Information Competition in Disruptive Media Markets: Investigating Competition and User Selection on Google

Abstract

Due to digitization, traditional media brands are facing hypercompetition. For one thing, media outlets offering journalistic content no longer compete just with each other but also with all sorts of content from various sources, such as corporate publishers. This particularly applies to the information space provided by Google and other search engines. This leads to the question of how traditional media brands prevail in this information space: do traditional media brands have a competitive advantage because users perceive their journalistic content as more valuable in terms of credibility and reputation? Accordingly, this study investigates competition and user selection on Google. Drawing on a representative, experimental selection study of the German-speaking Swiss population (N = 1,100), search engine selection behavior was investigated. Results show that selection preferences do not differ between traditional media brands and other competitors, such as corporate publishers. This poses a major challenge for media brands. However, credibility and reputation significantly influence selection preferences. Thus, media managers should focus on effective branding activities in order to maintain a strong position in the digital information market.

Keywords:

• Media markets
• competition
• media brands
• user selection
• competitive information space
• blockchain

Introduction

Dwindling advertising revenues, diminishing levels of readership, and a decreasing willingness to pay by online consumers has led to a multi-faceted erosion of the economic foundations of media brands that offer journalism (Currah 2009; Doyle 2013; Nielsen 2019; Siles and Boczkowski 2012). In addition, new and highly professionalized non-media competitors have entered the media market and intensified competition in the online information market (Picard 2006). The once-solid position and information sovereignty of traditional media brands within media markets is increasingly being challenged as new competitors, such as corporate publishers, “compete for audiences” (Baetzgen and Tropp 2015, 137). Consequently, media brands are entering a state of “hypercompetition” (Hollifield 2006, 63), where a variety of content providers “engage in activities once carried out only by media firms” (Picard 2006, 32).

In addition to these market-side transformations, a fundamental shift in media usage has taken place: not only have media and information usage largely shifted to the online sphere (Newman et al. 2019), but search engines—first and foremost Google—have also become the most important source for the acquisition of information online (Segev 2010). Search engines are powerful drivers of hypercompetition; in this context, different types of information offerings (sources) compete for the user’s attention and clicks. However, virtually no empirical studies examine the competitive situation on search engines for different thematic segments of the online information market (Krebs et al. 2021; Russi et al. 2014).

The presented study addresses this research gap. We use the term competitive information space to indicate how countless communicators and a plethora of content compete for users’ attention (Xiang and Law 2013). By focusing on the thus named competitive information space on Google, we investigate information competition and user selection for two socially relevant topics. More precisely, we examine underlying brand-related factors that influence user selection preferences in search engine environments. The research questions guiding this study are:

RQ1: Which types of information offerings are most likely to be selected by users?

RQ2: Which brand-related factors influence user selection preference in search engine environments?

In order to answer these questions, an innovative methodological approach was needed to capture the dynamic competitive conditions present in search engine environments. For this purpose, a specially developed web scraper was programmed that automatically extracts real search results for specific search queries. These results then served as the material basis for the experimental online survey. In this survey, the selection behavior of 1,100 respondents in response to a real search engine selection scenario was investigated.

First, this article describes how existing media markets are being disrupted in times of digitization and are transitioning to a market conceptualization that is increasingly dissolved. Second, the significance of search engines in the constitution of a competitive information space online is discussed, and empirical findings on influential factors within search engine environments are presented. Third, the role of media brands as a focal point of orientation in ever-more liquid media markets is discussed. Given this basis, specific hypotheses were developed. After the methodological approach is outlined, the analyses and results are presented and discussed.

The Significance of Search Engines: Toward a Competitive Information Space

As Picard (2006) outlines, the Internet is a new competitive environment, where a wide variety of companies can enter the media market and “engage in activities once carried out only by media firms” (32). Baetzgen and Tropp (2015) underline this development, demonstrating that in modern societies, a new interrelation between consumer brands and media is evolving in such a way that consumer brands are “becoming involved in media content production” (135).

In this context, researchers often speak of a collapse between brand and media management that blurs the boundaries between once solidly defined concepts such as information, entertainment, and advertising (Bentele, Hoepfner, and Liebert 2015; Koch and Obermaier 2014; Koch, Obermaier, and Riesmeyer 2020). This often results in the production of new formats, such as “brand-owned media” (Baetzgen and Tropp 2015, 138). With the emergence of such new formats and other differentiations, such as corporate publishing (Bentele, Hoepfner, and Liebert 2015), the conditions of the media market are changing drastically, as new media formats “will compete for audiences” (Baetzgen and Tropp 2015, 137).

Particularly against the backdrop of this blurring of activities, it is of vital importance to distinguish between traditional media companies and their non-media competitors, despite—or because—common definitions do not exist. When speaking of traditional media companies, we refer to a common understanding that is “based on the idea of a publisher or broadcaster producing or aggregating, bundling and distributing content” (Hess 2014, 3), with primarily journalistic competencies in creating, curating, editing, and providing content. Even though it must be argued that non-media competitors (such as corporate publishers) cannot be clearly defined, it can still be recognized that those offerings lack the central preconditions for critical and independent journalism (Altmeppen 2006, 201–208; Krebs and Lischka 2019).

