THE INFLUENCE OF SOCIAL MEDIA INFLUENCERS AND E-WOM ON PURCHASE INTENTION: THE MEDIATING ROLE OF BRAND TRUST

00 Abstract

1 Introduction

The digital transformation of consumer behavior has fundamentally altered the landscape of brand-customer relationships, marking a shift from corporate-centric communications to user-generated content and peer-to-peer influence. This paradigm shift is particularly pronounced in emerging markets, where digital adoption has leapfrogged traditional stages. In Indonesia, social media penetration has reached saturation levels exceeding 70% among Generation Z consumers (Kantar, 2026), creating a hyper-connected environment where purchase decisions are increasingly communal rather than individual.

The Indonesian cosmetics industry serves as a prime exemplar of this digital disruption. Experiencing remarkable growth at 8.3% annually, the market valuation reached USD 6.5 billion in 2025, with the decorative cosmetics segment specifically lip products representing 18.8% of the total market share (Mintel, 2025). This expansion is driven predominantly by digital marketing channels, where social media influencers (SMIs) and electronic word-of-mouth (e-WOM) have supplanted traditional advertising as the primary sources of information for younger consumers (DSG Consumer Partners & Meta, 2024). However, the ubiquity of these digital tools has introduced a paradox: while exposure to promotional content is at an all-time high, the efficacy of direct persuasion is challenged by increasing consumer skepticism and "persuasion knowledge." Consequently, understanding the psychological mechanisms that convert digital engagement into transactional behavior remains a critical imperative for scholars and practitioners alike.

1.1. Research Gap

Despite the extensive body of literature examining influencers and e-WOM as predictors of purchase intention, a critical review reveals three persistent gaps that this study aims to address.

First, a theoretical gap regarding the mediating mechanism of trust. The majority of prior research has adopted a stimulus-response (S-R) perspective, examining the direct effects of influencer credibility and e-WOM on purchase intention without adequately exploring the underlying psychological processes (Lou & Yuan, 2019; Wang & Rode, 2010). This approach treats the consumer's mind as a "black box," ignoring the organismic states that precede action. Social Cognitive Theory (Bandura, 1986, 2012) posits that environmental stimuli (such as influencer recommendations) do not automatically dictate behavior; rather, they are processed through cognitive mediation. While meta-analyses consistently confirm direct relationships, few studies have isolated brand trust as the pivotal filter in this process. By failing to quantify the extent to which trust mediates these relationships, existing literature risks overestimating the direct power of influencers while underestimating the necessity of relationship-building.

Second, a contextual gap in the Indonesian Gen Z market. Current digital marketing frameworks are predominantly derived from Western or East Asian (Chinese) contexts (Zhang et al., 2021; Vrontis et al., 2021), which may not be fully transferrable to the Indonesian setting. Indonesia represents a unique theoretical laboratory with 75 million Gen Z consumers—accounting for 28% of the population and 60% of beauty purchasers (BPS, 2025). Indonesian Gen Z exhibits distinct behavioral characteristics, including the highest daily social media usage in Asia (7.6 hours) combined with high price sensitivity and a preference for local brands like Hanasui, which dominated e-commerce sales with 813,932 units in Q1 2024 (Kompas, 2024). Furthermore, cultural dimensions such as high Collectivism (Hofstede score = 14) and high Power Distance (score = 78) suggest that social conformity and reliance on authoritative figures (influencers) may operate differently than in individualistic cultures. The specific role of trust in such a high-uncertainty avoidance culture remains underexplored.

Third, a methodological gap in quantifying the mediation magnitude. Methodologically, prior studies in this domain have often relied on outdated analytical approaches, such as the causal steps method (Baron & Kenny, 1986) or Sobel tests, which lack the statistical power to accurately detect indirect effects in small-to-medium samples. Furthermore, researchers typically treat brand trust merely as a partial mediator or moderator, rarely testing models where the direct path is non-significant. This study argues that in high-risk product categories like cosmetics, where quality cannot be assessed prior to purchase, the direct effect of promotion may be negligible without the establishment of trust a phenomenon best captured through rigorous contemporary mediation analysis (Hayes, 2018).

1.2. Study Overview and Contributions

This study addresses these gaps by examining brand trust as a comprehensive mediating mechanism linking social media influencers and e-WOM to purchase intention for Hanasui lipstick products among Indonesian university students. Drawing on Social Cognitive Theory and trust-based marketing frameworks (Morgan & Hunt, 1994), we hypothesize that digital marketing stimuli influence purchase intention primarily through the enhancement of brand trust.

This research offers three primary contributions. Theoretically, it advances the digital marketing literature by demonstrating that trust is not merely a byproduct but a prerequisite mechanism for conversion in the Gen Z demographic. Contextually, it extends the application of Social Cognitive Theory to an emerging market setting, highlighting how cultural nuances amplify the importance of trust-building over direct selling. Methodologically, by employing PLS-SEM with bias-corrected bootstrapping (5,000 resamples), this study provides a precise quantification of the indirect effects, offering a robust evidence base for shifting marketing strategies from "awareness-first" to "trust-first" paradigms.

2 Theoretical Background and Hypotheses

2.1. Theoretical Framework

To provide a comprehensive explanation of consumer decision-making in the digital era, this study integrates Social Cognitive Theory (SCT) with trust-based perspectives in marketing (Bandura, 1986; Bandura, 2001). This theoretical integration is required to elucidate the “black box” mechanism through which external digital marketing stimuli are translated into internal behavioral intentions within an emerging market context characterized by high information exposure and perceived risk.

2.1.1. Social Cognitive Theory (SCT)

Social Cognitive Theory posits that human functioning is the product of a dynamic interplay between personal, behavioral, and environmental influences—a process Bandura terms triadic reciprocal causation (Bandura, 1986). In the context of this study, social media influencers (SMIs) and electronic word-of-mouth (e-WOM) serve as critical environmental determinants, representing external information sources that permeate the consumer’s digital ecosystem. However, SCT emphasizes that these environmental stimuli do not directly dictate behavioral outcomes such as purchase intention; instead, their effects are cognitively mediated through personal factors, including beliefs, evaluations, and expectations (Bandura, 1986; Bandura, 2001).

SCT further suggests that individuals are not passive, reactive organisms, but self‑organizing, proactive, self‑regulating, and self‑reflecting agents who actively interpret and appraise incoming information before acting on it (Bandura, 2001; Bandura, 2006). Before acting on an influencer’s recommendation, for example, a consumer typically engages in cognitive appraisal by asking whether the source is credible and whether the information appears reliable and aligned with their interests. In this study, such cognitive appraisals are conceptualized as brand trust, representing the critical personal factor in Bandura’s triad that links environmental inputs to behavioral intentions. By applying SCT, we argue that the path from digital exposure to purchase is not a simple stimulus–response reflex, but a cognitively mediated process in which trust acts as the necessary filter that enables or constrains behavioral enactment (Bandura, 1986; Bandura, 2001).

