Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Interactive Advertising

ISSN: (Print) 1525-2019 (Online) Journal homepage: https://www.tandfonline.com/loi/ujia20

Search Engine Advertisement Design Effects on


Click-Through Rates

Geoffrey Atkinson, Carl Driesener & David Corkindale

To cite this article: Geoffrey Atkinson, Carl Driesener & David Corkindale (2014) Search Engine
Advertisement Design Effects on Click-Through Rates, Journal of Interactive Advertising, 14:1,
24-30, DOI: 10.1080/15252019.2014.890394

To link to this article: https://doi.org/10.1080/15252019.2014.890394

Published online: 26 Mar 2014.

Submit your article to this journal

Article views: 1455

View related articles

View Crossmark data

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ujia20
Journal of Interactive Advertising, 14(1), 24–30
Copyright C 2014, American Academy of Advertising

ISSN: print / 1525-2019 online


DOI: 10.1080/15252019.2014.890394

Search Engine Advertisement Design Effects


on Click-Through Rates
Geoffrey Atkinson, Carl Driesener, and David Corkindale
University of South Australia, Adelaide, Australia

and that the success should be measured through the relevant


This article reports the relationship that specific elements of return rate, the click-through-rate (CTR). However, academic
search engine advertisements (SEAs) have with click-through rates research into how the elements of SEAs affect the CTR is still
(CTRs) within Google AdWord Campaigns. In total, 1,880 separate in its infancy (Rutz and Trusov 2011). The findings of our study
advertisements with more than 57 million impressions and 185,000
click-throughs were analyzed. The main finding is that some el-
are based on SEAs for one company in the automotive retail in-
ements have both positive and negative relationships with CTRs dustry in Australia. Our findings contribute to both the general
depending on where they are placed within the advertisement. An understanding of the possible effects of elements present within
implication is that the section of the advertisement in which an advertisements and also provide managers in the SEA industry
element appears should be taken into consideration when devising with a method for determining the effectiveness of SEAs on
SEAs and when evaluating the relationship between elements and
CTRs. We found the elements with mostly a positive effect on CTRs
CTRs from their accumulated data.
were brand, value “puffery,” and price; and those with a negative The specific elements we studied for their impact on
effect were promotions and questions. Limitations of the study and CTRs were selected from the limited research literature on
future research opportunities are given. SEAs, claims and generalizations in industry publications, and
Google’s recommendations. They can be set out under four
Keywords click-through rate (CTR), design elements, ad creative, broad headings: brand related, calls to action, marketing ap-
search engine, advertisement peals, and punctuation. In this study, brand can be broken down
into company, which was a car retailer brand; the manufacturer,
Small, four-line advertisements are often placed on the pe- which was the manufacturer of the car; and the product, which
riphery of the organic results pages for a search engine inquiry. is the model of the car brand. Marketing appeals included the
This paid search engine advertisement (SEA) activity was pre- following elements: keyword, choice (e.g., range offered), price,
dicted (Goodwin 2012) to have grown worldwide by 27% in promotions, and value puffery. Punctuation was represented by
2012 to $19.5 billion, which represents some 44% of all pre- use of exclamation points and use of capitalization.
dicted online advertising expenditure, the other main display We provide an overview of what is generally known or rec-
advertising activity being banner advertisements. In Australia, ommended about the effectiveness of such elements as the four
for example, the total online advertising expenditure is now just noted before outlining the data and method of analysis used
greater than that for newspapers and almost equal to that for in our study, which is to explore the effects of each element on
free-to-air TV (IAB and PWC 2013). Chandon, Chtourou, and CTRs.
Fortin (2003) contend that much advertising on the Internet is
a form of direct marketing, likening it to consumers’ response
A SELECTIVE LITERATURE REVIEW
to coupons in print media or offers received via direct mail,
A relatively recent study on the relationship between the
SEA content and commercial outcome was based on the ring-
Address correspondence to David Corkindale, University of South tone sales industry (Rutz and Trusov 2011). Rutz and Trusov’s
Australia, City West, North Terrace, Adelaide, South Australia 5000,
Australia. E-mail: david.corkindale@unisa.edu.au (2011) investigation examined the ways of collecting, coding,
Geoffrey Atkinson (BA, Honours, University of South Australia) and analyzing available data. Their analysis focused on the ef-
research associate, Ehrenberg-Bass Institute for Marketing Science, fects on CTRs in three categories of elements in SEAs: the
University of South Australia Business School. attractiveness of the headline, the creation of “conviction or
Carl Driesener (PhD, University of South Australia), senior research desire,” and calls to action.
associate, Ehrenberg-Bass Institute for Marketing Science, University
of South Australia Business School. They found that calls to action and the use of the search key-
David Corkindale (PhD, Cranfield), professor of marketing, Uni- word in the title or headline have a positive effect on CTRs. They
versity of South Australia Business School. also had indications that the price of an item in an advertisement

