The different effects of content and placement of Search Engine
Advertising on click and purchase intention across countries.
Author: Minji Oh
Student Number: 11375582 Date: June/23/2017
MSc. in Business Administration -Digital Business Track Supervisor: Abhishek Nayak
Statement of originality
This document is written by Student, Minji Oh, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
Abstract ... 4
1. Introduction ... 5
2. Literature Review... 9
2.1. Types of Search Engine Advertising ... 9
2.1.1. Effectiveness ... 9
2.1.2. Content ... 11
2.1.3. Position of Search Engine Advertising ... 14
2.1.4. Cultural aspects ... 15
3. Research Design and data ... 21
3.1. Independent Variables ... 21 3.2. Dependent Variables ... 23 3.3. Moderators ... 25 3.4. Experiment subjects ... 25 3.5. Descriptive data ... 26 4. Result ... 28
4.1. Click intention on the conditions ... 28
4.2. Purchase intention on the conditions ... 31
4.3. Title influence ... 34
5. Discussion ... 35
5.1. Managerial implications... 36
5.2. Limitations and future research ... 37
6. Conclusion ... 39
Abstract
This study attempts to reveal the effect of culture on the effectiveness of Search Engine Advertising. By applying different message types (i.e. statistical and expert evidence) and controlling for placement (i.e. top or non-top placement), the study finds that the cultural dimensions Power Distance and Uncertainty Avoidance affect the click and purchase intention of distinguished markets differently. The experiment compares consumers in the Netherlands and South Korea, two countries that have significantly different cultural dimensions, as measured by Hofstede (2001, 2010). By analyzing the data with chi-square and logistic regression, the author finds that both markets show similar effectiveness for both statistical and expert evidences placed on top. This is because the effect of the culture on click and purchase intention is reduced due to the strong relationship between top placement and click and purchase intention. However, if the evidence types are not placed on top, the click and purchase intention vary between countries. The Korean market clicks on the statistical or expert message more when it is not placed on top than the Dutch market. This shows that the Korean market, with higher Power Distance and Uncertainty Avoidance, is more sensitive to the types of advertisements than the Dutch market, proving the effect that culture has on Search Engine Advertising.
Keywords: Search Engine Advertising, Top Placement, Content, Cultural Dimensions, Click
1. Introduction
Worldwide Internet access has drastically increased over the last decade. Around 82% of households in OECD countries have Internet access at home in 2016 (Internet World Stats, 2017). Because websites exist for just about everything, it is easy to find customers with specific types of interests online. Furthermore, tracking software used by companies such as Google, makes it easy to target consumers with advertisements of products that they are or have previously been looking for. This creates a belief that online marketing can be highly effective, resulting in increasing online advertising expenditures. For 2017, the worldwide expenditures are expected to be 229 billion dollars (Statista, 2017).
In line with the practitioners’ attention on online marketing, several papers were written about a large diversity of online marketing methods such as online banners, emails, blogging, and social media in order to reveal their effectiveness. Yet, the focus on this field has mainly been on online banner advertising (Manchanda et al 2006; Möller & Eisend, 2010). This may be because it was one of the first popular online marketing methods.
Among the online marketing tools that have been gaining popularity, Search Engine Advertising is growing fiercely compared to the other online marketing methods. Search Engine Advertising totals revenues of $16.3 billion, accounting for 50% of total online marketing revenue in the first half year of 2016 (PwC, 2016). In spite of the large amount of investment and respective revenues from the companies or practitioners, the studies regarding Search Engine Advertising has gained less attention. (Ghose and Yang 2009, Rutz and Bucklin 2011).
This particular form of online marketing is defined by Ghose and Yang (2009, p. 1605) as “where advertisers pay a fee to Internet search engines to be displayed alongside organic (i.e. non-sponsored) Web search results”. Among the topics studied in the Search Engine Advertising field, the effectiveness of a high position of advertisements in Search Engine
results has gained traction. For example, Ghose and Yang (2009) show that highly-ranked advertisements in a Search Engine, attract higher click through rates. Yet, it leaves us with still some critical questions. If the top placement matters, how strong is the effect of it? And how does the quality of the advertisement affect click through rate of advertisements placed on top?
Haans et al (2013) highlight that studies about the content of the Search Engine Advertisement has been scarce, however, his study did show that it has effects on the effectiveness of Search Engine Advertising. The importance of the content of Search Engine Advertisement messages is currently underrepresented in the literature, even though it may have important implications in relation to other factors in the field. For example, it could give lower placed advertisements a boost in effectiveness, diminishing the strength of the top placement argument.
At the same time, little empirical research has been done to reveal the impact of cultural differences on Search Engine Advertising. As the market is becoming more globalized, companies and academics alike have shown the importance of culture in marketing activities. This is shown in for example product differentiation between cultural markets, the use of different television advertisements or the staring of national celebrities in advertisements in different countries. However, companies that focus on online marketing have treated the target market as a huge cluster of people without considering their cultural background such as values, education, and customs. For example, when looking at Duolingo -a well-known online language learning platform – in a Search Engine, their search result describes the company and how their service works. It tries to give persuasive messages to potential or existing consumers to lead them to click. However, this message does not differ between different languages, meaning that it shows the same, translated message when
searching in the Korean, Spanish, or Dutch Google website. This approach is ironic because cultural segmentation is a greatly acknowledged and discussed factor in offline marketing.
Thus, this paper is designed to answer the question “How does content and placement of Search Engine Advertising affect consumer’s click intention and purchase intention, and whether and if so, how cultural differences intervene with this?”.
By answering this question, the thesis will contribute in both an academic and a practical way. Academically, it will give new insights into important factors of Search Engine Advertising, this contribution is threefold: firstly, the thesis will show the importance of culture in the Search Engine Advertising; secondly, it will prove that the content of the advertisement greatly affects effectiveness; and finally, it will determine whether the effect of top placement is still a relevant factor when considering content in Search Engine Advertising. Furthermore, this study will provide an integrative view of the various factors that affect the effectiveness of Search Engine Advertising. Numerous studies have focused on single factors affecting the effectiveness. In contrast, this study combines factors and tests whether these the factors remain relevant in combination.