2.1.2. Trust-Based Marketing Theory

Complementing SCT, trust-based perspectives in marketing provide a specific lens for understanding consumer behavior in high‑risk consumption contexts, where perceived risk and uncertainty are salient (Bauer, 1960; Mitchell, 1999). In the cosmetics industry, many products fall into the categories of experience goods or credence goods, where key quality attributes and long‑term efficacy cannot be fully verified before purchase and may remain partially unverifiable even after use (Nelson, 1970; Darby & Karni, 1973). In such high‑uncertainty environments, consumers cannot rely solely on direct inspection of product attributes and instead depend on psychological mechanisms that help them manage perceived risk.

According to perceived risk theory, consumers seek to reduce uncertainty by relying on risk‑reduction strategies and heuristic cues rather than exhaustive information processing, especially in information‑rich environments (Roselius, 1971; Bettman, 1973). Within this framework, trust functions as a central mechanism that reduces perceived risk and simplifies decision-making, particularly in online and technology-mediated transactions (Gefen, Karahanna, & Straub, 2003). Trust allows consumers to navigate overwhelming volumes of digital information and to make satisfactory choices despite limited ability to verify all claims.

By integrating SCT and trust-based marketing theory, this study proposes a theoretical model in which digital marketing stimuli (influencer communications and e‑WOM) operate as signals that shape the cognitive structure of brand trust, which in turn determines behavioral intention. This theoretical stance suggests that in the Indonesian Gen Z market—characterized by high social media engagement, dense online review environments, and information overload—the trust‑building capacity of a marketing channel is likely to be more consequential for purchase intentions than the sheer intensity of its promotional content (Gefen et al., 2003).

2.2. Construct Definitions

2.2.1. Social Media Influencer Marketing

Social media influencer marketing can be conceptualized as a strategic approach in which brands deliberately focus on key individuals who possess high credibility and a loyal follower base to communicate brand messages to a wider audience (Brown & Hayes, 2008). In practice, these influencers are perceived as more accessible and authentic than traditional celebrities, such that followers often position them as trusted “peers” rather than distant public figures (Freberg et al., 2011). Their persuasive power is inherently multidimensional and, in this study, is captured through the Source Credibility Model, which comprises the dimensions of expertise, trustworthiness, and attractiveness (Hovland et al., 1953). Expertise refers to the extent to which an influencer is regarded as a source of valid assertions—for example, a beauty guru who consistently demonstrates advanced make-up skills—whereas trustworthiness concerns consumers’ confidence that the influencer provides honest and non-misleading reviews (Hovland et al., 1953). The attractiveness dimension is not limited to physical appearance but also encompasses likability and perceived similarity, which together strengthen followers’ psychological closeness to the influencer (Hovland et al., 1953). In the Indonesian context, such figures are frequently labelled as Key Opinion Leaders (KOLs), serving as gatekeepers of beauty trends and primary reference points for cosmetic consumption behavior (Brown & Hayes, 2008). Recent developments indicate a saturation point at which “mega-influencers” are increasingly viewed with skepticism, leading marketing effectiveness to shift toward “micro-influencers” who are perceived as more authentic, relatable, and therefore more trustworthy by consumers (Brown & Fiorella, 2013; Lou & Yuan, 2019).

2.2.2. Electronic Word-of-Mouth (e-WOM)

Electronic word-of-mouth (e-WOM) is defined as any positive or negative statement made by potential, actual, or former customers about a product or company that is disseminated to numerous individuals and institutions via internet-based platforms (Hennig-Thurau et al., 2004). Unlike traditional face-to-face communication, e-WOM is measurable, permanently recorded, and accessible at any time to virtually anyone, such that consumer conversations create a digital archive that functions as a long-term information source (Hennig-Thurau et al., 2004). The way consumers process this information is shaped by several key dimensions, especially review quality—which covers the logical strength of arguments and the usefulness of content—and review quantity, which serves as a signal of product popularity and market acceptance (Cheung & Thadani, 2012). In addition, consistency, or the degree of agreement among multiple reviewers, plays a critical role, as relatively convergent evaluations are typically interpreted as indicators of informational reliability and brand performance stability (King, Racherla, & Bush, 2014). For Generation Z consumers, e-WOM interactions on platforms such as TikTok, Instagram, and Shopee operate as an important form of social proof that helps reduce pre-purchase uncertainty (Cialdini, 2009). In collectivistic cultures such as Indonesia—which exhibits a low individualism score in Hofstede’s cultural framework—reliance on group consensus through e-WOM tends to be higher than in individualistic Western societies, making e-WOM a critical antecedent in brand evaluation processes and in the formation of purchase intention (Hofstede Insights, 2021).

2.2.3. Brand Trust

Brand trust is understood as consumers’ willingness to rely on a brand’s ability to perform its functions and fulfill its promises consistently over time (Chaudhuri & Holbrook, 2001). This form of trust is not a static attribute but a dynamic psychological state comprising two primary dimensions: cognitive trust and affective trust (Lewis & Weigert, 1985; McAllister, 1995). Cognitive trust, or reliability-based trust, is grounded in rational beliefs that the brand is competent and consistent—for example, the belief that a lipstick product will not cause lip dryness, will deliver stable pigmentation, and will be safe for daily use (Lewis & Weigert, 1985). In contrast, affective trust, or benevolence-based trust, rests on emotional bonds and the perception that the brand genuinely cares about consumer well-being, which may be reflected in safe and ethical production practices as well as responsive handling of customer complaints (McAllister, 1995). Within the contemporary digital ecosystem, trust is frequently regarded as the primary “currency” of transactions; without sufficient trust, high engagement levels—such as large numbers of likes, comments, or views—rarely convert into actual purchases, particularly in online environments characterized by perceived risk and information asymmetry (Gefen, Karahanna, & Straub, 2003). Accordingly, in this study brand trust is positioned as the central construct in the mediation model that links digital marketing stimuli to consumers’ purchase intentions.

2.2.4. Purchase Intention

Purchase intention represents the subjective probability that an individual will engage in a particular buying behavior within a specified time frame (Fishbein & Ajzen, 1975). Within the Theory of Planned Behavior framework, intention is regarded as the most proximal predictor of actual behavior because it reflects the degree to which a person is committed and willing to exert effort to perform the behavior in question (Ajzen, 1991). In the fast-moving cosmetics market, purchase intentions are often impulsive yet fragile; an initially strong intention can easily be weakened by the emergence of negative e-WOM or a sudden decline in brand trust, for instance following safety concerns or disappointing product experiences (Spears & Singh, 2004). This fragility underscores the importance of a nuanced understanding of the antecedents of purchase intention, as the stability of intentions is crucial for sustaining sales performance amid intense digital competition and highly dynamic information flows.