24
SEARCH ENGINE AD DESIGN EFFECTS 25

had some influence on where in the ranking the advertisement study. The overall profitability of an SEA campaign will be a
appeared, that is, how high it is positioned from top to bottom major criterion for identifying a successful advertisement and
along the side of the results page. In the electronic appendix to factors such as the conversion rate of those who do click an
their article, Rutz and Trusov (2011) also recommend the use of advertisement, and the cost of buying the keyword for it are
attention-grabbing content, such as promotions, discounts, and part of this overall evaluation (Nabout and Skiera 2012). The
“rewards for reading.” The broad elements in our study taken strategy that advertisers should follow when bidding for access
from Rutz and Trusov’s work are calls to action, promotions, to search terms is a specific area of online advertising research
price, use of questions, and use of keywords. (see, for example, Baye, Gao, and Morgan 2011; Yao and Mela
The use of questions in the headline for SEAs was studied 2011; Zhang and Feng 2011) and beyond the scope of our study.
by Turnbull and Bright (2008), who observed that including a However, the necessary first step in assessing the overall success
question in the ad headline generally was associated with under- of an advertisement is its achievement of an initial response, the
performing CTRs. A separate study conducted on the branding CTR. The elements used in the advertisement and their location
effects of keywords noted that advertisements with the best per- in it and how these influence the initial response, the CTR, is
forming CTRs had both the same brand name used in the key- what our study examines.
word as present in the advertisement (Jansen, Sobel, and Zhang The position that an advertisement attains on the search re-
2011). Based on these two studies we included the following sults page has been found to influence CTR (Agarwal, Hosana-
elements in our study: use of questions, use of keywords, and gar, and Smith 2008; Animesh, Viswantathan, and Agarwal
mention of brands. 2011; Bar-Ilan, Mat-Hassan, and Levene 2006; Rutz and Trusov
Industry publications also suggested factors in SEAs that can 2011). Google actually uses past CTR performance in its calcu-
influence CTRs, but the claims made for their effectiveness were lation of how much to charge an advertiser for a given position,
rarely backed by evidence. To benefit from industry experience or rank, on the results page. As such, it was planned in our study
we included some elements most often cited in industry publi- to control for the rank that advertisements achieved, but in the
cations as being influential on CTRs: punctuation used in the initial analysis of our particular data set we found very little
headline and other lines, value puffery, and the offer of a choice. correlation between position, or rank, and CTRs, and so this
Google, an authority on sponsored-search research, provides was excluded from our subsequent investigation.
advice on “creating successful advertisements” (Google 2012). Following is the list of elements we examined for their re-
We have included the following elements from this advice in our lationship with CTRs. To help with the understanding of the
study: highlight the business’s unique features; include prices, meaning of these elements, we give an illustration or elabora-
promotions, and exclusives; give a call to action; include key- tion of each:
words; use strong brands; and use punctuation (such as title
• Call to Action: Advertisement asks the consumer to do
case).
something. Examples: “Enquire Today!” and “Come
Other studies on content in SEAs provide comparisons
compare the prices now!”
among differing content tactics. The effects of positive and neg-
• Price: Advertisement includes a specific price. Exam-
ative framing of proposals have been examined, with positive
ple: “$24,990”
framing being beneficial in low-involvement purchases (Yoo
• Choice: Gives the consumer more than one option.
2011). A comparison between commercial (value offers) and
Examples: “Select from our wide range of vehicles”;
descriptive (product features) found that descriptive content per-
“Large range available”
formed best, although price-sensitive readers do not discrimi-
• Keyword: Advertisement includes the exact search
nate between the two; likewise, quality-based ads outperformed
keyword used by the searcher.
price-based ads (Animesh, Viswantathan, and Agarwal 2011;
• Promotion: Advertisement includes a specific offer.
Gauzente and Roy 2012). Finally, argument types have been
Examples: “Purchase online today to receive $100 cash
compared that investigated the effect of expert advice, statisti-
back”; “Sale ends soon!”
cal evidence, and causal evidence, finding that expert advice and
• Value puffery: Advertisement makes an unspecified
statistical evidence create greater CTRs (Haans, Raassens, and
claim to value. Example: “Great price on cars!”
van Hout 2013). This is not an exhaustive list of all elements
• Punctuation: Advertisement uses forms of punctua-
that could be included in our study, and this is discussed in the
tion to attract attention, including excessive capitaliza-
Limitations section.
tion of words within a headline and/or text. Example:
In SEAs, it is assumed that the audience to whom an adver-
“Get a Great Deal on Toyota”
tisement is exposed is a relevant one as a result of the search
term (keyword) that a searcher has used. What the advertiser
has to offer and what the search engine actually displays is a METHODOLOGY
result of the interplay of these two factors. Including the key- The research aim was to look for the quantitative relation-
word in an advertisement is, therefore, likely to be an important ships between elements used in SEAs and the CTRs associated
influence on CTRs and should be one of the elements in our with them. Data for this were supplied to us by a company that
26 G. ATKINSON ET AL.

