upa - home page JUS - Journal of usability studies
An international peer-reviewed journal

Text Advertising Blindness: The New Banner Blindness?

Justin W. Owens, Barbara S. Chaparro, and Evan M. Palmer

Journal of Usability Studies, Volume 6, Issue 3, May 2011, pp. 172 - 197

Article Contents


Introduction

Internet advertising is a primary source of revenue for the majority of websites. In 2009, approximately $22.7 billion in advertising revenues were generated on the Internet (Interactive Advertising Bureau, 2010). Often, Internet advertising pays for web content and services such as videos, blogs, photos, social networking, or news stories, which otherwise would be paid for by web users through per-use or subscription fees (Prasada, Mahajan, & Bronnenberg, 2003). Search and banner advertising accounted for 69% of advertising revenue in 2009 (Interactive Advertising Bureau, 2010). Search advertising was the largest generator (47%) of advertising revenue. Second to search advertising was static banner advertising, which accounted for 22% of revenue. Search advertising remained the largest source of revenue even when animation (rich media advertising) is included (7%). A large component of search advertising is text advertising. Often in search engine results, text advertising is related to the searched topic and typically contains a title, a tag line containing a promotional message from the sponsoring entity and a URL to the entity’s website (AdWords, 2011). In addition to search engines, another common example of such text advertising is the Google advertising network (AdSense, 2011). Design decisions about the appearance of the advertisement are left largely to the producers of the website (see Figure 1).

Figure 1

Figure 1. Google advertising regions located on (A) smashingapps.com and (B) NYTimes.com.

Even though banner ads are a large source of revenue, users exhibit a phenomenon called "banner blindness" in which they ignore page elements that resemble banners while reading a web page (Benway, 1998). This phenomenon occurs with both large and small text, and salient elements not relevant to the task. A more recent and comprehensive examination of banner blindness was conducted by Burke, Hornof, Nilsen, and Gorman (2005). They found that when advertisement banners were present, search performance decreased, although differences in search time and mental workload between static and animated banners occurred only for the most salient banners (e.g., flashing and brightly colored). Interestingly, participants habitually avoided fixating on regions of the web page stimuli where banners were typically located or expected.

The location of the banner advertisements on a web page has been identified as a prominent factor in the degree of banner blindness (Albert, 2002; Benway, 1998; Burke et al., 2005; Granka, Hembrooke, & Gay, 2006; Mosconi, Porta, & Ravarelli, 2008). Albert (2002) found that banner advertisements located below the website branding/search area were viewed longer and by more participants than banners that were placed above the branding/search area. Granka et al. (2006) and Mosconi et al. (2008) both concluded that advertising placed outside the main content area led to decreased viewing of banners. Cooke (2008) found that participants ignored information located on the right side of the web page, which typically consisted of text and image-based links. Through a talk-aloud protocol, participants revealed that they did not look at that area of the page because they expected it to contain information irrelevant to their task, such as advertising. Readers tend to make initial or anticipatory eye movements towards the center of a text page (Rayner, 1988) or the top banner of a web page (Burke et al., 2005), but not towards the right side of a page, which may contribute to the high degree of banner blindness for that region.

The type of search a user performs on a web page also influences the degree of banner blindness. Yesilada, Jay, Stevens, and Harper (2008) asked participants to complete both search and “aimless” browsing tasks. During the search task, users located information to answer a specific question. During the aimless browsing task, users merely familiarized themselves with the web page. The result was that more banners were ignored in the search task than in the aimless browsing task. Similarly, Pagendarm and Schaumburg (2001) instructed users to search web pages in either a goal-directed or aimless fashion and found that the goal-directed group recognized fewer banners, recalled fewer details about banners, and recalled fewer of the advertised companies than the aimless browse group. Riegelsberger, Sasse, and McCarthy (2002) noted that the less directed the search task, the longer its duration, which may provide more opportunities for users to fixate on banner ads while aimlessly browsing.

Rayner (1998) noted that fixation duration can be used as a measure of cognitive load because it varies for different types of tasks such as visual search, scene perception, reading, etc. As cognitive workload increases, it is suggested that fixation duration also increases. Burke et al. (2005) found that users perform tasks with known targets (i.e., exact searches) faster and with fewer fixations than tasks with semantically defined targets (i.e., semantic searches). Their finding that users fixate more during semantic searches coincides with eye tracking differences in efficient and inefficient visual search tasks (Rayner, 1998; Wolfe, 1998). Semantic searches also yielded more diffuse gaze patterns that tended to include banner advertisements while exact searches yielded more focused gaze patterns that tended to ignore banner advertisements (Burke et al., 2005). Thus, the more goal-directed the search task, the higher the degree of banner blindness. Semantically defined targets do not allow for the guidance of visual attention (Wolfe, Cave & Franzel, 1989), leading to inefficient search processes characterized by more fixations and longer response times (Wolfe, Butcher, Lee, & Hyle, 2003). With exact search tasks, participants knew the exact target, which allowed them to guide their attention to the target more efficiently and thus disregard distracter links. In semantic searches, the exact target was unknown so participants had to engage in less efficient visual search behavior. Interestingly, reading tasks generated more fixations and shorter fixation durations than visual search and scene perception activities (Rayner, 1998). Rayner also noted that eye movements and attention are coupled tightly in complex tasks, such as reading, because it is more effective to move the fovea to the focus of attention than to move attention to the parafovea or periphery.