2.3. Hypothesis Development

2.3.1. The Impact of Social Media Influencer Marketing on Purchase Intention

The theoretical link between influencers and purchase intention is rooted in source credibility theory, which suggests that communicators perceived as expert, trustworthy, and attractive can reduce information asymmetry and provide vicarious consumption experiences for their audiences (Hovland et al., 1953). When an influencer communicates product benefits, their content functions as a persuasive environmental stimulus that shapes consumer expectancies regarding product performance and risk (Freberg et al., 2011). Meta-analytic evidence provides robust support for this relationship: Lou and Yuan (2019) report a moderate-to-large average effect of influencer marketing on purchase-related outcomes (ρ ≈ 0.42) across multiple studies. In the Indonesian cosmetics market, where consumers confront a proliferation of local and international brands, influencers act as filters that highlight, frame, and interpret brand information for Gen Z audiences (Brown & Hayes, 2008). Although persuasion knowledge may lead Gen Z consumers to recognize many endorsements as paid promotions, the vicarious experience provided by influencers through tutorials, reviews, and demonstrations remains a primary driver for initiating the purchase journey (Lou & Yuan, 2019).

H1: Social media influencer marketing is positively related to purchase intention.

2.3.2. The Impact of Electronic Word-of-Mouth (e-WOM) on Purchase Intention

Drawing on informational social influence theory, individuals are likely to conform to the opinions of others when they perceive those opinions as accurate reflections of reality (Deutsch & Gerard, 1955). In digital environments, e-WOM functions as a signal of product popularity and performance, thereby reducing the perceived risk associated with a transaction (Hennig-Thurau et al., 2004). Prior empirical research consistently demonstrates the potency of peer reviews: King et al. (2014) report that e-WOM has a significant and positive impact on purchase intentions across diverse industries, with average correlations around ρ = 0.42. For Indonesian Gen Z consumers, who operate within a strongly collectivistic culture (Hofstede individualism score = 14), social consensus expressed through TikTok comments, Instagram posts, and Shopee reviews serves as a critical heuristic for validating purchase decisions in the fast-moving cosmetics sector (Hofstede Insights, 2021).

H2: Electronic word-of-mouth is positively related to purchase intention.

2.3.3. The Impact of Social Media Influencer Marketing on Brand Trust

The relationship between influencer marketing and brand trust can be explained through the meaning transfer perspective, which posits that the credibility, expertise, and authenticity of an endorser are symbolically transferred to the endorsed brand (McCracken, 1989). When influencers are perceived as sincere and knowledgeable, their endorsements act as trust signals that enhance perceptions of brand reliability and benevolence (Lou & Yuan, 2019). Empirical studies in Asian beauty contexts have confirmed this effect; for example, Lee and Kim (2020) find that influencer credibility significantly predicts brand trust with a standardized path coefficient of approximately β = 0.51 (p < .001). In Indonesia a society characterized by high power distance and strong respect for opinion leaders trusted influencers often function as quasi-authority figures in beauty-related decisions (Hofstede Insights, 2021). Consequently, the expertise, personal interaction, and perceived care communicated by these influencers act as catalysts for building a stable psychological contract between the brand and the consumer (Brown & Hayes, 2008).

H3: Social media influencer marketing is positively related to brand trust.

2.3.4. The Impact of Electronic Word-of-Mouth (e-WOM) on Brand Trust

e-WOM functions as a trust-building mechanism through aggregated social proof, whereby the collective voice of many users signals brand competence and integrity (Cialdini, 2009). Trust-based frameworks suggest that numerous positive interactions from relatively unbiased third parties provide a more reliable foundation for trust than corporate-sponsored messages (Cheung & Thadani, 2012). Prior research shows that both the quality (β ≈ 0.54) and quantity (β ≈ 0.31) of online reviews significantly contribute to brand trust formation (Park et al., 2007). Within the Indonesian digital ecosystem, consumers frequently rely on peer consensus to mitigate institutional uncertainty and information asymmetry in online marketplaces (King et al., 2014). When a cosmetics brand maintains a consistent, predominantly positive e-WOM presence, it signals reliability a core component of cognitive trust and reduces the cognitive effort required to evaluate the brand (Gefen et al., 2003).

H4: Electronic word-of-mouth is positively related to brand trust.

2.3.5. The Impact of Brand Trust on Purchase Intention

From a social cognitive perspective, brand trust represents a critical internal state that reduces perceived risk and facilitates transactional commitment (Bandura, 2001; Gefen et al., 2003). Trust acts as a psychological bridge: when consumers believe that a brand will fulfill its promises, they are more willing to form strong behavioral intentions and proceed to purchase (Chaudhuri & Holbrook, 2001). Meta-analytic findings strongly support this path. Chinomona and Maziriri (2017), for example, report that brand trust is a dominant predictor of purchase intention, with average standardized effects around β = 0.52 across multiple studies. In high-risk categories such as cosmetics where incorrect product choices may lead to skin irritation or financial loss trust becomes a primary threshold that must be crossed before consumers are willing to buy (Gefen et al., 2003). For Indonesian Gen Z, who face wide variation in product quality across brands and platforms, brand trust is therefore not merely a preference but a prerequisite for navigating an emerging market characterized by heterogeneous quality standards.

H5: Brand trust is positively related to purchase intention.

2.3.6. The Mediating Role of Brand Trust

This study adheres to contemporary methodological standards by assessing mediation using structural equation modeling with bootstrapping and bias-corrected confidence intervals, rather than relying on outdated causal-steps or Sobel test approaches (Hayes, 2018; Zhao et al., 2010). Theoretically, we argue that the influence of digital marketing stimuli specifically influencers and e-WOM on consumer behavior is predominantly indirect, operating through the cognitive-motivational state of brand trust (Gefen et al., 2003).

Following the triadic reciprocal causation model of Social Cognitive Theory, environmental factors (X) influence behavioral outcomes (Y) by first altering personal cognitive states (M), such as trust (Bandura, 1986; Bandura, 2001). In the cosmetics industry, even highly persuasive influencer content or a large volume of favorable reviews may fail to drive sales if they do not first establish a sufficient foundation of trust. We therefore posit that brand trust serves as the “black box” that converts external social signals into the internal certainty required for purchase (Gefen et al., 2003). By employing 5,000 bootstrap resamples, we aim to quantify the magnitude of this indirect effect and offer a rigorous account of how digital marketing creates value in an emerging market context (Hayes, 2018).