The thesis will also provide a different perspective to practitioners. Many companies have taken a monotonous approach to Search Engine Advertising. As shown in the example of Duolingo, the same message is being translated into the local languages and listed in the Search Engine of the targeted markets. However, this paper will show if any cultural dimensions should be taken into account when doing Search Engine Advertising. It will show an example of how these messages could be customized to maximize the click and purchase intention. Thus, this will give a more concrete idea of how practitioners could lead potential or existing customers to more clicks on their advertisements in different cultural markets. This will eventually reduce the chances of making erroneous decisions in Searching Engine Advertising and maximize their online marketing effectiveness. In addition, if the
effectiveness of content is proven critical in comparison to top placement, it could decrease the percentage of the advertising budget being spent on top placement.
The structure of this paper is as follows. Firstly, existing literature is studied to gather relevant information and provide hypotheses. Then, the research design chapter explains how the experiment is designed by discussing each dependent and independent variable. Furthermore, it will show how the respondents are selected in detail. The following chapter presents the data analysis in which two main methods - chi-square and logistic regression - are used. Finally, the last chapter concludes and discusses the findings.
2. Literature Review
2.1 Types of Search Engine Advertising
Search Engine Advertising has become one of the most important online marketing channels and due to its rapid growth, is expected to be an important form of future online marketing (Jansen and Resnick 2006; Bucklin and Sismeiro 2009). In the field of Search Engine Advertising, the primary focus of studies has been on ad banners, sponsored advertisements and keywords. Manchada et al (2006, p.98) explain that “ad banners consist of graphic and textual content and contain a link to the advertiser’s Website”.
The difference between ad banners and sponsored advertisement is that the former method has bigger restrictions in terms of intrusiveness, as consumers have higher control over them. (Chandon et al., 2003). While ad banners are placed on side in a Search Engine, a usual sponsored advertisement is presented as part of the Search Engine results above the organic search results in general. For the purpose of this study, the term of Search Engine Advertising is limited to sponsored advertisement.
2.1.1. Effectiveness
Vakratsas and Ambler (1996) concluded that cognition, affect, and conation create the metrics to measure effectiveness of advertisements. These metrics are related to the following parameters that are widely used in online marketing: (1) impressions generated, (2) number or percentage of click-throughs, or (3) induced sales or conversion rates (Zenetti et al 2014, p.9).
Firstly, impressions generated refers to how many times consumers or potential consumers saw advertisements. If companies advertise frequently on a Search Engine, the probability of their advertisements getting seen by customers increases. In this way, the reach to consumers improves. This method is directly related to the Cost Per Mille (CPM), a
commonly used metric by practitioners, which is the cost of advertising per one thousand people who saw the advertisements.
The second method of the click through rate accounts for how many times these advertisements actually get clicked out of number of impressions, it is usually expressed in percentage. Usually companies pay to Search Engine providers such as Google or Yahoo for each click they get from advertisements which is in the same line as Cost per Click (CPC). The click-through rate is related to click intention because people should be interested in clicking on an advertisement before they actually click on it. This intention can be measured by directly asking consumers if they are interested in clicking on a specific ad, or by asking how likely it is to click on the advertisement with a Likert scale as conducted in the study by Gauzente (2010).
However, the validity of click through rate as a measurement for effectiveness is still discussed among scholars. Moe and Fader (2004) and Chatterjee et al. (2003) posit that this metric might be too broad to measure effectiveness in ad banners used in big markets. In contrast, Manchanda et al. (2006) revealed that banner advertising actually increases consumers’ purchase behavior. In spite of ongoing discussion, click through rate is widely used among marketers as it is a new effectiveness method that offline marketing cannot measure directly.
Lastly, the conversion rate shows the actual behavior of purchasing a product or service. This rate is calculated by dividing the number of clicks into the actual purchases. As revealed in the study of Hans et al (2013), a high click through rate is not always proportionally related to the conversion rate. People who click on an advertisement because of its attractiveness or other factors, do not always end up purchasing the offerings. Furthermore, the conversion rate is related to the purchase intention because people should have interest on a certain product of service in order to do actually decide to buy it. This is
revealed in the study conducted by Morwitz, et al (1992). They confirm that the purchase intention and the actual purchase (i.e. conversion), are related although the conversion rate is not always fully matched to purchase intention.
However, it has to be noted that studies quoted so far measure the direct effectiveness of Search Engine Advertising. Aside from direct effects, Search Engine Advertising might have indirect effect on consumer’s interaction with the brand in the future (Hollis, 2005). Indirect effects are also of importance in the field of Search Engine Advertising as it could affect consumers brand awareness or future behavior. Yet, it is not part of the scope of this study, as the purpose of this study is to find direct effects of advertisements.
2.1.2. Content
Lohtia et al. (2003) categorize the content of online advertisement in two ways. Although their study focuses on ad banners, the insights may be useful for sponsored results as well. The distinguished components are cognitive and affective. The former are advertisements with incentives such as special offers or promotions, while affective components have emotional aspects such as happiness, joy and fear. This categorization is similar to the content categorization as presented by Maheswaran et al. (1990). They claim that the types of messages in advertisements can be (i) attributes, (ii) benefits, or (iii) attributes and benefits. The cognitive components as distinguished by Lohtia et al. (2003) are comparable to benefits, as it offers benefits such as promotions and coupons to consumers, while affective components are similar to attributes, as these focus on the mental aspect.
While the topic of design of ad banners gains a lot of attention (Ghose and Yang 2010), the design of the sponsored advertisement lacks of empirical studies. Ad banners have both visual and textual elements that compose the advertisements, this gives a marketer more
freedom in designing the ad. The design of a sponsored advertisement, on the other hand, is mainly textual rather than visual. This decreases the design choices, but does not mean that design is of lesser importance for this type of advertisement.
For the studies about textual design in Search Engine Adverting, Haans et al. (2013) introduce a framework in which effectiveness of Search Engine Advertising is measured depending on distinguished messages types. This study has an outstanding approach to Search Engine Advertising as it emphasizes the content of the message. His study critically analyzes the difference in effectiveness of different messages in Search Engine Advertising. This contrasts to previous authors, that focused on measuring effectiveness of Search Engine Advertising in comparison to offline marketing. Haans et al. (2013) apply the persuasion theory and introduce four advertisement types proposed by Hornikx (2005, 2007). These four categories are anecdotal, expert, causal and statistical evidences. These are called evidence because each of them shows a type of proof for a product or service to convince consumers to use or purchase the offerings.
