondents remarked that contests with financial rewards organized by car dealers on social media motivated them to engage and generate eWOM (“Looking for freebies or contests”). In addition, others shared posts that provided information on car prices, promotions, offers etc. (“Share any offers”; “Share information on any promotions”). As such, for the NMPs who created these eWOM messages, they seemed to seek both informational gratification and a desire to gain financial benefits.

There were two interviewees who stated that the motivation for them to engage in eWOM was to verify the reliability of car information they had received from other sources (Blend different sources of information to get the truth”; “Share news to find out if it is fake or true”). Though this motivation was not highlighted frequently in our interviews, verification as a motivation to engage in eWOM was a relatively unexpected finding for us. This motivation may become more prominent as fake news and artificial intelligence-doctored pictures are getting increasingly rampant.

A motivation identified from bottom-up thematic analysis was that NMPs seek to gain authority in the subject matter of automobiles by sharing car-related news. To be a “go-to” person or influencer on anything to do with cars, NMPs must diffuse all kinds of related information. This includes information about brands that these individuals do not own.

“Financial Gains”, “Verification” and “Gain Authority” are not apparent motivators in older U & G literature. These gratifications are related to functions offered by digital medium and they emerged only when U & G theory is gradually being used to study new media (Hossain, 2019).

Our findings from this research have several important managerial and theoretical implications.

Many online platforms allow marketers to determine the type of people who will be exposed to their digital campaigns and messages. While such segmentation functions have been endorsed by social media owners as tools for more effective marketing, marketers should be aware that this could be a flawed assumption, at least when it comes to eWOM generation. Narrow segmentation forces them to focus only on MPs and risk missing out on NMPs who have important roles in diffusing marketing messages to reach previously untapped customers (Granovetter, 1973). These NMPs help to increase brand awareness and, consequently, reduce marketing expenses needed to amplify brand messages. Moreover, information from NMPs may sometimes be even more influential than that from stronger, direct ties (Steffes & Burgee, 2009). We, therefore, refer to such individuals as “Incidental Influencers” as eWOM generated by them may be more effective in converting sales. We will use this term instead of NMPs from here onward.

This study can form the foundation of new eWOM strategies for managers to adopt. Firstly, marketers should note that the gratification of informational needs ranks high among the motivations that drive these incidental influencers to engage in eWOM generation behavior. As such, managers may consider designing their social media activities to be richer in informational value. This can apply to a wide range of subjects, including product features, upcoming launches and brand stories. Our findings further show that many of them are especially eager to share interesting information about new technology or design breakthroughs. Hence, this is the type of informational tidbits that marketers should look out for and post on their social media channels whenever they are available.

From our findings, there are no discernible differences in motivations for those incidental influencers who create eWOM for premium brands and those who do so for entry-level brands. Aspiration as a gratification factor ranks equally high in both segments. In current practice, aspirational values are usually attached to premium brands, while marketing for entry-level brands often focuses on practicalities and price promotions. These marketing executions may be based on flawed assumptions, at least when it comes to eWOM. It is important for managers to recognize the aspirational value in entry brands, especially for high-value products. Social media posts for these brands can place a stronger emphasis on attractive visuals. They can also highlight new features, technology, and interesting design stories that can help to evoke ownership pride and emotions. All these can motivate incidental influencers to generate more eWOM.

Another motivation that has important practice implications is the need for verification. With the spread of fake news and AI-doctored communication materials, incidental influencers are sourcing and sharing authentic brand messages. This need is likely to continue growing. As such, brands should administer and attain verified badges on social media platforms such as Facebook, Instagram, and LinkedIn. In addition, links or icons leading to these platforms should also be prominently located on their official websites.

Besides practice implications, our research contributes to both eWOM and U & G literature. To the best of our knowledge, there are no studies done on the role and behavior of incidental influencers in both research fields. With their important function in spreading brand eWOM and bridging consumers across different social groups, it is critical that our study addresses these gaps in both eWOM and U & G literature.

Although research work on eWOM is expanding, there are limited studies adopting U & G as a theoretical framework. As discussed earlier, a core tenet of U & G is based on an active audience that puts media into self-gratifying uses. Interactive digital platforms are perfectly suited to be analyzed through the lens of U & G. Together with the use of participants who contributed actual eWOM, our study contributes to these methodological gaps in eWOM literature. Lastly, financial gains, verification and gaining of authority are gratifications that did not emerge from older U & G research. These three findings add to the theory and will be useful for future research on the digital medium.

Marketers are actively managing eWOM for their brands as digital platforms play an increasingly important role in defining brand impressions and perceptions. While businesses and scholars are often focused MPs in eWOM management, a counterintuitive hypothesis emerges that a substantial proportion of eWOM is created by NMPs, who may be more aptly described as “Incidental Influencers”. Our findings provide empirical evidence to support this hypothesis. Incidental influencers who create eWOM are motivated by aspirational, informational, social, entertainment, verification and financial needs. In addition, some of them spread eWOMs to gain subject authority among friends. Our findings expose the strategic flaws in overly narrow marketing segmentation, a tactic often used by marketers in their digital marketing campaigns. As such, they miss out on getting their brand messages to incidental influencers who help amplify brand awareness on a scale that MPs cannot achieve. Therefore, there is an imperative need to rethink and realign strategies to market realities.

Any research in eWOM is always going to be challenging due to large data sets. As such, there are various limitations in this study that should be highlighted.

First, we specifically selected the automotive industry in Singapore as our field of investigation due to data accessibility. In addition, we were only able to gather information on eWOM data, but not traditional WOM. Therefore, we can only infer that this phenomenon of incidental influencers creating eWOM may be applicable to other industries and offline WOM as well.

In Study 1, we manually noted the names of every “likes,” “shares,” and “comments” of each Facebook post from Bentley. However, some of the posts had eWOM engagements that ran into the thousands. As it was impractical to apply the same method, we removed duplicates using an average proportion derived from posts that were countable. While this might have somewhat affected the accuracy of our data in Study 1, we are confident that it should not change our conclusions that there are more eWOM creators than MPs.

Another limitation is that our research participants were all recruited from Facebook. While Facebook is one of the most popular social media platforms in Singapore, there may be potential differences in motivations if we had recruited interviewees from other social media platforms, such as Instagram.

This study sets new research directions for eWOM. Future research can focus on different social media platforms as well as other industries. This will be especially important for product or service categories that are highly dependent on eWOM. Examples will include movies, holiday destinations, and dining establishments. Researchers can also empirically examine eWOM spread by incidental influencers and measure their effectiveness in achieving specific business goals versus those generated by MPs. Besides commercial applications, future work can also explore the role incidental influencers may play in the spread of public messages, especially those related to government policies. There are often urgent needs to quickly spread these messages across communities and they may be a critical part in policy makers' communication strategies.