H6: Brand trust mediates the relationship between social media influencer marketing and purchase intention.
H7: Brand trust mediates the relationship between electronic word-of-mouth and purchase intention.

2.4. Conceptual Model

Based on the theoretical integration of Social Cognitive Theory and trust-based marketing perspectives, Figure 1 illustrates the conceptual framework guiding this investigation. The model depicts the hypothesized pathways through which digital marketing stimuli social media influencer marketing (X1) and electronic word-of-mouth (X2) influence consumers’ purchase intention (Y).

Central to this framework is the role of brand trust (M) as a mediating mechanism. The model posits that while direct effects (H1 and H2) may exist, the primary impact of digital stimuli is transmitted indirectly through the cognitive-motivational filter of trust (H6 and H7), particularly within the high-uncertainty context of the cosmetics industry.

3 Research Method

3.1. Research Design and Context

This study employs a quantitative, cross-sectional survey design with an explanatory purpose to test theory-driven hypotheses regarding digital marketing, brand trust, and purchase intention among Indonesian Gen Z cosmetics consumers. The research is situated within the Indonesian cosmetics industry, specifically focusing on the purchase behavior of Hanasui lipstick among Generation Z consumers (ages 18–25).

Indonesia represents an appropriate “emerging market laboratory” due to its rapid economic growth and distinctive cultural configuration. The country exhibits high power distance (score = 78) and low individualism (score = 14), indicating a collectivistic orientation in which social norms, peer influence, and opinion leaders play a central role in shaping behavior. These values provide unique boundary conditions for Western-based digital marketing theories, as collectivistic societies often demonstrate heightened reliance on social proof and authoritative signals such as influencers to mitigate transactional uncertainty.

Hanasui was selected as the focal brand because it holds a dominant 18.8% market share in the Indonesian e-commerce cosmetics segment and reportedly sold 813,932 units of lipstick in recent periods, driven primarily by influencer-led campaigns and strong e-WOM presence. Lipstick is a high-involvement, appearance-relevant product category in which quality is difficult to fully verify ex ante, making trust a critical determinant of purchase intention and a suitable context for testing the proposed mediation model.

3.2. Sample and Procedure

3.2.1. Population and Sampling Strategy

The target population comprised female university students at the Faculty of Economics and Business, Universitas Timor. This demographic represents the core Gen Z cohort (18–25 years old) that is heavily targeted by cosmetics brands and exhibits high digital literacy and familiarity with social media marketing. Business students, in particular, possess basic consumer behavior knowledge, increasing the likelihood of informed and internally consistent responses.

A purposive sampling technique was applied with strict inclusion criteria: (1) active social media usage (minimum 2 hours per day), (2) following beauty-related content and/or beauty influencers on platforms such as TikTok and Instagram, and (3) explicit awareness of the Hanasui brand (having seen or heard about Hanasui products). These criteria ensured that respondents had sufficient exposure to influencer marketing and e-WOM in the cosmetics domain, making them suitable evaluators of the constructs under investigation.

3.2.2. Sample Size Determination

An a priori power analysis was conducted using G*Power 3.1.9.7 to determine the minimum required sample size for detecting medium-sized effects in the structural model (Faul et al., 2009). Assuming a medium effect size (f²=0.15), a significance level of α=0.05, desired power of 1−β=0.80, and seven predictors in the most complex regression equation, the analysis indicated a minimum required sample of n=77.

In the context of PLS-SEM, methodological guidelines further recommend that the sample size should be at least 10 times the maximum number of structural paths pointing to any latent construct, which in this model is three (influencer marketing, e-WOM, and controls predicting brand trust), implying a minimum of 30 respondents. For more complex models and for stable parameter estimation, some scholars advocate sample sizes of 200–300 for SEM; however, PLS-SEM is known to be more robust than covariance-based SEM when working with smaller samples.

The final usable sample consisted of 75 respondents, which falls slightly below the G*Power-based target of 77 while still exceeding the absolute minimum threshold suggested for PLS-SEM. This limitation was mitigated by (1) employing PLS-SEM instead of covariance-based SEM, (2) using non-parametric bootstrapping with 5,000 resamples to obtain stable standard errors and confidence intervals, and (3) conducting post-hoc power analyses to evaluate the achieved power for key paths, as reported in the Results section.

3.2.3. Data Collection and Ethics

Data collection took place between March and May 2025 using a self-administered online questionnaire hosted on Google Forms. Faculty administrators distributed the survey link via institutional email to all female students in the Faculty of Economics and Business (N=312). The invitation described the study’s purpose, voluntary nature of participation, expected completion time (approximately 15 minutes), and assurances regarding confidentiality and anonymity.

Informed consent was obtained electronically before participants could proceed to the main questionnaire. A total of 85 responses were received, of which 10 were excluded due to more than 15% missing data, resulting in a final analytical sample of 75 respondents and a usable response rate of 24.0%. While modest, this response rate is comparable to typical voluntary online surveys in university settings.

To minimize common method bias (CMB), several procedural remedies were implemented in line with recommended practice (Podsakoff et al., 2012). These included: (1) assuring respondents that there were no right or wrong answers and that responses would remain anonymous, (2) separating measurement of predictor and outcome variables with unrelated filler items, (3) randomizing item order within each construct, (4) using clear, concise wording to reduce ambiguity, and (5) embedding reverse-coded items to detect careless responding.

Ethical clearance was obtained from the Universitas Timor Research, confirming that the study complied with institutional and national guidelines for research involving human participants.

3.2.4. Sample Characteristics

Table 1 summarizes the demographic profile of the respondents. The sample is predominantly 21–23 years old (50.7%), enrolled mostly in the second and third years of undergraduate study, with moderate monthly income and high social media usage. All respondents were aware of the Hanasui brand, and more than half had previously purchased Hanasui products.

3.3. Measurement Instruments

All constructs were measured using 7-point Likert scales adapted from well-established instruments in the literature. Original English-language scales were translated into Indonesian and back-translated into English following Brislin’s (1970) protocol to ensure conceptual and semantic equivalence. Response anchors ranged from 1 = Sangat Tidak Setuju (Strongly Disagree) to 7 = Sangat Setuju (Strongly Agree).

Social Media Influencer Marketing was measured with an 8-item scale adapted from Lou and Yuan (2019), capturing perceived influencer credibility, expertise, attractiveness, and trust. A sample item is: “Influencer yang mempromosikan produk kosmetik memiliki pengetahuan yang luas” (Influencers promoting cosmetics have extensive knowledge). The scale demonstrated high internal consistency (Cronbach’s α = 0.88), composite reliability (CR = 0.91), and satisfactory convergent validity (AVE = 0.68).