Firstly, anecdotal evidence shows how a particular person experiences a product. It highlights a specific case that a consumer has had with a product or service by showing a small story to other consumers. This is a type of personalized advertising but due to the same reason, it may have limited effects, as it can only be applied to a certain group of people. Furthermore, this type of message usually requires a lot of space to tell a story. This is possible in TV ads where the time limit is relatively spacious compared to Search Engine Advertising.
Secondly, expert evidences are those in which people with expertise in a particular field try to persuade consumers. Their attributes such as fame, prestige, or knowledge give credibility to the message. By leveraging an expert, customer’s trust in the product gets
affected and thus it encourages consumers to have a positive image of the product. Thus, it increases their click or purchase intention.
Thirdly, causal evidence shows the result of an occurrence. It illustrates potential effects of a certain product or service but it seldom has factual numbers. In other words, it shows what the expected outcome is, when a consumer decides to use a product or service.
Lastly, statistical evidence shows more concrete and factual numbers to potential customers. Showing a factual number could be very appealing as seen in the study of Haans et al (2013). This is could be explained by general behavior of people. People tend to prefer certainty and high chances of success over uncertainty and taking risks, as seen in the study of Kahneman et al. (1979). It also increases credibility of the message, which may convince consumers to click on the message or to use the product.
According to Haans et al (2013), these types of evidence used in the advertisement convince consumers differently. This is proven by the different click through rate and conversion rate for each type. The authors found that statistical evidence and expert evidence bring higher click-through rate compared to causal evidence. This shows that statistical and expert evidence types are very appealing to consumers and these evidence types lead them to click on the advertisements with the persuasive messages. However, the conversion rate was higher for expert evidences rather than the statistical evidences. Even though both of the evidences are appealing enough to be clicked, the persuasiveness level was different.
Despite its meaningful contribution to the literature, the study lacks some critical perspectives. Firstly, their basic premise is that consumers in the online market have similar tastes or perspectives towards advertisement types. The experiment was taken in a single country, the Netherlands, where people share similar values and thus it restricts the external validity of the study result. The study result could differ if the same study was taken in different countries. Secondly, the order of advertisements of which study subjects are asked
to read was not taken into account. The respondents for this study could have clicked the advertisements not only because it was appealing enough, but also because of their placement. As Ghose and Yang (2009) reveal, the placement of an advertisement is of high importance and cannot be neglected.
2.1.3. Position of Search Engine Advertising
Jerath et al. (2011) introduce the “Position Paradox” which implies that a well-known company with better performance in comparison to other inferior companies, achieve higher click through rates although their advertisement is placed below.
Even though this paradox contradicts to the studies revealed by Agarwal et al. (2008) in which the authors confirm that click through rate decreases with position, the position of advertisements is considered crucial to attract higher clicks. Likewise, Ghose and Yang (2009), revealed that higher ranked advertisements have a higher conversion rate. This rate decreases steadily as the advertisements are placed lower down in the search engine results.
The purpose of this thesis is to test whether the aforementioned factors (i.e. position and advertisement types) directly affect consumer’s click and purchase intention of an advertisement. The two most effective message types from the study of Haans et al. (2013), namely statistical and expert evidences are selected because the click through rate and conversion rate were the highest. Thus, the hypotheses that combine the placement factors and message types are as follows;
H1a. The click intention on the advertisements with either statistical or expert evidence placed on top, is significantly higher than for the same advertisement not placed on top.
H1b. The purchase intention on the advertisements with either statistical or expert evidence placed on top, is significantly higher than for the same advertisement not placed on top.
2.1.4. Cultural aspects
Culture helps the members of a society understand their environment and share common values and beliefs (Roth and Moorman 1988). Furthermore, it affects how people behave, what they eat, how they look at products or services, and so on. Thus, cultural differences are widely studied in different fields such as psychology, sociology, and business. These fields have deeply investigated culture as a phenomenon. For example, the business field studies culture because increasing globalization creates a large number of opportunities to expand to foreign markets for international companies. By studying the effect of cultural differences on customer preferences, international companies can reduce the risk they face by going abroad. Studies have shown that companies should take different marketing strategies in different cultural backgrounds and that the effectiveness of each method is significantly different among different cultural groups. An example of this is the empirical study carried by Choi and Miracle (2004). Their study shows that effectiveness of comparative advertising varies among countries. These findings are not only academically relevant, but are also being applied in practice. For example, comparative advertising is very common in the United States, where comparative advertising is very effective due to its cultural factors. But at the same time, it has been harder to find this type of advertisement in the Netherlands. This example shows that culture is considered crucial in offline marketing environments, but the attention to this phenomenon in online marketing is too little.
Cultural differences between countries can be divided into the framework presented by Ghemawat (2004). He created the CAGE framework (Cultural, Administrative, Geographical and Economical) in which he shows the distances between host and home
country of multinational companies. His findings show that the larger the difference between the home and host country on one or more of the distinguished dimensions, the more difficult it is to do business in the other country. These difficulties in doing business occur because they have to face differences in terms of culture, which affects preferences and ways of doing business. Likewise, Hofstede (2001, 2010) shows six disparate parameters of culture, namely, Power Distance, Individualism, Masculinity, Long-term Orientation, Uncertainty Avoidance and Indulgence. These parameters are measured for each country, and expressed in a 1 to 100 scale. His database shows that countries differ significantly on the aforementioned parameters. By combining the work of Ghemawat (2004) and Hofstede (2001, 2010), it can be concluded that each country requires a different way of doing business, and that it may be difficult for international companies to succeed abroad if they do not consider cultural factors.
For this thesis, Ghemawat’s CAGE framework is not applicable, because the boundaries between a host country and home country have to be clear. In online marketing, these boundaries are vague, because many companies start their operations without borders and launch online advertising at the same time in various countries. This makes it hard to consider which country is the home, and which country is the host country. Therefore, Hofstede’s cultural dimensions are applied instead. Hofstede’s approach analyzed each country as a whole, and gives a score for each of the distinguished parameters. Because companies do work in different countries, and search engines automatically redirect users to their country specific website, it is easy to apply a country-specific framework in practice. Therefore, a country-specific approach will be used for this experiment as well.
For this study, two of Hofstede’s cultural dimensions are used. These dimensions are Power Distance and Uncertainty Avoidance. These cultural dimensions are selected because they can be matched to the two types of evidence described before (i.e. expert and statistical evidence).
