ABSTRACT
Past research shows that advertising awareness is systematically higher among a brand's users than nonusers. This past research has been confined, however, to measures where a brand name forms part of the cuing material. The current authors' research across six different measures, which extends cues to execution and media prompts, shows the user bias in memory for advertising is not a measurement artifact. It is, in fact, a real phenomenon, occurring under a wide range of conditions. This has implications for creative design, branding, and pretesting, particularly with advertising that primarily aims to attract nonusers. It also has an impact on metrics for assessing global or cross-platform advertising.
MANAGEMENT SLANT
Brand users systematically remember advertising more than nonbrand users do.
This bias is irrespective of whether the brand is present or absent in the advertising awareness question.
To avoid biases from brand size or media audiences confounding awareness scores at the aggregate level, advertisers should examine brand users and nonusers separately.
If a brand's strategy is to grow by attracting new customers, then advertising should be pretested with nonusers to maximize cut-through to this harder-to-reach group.
Advertising-effectiveness measures provide tools to evaluate a brand's advertising. One such measure is advertising awareness, which identifies whether traces of a brand's advertising reside in consumers' memories. These measures capture the impact of advertising and can act as screening questions to identify respondents for subsequent measures, such as likability or message comprehension (Dubow, 1994; McDonald, 2000). Campaigns are lauded, changed, or abandoned based on the results from these intermediary measures.
Advertising-effectiveness measures also provide an early indication of advertising performance, particularly in environments where the relationship between advertising activities and sales is muddied by price promotions and other in-store activities. Correct measurement is crucial for decisions regarding investment in specific executions. In addition to the practical uses, the nature of the responses to advertising-effectiveness measures helps theorists understand how advertising influences consumers. These benefits to theory and practice are only accrued, however, when the measures are accurate and contextualized, so that one may appropriately interpret the results. Contextualization is important for advertising-awareness measures because many factors can impact the score achieved by an advertisement.
When measuring advertising awareness, researchers provide respondents with a retrieval cue, to trigger respondents' memories of a brand's advertising. There are many types of advertising awareness measures, with each type drawing on a particular cue and requiring a specific response (du Plessis, 1994a; McDonald, 2000). An example of a category cue is, “What soft drink brands have you recently seen advertising for?” By contrast, a brand cue asks specifically about advertising for a named brand: “Have you recently seen any advertising for Brand X?” An execution cue may involve showing stills, showing the entire advertisement, or describing it verbally (du Plessis, 1994a). The choice of cue and the difficulty of the response affect the absolute level of advertising awareness; broader cues and easier responses gain higher response levels (Brown, 1988; Romaniuk, Sharp, Paech, and Driesener, 2004; Romaniuk and Wight, 2009).
Users' past experiences with a brand also affect their response level. Brand users more likely will remember seeing their brand's advertising than nonusers of the brand; nonusers are defined as category users who are not current users of the advertised brand (Harrison, 2013; Rice and Bennett, 1998; Romaniuk and Wight, 2009).
This empirical generalization has been observed across multiple categories and countries but only for questioning approaches that use the brand name (Hammer and Riebe, 2006; Harrison, 2013; Romaniuk and Wight, 2009; Sharp, Beal, and Romaniuk, 2001, 2002). That is, the brand formed part of the cuing material in a brand-cued question like, “Have you seen any advertising for Coca-Cola?” Alternately, the brand was the answer to a category-cued question where respondents must retrieve the brand name, such as, “Which soft drinks brands have you seen advertising for?” These brand-centric approaches might enhance retrieval among brand users, who have stronger, more salient brand associations in their memory (Krishnan, 1996; Okechuku, 1992; Zinkhan and Muderrisoglu, 1985).
Other cues, such as showing the creative execution as the response cue, and verifying recognition exposure, are not reliant on brand name as a key advertising memory structure and therefore might not display the same user bias. To the authors' knowledge this has yet to be tested in the literature. Testing whether this empirical generalization holds under different cuing conditions allows researchers to understand if the user bias is a real reflection of differential attention/processing or if it is simply a measurement artifact that can be avoided with a different measurement approach.