Corporate publishing refers to the process and result of planning, producing, organizing, and evaluating corporate publications. It can include cross-media formats, such as blogs, newsletters, magazines, and other corporate media formats, that all serve the main goals of image-building and customer loyalty for a specific corporation (Bentele 2012; Bentele, Hoepfner, and Liebert 2015; Weichler and Endrös 2010). Due to digitization, such offerings have multiplied in recent years, which has not only resulted in a definitional dissolution of erstwhile solid concepts but has also fundamentally changed the competitive situation, especially in the online media market. One key consequence of this development, which is important from a media managerial perspective as well as for media economics scholars, is the question about the extent to which traditional media companies can face these new competitive conditions online, build up strong brands, and thus eventually survive creative destruction in times of digitization.

Within the digital sphere, a focal point has arisen where new competitive dynamics become particularly visible. Following Xiang and Law (2013), we argue that the changing competitive conditions in digital environments can best be illustrated and assessed by focusing on the information space offered by search engines like Google. Search engines are not a trivial environment in terms of where the growing liquidity of digital media markets manifests itself: after all, it is precisely on search engine results pages where different types of offerings convene, traditional boundaries between media products and providers blur, geographic limitations of dissemination are overcome, and 24-hour availability is enabled (Hollifield 2006; Krebs et al. 2021). Therefore, search engines constitute a dynamically changing digital space where different offerings compete for users’ attention and clicks. This is reflected in the increasing role played by search engine optimization (SEO) marketing for reaching, attracting, and engaging users and potential customers online (Enge, Spencer, and Stricchiola 2015; Xiang and Law 2013).

From a media user’s perspective, search engines have not only become an integral part of our daily personal and professional life but have also established themselves as the primary source of information searches online (Kammerer and Gerjets 2014; Segev 2010). As the most popular and most used search engine, Google has become the market leader and the key digital gatekeeper for information access (Machill and Beiler 2007; Machill et al. 2004; Segev 2010; Stark, Magin, and Jürgens 2014; Unkel and Haas 2017). Google enables users to find relevant options from among a huge abundance of information online that are considered to satisfy their current information needs (Kammerer and Gerjets 2014) in times of information overload (Eppler and Mengis 2004; Schmitt, Debbelt, and Schneider 2018).

Search engine users are a key factor in this regard, as they select offerings that fulfill their information needs and therefore, along multiple other criteria such as Google’s crawling and PageRank algorithms (Avrachenkov and Litvak 2006), co-determine competition. Albarran (2017) states: “Consumers play a pivotal role in the transformation of the media economy, not just in terms of their tastes and preferences, but also in their allocation of time (attention) and expenditures on media-related products and services” (183). This particularly manifests itself in the competitive environment that search engines engender, contouring a competitive information space. This space allows the assessment and evaluation of emergent dynamics within digital media competition: “Search engines can be used to gauge and assess emergent changes of the competitive information space” (Xiang and Law 2013, 534).

Some empirical studies have analyzed the composition of the information supply on Google (e.g. Diakopoulos et al. 2018; Magin et al. 2015, Puschmann 2019); however, almost no empirical studies have examined the competitive situation between different information providers on Google from a structural perspective. A study conducted by Krebs et al. (2021) demonstrates that the competitive structure for different thematic search queries on Google is still dominated by traditional journalistic news offerings. Nevertheless, these offerings are surpassed by non-profit offerings or corporate publishers in terms of the weighted ranking position on the search engine result page (SERP) (Krebs et al. 2021).

Media branding is considered to be one of the most important success factors for media companies, as it enables and guarantees a strong position in media markets in the long term (Bruno and Nielsen 2012; Siegert, Gerth, and Rademacher 2011). However, previous research has not been able to comprehensively address this question, as new conditions of the platformization of media consumption have not been considered. As a result, we assume that journalistic information offerings have a selection advantage and are thus more likely to be selected than other types of information offerings. In relation to the RQ1, and based on the above-mentioned considerations, we develop the first two hypotheses that will be tested by this experiment:

H1.1: Journalistic information offerings are most likely to be selected by users.

H1.2: Corporate information offerings are least likely to be selected by users.

Even if such descriptive analyses provide valuable insight into the competitive conditions among traditional journalistic offerings, corporate publishing, and other offerings on Google, they still remain limited by the absence of a user-centric perspective. In purely descriptive analyses, real user selection processes remain unconsidered; therefore, only an incomplete picture of potential differences in the competitive (dis-)advantages between different types of offerings can be drawn. Knowing the reasons why search engine users select certain offerings can offer key insights for strategic media management and branding decisions, thereby strengthening the strategic positioning of journalistic offerings in the digital market.