The first cultural dimension that will be used in this paper is Power Distance. It is defined as “the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally” (Hofstede 2001, 2010). A high score in this cultural dimension, results in the acceptance of strong subordinate-superior relationships. This results in a strong sense of hierarchy within societies. It also creates a situation in which people truly look up people with higher social status. Furthermore, it also results in a situation in which people may attempt to climb the hierarchy to attain respect. People who consider themselves as “subordinate” in a particular sector or field, and people who wish to climb within the hierarchy, will look for professional opinions or recommendations. As a result, people from a culture with a high score on the cultural dimension of Power Distance, will more likely to rely on experts’ opinion rather than people from countries that score lower on Power Distance. This means that a Search Engine result that contains expert evidence will be more appealing in countries with a high score on Power Distance. The hypotheses on this cultural dimension are differentiated to aforementioned measurements of effectiveness and include the factor of top placement.
H2a. Consumers’ click intention for advertisements containing expert evidence placed on top is positively related to the level of Power Distance of the consumers’ cultures.
H2b. Consumers’ purchase intention for advertisements containing expert evidence placed on top is positively related to the level of Power Distance of the consumers’ cultures.
H2c. Consumers’ click intention for advertisements containing expert evidence not placed on top is positively related to the level of Power Distance of the consumer’s cultures.
H2d. Consumers’ purchase intention for advertisements containing expert evidence not placed on top is positively related to the level of Power Distance of the consumer’s cultures.
The second cultural dimension, Uncertainty Avoidance, is “the extent to which the members of a culture feel threatened by ambiguous or unknown situations and have created beliefs and institutions that try to avoid these” (Hofstede 2001, 2010). Uncertainty is a situation in which the actor is unsure about the outcome of actions. Individuals in a culture with a high score on uncertainty avoidance, prefer to avoid these situations. According to Camerer and Weber (1992), people avoid opinion with missing probability information. This effect is called ambiguity effect. People do not want to deal with situations in which chances of outcomes are unclear. Thus, it could be concluded that access to information makes people more willing to take decisions. As stated by of Hofstede (2001, 2010), the extent to which people desire to rely on probability information differs among different cultures. Therefore, individuals from a culture with a high score on Uncertainty Avoidance, will be more likely to rely on probability information than people from a culture with a low score on Uncertainty Avoidance.
Statistical evidence as introduced by Haans et al (2013) is a type of probability information. Therefore, individuals from a culture with a higher score on Uncertainty Avoidance, will be more likely to click on an advertisement containing statistical evidence.
H3a. Consumers’ click intention for advertisements containing statistical evidence placed on top is positively related to the level of Uncertainty Avoidance of the consumers’ cultures. H3b. Consumers’ purchase intention for advertisements containing statistical evidence placed on top is positively related to the level of Uncertainty Avoidance of the consumers’ cultures.
H3c. Consumers’ click intention for advertisements containing statistical evidence not placed on top is positively related to the level of Uncertainty Avoidance of the consumer’s cultures. H3d. Consumers’ purchase intention for advertisements containing statistical evidence not
placed on top is positively related to the level of Uncertainty Avoidance of the consumer’s cultures.
These hypotheses are visualized as seen in Figure I.
Figure I. Visualization of the hypotheses.
Figure II summarizes the conceptual framework of this thesis. The click and purchase
intention for advertisements is directly related to the effectiveness of advertisement. The effectiveness is assumed to be influenced by the content and the position of the Search Engine Advertising. Therefore, the independent variables are the content and position of Search Engine Advertising while dependent variables are the click and purchase intention on advertisements. Moderators will intervene to affect the result of this research. These are the cultural dimensions Power Distance and Uncertainty Avoidance as previously explained.
Low High Cli ck In ten ti o n (% ) Power Distance
Click intention - Power Distance
Expert evidence placed on top Expert evidence not placed on top
Low High C li ck I n te n ti o n ( %) Uncertainty Avoidance
Click intention - Uncertainty Avoidance
Statistical evidence placed on top Statistical evidence not placed on top
3. Research design and data
To answer the research question of this paper, an online experiment is conducted. Two representative countries with different cultural dimensions are selected. Each group of respondents from these countries is asked to answer questions about their click and purchase intention depending on four conditions. These conditions are created by combining the two advertisement types (i.e. statistical and expert) and its placement (i.e. top and non-top). The detailed explanations for each variable are shown in the following sections. Following the clarification of variables, this chapter shows how data is collected and introduces the descriptive statistics.
3.1. Independent variables
Google has a tool called Google Adwords. This tool allows businesses to place their advertisement online. Even though companies can add display ads such as pictures or videos, the design of this study is limited to textual ads. The type of advertisement thus has three main components – headline, display URL and description. Figure III shows an example of what an advertisement in Google looks like.
Figure III. Example of components in Google Adwords Text Ad.
In addition, Google recently added a new function called extensions for this type of advertisement. With this function, companies can add certain functionalities such as calls, reviews, and location buttons that consumers can click in the search engine result. Some companies that desire to provide consumers with these practical and appealing tools, pay an
additional fee to get these extensions. To eliminate the effect of these features in the experiment, these functions are excluded.
The main message of the advertisement is the focus of this experiment. The two main components of the independent variables are the content and the placement. It is a 2x2 condition as each component contains two subcategories. The content of the advertisement message is divided into statistical and expert evidence types. And the placement is divided into placement on top and not on top.
Thus, in total, there are four conditions:
1. Statistical evidence and placed on top 2. Statistical evidence and not placed on top 3. Expert evidence and placed on top
4. Expert evidence and not placed on top
In the experiment, each of these conditions will be shown next to two neutral advertisements and two organic search results in order to show more realistic Search Engine results. The display URLs have the same message as in headline to reduce its impact on the result. Thus, only the condition with either statistical or expert evidence will be affect the click or purchase attention. Figure IV shows an example of what the experiment looks like for the first condition.
Figure IV. Example of the first condition
There is one possible restriction to this experiment. The title may have an influence on the click or purchase intention of the subjects. Some of the respondents might only click on a certain advertisement because they find the title to be attractive. To determine whether the title has an influence, respondents are asked to choose the most appealing advertisement title out of the five titles that are used in this experiment. The titles were shown randomly so the order has little impact on their decision. In the data analysis part, the effect of the titles is studied to see if these titles influence on the click and purchase intention
3.2. Dependent variables
As seen in literature review, there are two ways to measure click and purchase intention. The first option is to ask respondents to click on the most appealing advertisements. The second option is to ask respondents to express how likely it is to click on a certain advertisement. The second option is not used in this study for several reasons.
