How Mobile Advertising Works: The Roie of Trust in ...

How Mobile Advertising Works: The Roie of Trust in Improving Attitudes and Recall

SHINTARO OKAZAKI

How does trust affect consumer attitudes and recall in mobile advertising? This study

Universidad Autonoma

explores this question by conducting a large-scale "pseudo" mobile advertising

de Madrid [email protected]

campaign in Japan. Two "real" brands (one durable and one nondurable good) of major Japanese manufacturers were used as study stimuli. Using a push messaging

AKIHIRO KATSUKURA

Dentsu Inc. akihiro.katsukura@ dentsu.co.jp

service, both a campaign message and a subsequent questionnaire for each brand were sent to 40,000 opt-in mobile users. Five primary constructs were examined: brand trust, mobile advertising trust, attitude toward brand, attitude toward mobile

MAMORU NISHIYAMA

advertising, and mobile campaign recall. Findings suggest that a mobile campaign's

Dentsu Inc.

recall largely depends on perceptions of both the medium and the advertised content,

[email protected]

and that the effects of mobile advertising trust on attitude toward mobile advertising were stronger than those of other relationships. The path from attitude toward brand to mobile campaign recall was notably and statistically stronger for the durable good sample than for the nondurable good sample. Only in the durable good sample is attitude toward brand a mediating variable in linking attitude toward mobile advertising and mobile campaign recall. This study offers a basic but useful research framework for a mobile-based online survey.

RESEARCHERS AND PRACTITIONERS AGREE that the

most explosive growth in telecommunications has occurred in internet and mobile telephone services (Lehr and McKnight, 2003). Combining these two technologies, the internet-enabled mobile device has become one of the most promising advertising media around (Okazaki, 2006). For example, SMS (short message service) direct marketing is a popular campaign method worldwide, with consumer spending in the United Kingdom reaching 58 million messages per day in 2004 (Trappey and Woodside, 2005). The trend has migrated to the United States as well. McDonald's conducted an SMS-TV direct marketing campaign in conjunction with a popular song contest program, offering concert tickets and backstage passes. Coca-Cola executed a text message campaign for college students in which a number printed on a DOI: 10.2501/S0021849907070195

Diet Coke can could be used as an entry in the Coca-Cola Grand Sweepstakes Competition. According to Mullman (2006), as many as 81 percent of 18- to 2]-year-olds have mobile phones, and most of them are likely to participate in TV or radio polls, purchase ring tones, play games, and send text messages. On the other hand, SMS is not used in Japan as widely as in Western markets. Instead, an emailcompatible version of mobile mail is commonly used in NTT DoCoMo's i-mode, which has been cited as "by far the most successful and most comprehensive example of m-commerce today" (Sadeh, 2002, p. 5). While widely used for delivering digital content such as ring-tones/songs, news, weather reports, travel and traffic reports, games, standby displays, and animated video, this service allows users to transmit email of up to June 2007

L QF HDUERTISinG RESEHRCH 1 6 5

HOW MOBILE ADVERTISING WORKS

10,000 alphanumeric characters or 10,000 bytes (NTT DoCoMo, 2006). For promotional campaigns, i-mode offers "Message F (Free)," which is a push-type service delivering advertising exclusively to opt-in users. Advertisements are delivered to a designated "Message F" inbox, and users are exempt from the normal packet transmission charges. This Message F can be sent to specific regional, gender, or age groups, thus allowing firms to target messages to selected segments (D2 Communications, 2005). In addition, an image attachment service begun in July 2005 supports the transmission of images, logos, and other visual effects up to 8KB per message (192 X 192 pixels in JPEG or GIF). In 2006, major Japanese mobile carriers are beginning to launch mobile handsets that can receive terrestrial digital broadcasting signals in addition to conventional analog signals, thus enabling firms to take advantage of interactive presentations with static, animated, and/or broadcast images (NTT DoCoMo, 2005). Hence, the use of SMS as a reference point for mobile advertising may no longer provide a realistic impression. However, there is little, if any, empirical evidence as to the question of how mobile advertising works in a more enhanced environment. Our primary objective here is to close this research gap through an experimental study of mobile advertising campaigns. The study consisted of transmitting "pseudo" mobile advertisements of popular Japanese brands, one a durable good and the other a nondurable good, to opt-in users. The advertisements contained both textual and visual parts, which is more consistent with currently prevailing mobile technology that enables firms to use rich digital content rather than simple text-based messages. One week after sending the advertisements, we sent a structured questionnaire to the same users in an attempt to examine their level of