Factors Influencing User Selection in Search Engine Environments

User selection processes in digital search engine environments differ considerably from those in traditional media environments, as search engines not only provide users with a vast amount of different information offerings but also confront them with a “higher degree of uncertainty regarding the source and quality of the retrieved information” (Unkel and Haas 2017, 1850). In line with information processing and decision theory (Payne and Bettman 2004; Payne, Bettman, and Johnson 1993), internet searches can be described as a situation of high uncertainty in which a selection between innumerable alternatives is required. Search engine rankings help reduce this uncertainty (Kammerer and Gerjets 2014; Unkel and Haas 2017).

Typically, on search engine result pages, only a little information about the offering is visible (Kammerer and Gerjets 2014). From a theoretical point of view, it is assumed that individuals in such uncertain situations aim to maximize their outcome (e.g. meeting their need for information) while minimizing their cognitive effort, which particularly manifests itself in conditions of limited knowledge and time (Kammerer and Gerjets 2014; Payne and Bettman 2004). As Kammerer and Gerjets (2014) point out, “[I]n trying to find an optimal tradeoff between cognitive effort and efficient outcome, instead of a systematic, that is, thorough and complete knowledge based evaluation of all given information, individuals often evaluate information in a rather heuristic way” (178).

As a result, individuals rely on so-called heuristic cues when selecting between different alternatives on online media and particularly in search engine environments (see also Metzger 2007; Metzger, Flanagin, and Medders 2010; Wirth et al. 2007). Research in this area has identified several heuristic cues that (partially) explain user selection of search engine results, namely ranking, source reputation, and source credibility (Haas and Unkel 2017; Kammerer and Gerjets 2014; Knobloch-Westerwick et al. 2015; Krämer et al. 2018; Sundar 2008; Sundar, Knobloch-Westerwick, and Hastall 2007; Unkel 2019; Unkel and Haas 2017).

According to Metzger and Flanagin (2015), source credibility plays a decisive role in the selection of information offerings, especially (but not only) in the online sphere (Krämer et al. 2018). The central role of credibility judgment in information selection has been investigated by scholars for decades, yet results conflict and often contain confusing concepts and definitions of credibility (Self and Roberts 2019). Nevertheless, one common thread is the depiction of credibility as a subjective, ascribed characteristic that is perceived by media recipients and not as a preexisting trait of a specific media object itself (Unkel and Haas 2017). Sundar (2008) points out: “Credibility is classically ascertained by considering the source of information. If the attributed source of a piece of information is a credible person or organization, then, according to conventional wisdom, that information is probably reliable” (73). In line with this idea, we are interested in the role of source credibility ascriptions as cues “signifying expertise and trustworthiness of search results in search engine selection behaviour” (Unkel and Haas 2017, 1851).

Based on the above-mentioned findings and theoretical considerations, we develop two hypotheses that will be tested by means of this experiment:

H2.1: Brand credibility positively influences user selection preference.

H2.2: Brand reputation positively influences user selection preference.

Previous research has shown that selection processes on search engines are also strongly influenced by ranking, meaning the position of search results on a specific SERP (Kammerer and Gerjets 2014; Lewandowski 2012; Machill et al. 2004; Metzger, Flanagin, and Medders 2010; Pan et al. 2007; Schultheiß, Sünkler, and Lewandowski 2018; Unkel 2019; Unkel and Haas 2017; Wirth et al. 2007). These studies indicate that search engine users follow a habitual selection pattern: they predominantly select top-ranked results, as higher ranking seems to signal higher relevance (Lewandowski 2012). In addition, search engine users “rarely go to the second page of results” (Pan et al. 2007, 803) or leave the “visible area” (Lewandowski 2013, 188) of the SERP—that is, the area that is only reached by scrolling down (Unkel and Haas 2017). From a user’s perspective, it is often assumed that top-ranked search results automatically link to more credible information offerings (Hargittai et al. 2010), although this, in fact, should not be considered an indicator of credibility, since search engines have no underlying control mechanism for this purpose (Unkel and Haas 2017). Drawing on this, we derive the third hypothesis:

H2.3: The preset ranking position positively influences user selection preference.

Studies in this field have provided valuable insights into the influence of credibility cues and ranking position on selection processes in search engine environments. Nevertheless, they have not been able to take brand-related characteristics into account. This seems paradoxical, since research on media brands has decisively emphasized the role of brands in the selection process of media products and content (Krebs 2017; McDowell 2005).

The Role of Media Brands as Focal Points of Orientation in Hypercompetitive Markets

Media brands play a fundamental role in users’ selection processes in liquid media markets; they provide orientation, reduce risk and uncertainty, and prevent cognitive overload in an increasingly complex and nebulous digital market environment (Krebs 2017; McDowell 2005). Nevertheless, against the backdrop of intensifying competition between journalistic media offerings and other formats such as corporate publishing, it seems important to clearly distinguish media brands from other (e.g. corporate) brands. We follow the user-centered approach by perceiving

“a media brand as a construct carrying all the connotations of the audience in terms of the emotional, stylistic, cognitive, unconscious or conscious significations. These significations can refer to different levels in a media brand’s architecture, which typically consists of the corporate or channel brand as well as its sub-brands with genre, format, and persona brands” (Siegert et al. 2015, 1).

