The Effectiveness of Banner Advertisements:
Involvement and Click-through
by
Chang-Hoan Cho, Ph. D.
Assistant Professor
(402) 472-3042
ccho2@unl.edu
http://jet.unl.edu/coj/advt/faculty/ccho/
43 Avery Hall
Department of Advertising
College of Journalism and Mass Communications
University of Nebraska-Lincoln
Lincoln, NE 68588-0130
And
John D. Leckenby, Ph. D.
Everett D. Collier Centennial Chair
in Communication
(512) 471-1101
john.leckenby@mail.utexas.edu
http://www.utexas.edu/coc/admedium/
Department of Advertising
College of Communication
The University of Texas at Austin
Austin, Texas 78712
paper submitted to
The Association for Education in Journalism
and Mass Communication
(Session: Research)
The Effectiveness of Banner Advertisements:
Involvement and Click-through
Abstract
This paper explores the relationship between consumer's level of involvement and clicking of banner ads on the WWW. This study indicates that people in high-involvement situations are more likely to click a banner ad in order to request more information than those in low-involvement situations. Meanwhile, it is found that people in low-involvement situations are more likely to click a banner ad when it has a large size and dynamic animation. However, the size and the animation of the banner ad do not influence people's clicking of banner ads for people in high-involvement situations. This study measured a real click-through rate with the aid of online data collection technology called Cold Fusion.
Introduction
The Internet is one of the
fastest-growing media in terms of its users.
NUA Internet surveys estimate that 275 million people use the Internet
worldwide as of February 2000 (URL: http://www.nua.ie/surveys/how_many_online/). Along with this explosive growth of Internet
users in the world, Internet advertising is also experiencing the exponential
growth, with $3 billion spent online in 1999 and $8 billion projected to be
spent online advertising by the year 2002 (Jupiter
Communication 2000, at URL: http://www.jup.com).
The Internet has several distinguishing characteristics such as interactivity, irrelevance of distance and time, low set-up costs, targeting, global coverage, and ease of entry (Berthon et al. 1996; Zeff and Aronson 1999). Among these characteristics, interactivity is considered to be the key advantage of the medium (Rafaeli and Sudweeks 1997; Morris and Ogan 1996; Pavlik 1996). Even though the concept of interactivity has a long history, the Internet revivifies the discussion of interactivity because of its explosive growth since mid 1990s. There are many different ways of defining interactivity in the Internet (Flaherty 1985; Cook 1994; Rice 1984; Steuer 1992; Williams et al. 1988; Ariely 1998; Ha and James 1998). In addition, many different consumer activities can be classified as interactivity on the WWW (e.g., clicking, providing feedback, searching, etc.).
There are many different forms of advertising on the Web: e.g., banners, buttons, text links, sponsorships, target sites, interstitials, and more. Banner advertisements on the WWW began in October 1994, when AT&T first advertised on HotWired.com (Zeff and Aronson 1999). Since then, banners have dominated advertising on the WWW and have become the standard advertising format on the Web (Meland 2000). Even though there are other ways of finding out and arriving at target sites on the WWW, the banner advertisement click-through is believed to be the most common way to draw consumers into the target sites and thus engage them with a brand or product (Cho and Leckenby 1999). Accordingly, measuring advertisement banner click-through rates has already become important both for the advertiser and the Web site. In addition, the pricing method of online advertising is moving toward being based in click-through rates (Zeff and Aronson 1999; MediaPost 2000), and it is relatively easy to measure click-through rates with the aid of innovative technology. There are many known and unknown factors influencing people's clicking behaviors. Among these factors, this paper will focus on the effects of level of involvement on people's clicking behaviors.
How People Process
Advertising on the WWW
Many researchers have formulated different models of the stages, routes or hierarchy consumers go through when they are exposed to advertising messages (Krugman 1965; Ray, Sawyer, Rothschild, Heeler, Strong and Reed 1973; Houston and Rothschild 1978; Petty and Cacioppo 1981; 1983; 1986). These models are called hierarchy-of-effects models or how-advertising-works models. Traditional hierarchy-of-effects models assume that the first stage of the persuasion process is awareness through advertising exposure (Lavidge and Steinger 1961; Barry 1987; Preston 1982). In other words, the bottom line of advertising is to be noticed or advertising exposure (Lavidge and Steinger 1961; Barry 1987; Preston 1982). Here, advertising exposure is involuntary because individuals incidentally just happen to come across an ad in traditional media.