TABLE 1 vestigating. The advertisements were examined for the presence


How the Advertisements Were Categorized of these elements in each of their four sections: title, description
line 1, description line 2, uniform resource locator (URL), and
CTR category No. of ads CTR range (%) a fifth measure, “appeared at all.” The latter measure captured
High 628 0.56−7.64 whether any of the four advertising elements we are studying
Medium 633 0.15−0.55 appeared at least once somewhere in the advertisement.
Low 619 0.00−0.14 The elements were also coded for which section of the ad-
vertisement they occurred in, for example, the heading or line
Note. CTR = click-through rate. 2. We did not include examination of the effect of the presence
of the URL, as we found it was very highly correlated with
manages the search advertising campaigns for a range of large the presence of the company brand name. Our data included
companies in Australia. We were given the data on the CTRs the keyword associated with each advertisement only when it
for 2,494 advertisements, with more than 57 million impressions was “automatically inserted” (i.e., where the ad copy is scripted
and 185,000 clicks, by an automotive retailer in a city in Aus- to include the keyword used in the search). More than 50,000
tralia shown to searchers on Google via a personal computer for keywords were present in the data set’s campaign, however, and
automotive-related keywords during the period January 2008 coding all of these into meaningful categorizations would be
through July 2012. extremely difficult. More important, it was impossible to link
All advertisements that received no clicks were removed, as the ad content with the keyword used, as it would not be known
were all ads that appeared less than 25 times, as these would not which keyword an ad had appeared with; thus pairing the con-
represent mainstream search ads and could potentially produce tent with the keyword was also not possible. Consequently, we
outliers in the data. This reduced the number of advertisements were unable to identify the influence the keyword may have had
for our analysis to 1,880. The CTR for the purpose of our anal- on CTRs in our market category.
yses was derived from the following for each advertisement: Table 2 shows the frequencies of occurrence for the elements
examined in our study and in which section of the advertisement
CTR = (Total Clicks by Consumers/Total Impressions) × 100 they appeared; where data are not present there were no exam-
ples in the data set for that section. For instance, price was never
We subsequently observed that, across our set of advertisements, mentioned in an advertisement heading or title. The percentage
the distribution of values of CTRs was very positively skewed figures refer to the percentage of advertisements that contained
and not normally distributed. For subsequent analysis purposes the element in a position, for example, price was mentioned 228
the CTRs were categorized into equal thirds: high, medium, and times in body line 2 of 12% of all the advertisements that we
low bands, as shown in Table 1, and this became the dependent, examined.
categorical variable in our analyses. As described, we coded each advertisement for the elements
We needed to code each advertisement for the presence, or it contained and the sections of the ad in which they were lo-
absence, of the elements in the advertisements that we were in- cated. Three coders were used: a coordinator/instructor and two

