In the purchase process consumers often accept recommendations from others to make effective decisions with the least effort. There are two main recommendation sources: (a) human users and (b) artificial intelligence (AI) supported by data and algorithms. With the rapid development and application of AI, the form and content of advertisements can be tailored to individual users. Given this opportunity, providers of newsfeed advertising are increasingly adopting AI-driven personalized advertisements (Grewal et al., 2016). In particular, advertisements on newsfeeds have become an important form of social media advertising. Notably, as an integral interface component of newsfeed advertising, AI recommendation tags play a key role in visual presentation, interactive communication, and attention capture. However, prior research has largely focused on improving the efficiency of AI recommendations, with limited attention paid specifically to the design of recommendation tags.

 

AI recommendation tags are a crucial component of AI recommendation systems. Unlike physical tags, digital tags not only convey information through images, text, and colors, but also flexibly adjust parameters according to context, facilitating instant searching and sharing (Pentland et al., 1996). Recommendation tags can be categorized into explicit and implicit types based on the degree of information disclosure (Tran et al., 2022). Explicit recommendation tags directly reveal the basis of recommendation, such as “Eight friends have seen” or “From the fashion channel you follow.” By contrast, implicit recommendation tags suggest recommendation without clearly disclosing their basis, such as “Guess you like.” However, limited research attention has been paid to AI recommendation tags from a consumer perspective, particularly how different types of tags influence consumers’ perceived personalization, and how this process is shaped by recommendation fit and privacy concerns. This gap is particularly important because consumers often value the convenience of personalized recommendations while also being concerned about how their personal information is used.

 

We conducted three scenario-based experiments grounded in self-congruity theory to examine how explicit versus implicit recommendation tags in AI-powered newsfeed advertisements influence consumers’ advertising attitude. Self-congruity theory (Sirgy, 1982) suggests that consumers respond more positively when they perceive external cues as congruent with their needs and preferences. In the context of AI recommendation tags, compared to tags that are implicit, explicit tags make the basis of the recommendation more visible and may therefore strengthen consumers’ perception that the product being advertised matches their interests. Such perceived congruence with both the product being promoted and the advertising content is manifested in perceived personalization and is expected to enhance consumers’ attitude toward the advertisement (De Keyzer et al., 2015). Recommendation fit and privacy concerns further shape this process by influencing whether the consumer perceives such congruence as relevant or as intrusive (Awad & Krishnan, 2006). In our three experiments we examined the effects, mechanisms, and boundary conditions of AI recommendation tags from a consumer perspective, thereby providing a consumer-centered framework for understanding recommendation tags and their implications for perceived personalization and privacy. The theoretical model that is the basis of our three experiments is depicted in Figure 1.

AI recommendation tags serve as cues that explain why an advertisement is being shown to a consumer. In online settings, such cues reduce uncertainty and shape the viewer’s evaluation of advertising content (Chen & Dubinsky, 2003). Explicit tags directly disclose the basis of recommendation, whereas implicit tags rely on algorithmic inference and are less transparent (Batmaz et al., 2019). Because explicit tags make recommendation logic more visible and easier to understand, they are more likely to generate a favorable advertising attitude than are implicit tags or advertisements without a tag.

Hypothesis 1: Compared with advertisements with an implicit artificial intelligence recommendation tag and those with no tag, advertisements with an explicit artificial intelligence recommendation tag will lead to a more positive consumer advertising attitude.

Perceived personalization refers to the extent to which consumers perceive an advertisement as tailored to their needs and preferences (De Keyzer et al., 2015). Prior research has shown that compared to nonpersonalized messages, messages that are personalized are more likely to be remembered, preferred, and responded to positively (Noar et al., 2007). Because explicit tags provide clearer reasons for recommendation than implicit tags do, they may strengthen consumers’ perceived personalization, which will, in turn, enhance their advertising attitude.

Hypothesis 2: Perceived personalization will mediate the effect of artificial intelligence recommendation tag type on consumers’ advertising attitude.

Recommendation fit refers to the degree of match between the content of the recommendation and the consumer’s actual needs. Although recommendation cues can enhance perceived personalization, their effect depends on whether the consumer considers that the recommendation in the advertising is relevant (Deng et al., 2024; Xu & Chen, 2025). When consumers perceive that the fit is good, both explicit and implicit tags may appear appropriate; however, when the fit is perceived as poor, the greater transparency of explicit tags may help consumers understand more easily why they are being shown the advertisement, leading to a more favorable evaluation than they have for implicit tags.