If this user bias is evident irrespective of cuing/response material, it would suggest that category buyers have less salient memory structures for advertising promoting brands they do not use. This might help explain why nonusers are difficult to recruit to become brand users via advertising and suggests that “opportunity to see” might not be sufficient to determine the effectiveness of advertising for this group. If, however, this empirical generalization is a measurement artifact—and researchers are able to identify the conditions under which it is, or is not, observed—then measures can be selected or advertising awareness scores can be calibrated accordingly.
This study, therefore, is concerned with testing whether the empirical generalization that “brand users are more likely to remember seeing their brand's advertising than are the brand's nonusers” persists across six different advertising awareness measures that vary by cuing material and response required. It also tests the consistency of the quantitative relationship, which is important for developing robust empirical generalizations (Wind and Sharp, 2009).
LITERATURE REVIEW
The theoretical frameworks of human memory underpin the measure of advertising awareness. In particular, the associative network theories of memory (Anderson and Bower, 1980) help explain the cognitive processes that consumers use when retrieving memories of advertising exposure (Heckler, Keller, Houston, and Avery, 2014; Zinkhan and Muderrisoglu, 1985). Associative network theories represent memory as a network of nodes connected by associative links. Each of these nodes can act as a cue to retrieve linked material from memory (Anderson and Bower, 1980). This structure is the basis for memory processing.
The relevant literature describes two key parts of memory processing: “encoding” and “retrieval” (Tulving and Craik, 2000). Encoding refers to how information gets into a person's memory. Retrieval, which is most relevant to the current research, refers to the ability to access the stored information from long-term memory. The amount and type of existing knowledge that is already established in memory affects encoding ability (Craik and Lockhart, 1972). Stimuli that are familiar and personally relevant will activate deeper levels of processing than will less familiar or less meaningful stimuli (Celsi and Olson, 1988). Existing knowledge and established memory networks also affect how people pay attention to different aspects of information (Wyer, 2008).
Together, these aspects of memory processing suggest that the encoding of new brand information, such as that presented in an advertisement, will be easier for brand users than for nonusers. This is due to brand users having stronger and greater opportunity for linkages between the new brand information and information already in their memory (Okechuku, 1992; Romaniuk, Bogomolova, and Dall'Olmo Riley, 2012).
Once encoded in memory, the information is available for retrieval. The depth of encoding, the retrieval cue(s), and the presence of competitive links all affect the retrieval process (Nelson et al., 1993). Retrieval, therefore, generally is stacked in favor of brand users who would be expected to encode more deeply and to have fewer competitive brand links than would nonusers. Using a cue or response that involves the brand name further enhances a brand user's ability to retrieve the advertisement, which makes the bias under these cuing/response conditions unsurprising. What is unclear is how much residual bias, if any, remains when the advertising retrieval cues do not involve the brand name.
In this research, the authors examined six retrieval approaches commonly used by practitioners to capture advertising awareness:
top-of-mind recall;
unprompted recall;
brand prompted;
brand-plus-media prompted;
execution-prompted;
and execution-plus-media prompted.
The conditions also covered different media formats, types of execution prompts—visual or verbal—and allowed the testing of past findings as well as extension to new conditions.
The current empirical generalization extensively has been observed in the first three advertising awareness approaches listed above: top-of-mind recall; unprompted recall; and brand-prompted recall. The authors included these measures as a close replication and to model the relative strength of the bias for the measures new to this study.
Top-of-mind advertising recall, also known as spontaneous advertising awareness or first mention, records the first brand recalled as having advertised when someone is prompted with only a category cue (McDonald, 2000). This approach was found to have the strongest user bias, with users on average being 2.7 times more likely to recall their brand's advertising than were nonusers (Romaniuk and Wight, 2009). Bias ratios, such as the 2.7 referred to here, are used throughout the paper to provide a comparative figure across past studies and the current results. The bias ratio is calculated by dividing the brand-user retrieval score by the nonuser retrieval score, thus the score is interpreted as users are 2.7 times more likely to recall the brand's advertising than are nonusers.