acceptance. The study set out to examine five primary constructs: trust toward the advertised brand, trust toward mobile advertising, attitude toward brand, attitude toward mobile advertising, and mobile campaign recall. In what follows, we first critically review the extant literature on mobile advertising research, then establish the theoretical framework of the study. On this base, research hypotheses are proposed. Following a detailed explanation of the methodology, we present the study results and discuss the implications of the study while recognizing important limitations. ADVANCES IN MOBILE ADVERTISING RESEARCH

One of the benefits of a mobile-based campaign is its capacity to attract consumer attention and increase consumer responses through a "one-to-one dialogue" with customers (Kavassalis et al., 2003). This matches the characteristics of the mobile phone: a highly portable communication device with ubiquitous data transmission capability. In recent years, higher-capacity 3G technologies have responded to more specific information needs by offering always-on service with more interactive and location-based applications (Perlado and Barwise, 2004). So far, however, with the exception of some Asia Pacific countries, SMS has been the most common platform for mobile advertising (Kavassalis et al., 2003) and therefore the most studied application. Table 1 summarizes the recent advances in mobile advertising research. In a pioneering study, Barwise and Strong (2002) conducted a trial of permission-based SMS message advertising in the United Kingdom. On recruitment, respondents were paid cash incentives and received more than 100 messages in the six-week trial period. Al-

1 6 6 JDURRRL OF RDUERTISIRB RESERRCR June 2 0 0 7

most all respondents were satisfied or very satisfied. The study found that 81 percent read all messages, 63 percent responded or took action, and 17 percent forwarded at least one message. Surprisingly, as many as 84 percent of respondents were likely to recommend the service to their friends, whereas only 7 percent were likely to abandon the service. That research was later corroborated by Rettie, Grandcolas, and Deakins (2005), who conducted a study based on "real" mobile advertising campaigns that took place between October 2001 and January 2002. In total, the researchers conducted 5,401 telephone interviews regarding 26 different campaigns and found that the overall acceptability of SMS advertising was 44 percent, with an average response rate of 25 percent—much higher than email marketing. Acceptability was also significantly correlated with campaign interest, campaign relevance, and monetary incentives. Similarly, sporadic industry surveys report a rather optimistic blueprint. An experimental survey conducted by Ericsson indicates that 60 percent of consumers liked receiving mobile advertising, while Quios found that the level of recognition of mobile advertising was surprisingly high: 79 percent of participants recalled 60 percent of the advertising (Bames, 2002). In contrast, a study conducted by Tsang, Ho, and Liang (2004) reveals less enthusiastic attitudes toward mobile advertising among Taiwanese. Their structural model included both utilitarian and experiential factors affecting consumers' attitude toward permission-based SMS advertising. It was found that (1) consumers generally have negative attitudes toward mobile advertising unless they have been informed and have preconsented (i.e., opted-in) to the advertisements, and (2) there seems to be a direct and positive relationship between consumers' attitudes

HOW MOBILE ADVERTISING WORKS

TABLE 1 Prior Research on Mobile Advertising Sample

Methodology Gender

Authors

Topic

Population

Barwise and

Adoption of

General

Strong (2002) permission-

Response

(Male/

Research Size 500

Scale

Reliability

Age

Female)

Occupation

Rate

Survey

Type

Assessment

16-30

50%/50%

Unknown

100%

P&P"

Interval

CFA,'' a

consumers

scale

based push advertising Okazaki (2004)

Perceptions

General

of pull-type

consumers

800

n.a.=

n.a.