TABLE 2
Frequency and Location of the Advertising Elements
Appeared at all In heading In body line 1 In body line 2
Ad element n % n % n % n %
Price 238 13 0 0 10 1 228 12
Product brand 1,309 70 1,235 66 436 23 335 18
Manufacturer brand 1,147 61 1,019 58 354 19 277 15
Value puffery 1,051 56 16 1 745 40 306 16
Call to action 1,198 64 0 0 57 3 1,141 61
Keyword 799 43 799 43 0 0 0 0
Choice 521 28 0 0 369 20 152 8
Exclamation 1,382 74 0 0 11 1 1,371 73
Company brand 471 25 34 2 247 13 190 10
Title case 1,730 92 744 40 1,268 67 1,570 84
Promotion 334 18 41 2 153 8 140 7
Question 168 9 142 8 26 1 0 0
SEARCH ENGINE AD DESIGN EFFECTS 27

TABLE 3 a low CTR response, 40 with a medium level of CTR, and 58


Capitalization of Words in the Title Line, or Not, and CTR with a high response. Consequently, from the data in the table
Level we see that the use of uppercase letters or capitalization of title
or headline does not appear to be associated with CTR.
CTR Level Next, in order to determine if an element like capitalization
Title capitalization Low Medium High Total is associated with a higher CTR, odds ratios were calculated
to indicate the likelihood of a given ad content eliciting greater
Not present 52 40 58 150 CTRs. For example, the title case in a particular line location was
Present 567 593 570 1,730 calculated as being more or less likely to be associated with high
Total 619 633 628 1,880 or low CTR. The greater the number above 1, the stronger the
Note. CTR = click-through rate. relationship. An example of this is calculated as follows (using
Table 3 as reference): Low CTR is calculated as having odds of
10.9 of occurring (567 present/52 not present); high CTR has
research assistants. The coordinator coded the entire data set, odds of 9.8 of occurring (570 present/58 not present). We note
and the research assistants each coded half of the data set. The that categories vary only very slightly in size. Comparing high to
intercoder reliability was tested, and any inconsistencies found low, we find 0.9 odds of title case to be in high CTR rather than
were reassessed by the coordinator. The elements were coded low CTR (9.8/10.9); title case, when present is 1.1 (10.9/9.8)
as either being present (coded 1) in each line of the ad or not times more likely to be in low CTR rather than high CTR. This
present (coded 0). These dummy variables were created for the means that when title case is used, it has a negative influence on
eight elements in our study. The most common form of punctu- CTR. However, this relationship was not statistically significant,
ation device that was used was that of capitalization; therefore, as the data showed little variation. In our results these odds will
its presence or absence was observed and coded to represent the always be presented as either a negative or positive relationship
punctuation element. with CTR. “Positive” in this context means being more likely to
be associated with high CTR, and “negative” means a likelihood
of being associated with a low CTR.
ANALYSIS METHOD
The data were analyzed initially using cross tabulation in
SPSS to identify the elements that appeared to be significantly FINDINGS
related to CTRs. This was then used as the basis of an odds ra- Table 4 gives a summary of the study’s results where scores
tios analysis. An example of the odds ratio analysis is presented ranged from −7.6 for call to actions in line 1 to +2.8 for price
in Table 3 for the examination of the effect of the headline/title in line 2. For each of the elements that may occur in an adver-
being in uppercase or lowercase. First, we can see that for 150 tisement, which are listed in the first column at left, we first
advertisements, no part of the title/headline contained an upper- show the strength of the relationship with CTR depending on
case letter; 52 that had no uppercase letter were associated with the locations of where it may occur in the advertisement, which