Meanwhile, advertising exposure in the Internet can be either involuntary or voluntary (Cho 1998). There are two current dominant forms of Web advertising: 1) the banner ad and 2) the target ad or linked site from the banner ad (Hoffman, Novak and Chatterjee 1995; Hoffman and Novak 1996a; 1996b). Depending on these two types of Web advertising, there are two different types of advertising exposure on the WWW: 1) involuntary exposure to the banner ad and 2) voluntary exposure to the target ad. When consumers just come across a banner ad incidentally, it is called involuntary exposure to the banner ad. During this involuntary exposure to banner ads, consumers have two choices-clicking the banners or not. If consumers click the banner ads to see the content of target ads, this is called voluntary exposure to target ads (Cho 1999).
Voluntary Exposure to the Target
Ad (Clicking the Banner Ad)
Clicking banners is a voluntary action for the purpose of seeing more detailed advertising messages by requesting more information. This voluntary exposure to advertising messages appears to be highly dependent on consumers' level of personal and product involvement. There exists extensive literature pertaining to various types of involvement in advertising research: personal relevance (Zaichkowsky 1985), elaboration (Petty and Cacioppo 1981); staged process (Ray, Sawyer, Rothschild, Heeler, Strong and Reed 1973; Greenwald and Leavitt 1984); Bridging connections (Krugman 1965), personal/internal state (Mitchell 1979; Cohen 1983; Andrews, Durvasula and Akhter 1990), cognitive and affective involvement (Park and Young 1983); involvement with the product--situational and/or enduring involvement (Houston and Rothschild 1978; Celsi and Olson 1988; Laurent and Kapferer 1985; Richins and Bloch 1986).
1) High Involvement
The literature on involvement with the product suggests that when consumers are highly involved with a product, they tend to be very receptive to most information related to that product and thus pay more attention to ads for information (Bloch, Sherrell and Ridgway 1986; Houston 1979; Lehmann 1977). Likewise, consumers are more likely to conduct greater information search to obtain knowledge when they are highly involved with the product (Hirschman and Wallendorf 1982; Beatty and Smith 1987; Zaichkowsky 1985).
Applying this to Web advertising, in high involvement situations, consumers are more likely to request more information by clicking banners in order to see detailed ad content than are consumers in low-involvement situations. Based on this rationale, the following hypothesis can be postulated:
H1: People in high-involvement
situations are more likely to click a banner ad than are those in
low-involvement situations.
2) Low Involvement
There exist many research studies on the interaction effect of level of involvement and various advertiser-controlled stimuli on advertising effectiveness: e.g., fear appeal (Tanner, Hunt and Eppright 1991), conclusion-drawing (Kardes 1988), comparison (Gotlieb and Sarel 1991), message repetition (Schumann, Petty and Clemans 1990), central and peripheral (Petty and Cacioppo 1983), and more. According to Petty and Cacioppo's Elaboration Likelihood Model (1983), consumers in low-involvement situations have low elaboration to process advertising messages and thus engage in the peripheral route to persuasion, while consumers in high-involvement situations have high elaboration and engage in central-route processing. In other words, when consumers are not highly motivated to process further ad content, they do not want to engage in message-related thinking; rather they are more likely to focus on available peripheral cues (Petty and Cacioppo 1986).
Applying this to Web advertising, consumers are less likely to request more information, i.e., less likely to click banners to see more detailed information when they are in low-involvement situations. However, they follow another route to clicking banners--the peripheral route to voluntary exposure. According to Petty and Cacioppo (1983), when consumers are not highly motivated to process further ad content, they do not want to engage in message-related thinking; rather they are more likely to focus on available peripheral cues. Applying this to Web advertising, it can be inferred that the favorability of peripheral cues may influence clickability of banner ads in low-involvement situations. In the case of banner ads, attention-getting or curiosity-generating peripheral cues would be novelty- or contrast-related components of banner ads, such as 1) large-sized banner, 2) bright colors, and 3) attention-getting animation.