Students,

8.0%

P&P

office workers,

7-point

CFA,

a

CFA,

a

Likert

housewives.

advertising

professionals.

platform

etc. Tsang, Ho,

Adoption of

Mobile

and Liang

SMS

internet

(2004)

advertising

users

Rettie,

Adoption of

General

Grandcolas.

SMS

consumers

and Deakins

advertising

430

5,401

Under 30

Students

(85%)

(60%)

Unknown

Unknown

88.3%

P&P

7-point Likert

Unknown

25%

Phone

Ordinal

n.a.

7-point

CFA

interview

(2005) Wu and

Adoption of

Wang (2005)

m-commerce

consumers

(including

-H mobile banking

advertising)

users + students

Wang and Liao (2007)

General

Satisfaction in Mobile m-commerce

850

Under 39

Unknown

(82%)

Students,

36.7%

P&P

office workers.

Likert

etc.

116

site users

Under 4 1

76%/24%

Unknown

n.a.

P&P

(average:

7-point

CFA,

a

Likert

24)

(including advertising) 255

59%/41%

Lin and

Loyalty in

University

Students,

n.a.

Wang (2006)

m-commerce

students

high-firm

(quota

(including

-^ general

employees.

sample)

advertising)

consumers

and insurance

18-39

P&P

7-point

CFA

Likert

employees ''i^&P = paper-and-pencil survey ''CFA = confirmatory factor analysis 'n.a. = not available

and behavior. The authors noted further that consumers' intention to accept mobile advertisements is affected by incentives. The use of sophisticated statistical analysis dis-

tinguishes this study from other empirical studies carried out in recent years. Much less attention has been paid to web-based "pull" advertising. Okazaki

(2004) examined the factors influencing consumers' motives to click text banner advertisements in the i-mode mobile advertising platform known as "Tokusuru

June 2 0 0 7 JDUIlllIlL OF HDUERTISinG RESEimCH 1 6 7

HOW MOBILE ADVERTISING WORKS

Menu." This platform is included in an official i-mode menu, which enables subscribers to freely access the promotional information delivered by various companies. He found that three constructs— content credibility, infotainment, and irritation—affected the formation of attitudes toward wireless advertisements, which in turn determined tbe level of intention to click the advertisements. Interestingly, tbe demographic analysis revealed that tbe unmarried working youth segment has a higher propensity to access such pull mobile advertisements. THEORETICAL FRAMEWORK Trust as a factor determining mobile advertising acceptance

Traditionally, the relevance and credibility of advertisements have been considered important mediators of advertising effects (MacKenzie and Lutz, 1989). In particular, perceived trust can be seen as consumer feedback, reflecting individuals' determination of whether the imparted information is true or false. Trust becomes crucial when consumers must make decisions or take action on the basis of uncertain information. On the internet, however, it may be seen as more difficult to distinguish between more and less trustworthy information because, as is not the case with traditional media, much of the content of online information is not subject to governmental or ethical regulation (Eastin, 2001). In tbis light, Hoffman, Novak, and Peralta (1999) claim that a fundamental mistrust by customers of online business lends support to the view of some separation between brand identity and brand experience. Tbis lack of faith in the unregulated flow of information presents a similar problem for those seeking information via mobile devices. In this vein, Sadeh (2002) points out that the success factors associated with the i-mode m-commerce portal are precur168

sors of future internet business models, in wbicb value will be created through convenience, ease of use, and compelling content tbat users will be willing to pay for. Mobile users are likely to choose to open mobile advertisements out of curiosity, but they must then decide whether to further access mobile campaign sites. Such a decision must be made mainly on tbe basis of trust in the advertisement and in the advertised brand. Therefore, trust should be conceptualized as consisting of two different constructs, brand trust and mobile advertising trust, both of which affect users' choice to open push mobile advertising. The users are likely to do so only when they perceive both the medium and the content to be nondeceptive and trustworthy. Trust as a factor determining online branding effectiveness