Hypothesis 3: The fit of an artificial intelligence recommendation tag will moderate the relationship between the tag type and consumers’ perceived personalization, such that the indirect effect of tag type on consumers’ advertising attitude through perceived personalization will vary across levels of fit of the recommendation.

Hypothesis 3a: When the recommendation fit is good, the effect of artificial intelligence implicit and explicit recommendation tags on consumers’ perceived personalization will not differ significantly.

Hypothesis 3b: When the recommendation fit is poor, compared with an implicit artificial intelligence recommendation tag, an explicit artificial intelligence recommendation tag will lead to higher consumer perceived personalization.

Privacy concerns refer to individuals’ worries about the collection and use of personal information by service providers (Awad & Krishnan, 2006; Hong & Thong, 2013). Although personalization of advertising can improve relevance, it may also trigger discomfort and perceived risk when consumers become sensitive to privacy issues (J. L. Hayes et al., 2021; McKee et al., 2024). Accordingly, the positive effect of perceived personalization on advertising attitude could weaken or even become negative when privacy concerns are high.

Hypothesis 4: Privacy concerns will moderate the relationship between consumers’ perceived personalization and advertising attitude, such that the indirect effect of artificial intelligence recommendation tag type on advertising attitude through perceived personalization will become weaker when privacy concerns are high.

Ethical approval for the research was obtained from the Social Science Ethics Committee of Jinan University before data collection for the three studies commenced (approval number SSE184-2024).

In Study 1 we tested the main effect of AI recommendation tag type on consumers’ advertising attitude and examined the mediating role of perceived personalization.

We conducted a two-stage pretest to select the recommendation tags used in the main experiment. In the first stage, we introduced 63 Master of Business Administration students in the School of Management at Jinan University (23 men and 40 women; M_age = 23.2 years, SD = 5.12, range = 19–38) to the context of newsfeed advertising and provided brief definitions of explicit and implicit recommendation cues based on prior research (Jawaheer et al., 2010). We defined explicit cues as recommendation tags that directly reveal the basis of recommendation, whereas implicit cues are tags that present a recommendation without clearly disclosing its basis. We then showed the students six recommendation tags in random order as possible explanations for why a user might see a recommended advertisement in a newsfeed: “Guess you like,” “Recommended for you,” “Selected for you,” “Followed channels,” “Current events” (trending or topical content), and “15 friends have seen.” We asked the respondents to select all the tags that they perceived as explicit recommendation cues. The results are shown in Table 1.

On the basis of the first-stage results, we conducted a second-stage confirmation test with adult participants recruited through the Wenjuanxing online survey platform. The surface wording of the selected tags was slightly adapted across the confirmation test and in the main experiments to better fit platform-style recommendation language and the specific advertising context, but the underlying recommendation bases were kept consistent. Responses in this second-stage confirmation test were considered valid if participants completed the questionnaire, provided a clear explicit/implicit recommendation classification judgment for the assigned tag, and did not submit duplicate or obviously invalid responses. We received valid responses from 189 participants (75 men, 114 women), who were evenly assigned to one of three AI recommendation tag conditions: “From the channel you follow,” “Your 15 friends may have viewed this,” and “Guess you like.” After we had informed participants of the definitions of explicit and implicit recommendation, we asked them to judge whether the tag represented an explicit or implicit recommendation. The result of a chi-square test of independence showed a significant association between tag type and recommendation judgment, χ²(2) = 45.04, p < .001. Results set out in Table 2 show that “Your 15 friends may have viewed this” and “From the channel you follow” were more frequently classified as an explicit recommendation, while “Guess you like” was more frequently classified as an implicit recommendation. These results suggest that participants were generally able to distinguish between explicit and implicit AI recommendation tag types.

We recruited 120 Master of Business Administration students from the School of Management at Jinan University to participate in the experiment in exchange for course credit. After excluding invalid responses, we retained 115 valid survey forms from 67 men and 48 women (M_age = 25.13 years, SD = 3.59, range = 23–36). A post hoc power analysis conducted with G*Power 3.1 (Faul et al., 2009) indicated that, with three groups, an effect size of f = 0.40 and α = .05, the obtained sample size provided sufficient statistical power (1 − β = .97), exceeding the conventional threshold of .80. Therefore, the final sample size was adequate for hypothesis testing.