TABLE 4
The Strength and Nature of Relationship Between Ad Elements by Location
Title Description line 1 Description line 2 Total
Ad element Strength Relationship Strength Relationship Strength Relationship Strength Relationship
Price 1.0 Negative 2.8 Positive∗∗∗ 2.7 Positive∗∗∗
∗∗∗
Product brand 2.0 Positive 1.7 Positive∗∗∗ 1.4 Positive∗∗ 1.7 Positive∗∗∗
Manufacturer brand 1.9 Positive∗∗∗ 1.5 Positive∗ 1.7 Positive∗∗∗ 1.7 Positive∗∗∗
Value puffery 1.1 Positive∗ 1.0 Negative 1.6 Positive∗ 1.3 Positive∗
Call to action 7.6 Negative∗∗∗ 1.6 Positive∗∗∗ 1.3 Positive
Keyword 1.1 Positive
Choice 1.4 Positive 1.8 Negative∗∗ 1.0 Positive
Exclamation 1.1 Positive 2.6 Negative 1.0 Neutral 1.0 Negative
Company brand 1.6 Negative 1.5 Positive 1.8 Negative∗∗ 1.1 Negative
Capitalization 1.3 Negative 1.2 Positive∗∗ 1.2 Negative 1.1. Negative
Promotion 3.6 Negative∗∗ 1.3 Negative 1.8 Negative∗∗ 1.8 Negative∗∗∗
Question 4.1 Negative∗∗∗ 1.9 Negative 3.7 Negative∗∗∗

p = 0.05; ∗∗ p = 0.01; ∗∗∗ p = 0.001.
28 G. ATKINSON ET AL.