Mylonopoulos and Doukidis (2003) argue tbat mobile advertising via email or SMS is effective in enhancing brand awareness and customer loyalty. Hence, tbere is strong evidence that many firms actually use mobile advertising for branding purposes. Okazaki (2005) interviewed 53 senior executives of multinationals operating in European markets and found that tbe creation and enhancement of brands are tbe most important motives for multinational corporations to adopt mobile advertising in international markets. This is consistent with Sultan and Rohm (2005), who argue, "Mobile marketing creates new opportunities for companies to form or shift consumer attitudes toward a brand through the use of value-added content" (p. 85). There is evidence that a growing number of firms actually use mobile advertising in brand promotion. For example, McDonald's offers three different types of mobile coupons on three different platforms, from the basic text-only SMS ver-

DFflDOERTISlOGRESEeRCH June 2 0 0 7

sion to the graphically rich version—that is, fully trackable and redeemable, all without any point-of-sale hardware or software requirements (iMedia Connection, 2005a). DreamWorks Home Entertainment used a viral campaign for the DVD of the film The Rin^; 2 that operated on mobile phones and email accounts. Visitors to the campaign site were invited to "scare their friends" by entering their email address and mobile telephone number. Tbe site then sent them an email, inviting them to click on a link and watch the video (iMedia Connection, 2005b). Adidas enables consumers to download popular athletes' pbotos on a mobile site and digitally superimpose their own pbotos on those images (Sultan and Rohm, 2005). In Japan, Kirin MC Danon Waters Co. has launched a campaign for Volvic mineral water in wbicb consumers enter a sweepstake contest in mobile phones with a serial bottle number (Senden Kaigi, 2004). P&G sends a sample of the Pantene brand shampoo to users who respond to pull mobile advertising in i-mode (Senden Kaigi, 2004). The success of these branding examples, however, appears to depend on the distance between brand identity and brand reputation. That is, tbe more consumers trust the brand, the smaller tbe incongruity between what firms intend to establish and what consumers identify with the brand (Jevons and Gabbott, 2000). Hence, in tbe context of mobile advertising, trust plays a crucial role in obtaining desired consumer responses. Attitude formation and mobile campaign recall

An attitude toward an object can be dofined as an individual's internal evaluation of it on tbe basis of his or her beliefs (Fishbein and Ajzen, 1975). In other words, beliefs determine tbe basic form of the attitude, which in turn triggers behavioral

HOW MOBILE ADVERTISING WORKS

intentions (Davis, 1993). Thus, exposure to an advertising message for a specific product first influences one's beliefs, which then mediate the attitude toward the product. Subsequently, behavioral intention is formed as a consequence of this attitude formation (Fishbein and Ajzen, 1975). A mobile advertisement can be seen as tl marketing stimulus in an interaction between an advertiser and a consumer that is mediated by a mobile handset. The level of exposure to mobile devices varies according to consumers' needs and wants, but it is likely that they have formed fnvorable or unfavorable attitudes, irrespective of their prior knowledge (Lu, Yu, Liu, and Yao, 2003). Research suggests that when consumers form positive attitudes toward the advertisement, they are more likely to demonstrate higher mobile campaign recall. However, a company's ultimate goal of mobile advertising should not be the formation of favorable attitudes toward mobile advertising, but the formulation of an attitude toward the advertised brand. In this vein, Delgado and Munuera (2001) have pointed out that trust is one of the most important factors affecting tbe creation of brand value. The literature has long neglected tbe importance of "brand trust," despite empirical and theoretical evidence (Hess, 1995; Seines, 1998), but researchers have become increasingly interested in establishing interactions between a consumer and a brand as a longterm relationship (Delgado and Munuera, 2001). Other research views trust as a determining factor in developing positive or favorable attitudes, resulting in a commitment to a certain brand as the maximum expression of a successful relationship between it and the consumer (Fournier, 1998). Moreover, trust leading to favorable attitudes could, in turn, influence the intention to engage in repeat purchases in internet commerce (Gefen,

Karahanna, and Straub, 2003; Kim and Benbasat, 2003), including m-commerce (Lin and Wang, 2006).