Firstly, the expressed scale does not directly show if the respondents will click on the advertisements or not. For example, if a respondent says that it is 3 out of 5 likely to click or
purchase on a product of service, it is not clear if this person will actually click or purchase or not. The value that each person gives on a scale could vary across people. Some people might click or purchase even with 3 level out of 5 while others might click or purchase when only their expectations are fully matched. Secondly, by asking how likely they would click or purchase a product or service, they are indirectly advised to read the conditions more carefully. However, this study attempts to measure the actual behavior of respondents and how this differs depending on the types of evidence and its placements in Search Engine results. If a person is advised to read every condition carefully, the effect of top placement could be disrupted. Thus, the careful consideration of a score for each advertisement could lead to inaccurate findings. Lastly, this method could make the experiment tiresome for the respondents. For each question, respondents will see five options – three Search Engine Advertising results, and two organic search results. If respondents have to evaluate the scale for each option, the experiment will last longer, and the chance of respondents stopping in the middle of the experiment will increase. In conclusion, the first option of directly asking for the click intention is used for this study.
The experiment asks each respondent for click intention first, because it is more likely that people quickly scan the Search Engine results when selecting which option to click. After the first question, respondents are asked to rank the advertisements from the most to least appealing to see whether each condition is positioned in high rank or not.
Lastly, the purchase intention of each condition is tested. The purpose of this is to see how the actual purchase intention differs to the click intention and if it does, how much it differs. The purchase intention is tested in the same way as the click intention was measured.
Thus, experiment subjects are asked to answer the following questions: 1. Which advertisement(s) would you click? (multiple answers are possible)
2. After carefully reading each advertisement, rank them from most to least appealing. 3. Whose service are you more likely to use? (multiple answers are possible)
4. Which one is the most appealing title?
3.3. Moderators
The selected countries have distinct cultural dimensions according to Hofstede. As explained before, the focus lays on the two cultural dimensions, namely: Power Distance and Uncertainty Avoidance.
Figure V. Comparison between the two representative countries.
Figure V shows the differences of scores in Power Distance and Uncertainty Avoidance
between the Netherlands and South Korea. This Figure shows that South Korean people are more likely to avoid uncertainties than the Dutch. Furthermore, South Korean people score higher in Power Distance than the Dutch.
3.4. Experiment subjects
Geert Hofstede’s database is used to find cultural differences between South Korea and Netherlands. These countries are chosen because they can be used to represent two of the
38 53 60 85 0 10 20 30 40 50 60 70 80 90
Power distance Uncertainty Avoidance Comparison between the NL and KR
biggest markets in the world, with very distinct cultures: East Asia, and Europe. The accessibility of respondents for the researcher also plays a role in the country selection.
Approximately eighty people from two representing countries are invited to this experiment. The experiment is online and the experiment subjects are asked to answer questions for two conditions. Nationality is the only factor considered when selecting experiment subjects. In other words, experiment subjects are selected randomly without differentiating by gender, age, or other factors.
For the experiment, respondents are asked to imagine in a situation in which they look for an online language course. This particular service type is chosen mainly because of two reasons. Firstly, the characteristics of online education pull people with high involvement as it requires motivation and relatively high investment of money and time. People will consider more carefully when choosing a course. Secondly, this type of service is not restricted to a certain group of people. If people are interested in this type of service, they will not differ in educational background, gender or age.
3.5. Descriptive data
The total number of respondents is 193, consisting of 91 Korean people and 102 Dutch people. 39% of respondents are men while 60% are women. The rest did not specify their gender. 57% of the respondents have a research university or postgraduate degree. Furthermore, most (61%) people are under 30 years old. The data was collected during 30 days between April and May of 2017 via internet.
The participants were invited to answer a combination of two conditions of either i) statistical-top (condition I) and expert-non-top (condition IV) or ii) statistical-non-top (condition II) and expert-top (condition III). Some people (n=17) dropped during the experiment after completing one of the two asked conditions. However, if the respondent
completed one of the two conditions fully, the data for the answered condition was still used for the analysis. Thus, in total, the number of respondents for each condition is 89, 104, 93 and 83 respectively, summing up to 369 cases (condition I = statistical, top; II = statistical, non-top; III = expert, top; and IV =expert, non-top).
4. Result
4.1. Click intention on the conditions
First of all, crosstabulations are used to run an exploratory analysis to find any differences between groups in the dependent variable (i.e. click and purchase intention) for each condition, depending on independent variables (i.e. evidence types and placement)
4.1.1. Condition I: Statistical evidence placed on top. (n=89)
A chi-square test for association was conducted between markets and their click intention on condition I. The expected cell frequencies of clicks on the condition for each market were 18.2 for the Dutch market (n= 49) and 14.8 for the Korean (n=40). This number did not differ strongly with the actual frequencies which results in no significant association between the market and the click intention on the condition I (χ2(1) = .653, p= .419).
4.1.2. Condition II: Statistical evidence not placed on top. (n=104)
A chi-square test for association was conducted between markets and their click intention on the condition II. The actual cell frequencies were 12 for the Dutch (n= 53) market and 24 for the Korean (n=51) while expected cell frequencies were 18.3 and 17.7 respectively. Thus, there was a statistically significant association between the market and the click intention on the condition II (χ2(1) = 6.843, p = .009), showing a strong association between the variables (φ = 0.257, p = .009).
4.1.3. Condition III: Expert evidence placed on top. (n=93)
The same test was used to see the association between this condition and the market. The expected frequencies for each market were 9.8 (Dutch market, n=48) and 9.2 (Korean market, n= 45). The actual number did not show any difference, as shown by χ2(1) = 0.10, p = .921. (φ = -.010, p = .921). This leads us to the conclusion that there is no significant association between the market and the click intention on condition III.
A chi-square test is used to see the association between the condition IV and the market. The expected frequencies for clicks were 10.2 for the Dutch market and 7.8 for the Korean market. Interestingly the actual number were 6 (total n= 47) and 12 (total n= 36) respectively. The result shows a strong association between the variables with χ2(1) = 5.078,
p = .024. (φ = .247, p = .024).
Interestingly, the conditions that are placed on top for each market received fairly equal number of clicks which confirms the study conducted by Ghose and Yang (2009). The advertisements that are placed on top, without strong influence of cultural differences, get the highest number of clicks. However, the advertisements containing one of the aforementioned evidence types (i.e. statistical and expert) that were not placed on top had significant differences between the countries. While Korean market clicked advertisements with statistical or expert messages, the Dutch market did not. This shows that the Korean market is more sensitive to the message of the advertisement than the Dutch market.