 

We employed a one-factor, three-condition between-subjects design, in which participants were randomly assigned to the explicit-tag, implicit-tag, or no-tag condition. The experimental context was a simulated social media platform modeled after WeChat, and the advertised product was a fictitious jeans brand (“Ludas”) modeled on the Levi’s® brand. We selected jeans because they are a relatively gender-neutral product and suitable for a broad consumer group.

 

Participants were told that they would view a screenshot from a social media platform. In the explicit-tag condition, the advertisement was tagged “Eight friends have seen.” In the implicit-tag condition, the advertisement was tagged “Guess you like.” In the no-tag condition, no recommendation tag was displayed. All other elements of the advertisement were identical across conditions.

 

After they had viewed the advertisement, participants completed measures of advertising attitude and perceived personalization in random order. Perceived personalization was measured using a three-item scale adapted from Kim and Han (2014). We made minor contextual wording changes to fit the newsfeed advertising context. The items we used were “I think the content of the ad is personalized,” “I think the ad is customized for me,” and “I think the ad is targeted at me” (α = .80).

 

Advertising attitude was measured with a four-item scale adapted from Li and Ali (2020). Adaptations included minor contextual wording changes and the use of a 7-point Likert scale (1 = strongly disagree to 7 = strongly agree) to keep the response format consistent with the other measures in this study. The items in our study were “I like this ad,” “I think this ad is good,” “I find this ad appealing,” and “I have a favorable attitude toward this ad.” Cronbach’s alpha in the current study was .85.

 

As manipulation and attention checks, participants reported whether they had seen the same advertisement before and whether they had noticed the recommendation tag, and briefly described it in their own words. They then provided demographic information and received the promised course credit.

A one-way analysis of variance (ANOVA) revealed a significant effect of AI recommendation tag type on advertising attitude, F(2, 112) = 7.99, p < .001. Advertising attitude was more positive in the explicit-tag condition (M = 4.74) than in either the implicit-tag condition, M = 3.84, t(112) = 4.88, p < .001, or the no-tag condition, M = 3.73, t(112) = 3.97, p < .001. The no-tag and implicit-tag conditions did not differ significantly (p = .93). Thus, Hypothesis 1 was supported.

 

A one-way ANOVA also showed a significant effect of AI recommendation tag type on perceived personalization, F(2, 112) = 9.09, p < .001. The sample mean for perceived personalization was higher in the explicit-tag condition (M = 4.76) than in either the implicit-tag condition, M = 3.66, t(112) = 4.29, p < .001, or the no-tag condition, M = 3.68, t(112) = 3.75, p = .001, and the mean for the implicit-tag and no-tag conditions did not differ significantly (p = .99). Table 3 sets out the means, standard deviations, and correlations between advertising attitude and perceived personalization for each AI recommendation tag condition.

 

Mediating Effect Analysis

In Study 2 we examined whether recommendation fit moderated the effect of AI recommendation tag type on perceived personalization and, consequently, the indirect effect on advertising attitude.

The survey was hosted on the Wenjuanxing platform and distributed via WeChat for an online scenario-based experiment. We collected responses from 246 adults recruited from the general population. Participation was not restricted by gender, occupation, or student status. The only context-related criterion was that participants were regular WeChat users, because the experimental scenario was modeled after a WeChat newsfeed advertising context. After excluding invalid responses, including those from participants who reported having seen similar advertisements before or failed the manipulation check, we retained 218 valid forms from 110 men and 108 women for analysis (M_age = 28.35 years, SD = 8.19, range = 18–55). A post hoc power analysis conducted with G*Power 3.1 (Faul et al., 2009) indicated that, with six groups, an effect size of f = 0.40 and α = .05, the obtained sample size provided sufficient statistical power (1 − β = .99), exceeding the conventional threshold of .80. Therefore, the final sample size was adequate for hypothesis testing.