Unprompted advertising recall also draws on the category as the primary cue for retrieval (McDonald, 2000). This metric widens the criteria to include any response, however, rather than just the first response. Previous studies have found that when using unprompted advertising recall, users were 1.6 to 2.5 times more likely to recall their brand's advertising than were nonusers (Hammer and Riebe, 2006; Romaniuk and Wight, 2009; Sharp et al., 2001, 2002).
Brand-prompted advertising awareness draws on both the category and the brand as cuing material (McDonald, 2000). This is a cognitively easier retrieval method than top-of-mind and unprompted recall methods, as the addition of the brand cue provides stronger links with stored advertising associations in the consumers' memory (McDonald, 2000; Romaniuk and Wight, 2009). Across two studies, the user bias ratio was found to be 1.7 and 1.3, respectively, which is lower than top-of-mind and unprompted recall measures (Hammer and Riebe, 2006; Romaniuk and Wight, 2009).
A subset of articles draws on linear regression modelling to quantify the relationship between user and nonuser awareness scores. These results suggest a strong relationship between user and non-user advertising awareness, with Romaniuk and Wight (2009) reporting an adjusted R2 of 0.93 (p < 0.001), whereas Sharp et al. (2001) reported an adjusted R2 of 1.83 (p < 0.001). As a consequence, a strong relationship between user and nonuser memory for advertising is generally expected.
The three approaches described above access respondents' memory network through an indirect link between the brand or category and the advertising, rather than directly through an advertising cue such as showing the execution. The next three advertising awareness measures provide more direct links to advertising memories.
Brand-plus-media-prompted advertising awareness is where respondents are provided with the category, brand, and media used to deliver the advertisement—such as television, radio, magazines, and so forth—as cues to test memory for advertising. A previous study measured the user bias for this cuing approach across 160 different brand contact touch-points, including paid, such as television advertising; owned, like the brand website; and earned touch-points, such as friends and family recommendations (Harrison, 2013). Paid touch-points, which are the most directly relevant to the current authors' research, had the lowest user bias, ranging from 1.3 to 1.7, with one exception of 2.1 for advertising observed on mobile phones (Harrison, 2013).
Execution-prompted advertising awareness uses the specific creative execution as the cue to trigger respondent memory (McDonald, 2000). This is recognition rather than a recall task, and therefore it gains higher response levels because of the ease of the matching task (du Plessis, 1994a; Stapel, 1998). This measure is not dependent on the brand or the category as an entry point into the consumer's memory; rather the respondent may see or hear a description of an advertising execution without the brand present (du Plessis, 1994a). The provision of creative-based cuing information should make it easier for nonusers to remember the exposure, by making accessible memories of the advertising that might be inaccessible under brand-centric cuing approaches (Tulving and Pearlstone, 1966). To the current authors' knowledge, the only evidence of user bias effects for this awareness approach was a bias ratio of users being 1.2 times more likely to remember exposure than nonusers (Hammer and Riebe, 2006).
Execution-plus-media-prompted advertising awareness, the final measure tested, is where respondents are asked whether they remember seeing a particular execution, but with the additional cue of a specific media to guide retrieval, such as, “Have you seen the following advertising on TV?” The addition of media should narrow the memory search criteria and lead to a more accurate reflection of exposure, thereby further decreasing the user bias. To the current authors' knowledge, however, no previous studies have tested this approach.
Thus the following research questions were proposed:
RQ1: Do all six types of advertising awareness measures display a user bias in responses?
RQ2: Is the user bias lower or higher for any specific type of cuing material?
RQ3: Is the user bias lower when the brand is not part of the cue or response material?
METHODOLOGY
This study took an empirical generalization approach as advocated by Wind and Sharp (2009). Data from multiple executions for different brands were examined in different categories and countries. Close replications of prior studies were first conducted using category and category-plus-brand cue approaches. This was followed by differentiated replications drawing on the different cuing approaches of brand-plus-media-prompted, execution-prompted, and execution-plus-media-prompted advertising awareness. The relationships were modeled across different types of cuing approaches to test for generalizability and for boundary conditions (as per Lindsay and Ehrenberg, 1993).