There have been extensive literature on the effects of advertising design factors on consumer processing of advertising messages: for example, picture size (Rossiter and Percy 1980; Hendon 1973; Holbrook and Lehmann 1980), commercial length (Rethans, Swasy and Marks 1986), number of exposure (Nickerson 1968; Robinson 1969), and so on. Among these factors, the size of the stimulus is believed to one of important factors that can affect attention. Obviously, larger ads are more likely to be noticed than smaller ones. According to Rossiter and Percy (1980), the increase in attention is in proportion to the square root of the increase in space. Therefore, a larger banner ad should work as a better peripheral cue to draw low-involved people's attention than will a smaller banner ad. Similarly, dynamic animation on banner ads will also work as a good peripheral cue to draw low-involved people's attention. This reasoning leads to the following two hypotheses
H2.1: In low-involvement
situations, people are more likely to click a banner ad when it has a larger
size than average banner ads.
H2.2: In low-involvement situations, people
are more likely to click a banner ad when it has dynamic animation than when it
has no dynamic animation.
However, in high-involvement situations, the size and
animation of banner ads do not make any difference in the clicking of banner
ads.
H2.3: In high-involvement situations, the size of banner
ads makes no difference in clicking of banner ads.
H2.4: In high-involvement situations, dynamic animation
makes no difference in clicking of banner ads.
Methodology
This study employed a between group experimental design. According to Preston (1985), the perfect advertising effectiveness measure should be related to the actual behavior. Following this suggestion, this study measured the real click-through rates of banner ads, which is a measure of actual clicking behavior. The survey was conducted online using Web database technology called Cold Fusion, where responses on each survey item including click-through were automatically transmitted to a Microsoft Access database file located at the server.
According to Mitchell (1986), professionally developed ads rather than mock ads are encouraged to be used in experimental research in order to elicit a more natural response from the subjects. Following this suggestion, professionally developed Web sites and banner ads were used in this study.
A total of six banner ads were used, two different versions for each of three products. This study had two subject groups. Each experimental group was exposed to a set of three banner ads. Three banner ads were selected, based on the four popular product categories of Web advertising, which included financial services, consumer brands, retailers, and travel-related products (WebTrack 1998). All available banner ads for these four product categories were collected from the three most popular search engines on the WWW, i.e., Yahoo, Infoseek, and Excite, and then three banner ads were randomly selected using simple random sampling. These three banner ads included the American Express Card (financial services), Kodak film (consumer brands), and American Airlines (travel-related products).
Each banner ad was placed at the top of either Infoseek (URL: http://www.infoseek.com). Two banner ads for the same product were linked to the same target ad (linked site from the banner ad). Table 1 summarizes the three experimental banner ads for each experimental group.
There were two different versions of American Express banner ads that conveyed different messages. The same American Airlines banner ad with two different sizes were used for different experimental groups: i.e., the larger (468 by 60 pixels) banner was just a blow-up of the smaller (390 by 50 pixels) one. The same Kodak banner ad also had two variations: i.e., the static banner had only one image frame which individual texts and images were dumped into while the animated banner showed 3 individual frames one by one. A different version of online questionnaire was assigned to each subject group, based on its experimental materials.
An electronic recruiting message for the survey was
distributed via postings in various discussion LISTSERV lists. The LISTSERV lists were selected from CataList, the catalog of LISTSERV lists
(URL: http://www.lsoft.com/catalist.html).
This Web site provided 21,003 public LISTSERV lists on the Internet at
the point of the study, 1999. Among
these LISTSERV lists, education- , Internet-, advertising-, and
marketing-related LISTSERV lists were selected at the researchers' discretion
(the researchers believed that the discussion subjects of these LISTSERVs were
relevant to the current study). The
study had a total of 817 participants (409 for the 1st experimental
group and 408 for the 2nd experimental gorup). To recruit these 817 subjects, the
researchers posted recruiting messages on a total of 165 LISTSERV lists. For the purpose of increasing the response
rate of the survey, the researchers provided a small financial incentive in the
form of a sweepstake for the survey participants. A monetary incentive of $100 was given to each of 10 randomly
selected survey participants.
Procedure
The online survey consisted of three parts. In Part I, each subject's level of involvement with three product categories was measured as a pre-banner-exposure measure. In Part II, each subject was exposed to the three experimental stimuli. First, each subject was exposed to the very first banner ad and homepage (the American Express banner ad located at the top of the Infoseek site). Here, all hyperlinks were made dead at this point so that the subject could not click any hyperlinks. Then, each subject was asked first whether he/she was exposed to the banner ad previously or not. This was to control previous exposure to the banner ad. If the subject was previously exposed to the banner ad, he/she was asked to move on to the next section (the second experimental stimulus). Only the subject without previous exposure to the banner ad confronted two options: clicking the banner ad or not. If the subject clicked the banner ad, he/she was exposed to the linked target ad. Each subject followed the same procedure for the remaining two experimental stimuli.