 

In Study 2 we employed a 3 × 2 between-subjects design, with AI recommendation tag type (explicit-tag vs. implicit-tag vs. no-tag condition) and recommendation fit (high vs. low) as the two factors. We followed prior research (Massar & Buunk, 2013) and  manipulated recommendation fit by matching or mismatching the advertised product with participants’ gender. Specifically, as advertising stimuli we used a fictitious razor brand (“Buley”), modeled after the Gillette brand, and a fictitious eye-cream brand (“Cekeo”), modeled after the Lancôme brand. Participants in the high-fit condition viewed the gender-congruent product, whereas those in the low-fit condition viewed the gender-incongruent product.

 

Participants first reported demographic information and, as a context-familiarity check, indicated whether they used WeChat on a daily basis. They were then assigned to either the high-fit or low-fit condition based on gender and were randomly shown one of three versions of the advertisement that varied in AI recommendation tag type. The stimulus design was otherwise identical to that used in Study 1. After viewing the advertisement, participants completed the same measures of advertising attitude and perceived personalization.

 

In addition, in Study 2 we measured brand familiarity and product expertise as control variables, because prior familiarity with the brand and prior knowledge of the product may influence consumers’ evaluations of an advertisement. We assessed brand familiarity with three items adapted from Kent and Allen (1994). The adaptations were made so that the scale would fit the fictitious brand stimuli used in our study without changing the construct being measured. A sample item is “I am familiar with this brand.”

 

We assessed product expertise with five items adapted from Flynn and Goldsmith (1999) to refer to the product category shown in the advertisement, such as “…this type of eye cream,” or “…this type of razor,” depending on the advertisement each participant viewed. A sample item is “I know a lot about this type of eye cream/razor.” All items were rated on a 7-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree.

A two-way ANOVA revealed significant main effects of AI recommendation tag type, F(2, 212) = 5.88, p < .01, and recommendation fit, F(1, 212) = 6.18, p < .01, as well as a significant interaction between AI recommendation tag type and recommendation fit, F(2, 212) = 4.81, p < .01. To decompose the significant interaction, we conducted simple effects analyses and follow-up independent samples t tests within each recommendation fit condition. Under the condition of high recommendation fit, perceived personalization did not differ significantly between implicit tags (M = 3.99) and explicit tags (M = 3.95), t(70) = 0.14, p = .893. However, both explicit and implicit tags elicited significantly higher perceived personalization than the no-tag condition did (M = 3.31). Specifically, the difference between the explicit-tag and no-tag conditions was significant, t(73) = 2.14, p < .05, and the difference between the implicit-tag and no-tag conditions was also significant, t(69) = 2.26, p < .05. Under the condition of low recommendation fit, explicit tags (M = 3.54) resulted in significantly higher perceived personalization than implicit tags did, M = 2.66, t(70) = 2.76, p < .05. There was no significant difference between the explicit-tag and no-tag conditions, M = 3.47, t(73) = 0.22, p = .827. A supplementary comparison further showed that the no-tag condition yielded significantly higher perceived personalization than the implicit-tag condition did when the fit was low, t(69) = 2.42, p < .05. These results remained statistically consistent after controlling for brand familiarity and product expertise, indicating the robustness of the findings. Table 4 shows the means, standard deviations, and correlations between advertising attitude and perceived personalization for each experimental condition.

Moderated Mediation Analysis

Building on the first two studies, in Study 3 we examined whether privacy concerns moderated the effect of consumers’ perceived personalization on their advertising attitude.

The survey was hosted on Sojump.com, one of the largest online survey platforms in China. Participants were adult members of the general population and participation was not limited to students or any specific occupational group. We received 160 surveys and retained 139 valid forms for analysis, yielding an effective return rate of 86.8% (73 men and 66 women; M_age = 32.47 years, SD = 6.59, range = 18–53). The power value estimated by G*Power 3.1 was 0.98 (number of groups = 6, effect size = 0.40, α = .05, sample size = 139), exceeding the conventional threshold of .80 and indicating sufficient statistical power.

 

To provide a causal test of the proposed second-stage boundary condition, we manipulated privacy concerns rather than treating this variable solely as an individual difference. Accordingly, we adapted the privacy-risk scenario used by Angst and Agarwal (2009) by replacing references to personal information in information systems with references to internet companies’ collection and use of consumers’ online behavioral data for personalized advertising recommendations in a Weibo advertising context, so that the manipulation would fit the AI-based personalized advertising setting of our study. Specifically, participants in the high-privacy-concern condition read a passage describing how internet companies collect consumer data from multiple sources and may be subject to hacking and data breaches. Participants in the low-privacy-concern condition read neutral content unrelated to privacy, describing everyday activities such as work, sports, and entertainment.