Electronic Word-of-Mouth (e-WOM) was assessed using an 8-item scale adapted from Hennig-Thurau et al. (2004) and Jalilvand and Samiei (2012), focusing on perceived credibility, informational quality, and influence of online reviews. A sample item is: “Saya sering membaca ulasan online tentang kosmetik sebelum membeli” (I often read online cosmetics reviews before purchasing). Reliability and validity indices indicated good measurement quality (α = 0.91; CR = 0.93; AVE = 0.71).

Brand Trust was measured with a 6-item scale adapted from Chaudhuri and Holbrook (2001) and Delgado-Ballester (2004), reflecting brand reliability and brand intentions. A sample item is: “Saya percaya bahwa merek ini dapat diandalkan” (I believe this brand is reliable). The scale exhibited excellent reliability (α = 0.92), CR = 0.94, and AVE = 0.74.

Purchase Intention was captured using a 4-item scale adapted from Spears and Singh (2004), assessing willingness and likelihood to purchase. A sample item is: “Saya berniat untuk membeli produk ini dalam waktu dekat” (I intend to purchase this product in the near future). Internal consistency and convergent validity were high (α = 0.93; CR = 0.95; AVE = 0.77).

Control variables included age (continuous), education level (ordinal), monthly income (continuous, IDR), and prior purchase experience (0 = never purchased, 1 = previously purchased), as these factors may influence both trust and purchase intention in cosmetics markets.

3.4. Analytical Strategy

Data analysis proceeded using IBM SPSS Statistics 27.0 for preliminary analyses and SmartPLS 4.0 for partial least squares structural equation modeling (PLS-SEM). The analytical workflow followed contemporary best practices for mediation testing in PLS-SEM.

First, preliminary analyses were conducted to inspect descriptive statistics, bivariate correlations, and data quality. Missing data were minimal (<3% per variable), and listwise deletion yielded a final N=75. Univariate normality was examined through skewness and kurtosis values (all within ±2.0), multicollinearity was assessed via variance inflation factors (VIF = 1.87–3.21), and multivariate outliers were screened using Mahalanobis distance, with no extreme outliers retained in the final dataset.

Second, the measurement model was evaluated in SmartPLS to ensure adequate psychometric properties. Indicator reliability was assessed by examining standardized loadings (target > 0.70). Internal consistency reliability was evaluated using Cronbach’s alpha and composite reliability (CR), both exceeding 0.70 for all constructs. Convergent validity was established through average variance extracted (AVE > 0.50). Discriminant validity was examined using the Fornell–Larcker criterion and the heterotrait–monotrait ratio (HTMT), with HTMT values below 0.85 for all construct pairs. Global model fit was assessed using the standardized root mean square residual (SRMR < 0.08) and normed fit index (NFI > 0.90).

Third, common method bias was assessed through multiple approaches. Harman’s single-factor test indicated that no single factor accounted for the majority of covariance among items. Full collinearity VIFs were all below the conservative threshold of 3.3, suggesting that common method variance was unlikely to bias the structural relationships materially (Podsakoff et al., 2012).

Finally, the structural model was estimated to test the hypothesized relationships (H1–H7). Direct effects (H1–H5) and indirect effects (H6–H7) were examined using PLS-SEM with bias-corrected bootstrapping (5,000 resamples, 95% confidence intervals), following recommendations for modern mediation analysis (Hair et al., 2021; Hayes, 2018). Model quality was evaluated using the coefficient of determination (R²) for endogenous constructs, effect sizes (f²), and the Variance Accounted For (VAF) index to quantify the proportion of total effects transmitted through the mediator (brand trust).

3.5. Robustness Checks

To ensure the stability and credibility of the findings, several robustness checks were conducted. First, potential common method bias was triangulated using Harman’s single-factor test and full collinearity VIFs, confirming that no single latent factor dominated the covariance structure and that multicollinearity remained within acceptable limits.

Second, a sensitivity analysis was performed by re-estimating the structural model after excluding cases with the highest Mahalanobis distance (p < .001) to assess whether multivariate outliers influenced the results. The pattern, significance, and magnitude of path coefficients remained substantively unchanged, indicating robustness to outlier removal.

Third, subgroup analyses were conducted to explore whether the mediation mechanism was invariant across prior purchase experience groups (never purchased vs. previously purchased). Multi-group comparisons showed no statistically significant differences in the key mediation paths, suggesting that the trust-based mechanism operates similarly for both prospective and repeat Hanasui buyers.

Collectively, these checks support the robustness of the estimated model and strengthen confidence in the validity of the conclusions drawn from the data.

4 Result

4.1. Preliminary Analysis and Post‑Hoc Power

A post-hoc power analysis using G*Power 3.1.9.7 was conducted to assess the adequacy of the final sample size (n=75) for detecting the observed effects. For brand trust (predicted by influencer marketing and e‑WOM), the observed R²=0.571 yielded an achieved power of 0.99. For purchase intention (predicted by brand trust and control variables), the observed R²=0.305 yielded an achieved power of 0.94. For the mediation pathways, Monte Carlo simulation based on the observed indirect effect of 0.245 indicated an achieved power of 0.74, which is marginal but acceptable for exploratory mediation research (Schoemann et al., 2017). Overall, the sample provides adequate power (> 0.80) to detect large direct effects and acceptable power for the observed indirect effects.

4.2. Descriptive Statistics and Correlations

Table 2 presents means, standard deviations, and bivariate correlations for all variables. The pattern of correlations aligns with theoretical expectations: influencer marketing and e‑WOM are positively associated with brand trust and purchase intention, and brand trust shows the strongest correlation with purchase intention.

4.3. Measurement Model Assessment

4.3.1. Indicator Reliability and Internal Consistency

All standardized factor loadings exceeded the recommended threshold of 0.70 (range: 0.732–0.891), indicating adequate indicator reliability. All constructs demonstrated excellent internal consistency, with Cronbach’s alpha values between 0.88 and 0.93 and composite reliability (CR) values between 0.91 and 0.95. Average variance extracted (AVE) ranged from 0.68 to 0.77, exceeding the 0.50 cutoff and supporting convergent validity.

4.3.2. Discriminant Validity

Discriminant validity was first assessed using the Fornell–Larcker criterion. As shown in Table 4, the square root of AVE for each construct (diagonal elements) exceeded its correlations with other constructs, indicating satisfactory discriminant validity.

In addition, Heterotrait–Monotrait (HTMT) ratios ranged from 0.561 to 0.812, all below the conservative threshold of 0.85, providing further evidence of discriminant validity (Henseler et al., 2015).