Description of Datasets
The analysis involved 26 different datasets (See Table 1). The datasets were collected as part of industry studies—typically continuous brand and advertising trackers—for the purpose of evaluating advertising. They were also subject to the same screening and quality criteria used in practice. The sampling frames included category users only, with samples collected via a variety of methods including face-to-face, telephone, and online panels. The datasets comprised a wide range of conditions with 101 executions, 88 brands, 18 categories, 10 countries (Australia, China, India, Portugal, Russia, South Africa, Spain, Taiwan, Turkey, and United Kingdom) and five media formats (television, print, radio, outdoor, and online). The use of such a diverse set of data, collected under different conditions, contributes to the robustness of the research as it suggests that the findings are not due to any specific study circumstances. Further, as the data was drawn from real-world surveys, it provides confidence for advertising practitioners that the findings of this study will be highly likely to be replicated in their own survey data.
Depending on the scope of the data, multiple advertising campaigns were tested for some brands. When multiple tests of similar executions from the same campaign were present, the sample was narrowed to only one execution to avoid duplicate responses. Respondents were asked a range of questions for each of the six advertising awareness measures (See Table 2). Across the 26 datasets, 247 brand-level observations were measured across the six different conditions.
Classifying Brand Usage
In each dataset, respondents were first screened to ensure that they were users of the category; they then later answered questions relating to their use of brands within a specific time period. These time periods varied depending on the category and were designed to reflect the broader concept of being a current customer. This meant shorter time frames for frequently bought categories and longer time frames for less frequently bought categories. For soft drinks, for example, a time frame of “bought or consumed in the past 3 weeks” was used, whereas “currently holds products with the brand” was used for financial services. The authors acknowledge that for these repertoire categories light users may be categorized under the nonuser group and thus the distinction is between heavy versus light or nonusers, as opposed to clearly defined user and nonuser groups. This approach is consistent with past research in the field.
The primary analysis involved cross-tabulations between the brand usage and advertising awareness measures, with Pearson's chi-square statistics used to determine the statistical significance of the result. Statistical testing is considered less important for replication and extension studies when multiple sets of data are investigated (Ehrenberg, 1990; Lindsay and Ehrenberg, 1993). Accordingly, the directions of those results not quite passing conventional statistically significant thresholds were also examined.
RESULTS
The current study's results show that across each of the six different measures tested brand users typically more likely will remember advertising for their brand than are nonusers. In 177 of the 247 observations, or 72 percent, these differences were statistically significant at p < 0.05; a further 21 percent trended in the same direction, with 51 such observations. This provides a total of 92 percent with statistically significant or trending results in favor of users being more likely to remember their brand's advertising exposure than nonusers. This result is consistent with past literature and demonstrates that the empirical generalization holds under this broader range of cues.
Summary of Awareness Measures and Examples of Questioning Techniques
This result addresses RQ1; the user bias is evident across all six types of advertising-awareness measures.
Drawing on the median, to minimize outlier influence, the findings also show that the average user-response level was 18 percentage points higher than the average nonuser–response level (See Table 3). The difference in median scores between users and nonusers was highest for brand-prompted advertising awareness at 21 percentage points and lowest for brand-plus-media-prompted advertising awareness at 8 percentage points. This variation is explored further in RQ2 and RQ3.
Effect of Cuing Material On the Level of Bias
Having determined that there are significant differences between user- and nonuser–awareness scores, the authors used multivariate regression modeling to determine whether the different types of cues included in the awareness measure explained the variance in the scores. To explore the reasons for the variance, the measures were decomposed using three key characteristics that could affect the user bias levels. These characteristics were
Brand forms part of the cuing material
Having the brand present should lead to an increased bias toward brand users as it provides a direct link to existing brand knowledge in the consumer's memory.