 

We employed a 3 (AI recommendation tag type: explicit-tag vs. implicit-tag vs. no-tag condition) × 2 (privacy concerns: high vs. low) between-subjects design. To confirm their familiarity with the Weibo social media advertising context, participants were first asked whether they used Weibo daily, and then received the privacy-concerns manipulation task. They were randomly shown a screenshot of one of three Weibo advertisements that differed according to the type of AI recommendation tag. To better fit the Weibo advertising context, the wording of the recommendation tags was slightly adapted while preserving the explicit–implicit distinction used in Studies 1 and 2. In the explicit-tag condition, the advertisement was tagged “From the fashion channel you follow”; in the implicit-tag condition, it was tagged “You may also like”; and in the no-tag condition, there was no recommendation. In this study we again used the fictitious jeans brand of Ludas. After the participants had viewed the advertisement, they completed the measures of advertising attitude and perceived personalization in random order. The rest of the procedure was consistent with that used in Study 1.

To rule out the alternative explanation that privacy concerns moderated the effect of AI recommendation tag type on perceived personalization, we conducted a two-way ANOVA with perceived personalization as the dependent variable. The main effect of AI recommendation tag type was significant, F(2, 133) = 9.33, p < .01, whereas the main effect of privacy concerns and the interaction effect were not significant (p > .10). The mean for perceived personalization was higher in the explicit-tag condition (M = 5.14) than in either the implicit-tag condition, M = 4.42, t(133) = 3.29, p < .001, or the no-tag condition, M = 4.26, t(133) = 3.48, p = .001, and the mean for those in the implicit-tag and no-tag conditions did not differ significantly. Table 5 sets out means, standard deviations, and correlations between advertising attitude and perceived personalization for each experimental condition.

Moderated Mediation Analysis

We examined how AI recommendation tag type influences consumer attitudes toward newsfeed advertising through perceived personalization. Across three experiments, we found that explicit AI recommendation tags generated a more positive advertising attitude than implicit tags did, perceived personalization mediated this effect, and recommendation fit and privacy concerns shaped the process. These findings extend prior research on AI recommendation tags and personalized advertising by showing that consumer responses are influenced not only by recommendation outcomes, but also by how recommendation cues are disclosed. Consistent with self-congruity theory (Sirgy, 1982), explicit tags may enhance consumers’ attitude toward an advertisement by strengthening perceived personalization. Specifically, consumers may infer that the advertisement promotes a product that fits their interests and preferences.

 

From a practical perspective, managers should not only improve recommendation accuracy on their platform but should also carefully design how recommendation cues are disclosed. Explicit tags can enhance consumers’ advertising attitude by strengthening their perceived personalization, but this advantage depends on the fit of the recommendation tag, and the advantage may weaken when the consumer’s privacy concerns are high.

 

This study has several limitations. First, the samples comprised mainly relatively young adults, which may limit generalizability to other groups. Second, the scenario-based experiments may not fully capture real-world advertising environments. Third, we focused on advertising attitude; in future studies, researchers could examine other downstream responses, such as engagement, avoidance, or trust-related outcomes.

Angst, C. M., & Agarwal, R. (2009). Adoption of electronic health records in the presence of privacy concerns: The elaboration likelihood model and individual persuasion. MIS Quarterly, 33(2), 339–370. https://doi.org/10.2307/20650295

 

Awad, N. F., & Krishnan, M. S. (2006). The personalization privacy paradox: An empirical evaluation of information transparency and the willingness to be profiled online for personalization. MIS Quarterly, 30(1), 13–28. https://doi.org/10.2307/25148715

 

Batmaz, Z., Yurekli, A., Bilge, A., & Kaleli, C. (2019). A review on deep learning for recommender systems: Challenges and remedies. Artificial Intelligence Review, 52(1), 1–37. https://doi.org/10.1007/s10462-018-9654-y

 

Chen, Z., & Dubinsky, A. J. (2003). A conceptual model of perceived customer value in e-commerce: A preliminary investigation. Psychology & Marketing, 20(4), 323–347. https://doi.org/10.1002/mar.10076

 

De Keyzer, F., Dens, N., & De Pelsmacker, P. (2015). Is this for me? How consumers respond to personalized advertising on social network sites. Journal of Interactive Advertising, 15(2), 124–134. https://doi.org/10.1080/15252019.2015.1082450