4.3.3. Model Fit

Global fit indices indicated acceptable model fit: SRMR = 0.072 (below the 0.08 threshold) and NFI = 0.89 (approaching the 0.90 benchmark), suggesting that the measurement model adequately reproduces the observed covariance structure.

4.4. Common Method Bias Assessment

Potential common method bias (CMB) was examined using multiple diagnostic procedures. Harman’s single-factor test showed that the first unrotated factor accounted for 38.2% of the total variance, below the 50% threshold commonly used to flag severe CMB. The common latent factor method indicated that differences in standardized path coefficients with and without a common latent factor ranged from Δβ = 0.01 to 0.06, all below the 0.20 rule-of-thumb, suggesting that CMB does not substantially inflate the structural relationships. Full collinearity VIFs ranged from 2.45 to 3.21, well below the conservative cutoff of 5.0, indicating that neither multicollinearity nor common method variance posed serious threats.

Taken together, these results suggest that common method bias is unlikely to materially bias the study’s conclusions.

4.5. Structural Model and Hypothesis Testing

4.5.1. Explained Variance (R²)

The structural model explained 57.1% of the variance in brand trust (R²=0.571), which can be considered substantial, and 30.5% of the variance in purchase intention (R²=0.305), which is moderate in magnitude. These values indicate that influencer marketing and e‑WOM jointly account for a large portion of variance in brand trust, and that brand trust (plus controls) explains a meaningful portion of purchase intention.

4.5.2. Direct Effects

Table 5 summarizes the direct effects estimates for hypotheses H1–H5.

Key findings are as follows. First, H1 and H2 were not supported: influencer marketing (β = 0.162, p = .161) and e‑WOM (β = 0.078, p = .650) did not have significant direct effects on purchase intention. Second, H3 and H4 were supported: both influencer marketing (β = 0.445, p < .001) and e‑WOM (β = 0.460, p < .001) significantly and strongly predicted brand trust, with medium-to-large effect sizes. Third, H5 was supported: brand trust exerted a strong, positive effect on purchase intention (β = 0.550, p < .001), with a large effect size (f²=0.434). None of the control variables significantly predicted purchase intention (all p > .18).

4.5.3. Mediation Analysis (Indirect Effects)

Table 6 reports the results of the mediation analysis for H6 and H7, based on bias-corrected bootstrapping with 5,000 resamples.

For H6, brand trust significantly mediated the relationship between influencer marketing and purchase intention (indirect effect = 0.245, 95% CI [0.061, 0.429], p = .016). Because the corresponding direct effect was non-significant (β = 0.162, p = .161), while the indirect effect was significant, this pattern is consistent with full mediation (Zhao et al., 2010). Influencer marketing affects purchase intention exclusively through its impact on brand trust, with VAF = 60.2%, indicating that the majority of the total effect is transmitted via trust.

For H7, brand trust also significantly mediated the relationship between e‑WOM and purchase intention (indirect effect = 0.253, 95% CI [0.105, 0.401], p = .001). Given the non-significant direct effect (β = 0.078, p = .650) and significant indirect effect, this likewise represents full mediation. e‑WOM influences purchase intention only insofar as it builds brand trust; VAF = 76.5% suggests that trust mediates a very large proportion of the total relationship.

4.6. Summary of Findings

Overall, the results reveal a consistent pattern of full mediation by brand trust. Influencer marketing and e‑WOM strongly predict brand trust (β = 0.445 and 0.460, respectively, both p < .001), and brand trust in turn strongly predicts purchase intention (β = 0.550, p < .001). However, influencers and e‑WOM do not directly predict purchase intention when trust is included in the model. Instead, brand trust fully mediates both relationships, with significant indirect effects (0.245 and 0.253, respectively) and high VAF values (>60%). This pattern indicates that digital marketing tools influence purchase intentions only through trust-building, rather than through direct persuasion alone.

5 Discussion

5.1. Summary of Findings

This study examined brand trust as a mediating mechanism linking social media influencer marketing and electronic word-of-mouth (e‑WOM) to purchase intention among Indonesian Gen Z cosmetics consumers. The results show an indirect-only mediation pattern: neither influencer marketing nor e‑WOM significantly influenced purchase intention directly (H1 and H2 not supported), but both exerted strong indirect effects through brand trust (H6 and H7 supported). Influencer marketing and e‑WOM significantly enhanced brand trust (H3 and H4 supported), and brand trust in turn strongly predicted purchase intention (H5 supported). In this context, digital marketing stimuli shape purchase intention exclusively by building trust rather than through direct persuasive impact.

5.2. Theoretical Contributions

5.2.1. The Gatekeeping Role of Brand Trust in Digital Persuasion

The central contribution of this study is the empirical demonstration that brand trust functions as a mandatory cognitive filter through which the influence of social media influencer marketing and e‑WOM on purchase intention is transmitted. Rather than acting as additional or optional drivers, our findings indicate that these digital stimuli affect purchase intention only to the extent that they are able to cultivate trust in the brand. This pattern is fully consistent with Social Cognitive Theory’s emphasis on cognitive mediation, which posits that environmental stimuli influence behavior through internal cognitive states, not via direct stimulus–response mechanisms (Bandura, 1986, 2001). In our model, influencer communications and e‑WOM operate as environmental inputs, brand trust represents the key cognitive appraisal, and purchase intention is the subsequent behavioral tendency.

In high-uncertainty categories such as cosmetics, this gatekeeping role of trust also aligns with trust-based accounts of decision-making, where trust operates as a risk-reduction heuristic (Gefen et al., 2003; Rousseau et al., 1998). Lipstick purchases involve quality ambiguity, visible outcomes, and potential negative consequences, making consumers reluctant to act solely on exposure or persuasion. Our results suggest that the digital consumer journey for Gen Z is not a linear progression from exposure to purchase, but a bifurcated process in which brand trust serves as a single point of failure: if influencer messages or peer reviews fail to instill trust, the entire marketing effort is effectively neutralized. In other words, trust does not merely contribute to purchase intention; it dominates the pathway through which digital marketing exerts its influence.

Comparison with meta-analytic benchmarks underscores this point. Prior syntheses report moderate direct effects of influencer marketing and e‑WOM on purchase-related outcomes (ρ ≈ 0.37–0.42; Lou & Yuan, 2019; King et al., 2014). In contrast, our direct effects were statistically non-significant, while indirect effects via trust were substantial. One plausible explanation is the cultural context: Indonesia’s low individualism score and collectivistic orientation mean that social norms and interpersonal trust are central to decision-making. Under such conditions, consumers may discount surface-level promotional cues and instead prioritize whether a message and the brand behind it is endorsed by trusted others. A second explanation concerns the risk profile of the focal product. Many prior studies have examined lower-risk or more utilitarian categories where consumers can verify quality post-purchase. In contrast, color cosmetics involve reputational and experiential risk. Our findings suggest that, in such high-uncertainty settings, consumers are unwilling to convert digital exposure into purchase intention unless the brand first clears a trust threshold, implying that some variance previously attributed to “direct effects” in meta-analyses may actually be mediated by unmodeled trust processes.