Advertising execution forms part of the cuing material
Having the execution shown visually with images or described verbally should reduce the user bias, as the execution provides direct links to existing knowledge of the advertising. The reduction in the user bias should be even more evident for visual execution cues, which are richer in stimuli than verbal descriptions.
The media type of the advertising
Having a media specified should reduce the user bias as it provides narrower search criteria in memory. Advertising for visually rich media, such as television, should lead to an even lower user bias than for less visually rich media, such as radio.
Summary of Median Results Across Awareness Measures (Ordered by Bias Ratio)
Nonuser awareness is the dependent variable, and the conditions of brand present, visual or verbal execution prompts, and different media types were added as dummy covariates. The overall adjusted R2 = 79%, F = 105.4, p < 0.001. The variance explained is quantitatively in line with past studies that focus on a narrower range of advertising awareness measures (Romaniuk and Wight, 2009; Sharp et al., 2001) and shows user awareness level is the strongest predictor of nonuser awareness (β = 0.78, p < 0.01).
Also significant are the two execution types, verbal and visual (Verbal β = 0.21, p < 0.01; Visual β = 0.12, p < 0.05). This finding suggests that adding an execution cue to the retrieval approach reduces the difference between user and nonuser response levels, as the user awareness score plus execution prompt predicts more accurately the nonuser awareness score. Although it appears that a verbal execution description is a stronger predictor, the overlapping confidence intervals with visual execution means that it is impossible to draw this conclusion. There was no significant contribution to explaining nonuser awareness scores from including either media type or brand name in the cue. Further, the confidence interval for these variables crosses zero, indicating that in some observations the predictor has a negative relationship, whereas in others it has a positive relationship. In addressing RQ3, this result shows that using the brand name or media type as a cue may affect the overall response level but does not lead to nonusers being more or less likely to remember advertising.
DISCUSSION
The current authors' research further develops understanding of how past experiences with a brand influences attention to advertising and processing it into memory. This study finds a systematic pattern: Among current users of the category, a brand's users more likely will remember seeing that brand's advertising than are nonusers of the brand. This finding, tested across 247 brand-level observations, holds across six different advertising awareness approaches, under conditions of there being a brand or no brand present in the cuing or response materials, two different types of execution prompting, and five different media.
The results show that, irrespective of whether the brand is present or absent in the advertising awareness question, brand users systematically remember advertising for that brand more than nonbrand users. This removes any prior concerns about the bias being merely a methodological artifact that potentially could be eliminated through changing the advertising awareness questioning approach. Rather, these results show that this brand user bias of heightened memory for a brand's advertising is a universal aspect of advertising response patterns.
The results also provide evidence that the reported differences between advertising recall measures largely are due to scaling effects. The higher bias ratio for top-of-mind advertising awareness is, for example, a reflection of the low response levels for both brand users and nonusers alike. It is the lower base of nonusers that inflates the bias when expressed as a ratio.
Overall Regression Coefficients Results for the Estimation Of Nonuser Awareness Levels
This is also why the regression results across all six measures are important, as they show the bias is not abnormally high as user and nonuser awareness scores are correlated. The measures themselves therefore do not enhance or mitigate this bias; conversely, each measure simply reflects a bias that is inherently present in buyer memory to a different degree depending on the measure's difficulty. This finding also supports the conclusion of a 2009 study that nonusers need additional prompting or cuing material to remember exposure (Romaniuk and Wight, 2009). Adding an execution prompt, either verbal or visual, reduces the user bias by making it easier for everyone, but particularly for nonusers, to remember the exposure. This demonstrates that more nonusers are able to remember the advertising when cuing material specific to the advertising is provided, rather than just the category or brand. This finding also supports the conclusion that the user bias is a real phenomenon and not a measurement artifact.