H6:

Attitude toward brand directly and positively affects mobile campaign recall.

RESEARCH HYPOTHESES

H7:

Tbe relationship between attitude toward mobile advertising and mobile campaign recall is mediated by attitude toward brand.

On the basis of tbe preceding arguments, tbis study proposes to examine trust in tbe advertised content and in the advertising medium, given that brand trust and mobile advertising trust affect favorable attitudes toward tbe advertisement and tbe brand, respectively. Both attitudes, in turn, are hypothesized to directly and positively influence the level of mobile campaign recall. Tbus, the following hypotbeses are suggested: HI:

Mobile advertising trust directly and positively affects attitudes toward mobile advertising.

H2:

Brand trust directly and positively affects attitudes toward brand.

H3:

Mobile advertising trust directly and positively affects attitude toward brand.

H4:

Attitude toward mobile advertising directly and positively affects attitude toward brand.

H5:

Attitude toward mobile advertising directly and positively affects mobile campaign recall.

At the same time, we posit tbat attitude toward brand would act as a mediating variable in linking attitude toward mobile advertising and mobile campaign recall. This is because the growing use of mobile advertising suggests that practitioners may intuitively believe that favorable perceptions of the advertising will lead to favorable perceptions of the brand advertised, which could achieve their ultimate goal: a higher level of mobile campaign recall. This suggests the following and final hypotheses:

Figure 1 shows our research model. METHODOLOGY "Pseudo" mobile campaign

This study was organized into two phases: (1) transmission of the pseudo mobile campaign, and (2) questionnaire survey via mobile device (Figure 2). In tbe first phase, we sent mobile advertisements that contained pseudo-campaign messages with hyperlinks to access further campaign information on the mobile site. D2 Communications (D2C), tbe largest mobile advertising agency in Japan, collaborated in tbis study by creating the pseudo advertisements. It offered its "Message F," a push service that delivers textual and visual information from advertisers to opt-in users only. Tbis service bas been successfully used as a campaign tool by many firms, such as P&G, Sapporo Beer, and Panasonic (Senden Kaigi, 2004). In addition, we contacted two large Japanese manufacturers, both listed in tbe first section of tbe Tokyo Stock Exchange, who agreed to collaborate in the experiment and allow us to use their most popular brands as real stimuli in tbe campaign. One company manufactures a sophisticated portable audio player (i.e., durable good), whereas the other makes a puffed com snack (nondurable good). Both brands are firmly established in tbe Japanese market. After designing textual and visual information for tbe mobile advertisements, we created a promotional mobile site to which respondents could jump by clicking a banner saying "Please click here

June 2 0 0 7 JOORML DF RDUERTISIIID RESERRCR 1 6 9

HOW MOBILE ADVERTISING WORKS

for further information" in the advertisements (Figure 2). Again, as an incentive to participate in the campaign, we offered a free ring-tone and a present (a book coupon).

Questionnaire design

In tbe second pbase, which took place approximately one week after that transmission, we contacted the customers by sending another mobile message containing a structured questionnaire and asking for tbeir impressions and perceptions of

Figure 1 Research Model

Push Mobiie Advertising

structured Questionnaire

MelJI

Carl

Present for 10,000 people Sales campaign! Buy our Carl and win a sweepstake, because 10,000 people can get a mysterious picture book. Free screen image, applications, too... Please click here for

One week later

NTT DoCoMo: This message is free of packet transmission charges.

Invitation to the survey.

