This study explores the effect of ad variation effectiveness as a function of cultural differences, drawing from visual attention and encoding variability theory. Findings on HK and UK participants suggest that varying ad features may be most effective by changing those features that correspond to the intended consumer’s cognitive styles. Introduction Previous research suggests that showing varied ads can enhance brand recall, and result in more positive attitudes towards the brand (Unnava & Burnkrant, 1991). Following the encoding variability theory (Lee & Lee, 2016; Yaveroglu & Donthu, 2008), variations in encoded information improve recall because these variations provide additional memory representations for the target information, hence leading to more retrieval cues for the information to be recalled (Anderson & Bower, 1973). Transposed to an advertising context, each element of an ad (e.g., graphics, layout, backgrounds, products, brand logo, etc.) can serve as a retrieval cue for information recall. For instance, if a target product is embedded in two different backgrounds (e.g., a beer in a social event vs beach background), these two different contexts can serve as two different retrieval cues for the target product, compared to only one possible retrieval cue provided in same ad repetition. Since encoding variability theory relies on contextual changes and multiple memory paths to explain the ad variation effect, it may be affected by cultural difference (specifically, selective attention), which can act as a moderator of ad variation effectiveness. Western and East-Asian people have been found to have different visual attentional biases (e.g., Nisbett & Masuda, 2003), with Westerners being more attentive on focal objects (analytic perception) and Easterners focusing more on the whole picture (holistic perception). As such, the selective attention they paid towards the ad may enable them more susceptible to specific changes of an ad, thus increasing the available memory pathway towards brand information, i.e., brand recall. Brand attitudes are suggested to be formed through learning (e.g., Van Osselaer & Alba, 2000), i.e., consumers learn and process the information conveyed in an ad, which eventually affects their attitude towards a brand. This study therefore also tested the role of visual attention in moderating the effect of ad variation on brand attitude, since Eastern and Western customers learn differently due to their visual attentional styles. To examine the possible moderation of ad-variation effectiveness by visual attention in a cross-cultural context, this study exposed HK (Eastern) and UK (Western) participants with two different varied-ad executions, foreground-varied ads (ads that vary in focal object) and background-varied ads (ads that change in background features). Identical-ad execution was also included as a baseline. We hypothesized that, due to the difference in attentional bias in Eastern participants (holistic) and Western participants (analytic), different types of ad variations might results in different extents of effectiveness of the ad variation. Method A 2 (Culture: UK vs. HK) x3 (Ad type: identical vs. background-varied vs. foregroundvaried) between-subject factorial design experiment was used in this study. We used a fictitious Beer brand - Helga Brugge - for this study. In the identical-ad condition, the same target ad was repeated three times. We designed a target beer and two beer pints as foreground features and background features included a camping scene (in identical-ad condition), and a football stadium, camping scene and a social event scene (in background-varied condition). In foreground-varied condition, while keeping the background image constant (camping background), the foreground features differed, i.e., a bottle beer, a canned beer, and a bottle beer with two beer pins. See Figure 1 for the target ad stimuli. Participants were randomly assigned to one of the three ad type conditions and watched the 20 advertisement slides (3 target ads with 17 filler ads), each of which was presented singularly for 3 seconds. Filler ads were chosen from real ads that were not used anymore at the time of the study. They included various product categories other than beer such as toy, beverage, and cosmetics. After being exposed to the advertisement, participants were asked to answer a series of questions measuring a number of variables in the following order: ad recall and brand recall (Lee & Lee, 2016), brand attitude (MacKenzie, Lutz & Belch, 1986) and demographic information. Data was collecting online using Qualtrics. Both Chinese and English versions of instructions were back translated (Miracle and Bang, 2002). Results After excluding incomplete datasets and participants whose nationality was not British or Hong Kong, the final sample comprised 117 UK (78 female) and 108 HK (63 female) participants. Cross-cultural Cognitive Differences in Perception To confirm the underlying assumption that HK participants tend to have holistic visual perception and UK participants tend to have analytical visual perception, participants’ responses were coded to identify the type of foreground information (e.g., beers, pints, foams) and background information (e.g., sunset, camping). The coding processes were independently performed by two coders; Cohen’s κ analyses showed a moderate agreement (Sim & Wright, 2005) between the two coders' judgments on both foreground information (κ = .781, p < .001, 95% CI, .716 to .846) and background information (κ = .775, p < .001, 95% CI, .706 to .843). Two 2(Culture: UK vs. HK) x 3(Ad type: identical, background-varied and foregroundvaried) analysis of variance (ANOVA) were conducted. Results support our assumption: UK participants elicited more comments on foreground information (M= 1.56, SD= 1.20) than HK participants did (M= 1.05, SD= 1.03; F(1, 219)= 11.80, p= .001, =.051), and HK participants elicited more comments on background information (M= 1.69, SD= 1.19) than UK participants did (M= 1.09, SD= 1.03; F(1, 219)=16.22, p<.001, =.069). There was no statistically significant main effect on ad type or any statistically significant interaction effect. Brand Recall For UK participants, we hypothesized that their selective attention towards focal objects would enable them more susceptible to changes in the foreground-varied ad conditions, but not in the background-varied ad conditions. Chi-squared analyses revealed that UK participants’ brand recall scores in foreground-varied condition (M=89.2%) was significantly higher than those in both background-varied (M=53.7%): X^2 (1, N=78) = 11.80, p<.001, ϕ =.39, and identical conditions (M=43.6%): X^2 (1, N=76) = 17.5, p<.001, ϕ =.48. There was no difference in brand recall between identical and background-varied conditions. Conversely, we hypothesized that both background- and foreground-varied ads would be equally effective for HK participants due to their holistic attention towards both foreground and background objects. Multiple chi-squared tests revealed that, HK participants’ brand recall scores in both foreground-varied condition (M=59.5.%) and background-varied condition (M=58.3%) were significantly higher than that in the identical condition (M=34.3.%; foreground-varied vs. identical,X^2 (1, N=72) = 4.57, p=.032, ϕ =.25; background-varied vs. identical, X^2 (1, N=71) = 4.13, p= .042, ϕ =.24). The difference between the two varied ad conditions was not significant. Brand Attitudes We hypothesized that HK and UK participants’ brand attitudes also differ to different extents in the three conditions, following a similar pattern found in measuring brand recall. Consistent with this, independent-sample t-tests on UK participants revealed that the mean scores for brand attitude in both foreground-varied ads condition (M= 4.64) was higher than that in both identical ads condition (M= 3.71; t=-4.09, p<.001, d=0.94) and background-varied ads condition (M=3.98; t=3.49, p=.001, d=0.80). No significant difference was found between the identical-ad condition and background-varied condition. For HK participants, results of t-tests revealed that the mean scores for brand attitude in both foreground-varied condition (M= 4.18) and background-varied condition (M=4.33) were both significantly higher than that in identical ads condition (M= 3.53; foreground vs identical: t=-3.19, p=.002, d=0.51; background vs identical: t=4.87, p<.001, d=0.80). No significant difference was found between the two varied ads conditions. Discussion This research provides supporting evidence that visual attentional biases initiated by cultural differences can moderate the effectiveness of ad variation. Specifically, for UK participants, ad variation appeared to be effective only in foreground-varied ads. Conversely, both foreground-varied and background-varied ads were effective for HK participants. These findings have both theoretical and managerial implications. To the best of our knowledge, it is the first cross-cultural research in the domain of ad variation. It addresses a gap in the ad variation literature, by identifying the moderating effect that cultural differences can have on the ad variation effect. This opens up new research directions including considering other forms of cultural variations (e.g., language) and cognitive differences (e.g., reasoning styles) to better understand individual differences in the domain of ad-variation. This study also offers insights for international marketers looking at tailoring their advertising strategies for different target audiences to maximize ad- and cost-effectiveness. Besides varying features that correspond to the intended consumer’s cognitive styles, marketers could also consider priming consumers’ cognitive styles when determining advertising strategies, as previous research has shown that consumers’ cognitive styles are relatively malleabile (Lin & Han, 2009). For example, when executing foreground-varied ads, inserting them into an article that could induce an analytic cognition (e.g., bibliography programs about a successful life story of a person) would be an effective strategy. This study has two main limitations. First, it only included data from nationals of the United Kingdom and Hong Kong. Non-cognitive cultural differences such as language and geographical mobility can affect cognitive styles (e.g., Rhode & Voyer, 2015). Future research should replicate and expand findings by looking at more countries (e.g., US, South Korea), with increased level of control on relevant non-cognitive crosscultural factors. Second, this study only looked at one hedonic product category: beer. Future research can replicate and expand findings, by looking at different product categories (e.g., utilitarian products, such as toothbrushes).