After completing Part II, each subject was asked to continue with Part III of the online questionnaire, which asked several questions about his/her demographic information, i.e., gender, age, occupation, the purpose for Internet surfing, and average surfing hours. The participation for each subject took approximately 15 minutes.
Results
This study used a between-group experimental design. To eliminate the possibility that subjects in two groups are different enough to influence the effects of the treatment, the researchers compared the groups in terms of their demographic and Internet usage. The two groups (experiment and control group) were very similar in terms of age, gender, Internet-surfing hours, and the purpose for surfing the Internet. Table 2 shows the results of a series of t-tests and chi-squares to compare the two groups. Table 3 shows individual scale items used to measure level of involvement and reliability coefficient for the involvement measures. The Cronbach's alpha coefficient was .87, above the accepted level (.70).
The second set of hypotheses states that in low-involvement situations, people are more likely to click a banner ad when it has a larger size (H2.1) and animation (H2.2), while there is no effect of size and animation on banner clicking for high involvement people (H2.3 and H2.4). Chi-Square tests were conducted to check the relationship between the size / animation of the banner ad and click-through. Table 7 and 8 show the relationship between the size of the banner ad and banner clicking. As shown in Table 7, in low involvement situations, the click-through rate for the large American Airline banner ad (21.4%) was significantly higher than that for the small banner ad (9.3%). The results were statistically significant (p £ .01). However, as shown in Table 8, for high involvement people, there was no significant difference in click-through rates between large and small banner ads (16.7% vs. 13.9%) (p > .01). Therefore, both H2.1 and H2.3 are supported.
Table 9 and 10 show the relationship between the animation of the banner ad and banner clicking. As shown in Table 9, in low involvement situations, the click-through rate for the animated Kodak Film banner ad (25.0%) was significantly higher than that for the static banner ad (15.8%). The results were statistically significant (p £ .05). However, Table 10 indicates that for high involvement people, there was no significant difference in click-through rates between animated and static banner ads (29.1% vs. 27.0%) (p > .05). Therefore, both H2.2 and H2.4 are supported.
Other Interesting Findings
A simple correlation analysis was conducted to find the relationship between the average number of hours surfing the Internet of the respondents and the number of banner clicking. This analysis is conducted to check the assumption that people who surf the Internet more are less likely to click banner ads because they are tired of, or less curious about, the banner ads. The result is not in the direction of supporting this assumption because there was no correlation between the two variables (r = .003) (p > .05). Table 11 shows the relationship between the purpose for surfing the Internet and the average number of hours surfing the Internet per week. It shows that those who surf the Net for the purpose of entertainment spend more time surfing the Net (M = 16.3 hours per week) than those who surf the Net for information search (M = 11.6 hours per week). The result was statistically significant (t=3.49*, p £ .05).
The current study has its unique contribution at illustrating the usefulness of existing advertising theories in understanding how consumer process a new advertising form, Web advertisements. The current study explores how level of involvement influences people's clicking behavior. As predicted in H1, it was found that people in high-involvement situations were more likely to click banner ads in order to request more information than those in low-involvement situations. The implication of this finding is that advertisers can focus on people's level of product involvement when identifying target audiences for their Web advertising campaigns.
It was found that people in low-involvement situations were more likely to click large banner ads than small banner ads (H2.1), and animated banner ads than static banner ads (H2.2). However, the size and the animation of banner ads did not influence people's clicking of banner ads for people in high-involvement situations (H2.3 and H2.4). These results provide an important managerial and strategic implication for advertising practitioners, i.e., creative strategies focusing on "fat" or soft facts will be effective to make low-involvement people click on banner ads in the purpose of attracting and getting them more involved with the products. However, the current study did not look at the effects of hard facts or message arguments in banner ads. Hence, as a future research, it would be valuable to study the effectiveness of hard facts or message arguments on clickability of banner ads for high-involvement people.
A weakness of this study is that the samples are not representative of the general Internet users, since they were drawn from the pool of people who subscribed to discussion LISTSERVs. It is believed that people who subscribe to discussion LISTSERVs tend to be more active and heavier users of the Internet than do general Internet users. This can be a good explanation for the reason the average click-through rate of the two banner ads used in this study (18.3 %) is significantly higher than the average industry click-through rate (2.0 %). Therefore, it would be valuable to replicate the current study with the samples drawn from general Internet users other than LISTSERV subscribers.