The authors also found that including the media type or brand name cue in the awareness measure does not contribute to the regression model's ability to predict nonuser awareness scores, further to that explained by user awareness scores and the presence of a visual or verbal execution prompt. This may be due to the execution prompt implicitly revealing the media type by its inherent qualities. A video cue typically would suggest television advertising, compared with the starkly different look and level of detail a still image for print and outdoor advertising would imply, thus providing a media-type cue may not provide the respondent with any further memory prompts. For the brand-name cue, the nonsignificant result in the model may be due to the indirect link to a brand's advertising memories through the brand name, compared with the more direct link an advertising specific cue such as the execution would provide.
The fact that the user bias was still evident when using execution prompts, which are considered a recognition task, suggests that exposure to advertising does have less of an impact on nonusers (in line with Celsi and Olsen, 1988). This implies that opportunity to see might not be sufficient to determine effectiveness of advertising, as nonusers less likely will pay attention to or process a brand's advertising. All viewers with an opportunity to see therefore do not have equal processing, and thus response, capability.
The study's results also add to the literature highlighting the importance of separating user and nonuser responses in analysis to avoid misleading results and conclusions (such as in Romaniuk et al., 2012).
Furthermore, the findings shed some light on why it is difficult for advertising to expand category sales through recruiting new category buyers. Nonusers of a brand have a low propensity to remember advertising for brands that they do not use. It is then possible to extrapolate that someone for whom both the category and the brand have low relevance might be even less likely to remember advertising exposure, thereby diminishing its impact.
Implications for Practitioners
This study has important implications for research professionals and practitioners. For marketers assessing the effectiveness of a brand's advertising, the findings suggest the choice of advertising awareness measure matters, as more difficult measures will dampen responses from users and nonusers, but as a ratio, nonusers are particularly penalized. The authors recommend, therefore, an approach that is easy for all respondents to retrieve, such as an execution-based recognition task. Regardless of the measure used, practitioners should always be encouraged to separate out responses for brand users from nonusers, to account for the usage bias that is ubiquitous in all advertising awareness measures. This will avoid understating exposure levels and overstating the impact of advertising on those exposed, which will be composed of a larger proportion of brand users for more difficult measures.
Separating brand users from nonusers also will improve advertising-effectiveness studies in which advertising awareness measures act as screening questions for subsequent measures, such as advertising likability or message comprehension. In such approaches, only respondents who remember the brand's advertising are asked further questions, so these findings suggest that execution-prompted approaches will be most effective in capturing advertising exposure across both brand users and nonusers and therefore provide a full assessment of advertising performance.
This systematic difference between brand users and nonusers has important implications when marketers are assessing a global campaign, in which the brand varies substantively in market share across countries. All advertising awareness measures are shown to be biased to users, and therefore aggregate-level metrics may inaccurately imply a campaign is less successful in countries where market shares are lower, since the user bases are much smaller. This could lead marketers to make unnecessary modifications to campaigns to compensate for perceived lower advertising awareness.
An advertiser's interest in reaching nonusers or brand users depends on the chosen strategy for growth. If a brand's strategy is to grow by attracting new customers, as advocated by the vast majority of empirical studies (Anschuetz, 2002; Baldinger, Blair, and Echambadi, 2002; Romaniuk, Dawes, and Nenycz-Thiel, 2014), the current results suggest that the brand's advertising should be developed with nonbrand users, but who are category users, in mind. The advertising should be pretested on this segment to optimize message take-out, cut-through, and advertisement likability, so that it more likely will gain attention and break through clutter (Biel and Bridgwater, 1990; du Plessis, 1994b; Walker and Dubitsky, 1994).
Another approach is to draw on distinctive assets, which are any nonbrand name elements that trigger the brand in the minds of category buyers, to introduce the brand (Romaniuk and Nenycz-Thiel, 2014). This is to avoid the problem of nonbuyers switching off due to lack of personal relevance. A future area of creative research should examine the types of brand advertising that have higher or lower relevance to nonbrand users. It might be that a different creative style or message is needed to cut through to those who do not use the brand at all, or very often, compared with that needed to appeal to more regular brand users.