In line with the distinction drawn earlier between traditional media companies and other competitors, we additionally limit the definition of media brands to traditional journalistic media companies that, in contrast to their non-journalistic competitors, follow the primary purpose of journalistic actions.

From a user’s perspective, media brands are considered valuable, as they enable quality-related assessments prior to actual usage (Lis and Post 2013). In this view, media brands fulfill the central function of signaling the quality and the credibility of media products (Krebs 2017; McDowell 2005). In recent years, brand management research has often focused on the influence of brand-related constructs and dimensions on audience behavior, such as motivations, loyalty, or usage (Krebs and Lischka 2019; Krebs 2017; Lischka and Messerli 2016). For example, Lis and Post (2013) demonstrate that brand image influences selective media content consumption.

Other studies have shown how the image, attitude, usage, credibility, and loyalty of news brands are interconnected (Chan-Olmsted and Cha 2008; Oyedeji 2007); these studies are often grounded in the theoretical concept of customer-based brand equity (CBBE) (Oyedeji 2007; Oyedeji and Hou 2010). CBBE is an integrative branding construct that measures consumers’ attachment to a brand (Aaker 1991; Oyedeji 2007). It contains several dimensions, including brand loyalty, quality perception, brand associations, and brand awareness (Krebs 2017). Studies in this area have also focused on the influence of reputation on media product assessments (Chan-Olmsted, Cho, and Yim 2013). Generally speaking, brand management research has reinforced the influence of specific brand dimensions on audience behavior and perceptions (Krebs 2017). However, research examining the influence of brand-related factors on selection processes in search engine environments remains indeterminate, and further investigation is needed.

From a market perspective, it is assumed that media brands can stem the tide of intensified competition to the extent that a strong brand can raise entry barriers for new competitors with similar media products that are potential substitutes (Aaker 1996). Furthermore, branding research has shown a direct correlation between a strong brand and economic success (Keller, Apéria, and Georgson 2012; Lis and Post 2013). However, due to the platformization of media usage (Helmond 2015), serious doubts have arisen about the efficacy of brands. While media brands currently remain important for the advertising industry (Knuth, Kouki, and Strube 2013), early research in this context has indicated that brand allegiance seems to be declining in platformized digital environments—such as search engines—as brand memory decreases. Still, differences in memory between certain brands persist that can be explained by brand loyalty (Kalogeropoulos and Newman 2017). Based on the above, we assume that strong brand loyalty provides a selection advantage:

H2.4: Brand loyalty positively influences user selection preference.

Methodology

In order to answer the research questions, a representative, experimental online survey of the German-speaking Swiss population was conducted. In addition to the clear advantages of online surveys, such as speed and multimedia flexibility, this method allows certain technical possibilities for reproducing and simulating realistic selection scenarios on Google search result pages. After several introductory questions about their internet and search engine usage, respondents were asked to select search results from a reconstructed, manipulated Google search result page, according to their selection preferences. In order to prevent brand-related questions from influencing the selection task, the standardized survey was conducted after the selection experiment.

Stimulus Material and Selection Tasks

The technical reconstruction of the SERP followed a real scenario, using a web scraper (Mitchell 2018) specifically programmed in a preliminary study (Krebs et al. 2021). The web scraping software, programmed using the Python programming language (Donaldson 2008), enables automated collection of a virtually unlimited number of Google search results for various predefined search queries. The search queries were conducted using a cookie-free IP address on the Swiss edition of Google (google.ch).

Search results on Google for two predefined search terms were automatically collected and extracted into a dataset (see Appendix). The thematic delimitation of the search terms was based on the following considerations.

First, a topic was selected that had high social relevance not only in Switzerland but also beyond European borders: blockchain technology. Blockchain technology is currently regarded as a key technology that is expected to have a significant economic, political, and societal impact in the future (Swan 2015). Despite the increasing media focus, the phenomenon of blockchain technology seems to be relatively new; therefore, it is assumed that knowledge of the topic is still limited among the general population, and an impartial selection process can be expected.

Second, in order control for possible differences in offering structure and selection behavior, another topic was selected for high recognition and relevance from the user’s perspective: Wimbledon. This topic was derived from the Top three Google Trends (2018), which represented the search queries most entered by Google users in Switzerland in 2018. Therefore, high real user interest on this topic was confirmed.

The selection of the individual search results for the construction of the SERPs was carried out using systematic logic and considering representation, ranking position, and page number. In order to create as realistic a SERP as possible, 14 search results were selected from the Web scraping data that matched all of the following criteria: 1) occurred most frequently, 2) were ranked highest, and 3) were displayed on the first page of the SERPs. In total, all relevant information offering types—i.e. non-profit, traditional journalistic, online-only journalistic, special-interest journalistic, corporate publisher, and corporate website—were included (for the definitions of the offering type categories, see Table in Appendix).