Another explanation for these relatively high click-through rates of the sample banner ads used in this study would be that the sample banner ads were selected based on the four most popular product categories of Web advertising. In other words, we may say that the sample banner ads yielded higher click-through rates than the industry average because their product categories are very popular on the WWW. Moreover, a relatively small number of sample materials were used in this study, i.e., six banner ads. Therefore, it would be valuable to replicate the current study with an increased number of banner ads for more diverse product categories.
Another significant weakness of the current study is its creation of an artificial surfing environment because of its inherent experimental effects. The artificial setting, which limits external generalizability, means that subjects could not click any hyperlink within a web site other than clicking a banner ad. Thus, there was less chance for them to consider and interact with the web site, which would be the activity they would be engaged in the real world. This artificial setting was necessary to keep the study manageable and to confine the online survey to 15 to 20 minutes. In short, remedying the above weaknesses, it would be valuable to conduct the experiment in a more natural setting using more representative subjects, general Internet users.
Table 1
Experimental Stimuli (Banner Ads and Homepages)
for Each Experimental Group
|
1st Experimental Group |
2nd Experimental Group |
Table 2
The comparison of two experimental groups
|
Group I (n=409) |
Group II (n=408) |
Statistical tests |
|
|
Age |
35.2 |
34.2 |
t-value = 1.12 |
|
Gender (female/male) |
198 / 149 |
182 / 157 |
X2 = .79 |
|
Average surfing hours per week |
11.9 |
11.3 |
t-value = 1.27 |
p > .05
Table 3
Scales and Reliability Coefficient of Involvement Measures
|
Scale Items |
Cronbach's alpha coefficients |
|
Involvement (e.g., credit cards) I am interested in credit cards in general. Credit cards are important to me. I get involved with credit cards. Credit cards are relevant to me. I am going to use or apply for a credit card in the next six months. |
.87 |
Note: All items were measured on 5-point Likert scales with anchors of "strongly disagree (1)" and "strongly agree (5)."
Table 4
The relationship between level of credit card involvement
and clicking of American Express banner ad
|
|
Low Credit Card Involvement |
High Credit Card Involvement |
|
No Click
|
Frequency Row percentage Column percentage |
210 34.5% 89.7% |
398 65.5% 79.4% |
Click
|
Frequency Low percentage Column percentage |
24 18.9% 10.3% |
103 81.1% 20.6% |
Chi-Square = 11.84**,
d.f. = 1, p £
.01
Table 5
The relationship between level of camera
film involvement
and clicking of Kodak banner ad
|
|
Low Camera Film Involvement |
High Camera Film Involvement |
|
No Click
|
Frequency Row percentage Column percentage |
239 44.1% 79.7% |
303 55.9% 72.0% |
Click
|
Frequency Row percentage Column percentage |
61 34.1% 20.3% |
118 65.9% 28.0% |
Chi-Square = 5.56*,
d.f. = 1, p £
.05
Table 6
The relationship between level of airlines
involvement
and clicking of American Airlines banner ad
|
|
Low Airline Involvement |
High Airline Involvement |
|
No Click
|
Frequency Row percentage Column percentage |
205 33.1% 85.1% |
414 66.9% 84.7% |
Click
|
Frequency Row percentage Column percentage |
36 32.4% 14.9% |
75 67.6% 15.3% |
Chi-Square = .02,
d.f. = 1, p > .05
Table 7
The relationship between the size of a
banner ad
and clicking of the banner ad for
low-involvement people
|
|
Group I Small American Airlines Banner Ad |
Group II Large American Airlines Banner Ad |
|
No Click
|
Frequency Row percentage Column percentage |
117 57.1% 90.7% |
88 42.9% 78.6% |
Click
|
Frequency Row percentage Column percentage |
12 33.3% 9.3% |
24 66.7% 21.4% |
Chi-Square = 6.94**, d.f. = 1, p £ .01
Table 8
The relationship between the size of a
banner ad
and clicking of the banner ad for
high-involvement people
|
|
Group I Small American Airlines Banner Ad |
Group II Large American Airlines Banner Ad |
|
No Click
|
Frequency Row percentage Column percentage |
210 50.7% 86.1% |
204 49.3% 83.3% |
Click
|
|||