Knowing how users typically search web pages is beneficial for understanding why banner blindness arises. Grier, Kortum, and Miller (2007) proposed the EHS theory, which states that search occurs in three distinct stages: Expected location search, Heuristic search, and Systematic search. In the Expected location search, the user examines the probable location of the information where the expected location is derived from previous experiences. In the event that an expected location does not exist or the user was unable to locate the information in the expected region, the user will engage in a Heuristic search. Users employ several heuristics while scanning a page. First, the user examines the top-left corner of the main content in the web page. If the target is not located there, the user will examine areas of the main content that have medium saliency levels. Because of this heuristic, both high salience elements, such as advertisements, and low salience items that blend well with the page content are ignored. The last heuristic implies that information clusters around other similar information. If the target is not located with the heuristic search, the user engages in a Systematic search of the main content area of the web page and examines each element serially. If the search target is not found in the main content area, the user looks at other areas of the page. The cycle of heuristic and systematic searches repeat for each area until the search target is located, the user gives up, or all elements/areas on the web page have been inspected.

Few studies have examined the efficacy of text advertising and its relationship to the banner blindness phenomenon. Shrestha, Owens, and Chaparro (2009) explored the effect of target location and the congruency of text advertisements on a search engine results page. Results indicated that 79.1% of participants successfully located targets in the text ad region above the search results, while only 40.9% found targets in the text ad region to the right of the search results. Shrestha et al. found increased fixation counts on both search results and advertisements to the right of the search results, but these extra fixations did not necessarily lead to increased success for participants. In fact, participants fixated less often on search results and advertisements during trials where the target was located in the ads above the search results than when it was located in the ads to the right of the search results. However, they were more successful at locating ads above the search results. These findings indicate that target location was the primary factor influencing both fixation counts and task success and that increased fixation counts do not always correlate with lower banner blindness. However, when considering this finding, note that Nielsen (2003, 2007) argued that the concept of advertisement blindness does not apply to search advertisements because when users employ search engines, they are searching for a solution, and if an advertisement may provide a solution to their task, they will click on it.

Fox, Smith, and Chaparro (2009) examined the effects of color schemes and locations on text advertisement efficacy. Analysis of eye-tracking data indicated that ads located above the page content received significantly more fixations, but not increased recognition. Buscher, Dumais, and Cutrell (2010) examined how users interacted with search results during several types of search task, ad/results congruence, and text ad locations. They used fixation impact as their dependent measure, which is highly correlated with fixations as it is derived by the product of fixation count and dwell time. Overall, Buscher et al. found that advertisements located to the right of search results had a lower fixation impact than other areas. This finding did not occur with advertisements located above the search results where the fixation impact may have been modulated by ad congruency. When the search trial used advertisements that were congruent with the search results, fixation impact for ads located above the search results increased at the expense of fixations directed towards the search results themselves. When the ads were incongruent, fewer fixations were allocated to ads located above the search results, but more toward the search results. The fixation impact of ads to the right of search results did not differ across congruent or non-congruent ad conditions. This provides evidence that users were likely aware that the region to the right of the search results were advertisements and were subsequently 'blind' to the ads. All three studies (Fox et al., 2009; Shrestha et al., 2009; Buscher et al., 2010) indicated some degree of text advertising blindness in which users ignore text advertisements.

Summary

Previous literature has indicated that a mixture of factors contribute to advertising blindness, particularly with web banners. In the past, users have been less successful when completing tasks that rely on advertising. Participants recalled fewer details about areas or elements associated with advertising. Advertisement location has been shown to greatly affect the extent of advertising blindness. Participants have exhibited blindness to advertisements placed to the right of the content more than other locations on a web page. Areas that fall outside of the content region or that are associated with advertising, such as the region to the right of the content, received fewer fixations. Fixation order has also been influenced by whether a region is perceived as advertising. The degree of blindness has been affected by characteristics of web elements, where elements resembling banners are actively ignored except for the most salient banners. The type of task users complete, ranging from simple browsing to knowing the exact terms they are supposed to locate, has been shown to affect the extent of blindness as well. Directed tasks may lead to fewer fixations and shorter task durations. Tasks with increased cognitive workload may lead to larger fixation durations. Exploring this issue further, particularly with eye-tracking, is important because text advertising is becoming more prevalent as the highest revenue generating form of advertising currently used on the web.

>Purpose

The purpose of this study was twofold: (a) to establish whether text advertisements are susceptible to user “blindness” and (b) to determine the effects of search type (exact or semantic) and target location on the degree of blindness. The following was hypothesized:

Both standard usability measures and eye-tracking data were examined for each hypothesis.

 

Previous | Next