The selected search results formed the base for the subsequent construction of the SERP, on which the participants were asked to perform the selection tasks. Search results from the following brands were used for the constructed SERPs: Wikipedia (non-profit), Deloitte/Wimbledon (corporate publisher), SAP/Adidas (corporate website), Neue Zürcher Zeitung/Blick (traditional journalism), Watson (online-only journalism), BTC-ECHO/Tennis-Magazine (special-interest journalism), and Cognizant/Stubhub (Google Ad). With the exception of Google Ads, all search results were positioned on the SERP in a randomized order (see Figure in Appendix).

It should be noted that Google has recently introduced selectively presented news boxes to display additional journalistic offerings, such as “Top Stories” (Lurie and Mustafaraj 2019). However, these offerings depend heavily on the current news situation; therefore, in this study, this additional form of presentation was intentionally excluded. Additionally, in order to guarantee the highest possible external validity, fixed advertisements were integrated in the upper area of the SERP. In this setting, no manipulation or treatment check is required (O’Keefe 2003).

In the experiment, all participants were asked to perform two selection tasks (search terms: “blockchain” and “Wimbledon”) from among seven search results each, wherein they had to click on each search result in their preferred order (compulsory selection). To enhance external validity, seven search results were used for each of the constructed SERPs, as this represents a common number of search results on the first page.

After the selection task, participants were asked to answer specific brand-related questions about each search result. Finally, self-assessment questions about the participant’s selection behavior were asked using a standardized questionnaire.

Sample and Procedures

An external market research company (Respondi) was asked to distribute the online survey based on a representative quota sample (see Table 1) of the German-speaking Swiss population aged 18–69 years (M = 27.99). The size and composition of the sample (N = 1,100 respondents) allowed general conclusions to be drawn about the German-speaking Swiss population as a whole. A concentration check (“If you read this sentence, please select the answer option never”) and a motivation check (“How focused were you during the survey?”) were integrated, whereby non-concentrated and unmotivated participants were automatically screened out of the sample. Each participant was asked to make 14 selection decisions, which led to a final case number of N = 15,400. The online survey was conducted in September 2019.

Table 1. Demographic characteristics of participants (N = 1,100).

Characteristic		n	%
Gender	female	554	50.4
	male	546	49.6
Age (Years)	18–29	258	23.5
	30–39	219	19.9
	40–49	222	20.2
	50–59	232	21.1
	60–69	169	15.4
Level of Education	low	34	3.1
	low-medium	459	41.7
	medium	157	14.3
	medium-high	207	18.8
	high	243	22.1
Total		1,100	100

Measurements

The post-experimental questionnaire started with questions on sociodemographic information (gender: 1 = male, 2 = female; age = years) and educational level (low to high), in order to examine quota overabundance as well as to remove potential screen-outs at the beginning. Questions on media usage (M = 2.98, SD = 1.12), internet usage (M = 2.99, SD = 1.09), and search engine usage (M = 4.17, SD = 1.06) were adapted from Haas and Unkel (2017) and Unkel (2019) and were measured on a five-point Likert scale. Previous knowledge (M = 2.10, SD = 1.17) on the two topics and individual topic relevance (M = 3.51, SD = 1.45) were determined using three items derived from Unkel (2019) and were measured on a five-point Likert scale. The use of specific search engines (Google, Yahoo, Bing, others) was asked via multiple selections.

Individual brand awareness was measured for each brand on a single-item five-point Likert scale (1 = “I don’t know this brand at all” to 5 =“I know this brand very well”) adapted from Unkel (2019). Brand credibility was measured for each brand on a single-item five-point Likert scale (1 = “very non-credible” to 5 = “very credible”) derived from Metzger and Flanagin (2015). Brand reputation, based on Oyedeji (2007) and Lis and Post (2013), was measured for each brand via two items (“In my opinion, this brand has a good reputation” and “I have a positive attitude towards this brand”) using a five-point Likert scale (1= “I strongly disagree” to 5= “I totally agree”) taken from Krebs (2017); these items were later indexed into one variable. Brand loyalty, based on Aaker (1991), was measured with a single item (“I have a strong loyalty to this brand”) using a five-point Likert scale (1= “I strongly disagree” to 5= “I totally agree”) derived from Krebs (2017). Table 2 shows the intercorrelations among these brand variables.

Table 2. Intercorrelations among brand variables.

Variable	1	2	3	4
1 Individual brand awareness	–
2 Brand credibility	.23	–
3 Brand reputation	.28	.73	–
4 Brand loyalty	.35	.54	.61	–

Note. All are significant at p < .01.