are listed in the column headings across the table. For example, recommendations. In our study it was found to have a positive
mention of price in an advertisement’s title line did not occur influence when used in line 2 of an advertisement, but it had
in any of our sample advertisements, and so no strength of re- a very strong negative influence if used in line 1. A possible
lationship of price to CTR is cited for that location; it is left explanation is that a call to action would be more appropriate
blank. Mention of price when it is in line 1 of an advertisement at the end of an advertisement when all the other reasons for
was associated with a low CTR (the very low odds ratio value a consumer being interested in the offer have been presented.
of 1.0), meaning it had a negative effect, but this was not signif- However, the sample of advertisements upon which the finding
icant and so is presented in a grayed-out typeface. The number for this element was derived was relatively small, so the finding
of asterisks in the table uses the usual convention to show the is tentative.
statistical significance.
Value Puffery
The use of value puffery has generally been recommended in
DISCUSSION AND CONCLUSION
industry publications, and Google recommends including spe-
The following sections provide an overview of each adver-
cific product benefits and/or specific reasons why the consumer
tising element’s relationship with CTR as found in this study.
should buy an advertised item. In our study we found that the
Punctuation: Title Case nonspecific promise of a “deal” was mostly positive—except in
The use of uppercase letters for words in an advertisement line 1, where it had a negative effect, though not significantly.
is recommended by Google as it is thought to attract the eye of
Choice
the consumer. In our study, we found that not only was it the
most-used device but it seemed to be mostly associated with Choice has been recommended by trade publications as a
a nonsignificant, though negative, effect on CTR except when way of increasing CTRs. It was not used at all in the title or
used in line 1 of an advertisement. heading of any of our sample of advertisements and was rarely
used in line 2. When appearing in line 1, it had a positive effect
Brand on CTRs; thus it appears to be a useful tactic.
In prior research, product brand has been found to increase
CTR when used as a keyword by the searcher (Ghose and Yang Promotion
2010; Jansen, Sobel, and Zhang 2011) and when it is a famil- Promotion was found to be a negative influence in our study.
iar brand (Dahlen 2001). Its use in banner advertisements has This contradicts advice by Google, which encourages the use of
been suggested to depress CTR (Chandon, Chtourou, and Fortin “offers.” However, research on the use of “cash promotions” in
2003), but in our study we found it to be a positive influence banner advertisements has found that it has no effect on CTR
regardless of where it appeared in the advertisement. This may (Baltas 2003), and this somewhat corroborates our finding.
well be because it provides both the benefits of branding and
usually assists to identify what the company is offering. Price
We found that the manufacturer brand has a positive influence Including pricing, on the other hand, had the most positive
wherever it is used. In the case of the automotive industry, effect on CTR of the elements we examined, particularly when
product and manufacturer are usually well known and branded used in line 2 of advertisements. The inclusion of price has been
and so the positive association with CTR is not unexpected. This found to be positive in other SEA studies (e.g., Rutz and Trusov
could be for similar reasons as product brand: it helps the reader 2011); in addition, Google (2012) recommends including price.
identify what they are looking for. These findings are opposite of those for banner advertisements,
In our data the company brand was the retailer’s brand, which where mention of price can reduce CTR (Chtourou, Chandon,
was found to have a negative effect if used in most positions in an and Zollinger 2002) and absence of price can have a positive
advertisement; this was opposite of the effect of the other types effect on CTR (Rettie, Grandcolas, and McNeil 2004).
of branding cited. However, the URL almost always contained
Questions
the company brand name, so presence elsewhere in an adver-
tisement would essentially be duplication. We did find that URL In previous research, the recommendation regarding posing
use in line 1 seemed to be positive, though not significantly. questions in SEAs was to avoid its use (Turnbull and Bright
These findings represent an important extension of the previ- 2008). Our study confirms this finding, as the use of a question
ously identified benefits of branding, as it shows the effect can had a negative effect on CTRs regardless of placement. In other
occur for both the product brand name and the manufacturer’s media (such as press and TV), advertisements that start with a
brand name. question are quite common, as they are believed to attract atten-
tion and identify problems viewers may have that the advertiser
Call to Action wants to solve. It is possible that the use of questions is not
Call to action has been found to be slightly positively related effective in SEAs as consumers are using search engines to look
to CTR (Rutz and Trusov 2011) and is also one of Google’s for answers, not more questions.
SEARCH ENGINE AD DESIGN EFFECTS 29