For the further analysis, a logistic regression was conducted to predict the click intention of the advertisement that contains either statistical or expert evidence placed in top or non-top position (2x2). The evidence types (i.e. statistical and expert) and its placement (i.e. top or non-top) were set as predictors as well as two representative countries (i.e. South Korea and the Netherlands). The dependent variable was set as click intention of each condition.
The classification table in block 0 which presents the results without considering any coefficients (i.e. evidence, placement and market), shows the percentage of 28.7%. This number improves as the predictors are added. Interestingly, the evidence type and the country have significant effect on the click intention (p=.002 and p=.048 respectively) while the top placement of the condition does not show any significant result (p=.948). This implies that
conditions that are placed on top, are frequently clicked for each type of evidence and for both markets.
After adding the predictors, the classification error rate in the classification table improved from 29.2% to 45.8% (cut-off value =.200). This means that now it is possible to predict the result with 45.8% of accuracy. The cut-off value is adjusted from .500 to .200 because the conditions are presented in a search engine result where three advertisements and two organic search results are shown. The probability of clicking any result is thus .200.
The overall significance is tested by using chi-square. The –2log likelihood value from the Model Summary Table is 428.770. In this model, the chi-square has 3 degrees of freedom with the value of 13.796 and the probability of p<.005 (p=.003). This shows that the predictors have significant effect and create substantially a different model. In addition, Nagelkerke’s R2 is 0.053, indicating a relationship of 5.3% between the predictors and the prediction.
Table I shows the significant results. The Wald values are as follows: 9.560 for
evidence type (P=.002), .025 for top placement (P=.875) and 3.866 for the market (P=.049). This depicts that the placement of the conditions does not contribute significantly to the prediction while other two factors (i.e. the evidence type and the market) have significant effect on the result (i.e. click intention).
B S.E. Wald df Sig. Odds ratio
95% C.I.for Odds Ratio
Lower Upper Evidence .743 .240 9.560 1 *.002 2.102 1.312 3.365 Placement .037 .235 .025 1 .875 1.038 .655 1.645 Country -.461 .235 3.866 1 *.049 .630 .398 .999 Constant -1.110 .250 19.665 1 .000 .330 Note: Country is NL compared to KR.
Placement is top compared to non-top placement
Table I. Logistic Regression Predicting Likelihood of Click on the condition based on the evidence
type, placement and countries.
The Odds Ratio show the likelihood of change of the dependent variable in accordance to each of the independent variables. So, increasing one unit for evidence type (i.e. statistical) increases the odds by 2.102. This implies that the odds of clicking the condition is 2.102 times greater for statistical evidence as opposed to expert evidence. In contrast, the value that are less than 1.000 indicate decreased odds for an increase in one unit of the independent variable. Thus, increasing one unit for market (i.e. the Netherlands) decreases the odds by .630. It means that the odds of clicking on the condition is .630 times smaller for Dutch people in comparison to the Korean.
4.2. Purchase intention on the conditions
Crosstabulations are used to run an exploratory analysis to determine whether a relationship exists between the independent variables (i.e. evidence type, placement and countries) and the dependent variable (i.e. purchase intention) for each condition.
4.2.1. Condition I: Statistical evidence placed on top. (n=89)
A chi-square test for association was conducted between markets and their purchase intention on the condition I. The expected cell frequencies of use of service on the condition for each market were 20.4 for the Dutch market (n= 49) and 16.6 for the Korean (n=40). This number did not differ strongly with the actual frequencies which results in no significant association between the market and the intention of use of service on the condition I, χ2(1) = 1.051, p= .305.
4.2.2. Condition II: Statistical evidence not placed on top. (n=104)
A chi-square test for association was conducted between markets and their purchase intention on the condition II. The actual cell frequencies were 13 for the Dutch (n= 53)
market and 24 for the Korean (n=51) while expected cell frequencies were 18.9 and 18.1 respectively. Thus, there was a statistically significant association between the market and the purchase intention on the condition II (χ2(1) = 5.756, p = .016), showing a strong association between the variables (φ = 0.235, p = .016).
4.2.3. Condition III: Expert evidence placed on top. (n=93)
The same test was conducted to see the association between this condition and the market. The expected frequencies for each market were 7.2 (Dutch market, n=48) and 6.8 (Korean market, n= 45). The actual number did not show any difference, as shown by χ2(1) = 1.060, p = .303. (φ = -0.107, p = .303). This leads us to the conclusion that there is no significant association between the market and the purchase intention on this condition.
4.2.4. Condition IV: Expert evidence not placed on top. (n=83)
A chi-square test is used to see the association between the condition IV and the market. The expected frequencies for clicks were 11.3 for the Dutch market and 8.7 for the Korean market. Interestingly the actual number were 7 (total n= 47) and 13 (total n= 36) respectively. The result shows a strong association between the variables with χ2(1) = 5.018,
p = .025. (φ = .246, p = .025).
Like the click intention, the purchase intention between two representative countries differed only when the evidences are not placed on top. Both Dutch and Korean market showed their similar purchase intention when the evidence types are placed on top. However, their purchase intention is different when the evidence types are not placed on top. This confirms that Korean market is more sensitive to the type than the Dutch.
The same analysis of logistic regression is conducted to predict the purchase intention for each condition. The independent variables are set as evidence type, placement and the countries while the dependent variable is added as purchase intention.
The classification rate before entering any independent variables shows 29.3%. This number increases after adding predictors (i.e. evidence types, placement and countries) to 46.3% (cut-off value .20). Similar to the analysis with the click intention, the evidence type and the country have significantly strong relationship with the prediction (i.e. purchase intention). The evidence type and the country have p=.000 and p=.015 respectively while the top placement has p=.604. The result shows that there is no relationship between the purchase intention and placement of the advertisement. In other words, advertisements that are in the top do not necessarily lead consumers to purchase services of the advertisements.
The -2log likelihood value is 423.651 with the 3 degrees of freedom. The chi-square value is 22.497 with the probability of .000 (p<.001). This confirms that the independent variables have effect on the purchase intention. Furthermore, Nagelkerke’s R2 is .084 which shows 8.4% of relationship between the independent variables (i.e. evidence type, placement and countries) and the dependent variable (i.e. use of service).