Follow-up questions about trust in search engines’ rankings (M = 3.48, SD = 1.14) (Schultheiß, Sünkler, and Lewandowski 2018), search engine experience (M = 4.08, SD = 0.88) (Kammerer and Gerjets 2014), different search engine usage motives (Unkel 2019), and various search engine behavior self-assessments (Beiler 2005; Stark, Magin, and Jürgens 2014) were analogously measured on five-point Likert scales.

In line with Haas and Unkel (2017), ranking position was captured as a metric variable that indicates each search result’s position on the SERP. Correspondingly, users’ selected ranking preference (1–7) was automatically captured during the selection tasks as a metric variable that measured users’ selection order for each individual search result on the SERP.

Results

User Selection Preferences regarding Different Types of Information Offerings

To answer RQ1, we analyzed the types of information offerings that were most likely selected by the users. Table 3 shows users’ selection preferences by offering type for both search queries. A low selection-preference mean (M) indicates a high preference for selection, and vice versa. Two one-way analyses of variance (ANOVA) show that there were statistically significant differences in selection preference between different offering types for both search queries (for Blockchain: F(6,7693) = 31.971, p < .001, η² =.30; for Wimbledon: F(6,7693) = 82.546, p < .001, η² =.26).

Table 3. Users’ selection preferences by offering type and search query.

	Blockchain (n = 7,700)	Wimbledon (n = 7,700)
Offering type	Mblock (SD)	Mwimb (SD)
Non-profit	2.8a (1.7)	3.2b (2.1)
Corporate publisher	2.8a (1.7)	3.1b (2.0)
Traditional journalism	3.0a (1.4)	3.3b (1.6)
Online-only journalism	4.1 (1.5)	3.5 (1.5)
Corporate website	4.6 (1.7)	5.1 (1.4)
Special-interest journalism	4.8 (1.8)	3.8 (1.5)
Advertisements	5.9 (1.8)	5.9 (1.8)

Note. Shared subscripts (a, b) indicate no statistically significant differences between these groups (p >.05) according to the Games-Howell criterion. All other differences are statistically significant (p < .05 if not reported otherwise). A low selection rank mean indicates a high preference for selection, and vice versa.

With regard to effect sizes, relatively strong effects were apparent. Offering type explains a high proportion of variance in selection preference (30% resp. 26%). Overall, a clear pattern can be seen in selection preferences with respect to offering types; these do not substantially differ between the two search queries. Table 3 indicates that users preferred non-profit offerings, closely followed by corporate and traditional journalistic offerings, within their first three selection decisions.

Given the heterogeneity of variance and the large number of cases, post-hoc multiple comparison tests were conducted according to the Games-Howell criterion of significance (Games and Howell 1976). However, the Games-Howell criterion shows no differences between these three offering types with regard to selection preference (p >.05). Therefore, H4—stating that traditional media brands have a higher selection preference than other brands—must be rejected. Nevertheless, the results also show that traditional journalism is significantly preferred to other types, such as online-only journalistic offerings (p <.001). Both analyses clearly show that ads were selected last, despite always appearing at the top.

How Ranking Position and Brand-Related Factors Influence User Selection Preference

To answer RQ2 and to further investigate the influence of search results’ ranking position as well as that of brand-related factors on selection preference, two linear regression models were calculated (see Table 4). Control variables (e.g. sociodemographics, internet usage, search engine usage, individual topic relevance) were excluded from the models, due to insignificant effects.

Table 4. Summary of linear regression models for variables predicting selection preferences.

	Model 1 (n = 15,400)		Model 2 (n = 6,975).
Variable	β	SE	β	SE
Ranking position	−.290*	.008	−.093*	.014
Brand credibility			−.080*	.031
Brand reputation			−.090*	.035
Brand loyalty			.003	.022
R^2	.084		.038
△R^2			.046*

Note: Selection preference was coded from 1–7, where a low mean indicates a high selection preference. Ranking position was coded from 1–7. Brand Credibility, Reputation, and Loyalty were coded from 1–5 (low to high). ^*p < .001. Listwise deletion (respondents who did not know the brand at all) led to the final number of n = 6,795.

Model one tested the influence of the search results’ ranking position (1–7) on selection preference (Table 4). A high-ranking position means that the search result was placed at the top of the SERP, whereas a low-ranking position signifies a position at the bottom of the SERP. A low selection preference mean indicates a high selection preference, and vice versa. The model shows a significant effect of ranking position on selection preference; however, the analysis shows that ranking position had a significant negative effect on selection preference (β = −.290, p < 0.01, SE =.0.08). This means that users did not simply select search results in line with their given ranking positions, but rather opted for results that were located in the lower part of the SERP. Therefore, H2.3 must be rejected, given the negative effect of preset ranking position on selection preference.