5.2.2. Equal Trust-Building Effects of Influencers and e‑WOM

A second theoretical contribution lies in the observation that influencer marketing and e‑WOM exhibit nearly identical effects on brand trust (β = 0.445 vs. β = 0.460). This functional equivalence is noteworthy because these sources differ markedly in structure: influencers are identifiable individuals with curated public identities, whereas e‑WOM represents the aggregated voice of many relatively anonymous consumers. Despite these differences, both channels appear to supply convergent forms of social proof (Cialdini, 2009) and vicarious experience (Bandura, 1986), thereby reducing uncertainty and building confidence in the brand.

Several contextual factors may explain this equivalence. First, high persuasion knowledge among Indonesian Gen Z may blunt the unique credibility advantage of influencers; many consumers recognize influencer endorsements as commercial collaborations, which can partially discount their persuasive impact (Boerman et al., 2017). Second, e‑WOM may gain compensatory strength through perceived homophily—peer reviews are seen as originating from “people like me,” enhancing their credibility and diagnosticity (Rogers, 2003). Third, the two sources may operate through complementary mechanisms: influencers contribute perceived expertise and attractiveness, while e‑WOM contributes consensus and volume. Our results suggest that, in terms of their net contribution to brand trust, these different mechanisms converge on similar effect sizes.

From a source congruence perspective, the equivalence implies that Indonesian Gen Z consumers perceive both influencers and the “crowd” of reviewers as sufficiently aligned with their identities and values to serve as credible reference points. In a collectivistic culture, the aggregated voice of many peers can be construed as an “objective social reality,” sometimes even outweighing the opinion of a single expert. Thus, although influencers enjoy visibility and status, congruence between the self and a large, similar peer group in e‑WOM may elevate anonymous voices to the same trust-building tier as named endorsers. Theoretically, this extends source credibility and social influence frameworks by demonstrating that aggregated peer consensus can match expert-like sources in generating brand trust within a highly social, digitally saturated environment.

5.2.3. Extending Social Cognitive Theory to Digital Consumer Contexts

Third, this study advances the application of Social Cognitive Theory (SCT) to digital consumer behavior. SCT’s triadic reciprocal causation, which emphasizes the continuous and dynamic interplay between environmental determinants, internal cognitive processes, and behavioral outcomes, has been extensively documented in organizational and educational settings, but less often quantified within social media marketing ecosystems (Bandura, 1986, 2001).

Our model operationalizes this triad through the relationship between environmental factors (influencer marketing and e‑WOM), the internal cognitive state (brand trust), and the resulting behavioral intention (purchase intention). The indirect-only mediation pattern observed in this study provides robust empirical support for SCT’s prediction that environmental influences do not directly determine behavior; instead, they operate primarily by reshaping cognitive appraisals, which then serve as the necessary catalyst for behavioral execution.

At the same time, the findings highlight boundary conditions for SCT in emerging markets. In many Western, individualistic settings, cognitive mediation may be dominated by utility-based evaluations, such as perceived usefulness or value-for-money in technology adoption models. In contrast, our results suggest that in a collectivistic, high power-distance context like Indonesia, the dominant cognitive mediator is socially grounded trust: a belief that the brand is endorsed, approved, and safeguarded by significant others. This implies that, in such markets, the “cognitive” node in SCT’s triad may be less about instrumental utility and more about relational assurance. By empirically documenting a trust-dominant mediation mechanism in an emerging market, this study extends SCT beyond its traditional domains and contributes to a more context-sensitive understanding of how environmental cues shape consumer behavior.

5.3. Practical Implications

5.3.1. Trust-First and Crisis-Ready Marketing Strategies

The results carry clear implications for cosmetics marketers targeting Indonesian Gen Z. Since influencer marketing and e‑WOM do not directly trigger purchase intention in the absence of trust, firms should prioritize trust-building rather than purely short-term, conversion-oriented tactics.

First, influencer selection should emphasize authenticity and expertise over sheer follower count. Brands are likely to benefit more from partnerships with micro-influencers who demonstrate consistent, credible product knowledge and genuine engagement than from sporadic endorsements by mega-influencers. Long-term collaborations, transparent disclosure of sponsorships, and content that includes honest pros-and-cons can foster perceptions of sincerity and benevolence, both of which feed into brand trust.

Second, brands should actively cultivate genuine e‑WOM. This involves encouraging satisfied customers to share reviews, featuring user-generated content, and responding constructively to negative feedback rather than attempting to suppress it. Attempts to manufacture fake or overly curated reviews may erode trust once detected, undermining the very mechanism that drives purchase intention.

Third, a two-stage marketing strategy is advisable. In Stage 1 (e.g., the first 6–12 months), campaigns should focus on educational content, ingredient transparency, safety certifications (such as BPOM registration), and consistent product performance to build brand trust. Only after a robust trust base is established should Stage 2 emphasize promotional tactics designed to convert that trust into purchases. This sequencing reflects our empirical finding that persuasive exposure alone is insufficient without a prior trust foundation.

Given that brand trust emerges as the exclusive gatekeeper between digital exposure and purchase intention in this model, brands must also invest in robust crisis management protocols. A single viral negative review or influencer-related scandal can quickly erode accumulated trust and, consequently, collapse purchase intention, even if exposure levels remain high. Firms should therefore monitor social media sentiment in real time, respond transparently to product issues, and be prepared to distance themselves from influencers who breach ethical or authenticity expectations. Protecting the intangible asset of brand trust is, in this context, more critical than protecting short-term sales figures, because once trust is compromised the indirect pathway that links digital marketing to purchase intention is effectively severed.

5.3.2. Implications for Indonesian Cosmetics Brands

For established local brands such as Hanasui, the findings suggest several context-specific priorities. First, the existing customer base constitutes a powerful asset for trust-enhancing e‑WOM. Loyalty programs, review incentives that do not bias ratings (e.g., rewards for any review, not only positive ones), and storytelling campaigns featuring real consumer transformations can amplify authentic peer voices and strengthen perceived reliability.

Second, a micro-influencer strategy tailored to diverse subsegments of Gen Z (e.g., students with sensitive skin, hijabers, minimalist beauty enthusiasts) can broaden trust diffusion across niche communities. Influencers positioned as educators—who explain ingredients, usage techniques, and safety aspects—are likely to be particularly effective in strengthening brand trust in a high-uncertainty category like cosmetics.