B S.E. Wald df Sig. Odds ratio
95% C.I.for Odds Ratio Lower Upper Evidence .960 0.244 15.441 1.000 *.000 2.611 1.618 4.214 Placement -.057 .237 0.058 1.000 .809 .944 0.594 1.502 Country -.577 0.236 5.948 1.000 *.015 .562 0.353 0.893 Constant -1.113 0.252 19.465 1.000 .000 .328 Note: Country is for the NL compared to KR.
Evidence is statistical compared to expert
Table II. Logistic Regression Predicting Likelihood of Use of Service based on the evidence type,
placement and countries.
In the Table II, the Wald value for evidence type is 15.441 with p=.000 and for the market is 5.948 with p=.015. This shows that these two factors contribute the prediction while the placement does not. In terms of odds ratio, increasing one unit for evidence type (i.e. statistical) increases the odds by 2.611. This presents that the odds of use the service of a certain advertisement is 2.611 times greater for statistical evidence compared to the expert evidence. The odds ratio for country is 0.562. Thus, increased one unit for market (i.e. the Netherlands) decreases the odds by 0.562. It means that the purchase intention of a condition is .562 times smaller for Dutch market as opposed to the Korean market.
4.3. Title Influence on click intention and purchase intention
Logistic regression is used to measure if there is any effect on the dependent variables, namely click intention and purchase intention, depending on the title that respondents found the most appealing. The p-value between the titles and click intention is .437, which shows that there is no relationship between the factor and the dependent variable. In other words, the respondents’ click intention on each condition did not get influenced by title of Search Engine results. Likewise, purchase intention on each condition did not vary depending on the titles that respondents found the most appealing (p=.145). By including this analysis, it is shown that the description of Search Engine Advertising was the only factor that influenced respondents’ click and purchase intention.
5. Discussion
The experiment is conducted to answer the question of “How does content and placement of Search Engine Advertising affect consumer’s click intention and purchase intention, and whether and if so, how cultural differences intervene with this?”. To do so, two representative markets – The Netherlands and South Korea – are selected and individuals from each market are asked their click and purchase intention for four different conditions.
The comparison between the Dutch and the Korean market for each of four conditions shows that there are only significant differences on click and purchase intention when the conditions are not placed on top. In other words, advertisements with the statistical or expert evidence that are placed on top get fairly similar amount of click intention and purchase intention for both of the countries. This result is related to the study conducted by Ghose and Yang (2009) in which the advertisements placed on top receive high clicks compared to other advertisements placed below. Thus, if the appealing messages (i.e. statistical or expert evidences) is combined with the high position, the likelihood of being clicked is high in both countries despite cultural differences in Uncertainty Avoidance and Power Distance. This result means that the findings of Ghose and Yang (2009) are true. Top placement remains a strong factor in pulling click and purchase intention, even when simultaneously measuring the content of the advertisement.
Interestingly, the advertisement with statistical or expert evidences not placed on top, received significantly different click intention and purchase intention between the two markets. This outcome proves that the degree of Uncertainty Avoidance and Power Distance is positively related to click and purchase intention when the ads with either statistical or expert evidences are not placed on top (H2c, H2d, H3c and H3d). Thus, the Korean market which has higher level of Power Distance and Uncertainty Avoidance clicks more on statistical or expert evidence than the Dutch market when the ads are not placed on top.
Based on this discussion, the research question could be answered by stating that different types of evidence affect click and purchase intention differently. However, it has to be noted that when these messages are placed on top, the effect is diminished, as top placement remains the most important factor in determining click and purchase intention. Finally, culture is an important moderator in this relationship. The experiment shows that the Korean market is more sensitive for the type of message than the Dutch market which leaves important implications for future research and practice.
5.1. Managerial implications
As several studies already reveal, the top placement of advertisements is of importance for Search Engine Advertising. This leads to diverse companies to bid for top placement of their ads in exchange of high expenditures. However, as the name indicates, the top placement is for only one limited advertisement. The companies that could not get the top placement for their advertisements or lack of marketing budget should consider carefully how to persuade consumers to click on their advertisements and encourage them to use their products or services. This study shows that the markets with high or low level of Uncertainty Avoidance and Power Distance both frequently click on the top placements. Further proving that attempting to buy top placement positions in Search Engines is worth it for marketers.
However, the markets with different cultural dimensions have different performance when the advertisement is not placed on top. According to the experiment, the Korean market is more sensitive to the type of messages than the Dutch market. If budgets are limited, marketers should go for high positioned advertisements in countries with lower Uncertainty Avoidance and Power Distance. In countries with higher Uncertainty Avoidance and Power Distance, marketers could economize the marketing expenditures by placing their advertisements in lower positions. Markets with lower Uncertainty Avoidance and Power
Distance still are likely to consider to click and purchase services or products if the message is appealing enough.
5.2. Limitations and future research
Although click and purchase intention are related to the actual click through rate and conversion rate, the performance in this experiment and reality can differ. Due to the technical limitations for the author, the research is designed to ask consumers for their click and purchase intention directly. This question might have affected their answers by making the respondents think more carefully. Future research could attempt to develop an experiment in which the actual click though rate and conversion rate are measured among countries with different cultural background.
This study sheds light on the new field of effects of cultural differences in Search Engine Adverting. Yet, one of the limitations of this study is that it is tested only among two different countries. Further research could test with more countries that score differently on the cultural dimensions. This could study whether the value of cultural dimensions have linear relationship with click and purchase intention, or whether it has a different type of relationship.
Likewise, only two important cultural dimensions were tested for this study – Uncertainty Avoidance and Power Distance. However, Hofstede (2001, 2010) introduces four more cultural dimensions, namely, Individualism, Masculinity, Long-term Orientation and Indulgence. Further research could study the effect of these different dimensions, which might lead to more interesting insights in the effect of cultural dimensions on click or purchase intention. This would also allow for more concrete recommendations to practitioners and a deeper development of this new field in academia.
add more interactive elements called extensions such as videos, pictures, direct phone buttons and so on. Thus, it could be interesting to see if these functions have positive effects on click and purchase intention. This will enhance marketer´s insight further and broaden online marketing effectiveness, which could further improve practitioner’s decision-making ability.