Main Finding ied. Other finer details that could be explored are variations in
Our main finding is that all of the elements we examined in presenting price—for example, whether the price was “expen-
SEAs can have an influence on CTR response level, but their sive” or “cheap” compared to other advertisements. For brand,
impact can be dependent on the line of the advertisement in it would be valuable to explore whether the size of a brand has
which they appear. an effect on the ad’s association with higher CTRs. There is,
From our findings we can formulate an ideal advertisement hence, the opportunity for more focused studies to investigate
for the automotive retail industry. It would include the “identi- this element’s association with CTRs. Furthermore, our study
fication brands” of the product and the manufacturer in the title did not investigate the possible interactive effects of elements
(1), a boasting of the range on offer in line 1 (3), and a combina- in SEAs; for example, it could be that the effects of price and
tion of value puffery, call to action, and price in line 2 (4) with branding are enhanced when both appear together, but the de-
the retail (landing page’s) brand in the URL (2). An example of gree of enhancement may vary depending on where in the ad
an ideal advertisement is provided: each is located. This is a needed area for future research.
There are various advertising theories or principles that could
1 New Toyota Camry
help explain the results of our study. The following outlines
2 www.CarRetailerBrand.com.au
3 A Huge Range Of Camrys Available how future studies might expand on the preliminary results of
4 Get A Great Deal Now! Only $28,990 this study by utilizing such theories and associated appropriate
methodologies.
In our results it was found that a call to action performs best
LIMITATIONS
in line 2. While we did not investigate why this is the case, it is
Our study and its findings are based on the SEAs of one com-
possible that the hierarchical process AIDA (attention, interest,
pany, in one market, and in one country; also, one person created
desire, action) could help explain such a gap in our knowledge
most of the advertisements in one digital advertising agency, and
through carefully designed experimental research where ads are
the advertisements had a high level of repetition over the several
tested using different sequences of elements. In addition, we
years for which we had data. The data are for advertisements
could explore further questions in the same way; for example,
that were presented for searches done on personal computers.
does the use of branding in the title and in early sections of an
In Australia, 79% of CTRs for SEAs were on personal comput-
ad result in positive CTRs due to readers preferring an AIDA
ers in 2012 rather than on mobiles (Marin Software 2013). We
format for ads?
were not able to take into account the possible effects on CTRs
Further studies might also examine what has been found to
of brand size and reputation, competitive advertising activity
stimulate response in other advertising media. For example, the
and their ad’s ranking position on the page, or the full impact
likability of an ad has been found to be associated with subse-
of the use of search keywords. These considerations limit the
quent sales response in the Advertising Research Foundation’s
generalizability of our findings but indicate the areas of further
Copy Research Validation Project (Haley and Baldinger 1991).
research to be able to more fully explain and predict CTRs for
Respondents in an SEA research study could be asked what as-
particular SEAs. Our finding— that the position in which an ad-
pects of the text had aided in them liking the ad, which could then
vertising element is used may well influence the degree of CTR
be compared with CTR results to establish whether likability is
achieved—can alert future researchers in this area to include
positively associated with CTR.
this consideration in their work.
Finally, research could investigate whether there is any con-
Many elements could be used in SEAs, and we examined the
nection between findings from structural patterns in text, such as
effects of just some of them. Further suggestions of elements
Gutenberg’s Z (Lidwell, Holden, and Butler 2003), and CTRs.
that can influence CTRs made by SEA practitioners are person-
Such studies could reveal how the location of content is more
ification and positive/negative tone, but these were not included
effective in one section of an advertisement than another section.
in our study as they would have been difficult to code consis-
The effectiveness of the title and bolded keywords in attracting
tently. We would also expect that additional elements may be
attention could be examined via typographical hierarchy stud-
found in other market categories. We note differences between
ies where theory suggests certain types of content draws readers
some of our findings and those in the study by Rutz and Trusov’s
into the next sections of the text.
(2011) in the mobile ringtones category; for example, the use of
product brands is prevalent in SEAs for automobiles but rare in
ACKNOWLEDGMENTS
those for mobile ringtones.
The authors wish to thank and acknowledge Frank Grasso,
CEO, and the staff of e-channel Search (e-channel.com.au) for
FURTHER RESEARCH their provision of data access for this research.
Our exploratory study investigates only whether an element
was present or not in SEAs at a basic level. Many elements REFERENCES
can be broken down into more specific levels. For example, Agarwal, Ashish, Kartik Hosanagar, and Michael D. Smith (2008), “Loca-
the wording used to create a call to action can be widely var- tion, Location, Location: An Analysis of Profitability of Position in Online
30 G. ATKINSON ET AL.