 

Deng, S., Chen, R., & Wu, L. (2024). Strategic fit revelation with advertising. Decision Support Systems, 177, Article 114103. https://doi.org/10.1016/j.dss.2023.114103

 

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149

 

Flynn, L. R., & Goldsmith, R. E. (1999). A short, reliable measure of subjective knowledge. Journal of Business Research, 46(1), 57–66. https://doi.org/10.1016/S0148-2963(98)00057-5

 

Grewal, D., Bart, Y., Spann, M., & Zubcsek, P. P. (2016). Mobile advertising: A framework and research agenda. Journal of Interactive Marketing, 34(1), 3–14. https://doi.org/10.1016/j.intmar.2016.03.003

 

Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (3rd ed.). Guilford Press.

 

Hayes, J. L., Brinson, N. H., Bott, G. J., & Moeller, C. M. (2021). The influence of consumer–brand relationship on the personalized advertising privacy calculus in social media. Journal of Interactive Marketing, 55(1), 16–30. https://doi.org/10.1016/j.intmar.2021.01.001

 

Hong, W., & Thong, J. Y. L. (2013). Internet privacy concerns: An integrated conceptualization and four empirical studies. MIS Quarterly, 37(1), 275–298. https://doi.org/10.25300/MISQ/2013/37.1.12

 

Jawaheer, G., Szomszor, M., & Kostkova, P. (2010). Comparison of implicit and explicit feedback from an online music recommendation service. In P. Brusilovsky, I. Cantador, Y. Koren, T. Kuflik, & M. Weimer (Eds.), Proceedings of the 1st International Workshop on Information Heterogeneity and Fusion in Recommender Systems (pp. 47–51). Association for Computing Machinery. https://doi.org/10.1145/1869446.1869453

 

Kent, R. J., & Allen, C. T. (1994). Competitive interference effects in consumer memory for advertising: The role of brand familiarity. Journal of Marketing, 58(3), 97–105. https://doi.org/10.1177/002224299405800307

 

Kim, Y. J., & Han, J. (2014). Why smartphone advertising attracts customers: A model of Web advertising, flow, and personalization. Computers in Human Behavior, 33, 256–269. https://doi.org/10.1016/j.chb.2014.01.015

 

Li, C., & Ali, K. (2020). Measuring attitude toward the ad: A test of using arbitrary scales and “p < .05” criterion. International Journal of Market Research, 63(5), 620–634. https://doi.org/10.1177/1470785320949844

 

Massar, K., & Buunk, A. P. (2013). Gender differences in adolescent advertising response: The role of involvement and message claim. Psychology, 4(7), 547–552. https://doi.org/10.4236/psych.2013.47078

 

McKee, K. M., Dahl, A. J., & Peltier, J. W. (2024). Gen Z’s personalization paradoxes: A privacy calculus examination of digital personalization and brand behaviors. Journal of Consumer Behaviour, 23(2), 405–422. https://doi.org/10.1002/cb.2199

 

Noar, S. M., Benac, C. N., & Harris, M. S. (2007). Does tailoring matter? Meta-analytic review of tailored print health behavior change interventions. Psychological Bulletin, 133(4), 673–693. https://doi.org/10.1037/0033-2909.133.4.673

 

Pentland, A., Picard, R. W., & Sclaroff, S. (1996). Photobook: Content-based manipulation of image databases. International Journal of Computer Vision, 18(3), 233–254. https://doi.org/10.1007/BF00123143

 

Sirgy, M. J. (1982). Self-concept in consumer behavior: A critical review. Journal of Consumer Research, 9(3), 287–300. https://doi.org/10.1086/208924

 

Tran, T. P., van Solt, M., & Zemanek, J. E. (2022). How does personalization affect brand relationship in social commerce? A mediation perspective: An abstract. In F. Pantoja & S. Wu (Eds.), From micro to macro: Dealing with uncertainties in the global marketplace (pp. 535–536). Springer International Publishing. https://doi.org/10.1007/978-3-030-89883-0_143

 

Xu, Y., & Chen, L. (2025). Personalized recommendations and consumer trust: The role of perceived control and locus of control. Acta Psychologica, 261, Article 105936. https://doi.org/10.1016/j.actpsy.2025.105936