Model two tested the extent to which brand-specific factors influenced users’ selection preferences (Table 4). We found a significant effect of brand-specific factors on selection preference. The model shows that brand credibility significantly influenced selection preference (β = −.080, SE = .031), wherein the higher the brand credibility, the higher the selection preference and the lower the selection preference mean. A similar result was seen in terms of brand reputation; here, too, the model shows a significant effect of brand reputation (β = −.090, SE = .035) on selection preference, wherein the higher the brand reputation, the higher the selection preference and the lower the selection preference mean. These results are in line with hypotheses H2.1 and H2.2. However, no significant effect of brand loyalty on selection preference was found; therefore, H2.4 must be rejected. In this model, ranking position still had a significant effect on selection preference (β = −.090, SE = .035).

Overall, users preferred search results from a brand that was both credible and reputable to them, while loyalty to a brand did not seem to influence selection preference. In addition, users did not simply select search results according to their predetermined ranking position; rather, they preferred search results from the lower part of the SERP in their selection decisions.

Discussion

One general finding of this study is that an information offering’s corresponding brand is of vital importance in the selection of information offerings in search engine environments. This is reflected in the high effect sizes revealed by the analyses of variance. This finding brings hope not only to traditional journalistic media organizations but to every company with a strong brand, since it undercuts existing assumptions about the subordinate role brands play in search engine environments (e.g. Kalogeropoulos and Newman 2017). However, the analyses show that in search engine environments, users’ selection preferences do not differ between non-profit offerings, corporate publishing offerings, and traditional journalistic offerings; this remains true for both technology-related and trending topics. This finding has several implications for competition online.

There are signs that traditional journalistic media brands, which were previously characterized as focal points of orientation (Krebs 2017; McDowell 2005), might lose their one-time competitive advantage to other information offerings, such as non-profit offerings or corporate publishers. Competition between these offering types does seem to be greater, as users are indifferent in terms of their selection preferences in this respect. However, it is precisely non-profit offerings, traditional journalistic offerings, and corporate publisher offerings that have the highest selection probability; thus, these have a competitive advantage compared to other information offerings, such as specialized or online-only journalistic offerings and for-profit corporate websites. Surprisingly, even for specific topics such as blockchain technology, specialized media offerings were almost the least preferred; this suggests that such offerings likely address a very specific user segment.

These findings indicate that the content of traditional journalistic news brands seems to increasingly become indistinguishable for users, which at least partially supports the assumption of the possible substitutability of traditional journalistic content by new (non-media) competitors (De Waal and Schoenbach 2010; Dimmick, Chen, and Li 2004; Domingo and Le Cam 2014; van der Wurff 2011). In fact, this poses a considerable danger for traditional journalistic media brands, as new corporate competitors are increasingly professionalizing (e.g. Bentele, Hoepfner, and Liebert 2015). Especially against the backdrop of the extraordinarily fast and dynamically changing competitive situation in search engine environments, the erstwhile sovereignty of traditional journalistic information seems to be increasingly called into question.

Another interesting result revealed that search engine advertisements were largely ignored by users, given that they were clearly last among the users’ selection consideration sets. This finding contradicts previous research and conventional wisdom, coupled with the recent increase of the use of search engine marketing (Xiang and Law 2013). One explanation for this unexpected result could be that the ranking was completely randomized in our experiment, while in real Google rankings a number of purposefully selected factors play a decisive role. Nevertheless, it is unclear whether this result is an artifact of our experimental set-up; if not, the finding would undercut the effectiveness of search engine advertising (SEA). Hence, future studies should address this question.

Relevant influencing factors were investigated in this study in two multiple linear regression models, from which the following major findings emerge. First, in line with prior research in this domain (see e.g. Kammerer and Gerjets 2014; Lewandowski 2012; Machill et al. 2004; Metzger, Flanagin, and Medders 2010; Pan et al. 2007; Schultheiß, Sünkler, and Lewandowski 2018; Unkel 2019; Unkel and Haas 2017), the analyses show that search results’ ranking position significantly influences selection preference, despite the randomization of search results. However, contrary to most extant evidence, users did not simply choose according to the predetermined ranking order by Google: users actually scrolled down and preferred information offerings from the lower part of the SERP, indicating that they engage in an active, conscious selection process in which different options in their consideration set are weighed against each other. In addition, due to the small effect sizes of ranking position, other factors seem to predominate when it comes to selection preference.

These other factors were investigated in the second multiple linear regression model. The analysis revealed that credibility and reputation assessments of the search result’s brand have a decisive influence on the selection preference of the user, which is reflected in this factor’s higher explanatory power on effect size. Users are most likely to select search results whose associated brand they find most credible and reputable, even if the results are randomly distributed on the SERP. This supports existing findings on the relevance of credibility and reputation in search engine selection processes (e.g. Haas and Unkel 2017; Kammerer and Gerjets 2014; Knobloch-Westerwick et al. 2015; Krämer et al. 2018; Sundar 2008; Sundar, Knobloch-Westerwick, and Hastall 2007; Unkel 2019; Unkel and Haas 2017). However, loyalty to a brand did not have a significant effect on selection preference; this seems surprising, since from a branding perspective, generating customer loyalty is a central aim (McDowell 2005).