6. Conclusion
This thesis attempts to study if the click and purchase intention are affected by the content and placement of a Search Engine Advertising, and whether the effect differs among countries. To test the research question, an online experiment is conducted in which consumer’s click and purchase intention are asked for directly. The four conditions in the experiment were created by mixing two most effective message types – statistical and expert – proposed by Haans et al. (2013) and its placement (i.e. top and non-top placement). The experiment was tested among two different countries, namely South Korea and the Netherlands. These countries differ clearly on the two cultural values (i.e. Uncertainty Avoidance and Power Distance).
The result shows that the click and purchase intention vary across the countries only when the advertisements are not placed on top. The advertisements in top placement received fairly similar click and purchase intention for both of the countries. However, the click and purchase intention differed significantly when the advertisements were not placed on top. Korean people with higher Uncertainty Avoidance and Power Distance clicked more on the conditions than the Dutch.
This contribution of the study is threefold. Firstly, it shows that the content of an advertisement in a Search Engine matters. Secondly, it shows that culture is an important factor in Search Engine Advertising. Although cultural values are considered as an important factor in offline setting, so far this has often been neglected in online marketing, especially in Search Engine Advertising. Finally, this study provides an integrated view of important factors in the field. The experiment shows that despite the importance of the content of Search Engine Advertising, top placement remains of critical importance for effective Search Engine Advertising.
References
Bucklin, R. E., & Sismeiro, C. (2009). Click here for internet insight: Advances in
clickstream data analysis in marketing
doi:http://dx.doi.org/10.1016/j.intmar.2008.10.004
Camerer, C., & Weber, M. (1992). Recent developments in modeling preferences: Uncertainty and ambiguity. Journal of Risk and Uncertainty, 5(4), 325-370.
Chandon, J. L., Chtourou, M. S., & Fortin, D. R. (2003). Effects of configuration and exposure levels on responses to web advertisements. Journal of Advertising Research,
43(2), 217-229.
Chatterjee, P., Hoffman, D. L., & Novak, T. P. (2003). Modeling the clickstream: Implications for web-based advertising efforts. Marketing Science, 22(4), 520-541.
Choi, Y. K., & Miracle, G. E. (2004). The effectiveness of comparative advertising in korea and the united states: A cross-cultural and individual-level analysis. Journal of
Advertising, 33(4), 75-87.
Gauzente, C. (2010). The intention to click on sponsored ads—A study of the role of prior knowledge and of consumer profile. Journal of Retailing and Consumer Services, 17(6), 457-463.
Ghemawat, P. (2001). Distance still matters. the hard reality of global expansion. Harvard
Ghose, A., & Yang, S. (2009). An empirical analysis of search engine advertising: Sponsored search in electronic markets. Management Science, 55(10), 1605-1622.
doi:10.1287/mnsc.1090.1054
Haans, H., Raassens, N., & van Hout, R. (2013). Search engine advertisements: The impact of advertising statements on click-through and conversion rates. Marketing Letters, 24(2), 151-163. doi:10.1007/s11002-013-9226-5
Hofstede, G. (Ed.). (2001). Culture’s consequences: Comparing values, behaviors,
institutions, and organizations across nations (Second Edition ed.) Thousand Oaks CA:
Sage Publications.
Hofstede, G., Hofstede, G. J., & Minkov, M. (Eds.). (2010). Cultures and organizations:
Software of the mind (Revised and Expanded 3rd Edition ed.) McGraw-Hill.
Hollis, N. (2005). Ten years of learning on how online advertising builds brands. Journal of
Advertising Research, 45(2), 255-268.
Hornikx, J. (2005). A review of experimental research on the relative persuasiveness of anecdotal, statistical, causal, and expert evidence. Studies in Communication Sciences,
5(1), 205-216.
Hornikx, J., & Hoeken, H. (2007). Cultural differences in the persuasiveness of evidence types and evidence quality. Communication Monographs, 74(4), 443-463.
doi:10.1080/03637750701716578
Internet World Stats. (2017). Http://www.internetworldstats.com/stats16.htm [accessed on June 4th, 2017].
Jansen, B. J., & Resnick, M. (2006). An examination of searcher's perceptions of
nonsponsored and sponsored links during ecommerce web searching. Journal of the
American Society for Information Science and Technology, 57(14), 1949-1961.
doi:10.1002/asi.20425
Jerath, K., Ma, L., Park, Y., & Srinivasan, K. (2011). A ``Position paradox'' in sponsored search auctions. Marketing Science, 30(4), 612-627.
Lohtia, R., Donthu, N., & Hershberger, E. K. (2003). The impact of content and design elements on banner advertising click-through rates. Journal of Advertising Research,
43(4), 410-418.
Maheswaran, D., & Meyers-Levy, J. (1990). The influence of message framing and issue involvement. Journal of Marketing Research, , 361-367.
Manchanda, P., Dubé, J., Goh, K. Y., & Chintagunta, P. K. (2006). The effect of banner advertising on internet purchasing. Journal of Marketing Research, 43(1), 98-108.
Moe, W. W., & Fader, P. S. (2004). Dynamic conversion behavior at e-commerce sites.
Management Science, 50(3), 326-335.
Morwitz, V. G., & Schmittlein, D. (1992). Using segmentation to improve sales forecasts based on purchase intent: Which "intenders" actually buy? Journal of Marketing
Research, 29(4), 391-405. doi:10.2307/3172706
PwC. (2016). IAB internet advertising revenue report. ().
http://www.iab.com/wp-content/uploads/2016/04/IAB_Internet_Advertising_Revenue_Report_HY_2016___.pdf. [Accessed Jan. 30 2017]: PwC.
Roth, M. S., & Moorman, C. (1988). The cultural content of cognition and the cognitive content of culture: Implications for consumer research. NA-Advances in Consumer
Research Volume 15,
Rutz, O. J., & Bucklin, R. E. (2011). From generic to branded: A model of spillover in paid search advertising. Journal of Marketing Research, 48(1), 87-102.
Statista. (2017). Https://www.statista.com/statistics/237974/online-advertising-spending-worldwide/ [accessed on may 25, 2017].
Vakratsas, D., & ambler, T. (1999). how advertising works: What do we really know? journal of marketing, 63(1), 26-43.
Yang, S., & Ghose, A. (2010). Analyzing the relationship between organic and sponsored search advertising: Positive, negative, or zero interdependence? Marketing Science,
29(4), 602-623.
Zenetti, G., Bijmolt, T. H. A., Leeflang, P. S. H., & Klapper, D. (2014). Search engine advertising effectiveness in a multimedia campaign. International Journal of Electronic