Advertising Markets,” Journal of Marketing Research, 48 (6), 1057–73. Haley, Russell I., and Allan L. Baldinger (1991), “The ARF Copy Research
http://papers.ssrn.com/sol3/papers.cfm?abstract id=1151537. Validity Project,” Journal of Advertising Research, 31 (2), 11–32.
Animesh, Animesh, Siva Viswanathan, and Ritu Agarwal (2011), “Competing IAB and PWC (2013), “2012 Internet Advertising Revenue Full-Year
‘Creatively’ in Sponsored Search Markets: The Effect of Rank, Differen- Report,” http://www.iab.net/insights research/industry data and landscape/
tiation Strategy, and Competition on Performance,” Information Systems adrevenuereport.
Research, 22 (1), 153–69. Jansen, Bernard J., Kate Sobel, and Mimi Zhang (2011), “The Brand Effect
Baltas, George (2003), “Determinants of Internet Advertising Effectiveness: of Key Phrases and Advertisements in Sponsored Search,” International
An Empirical Study,” International Journal of Market Research, 45 (4), Journal of Electronic Commerce, 16 (1), 77–106.
505–13. Lidwell, William, Kritina Holden, and Jill Butler (2003), Universal Principles
Bar-Ilan, Judit, Mazlita Mat-Hassan, and Mark Levene (2006), “Methods for of Design, Beverly, MA: Rockport.
Comparing Rankings of Search Engine Results,” Computer Networks, 50 Marin Software (2013), “Mobile Search Advertising Around the Globe,”
(10), 1448–63. http://www.iabaustralia.com.au/uploads/uploads/2013-10/1382054400
Baye, Michael R., Cathy Gao, and John Morgan (2011), “On the Optimality daf59f00d5fda6ce1b6f0cbae937f0ec.pdf
of Clickthrough Fees in Online Markets,” Economic Journal, 121 (556), Nabou, Nadia, and Bernd Skiera (2012), “Return on Quality Improvements
340–67. in Search Engine Marketing,” Journal of Interactive Marketing, 26 (3),
Chandon, Jean L., Mohamed S. Chtourou, and David R. Fortin (2003), “Effects 141–54.
of Configuration and Exposure Levels on Responses to Web Advertise- Rettie, Ruth, Ursula Grandcolas, and Charles McNeil (2004), “Post-
ments.” Journal of Adverttsing Research, 43 (2), 217–29. Impressions: Internet Advertising Without Click-Through,” working paper,
Chtourou, Mohamed S., Jean L. Chandon, and Monique Zollinger (2002), “Ef- Kingston Business School, Kingston University, Cheltenham, UK, July.
fect of Price Information and Promotion on Click-Through Rates for Internet Rutz, Oliver J., and Michael Trusov (2011), “Zooming In on Paid Search Ads: A
Banners,” Journal of Euromarketing, 11 (2), 23–40. Consumer-level Model Calibrated on Aggregated Data,” Marketing Science,
Dahlen, Micael (2001), “Banner Advertisements Through a New Lens,” Journal 30 (5), 789–800.
of Advertising Research, 41 (4), 23–30. Turnbull, Don, and Laura F. Bright (2008), “Advertising Academia with Spon-
Gauzente, C., and Y. Roy (2012), “Message Content in Keyword Campaigns, sored Search: An Exploratory Study Examining the Effectiveness of Google
Click Behavior, and Price-Consciousness: A Study of Millennial Con- AdWords at the Local and Global Level,” International Journal of Electronic
sumers,” Journal of Retailing and Consumer Services, 19 (1), 78–87. Business, 6 (2), 149–71.
Goodwin, Danny (2012), “Search Ad Spending Could Hit $19.51 Billion Yang, Sha, and Anindya Ghose (2010), “Analyzing the Relationship between
in 2012,” Search Engine Watch, February 1, http://searchenginewatch. Organic and Sponsored Search Advertising: Positive, Negative, or Zero
com/article/2143093/Search-Ad-Spending-Could-Hit-19.51-Billion-in- Interdependence?,” Marketing Science, 29 (4), 602–23.
2012-Report. Yao, Song, and Carl F. Mela (2011), “A Dynamic Model of Sponsored Search
Google (2012), “Tips for Creating Successful Text Ads,” Advertising,” Marketing Science, 30 (3), 447–68.
https://support.google.com/adwords/bin/answer.py?hl=en-AU&answer= Yoo, Chan Y. (2011), “Interplay of Message Framing, Keyword Insertion and
1704392&topic=1713901&ctx=topic&path=1713898-1710534. Levels of Product Involvement in Click-Through of Keyword Search Ads,”
Haans, Hans A. J., Neomie Raassens, and Roel M. W. M. van Hout (2013), International Journal of Advertising, 30 (3), 399–424.
“Search Engine Advertisements: The Impact of Advertising Statements on Zhang, Xiaoquan (Michael), and Juan Feng (2011), “Cyclical Bid Adjustments
Click-Through and Conversion Rates,” Marketing Letters, 24 (2), 151–63. in Search-Engine Advertising,” Management Science, 57 (9), 1703–19.

You might also like