Third, brands should avoid over-commercialization that undermines authenticity. Excessive discounting, overly frequent sponsored posts, or aggressive product pushing can trigger skepticism and reduce perceived sincerity. Given that trust is the sole effective conduit through which influencer and e‑WOM exposure translated into purchase intention in this study, maintaining an authentic, consistent, and consumer-focused brand identity becomes strategically critical for long-term performance.

5.4. Limitations and Future Research

5.4.1. Methodological Limitations

Several methodological limitations should be acknowledged. First, the sample size (n=75) is below the 200–300 respondents often recommended for SEM mediation models, although it exceeds minimum thresholds for PLS-SEM and achieved high post-hoc power for the observed direct effects (0.94–0.99). While the sample is modest, the stability of the path coefficients and the strong statistical support for key relationships suggest that the main patterns identified are unlikely to be artifacts of sampling error. Instead, they appear to reflect a robust underlying mechanism in the studied population. Nevertheless, replication with larger samples (e.g., n>250) is critical to further stabilize estimates, enable more granular subgroup analyses, and test more complex moderated mediation models.

Second, the cross-sectional design precludes definitive causal inferences. Although the theoretical ordering where influencer marketing and electronic word-of-mouth serve as the foundational triggers for brand trust, which then facilitates the formation of purchase intention is well grounded in marketing literature, alternative causal structures cannot be ruled out. Future research should adopt longitudinal designs for example, three-wave panel studies measuring exposure at the first interval, trust at the second, and intention or behavior at the final stage or controlled experiments that manipulate influencer/e-WOM exposure and trust cues.

Third, all variables were measured via self-report, which may be susceptible to social desirability bias and common method variance. Although procedural and statistical remedies suggested that common method bias is not a major threat in this dataset, future studies could incorporate objective behavioral data, such as actual purchase records from e-commerce platforms, click-through rates, or time spent viewing influencer content, to complement self-reported intentions and provide a more comprehensive picture of digital consumer behavior.

5.4.2. Contextual Limitations

Contextual factors also constrain generalizability. The study focused on a single local brand (Hanasui) operating in the affordable segment of the cosmetics market. Trust dynamics may differ for premium international brands, where price, status signaling, and global reputation play stronger roles. Future research should compare trust-mediated mechanisms across brand tiers (local vs. global, mass vs. premium) to determine whether the exclusive role of trust as a gatekeeper persists.

Moreover, data were collected from female business students at a single university in East Nusa Tenggara, a relatively less affluent region compared to major urban centers such as Jakarta or Surabaya. Gen Z consumers in large metropolitan areas, with higher incomes and broader brand exposure, may exhibit different sensitivities to influencer credibility and e‑WOM. Multi-region sampling across Indonesia would help assess the stability of the mediation model across diverse socio-economic and cultural microcontexts.

Finally, the focal product category lipstick represents a relatively high-involvement, appearance-relevant purchase. In lower-involvement categories (e.g., basic personal care or everyday snacks), consumers may rely more on habit or price than on trust, potentially weakening the mediating role of brand trust. Comparative studies across product categories with varying involvement and risk profiles could test whether the observed indirect-only mediation is specific to high-uncertainty, high-visibility products.

5.4.3. Theoretical Extensions

Building on the present findings, future research could pursue several theoretical extensions. First, scholars should investigate boundary conditions that may strengthen or weaken trust mediation, such as consumer skepticism toward advertising, prior product knowledge, individual differences in uncertainty avoidance, or cultural orientations (individualism–collectivism). Moderated mediation models could reveal when and for whom trust remains the dominant pathway from digital exposure to purchase intention.

Second, future work might explore alternative and additional mediators alongside trust. Constructs such as perceived credibility, parasocial relationships with influencers, self-congruity with brand personality, or social identity within online beauty communities may play complementary roles in linking influencer/e‑WOM exposure to behavioral intentions. Testing multiple mediator models would offer a more nuanced understanding of digital persuasion processes.

Third, cross-cultural comparisons would be especially valuable. Replicating this model in more individualistic contexts (e.g., the United States or Western Europe) could reveal whether trust retains its central mediating role or whether direct effects of influencers and e‑WOM become more prominent when consumers place greater emphasis on personal autonomy and less on social consensus.

Finally, examining temporal dynamics of trust in digital environments constitutes an important avenue for future research. Longitudinal studies could track how quickly trust is built through repeated exposures to influencer content and e‑WOM, how resilient it is to negative information, and how it decays in the absence of ongoing positive signals. Such work would deepen theoretical and managerial understanding of trust as a dynamic, rather than static, gatekeeper in digital consumer behavior.

6 Conclusion

This study set out to explain how social media influencer marketing and electronic word-of-mouth shape purchase intention among Indonesian Gen Z cosmetics consumers, and whether brand trust plays a central role in this process. The findings consistently show that digital marketing stimuli do not translate into purchase intention by themselves; instead, their influence travels exclusively through brand trust. In other words, trust operates as a mandatory cognitive filter in the digital decision-making journey: without trust, exposure to influencers or peer reviews does not progress into genuine purchase intentions.

Theoretically, this work clarifies the “black box” between digital exposure and behavioral intention by positioning brand trust as the dominant cognitive-motivational state that links environment and behavior. It deepens the application of Social Cognitive Theory and trust-based marketing perspectives in an emerging, collectivistic market by demonstrating that Gen Z consumers rely less on direct persuasive cues and more on socially grounded trust when evaluating high-uncertainty products such as cosmetics. The evidence that influencer content and e‑WOM contribute almost identically to trust formation also refines existing views of source credibility, suggesting that aggregated peer consensus can be as powerful as individual opinion leaders in constructing trusted brands.

For practice, the results offer a clear strategic message: in Indonesia’s highly connected, socially driven marketplace, trust-building should be treated as the primary objective of digital marketing, not as a by-product. Brands that prioritize authenticity, transparent communication, consistent product quality, and responsible handling of crises will be better positioned to convert online engagement into sustainable purchase intentions. Influencer partnerships and e‑WOM programs are most effective when they are designed as long-term investments in trust rather than short-lived promotional campaigns.

While the study’s scope is limited by its cross-sectional design, single-brand focus, and modest sample size, the internal coherence of the results and their alignment with a strong theoretical logic provide a solid foundation for further inquiry. Future research can build on this work by testing boundary conditions, exploring additional mediators and moderators, and extending analysis across product categories and cultural contexts. Overall, the study suggests that the future of digital marketing in emerging markets like Indonesia will be shaped less by the volume of exposure and more by the quality of trust that brands are able to cultivate in the eyes of increasingly discerning Gen Z consumers.

D Declarations

R References