Conclusion and Outlook

Traditional journalistic media brands are facing intensifying competition, especially in online media markets (Chan-Olmsted 2006; Doyle 2013; Küng, Picard, and Towse 2008; Schlesinger and Doyle 2015). With the rise of advanced digital technologies, new non-media competitors are entering the online information market, resulting in hyper-competitiveness (Hollifield 2006). This development particularly manifests itself in the competitive environment created by search engines, establishing a so-called competitive information space. This leads to questions about the extent to which traditional media companies can face these new competitive conditions online, build up strong brands, and eventually survive creative destruction in times of digitization.

This study provides the first empirical assessment of the competitive information space on Google by investigating factors influencing user selection preferences. Through the use of a technically innovative approach to data collection involving a web scraping software, empirical evidence on selection preferences based on a real search scenario was generated. Insights into actual search engine user behaviors were gleaned from an experimental online survey.

Nevertheless, this study has several limitations, most notably that an experimental online survey is an artificial situation in which respondents pay more attention to certain tasks than they normally would. In addition, the respondents were forced to make selection decisions, which does not fully represent a realistic setting. Furthermore, only a pull-oriented snapshot of the actual information competition could be examined, focusing merely on the competitive conditions in search engine environments. Moreover, competitive situations on search engines are extremely dynamic, fluctuating, and increasingly personalized, which might lead to the emergence of filter bubbles (Zuiderveen Borgesius et al. 2016). However, recent literature has shown that filter bubbles are not as prevalent on search engines as expected (Puschmann 2019), which reinforces the external validity of this study.

Furthermore, the study is limited to a very specialized area of information, namely information offerings on two technology-related and trending topics in Switzerland. In addition, only a limited brand variety could be integrated into the selection tasks. Future research should address these very limitations. Follow-up studies should focus on different thematic information markets in order to better assess competition between journalistic, corporate, and other offerings, with respect to different user segments (e.g. gender, sociodemographics). In addition, a broader variety of brands should be considered. Brand-specific influencing factors should be further refined in order to create a more comprehensive picture of the extent to which branding activities might lead to competitive advantages for traditional media publishers in liquid media environments.

For news companies, one way to succeed in the hypercompetitive information space is not only to invest in search engine optimization for enhanced visibility (rather than search engine advertising) but also to build a strong brand that is perceived as most credible and reputable, in order to obtain and maintain a strong positioning in media competition online (see also Bruno and Nielsen 2012; Siegert, Gerth, and Rademacher 2011). Only then can an organization hope to position itself successfully and sustainably in the ever-more fluid digital information market, thereby eventually surviving creative destruction.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Appendix

Types of Information Offerings (Sources) of Google Search Results

Table

Type	Description	References
Traditional Journalism	News sites of traditional media companies (e.g. publishing houses, TV stations, radio stations) with primarily journalistic competencies to produce, curate, edit, and distribute news and entertainment content.	Hess (2014)
Online-Only Journalism	News sites of media companies in the online sector with primarily journalistic competencies to produce, curate, edit, and distribute news and entertainment content. Solely online offerings without a print or broadcasting counterpart. These purely digital players also work according to journalistic principles but focus particularly on generating reach.	fög (2018); Hess (2014); Goldapp (2016, 20)
Special-Interest Journalism	News sites of topically specialized media companies with primarily journalistic competencies to produce, curate, edit, and disseminate news and entertainment content. Journalistic offerings (often magazines) with a specific thematic focus—e.g. sports, fashion, technology, etc.	Bogart (2017); Hess (2014)
Corporate Publisher	Websites of commercial companies whose core business is outside the media industry that produce, edit, and disseminate content. Often, they publish informative, multimedia formats such as corporate blogs, with the primary goal of image cultivation and generating customer loyalty for the company.	Bentele (2012); Bentele, Hoepfner, and Liebert (2015); Weichler and Endrös (2010)
Corporate website	Websites of commercial companies with the goal of generating economic profits by selling products or services.	Maki, Evans, and Ghezzi (2015); Yaqub and Ghezzi (2015)
Non-Profit (NPO)	Websites of organizations with expertise and/or non-profit, supporting, or educational services that are not operated for profit-making purposes (e.g. foundations, institutions, universities, etc.).	Chumber, Huber. and Ghezzi (2015); Maki, Evans, and Ghezzi (2015)
Other	Other websites that cannot be classified in the above categories (e.g. forums, portals, social media).	Arif et al. (2018); Aslam, Gibbons, and Ghezzi (2017)

Example of stimulus material

Note. The orange elements were not visible to the users but rather serve to illustrate the stimulus material. The first element is the Google ad, which always appeared at the top for the users. Elements 2 to 7 were shown to the users in randomized order. Users were asked to select the results in their desired order. In this example, the types of information offerings are as ordered follows: (1) Google ad, (2) special-interest journalism, (3) corporate publisher, (4) online only journalism, (5) traditional journalism, (6) non-profit (NPO), (7) corporate website.