The user bias effect was consistent across media types. This finding—which showed the robustness of the generalization—also suggests that advertisers should understand the difference in the audience bases of specific media and the degree to which they proportionally attract brand users and nonusers. Social media audiences, for example, skew more toward heavier brand users (Nelson-Field, Riebe, and Sharp, 2012; Romaniuk, Beal, and Uncles, 2013). If comparing the advertising awareness of a social media campaign with one on television, television might be unfairly disadvantaged merely because it reaches a higher proportion of nonusers, thereby hampering its cut-through scores.
Only by comparing brand users and nonusers separately can an advertiser determine whether this is because of the different effectiveness of the advertising itself, or the different composition of the advertising's audience. This is a necessary task to avoid incorrectly attributing the lower results to the effectiveness of the creative. These implications for practice flow into media planning and highlight the importance of ensuring that media schedules reach nonuser groups, and media vehicles are not skewed to existing or heavier brand users.
LIMITATIONS AND FUTURE RESEARCH
The main limitation of this research relates to the use of secondary data. Although the use of secondary data provides a wealth of datasets, minimizes confirmation bias, and enables results to be generalized across varying conditions, the work is constrained to the design and data-collection techniques of the previous studies. A second limitation is that the choice of brand-usage variable was limited to those available within the existing datasets.
Finally, whether or not the respondents had the opportunity to see the advertisement cannot be verified, unless done so in primary collection. This likely will be an important consideration for more targeted media, where the reach of nonusers might be systematically lower, such as for Facebook fan pages as described by Nelson-Field et al. (2012). Although these may be perceived as limitations, these conditions exist in real-world studies, and the consistency of the current study's findings in spite of these limitations highlights the robustness of the empirical generalization. For the findings to continue to be beneficial for marketing science and to see how far the generalization extends, further testing in a wider range of countries and media should be conducted.
The finding that the user bias phenomena are present in all advertising awareness measures opens up questions of why this occurs and whether it can be addressed by creative elements:
Is it that brand users pay more attention to advertising? if so, what type of attention?
Is it that the base of knowledge that brand users have that makes the memory of new exposures processed more deeply and therefore easier to retrieve?
Another avenue to explore is the role of the prominence of the brand. For example, if the brand is shown earlier in a television/video advertisement, do nonusers switch off their attention and therefore process the rest of the advertising less? In that case there may be differences in brand user–nonuser biases based on the quality of branding in the execution.
Empirical research into brand growth shows that attracting new users is vital for packaged goods brands to grow (Anschuetz, 2002; Baldinger et al., 2002; Romaniuk et al., 2014).
Therefore, a useful stream of research would be to determine whether it is possible to design advertising that is equally effective among brand users and nonusers. This may be a difficult task given that for nonusers, advertising needs to prompt trial, whereas for users, it should aim to prompt repeat purchase. Regardless, some level of processing is necessary to prompt either of these responses, therefore it is important to conduct research to determine whether advertising creative factors can generate equal awareness scores among brand users and nonusers. Further studies could delve beyond advertising awareness into additional measures of effectiveness, such as likability or message comprehension (Dubow, 1994; McDonald, 2000). It also would be useful for researchers to explore whether there are systematic differences between users and nonusers scores on these more complex measures. These are all potentially valuable areas of research that will help advertisers get more return on investment from their advertising expenditure.
ABOUT THE AUTHORS
Kelly Vaughan is a research associate at the Ehrenberg-Bass Institute. Her research focuses on measuring advertising effectiveness and consumer-based brand equity, in particular, mental availability.
Virginia Beal is senior researcher at the Ehrenberg-Bass Institute. Having held senior research roles at the BBC, TIME Inc, News International, and Network Ten, Beal's research focuses around advertising effectiveness and media scheduling.
Jenni Romaniuk is research professor and associate director (international) at the Ehrenberg-Bass Institute. Her specialist areas of expertise are brand strategy and metrics, distinctive assets, mental availability, advertising effectiveness, and word of mouth. Romaniuk is also author of How Brands Grow Part 2, available at Oxford University Press, and she is co-executive editor of the Journal of Advertising Research.
- © Copyright 2016 The ARF. All rights reserved.
REFERENCES
Vol 56 Issue 3