Figure 2 Research Design 1 7 0 JDDeRHL DF RODEHTISIRG RESERRCR June 2 0 0 7

Explanation of the incentives for participants

Question items to measure five constructs and demographics

HOW MOBILE ADVERTISING WORKS

After designing textuai and visual information for the mobile advertisements, we created a promotional mobile site to which respondents could jump by clicking a banner saying "Please click here for further information" in the advertisements.

mobile campaigns. As an incentive, we offered a free ring-tone and screen-image downloads. Because the review of survey literature indicated that the choice of response format is critical in online surveys (e.g., Llieva, Baron, and Healey, 2002; Schaefer and Dillman, 1998; Sheehan, 2001; Sheehan and McMillan, 1999), we decided to use a multichotomous response format, with a level of scale reliability properly assessed via multivariate analysis (NeteI'neyer, Bearden, and Sharma, 2003). The questionnaire consisted of two parts. In the first, we asked questions about the respondents' demographic information, such as gender, age, and occupation, along with media perceptions and media usage habits (i.e., frequency of use of traditional and nontraditional media). We assessed the demographic variables and media perceptions using a categorical scale and the media usage habits with an ordinal scale. In all cases, respondents were asked to check the appropriate box to indicate their answer (Figure 2). In the second part, we included five questions that corresponded to the five constructs: brand trust, mobile advertising trust, attitude toward brand, attitude toward mobile advertising, and mobile campaign recall (i.e., unaided recall). The majority of the scale items were adopted from existent e-commerce literature and modified into a mobile context. All constructs were assessed using a multiple-item measure of 5-point seman-

tic differential scales, with 3 ("cannot answer/determine") as an anchoring point. All surveys were completed by the end of July 2005. RESULTS Response rate

For each type of product, 40,000 pseudocampaign messages were sent to the opt-in users. Thus, 80,000 messages were sent in total. The respondents were randomly chosen from the customer database in the greater Tokyo Metropolitan District area, including three nearby prefectures. The click-through rate was estimated at approximately 17 percent for the durable good sample and 13 percent for the nondurable good sample. Thus, approximately 6,800 and 5,200 consumers, respectively, actually "checked" the link. In total, we received 1,335 usable responses for the audio player and 1,899 for the com snack, which made the response rate, based on the total clicks, approximately 19.6 and 35.5 percent, respectively. All the surveys were completed by the end of July 2005.

The proportion of female respondents outweighed their male counterparts, while the differences across the samples were not statistically significant (p = .423). Neither did the occupational patterns differ across the samples (p = .071). However, for age, a chi-square test detected significant differences between the two brands {p = .004). Participants under 19 years old responded more for the audio player, while those between 30 and 34 years old responded more for the corn snack. Although the sample was not stratified according to age, it was somewhat expected to receive a "reasonable" number of responses from older age groups. However, our results were consistent with prior research in that people over 40 years old hardly replied to our survey invitations. Measurement assessment

Respondent profile

The research model was tested by structural equation modeling (SEM), which is an approach to modeling complex dependencies among unobserved or hypothesized variables, such as perceptions, beliefs, and attitude. With SEM, any variable can serve as a predictor of any other variable, whether it is observed or hypothesized. By simultaneously evaluating dependencies among variables, SEM extends traditional modeling techniques by providing additional insight into the underlying nature of relationships. Thus, this approach is considered to be appropriate to test causal relationships between the five constructs of brand trust, mobile advertising trust, attitude toward brand, attitude toward mobile advertising, and mobile campaign recall. We used AMOS 5.0 to conduct all the analyses.

Table 2 summarizes the major demographics of the respondents for each sample (durable and nondurable goods) in terms of gender, age, and occupation. Generally speaking, the respondents of both samples exhibited very similar characteristics.

According to the recommendations made by Anderson and Gerbing (1988), we first tested the measurement model by conducting a confirmatory factor analysis. This is a necessary procedure in establishing measurement reliability and validity.

June 2 0 0 7 JOyRllIlL OF HDUERTISIIIG RESEHRCH I 7 i

HOW MOBILE ADVERTISING WORKS

TABLE 2

-96, Comparative Fit Index (CFI) = .95,

Characteristics of the Respondents (%)

incremental Fit index (IFI) = .95, and RootMean-Square Error of Approximation

Durable Good

Nondurable Good

(W = 1,335)

(/V = 1,899)

(RMSEA) = .055. All indicated an accept;f^ Test

_

able fit of the model. Further, all the items loaded onto the corresponding factors with

1^ .

A-^ R

AO o

....i:^f^^^

56.4

57.8