Empirical Analysis of Social Media Influencers’ Effect on Consumer Purchase Intentions and Behavior

1. Introduction

Considering the substantial expansion of social media in recent decades, social media influencers (SMIs) are playing an increasingly vital role in how consumers discover, evaluate, and purchase products and services [1,2,3,4]. SMIs are digital content creators who have cultivated substantial, engaged followings through strategic self-presentation and niche expertise, and who monetize their audience through brand partnerships [5,6]. This phenomenon has changed established marketing models by placing SMIs as intermediaries between brands and consumers [7,8]. Consequently, the influencer marketing industry grew to US$21.1 billion in 2023, more than three times its size in 2019 [9]. This high growth in revenue underscores the growing capacity of influencer marketing to deliver strategic marketing outcomes [10].

As content creators with huge followings on social media platforms, SMIs offer brands a strategic opportunity to reach a broad audience [5,11]. A growing body of research on SMIs has emphasized the benefits, including better brand awareness [12], effective consumer engagement [13], and higher sales conversions [2]. The literature also shows that SMIs influence consumer purchase intentions [7,14,15]. While robust evidence from recent meta-analyses found a significant impact of SMIs on purchase intention [13], translating intention into actual sales requires further research. Relatively limited studies assess the conversion of purchase intentions to actual purchases [16,17]. This gap between intention and actual behaviour is a significant challenge. Moreover, research suggests a positive correlation between trust in an SMI and purchase intention [15,18]. However, there are instances where trust in SMIs does not influence purchase intention and actual purchase [19].

To address these gaps, we investigate the following question: To what degree do SMIs drive consumer purchase intentions and purchase behavior, and what is the role of trust? The specific objectives of this study are to assess (1) the effect of SMIs on purchase intention, (2) how SMIs influence purchase behavior, (3) the role of trust in the process. To achieve the study objectives, we developed a conceptual model and hypotheses based on the literature. We then validated model and tested the hypotheses using partial least squares structural equation modeling (PLS-SEM) of survey data from 232 respondents in the UK. The study findings hold implications for brands seeking to optimize their influencer marketing strategies and contribute to the academic discourse in this rapidly developing research field.

The rest of the paper is organized as follows. We first provide a background to the study through a discussion of the prospects and challenges of influencer marketing and the theory of planned behavior (TPB). We then present the research model and hypotheses. Next, we describe the study methodology before we present the study’s findings. We discuss the findings and assess their implications. We conclude after a discussion of the study's limitations, and suggestions for future research are offered.

2. Background

2.1. Influencer Marketing: Prospects and Challenges

A symbiotic relationship exists between SMIs and social media, where SMIs utilize their substantial following to enhance their influence. At the same time, engaging and niche-oriented content created by SMIs enhances the user experience on social media [20]. Thus, SMIs have become invaluable to contemporary marketing, providing brands with opportunities to connect with consumers [21].

The literature stresses the strategic benefits SMIs offer organizations, including better customer engagement, improved brand recognition, and the ability to influence consumer perceptions positively [22]. By fostering trust and building a network, SMIs function as credible third-party endorsers, enabling two-way interaction and strengthening brand loyalty [2,7]. Brand partnerships with SMIs whose values align with a brand can enhance brand relatability and image, thereby driving consumer engagement [23,24]. This synergy has increased sales as SMIs present products or services that resonate with their followers. Furthermore, the strategic engagement of SMIs extends a brand’s market reach by leveraging their follower networks to reach a broader demographic [23]. The efficacy of SMIs transcends mere reach and visibility metrics. Their ability to shape consumer purchasing decisions may be attributable to their authenticity, relatability, and the trust they cultivate within their follower base. These characteristics have an advantage over traditional marketing, which often lacks the personal touch that SMIs possess [10].

However, while influencer marketing offers several benefits, it also presents critical challenges. One significant issue is the variability in content quality across these platforms, which ranges from credible endorsements to misleading or false information [25]. Furthermore, misinformation, unethical practices, and negative psychological impacts demand stronger regulatory frameworks and ethical standards for SMIs [26,27].

2.2. Theory of Planned Behavior, Purchase Intention and Buying Behavior

The theoretical underpinning of the research is TPB, a well-established framework for studying the psychological factors of purchase intention and behavior [28]. This theory provides a valuable perspective on how SMIs affect consumer purchase intention and behavior [29,30]. TPB argues that there are three antecedents of purchase intention: attitudes, subjective norms, and perceived behavioral control. Attitudes measure a consumer's summary evaluation toward a product (either positive or negative). For example, an SMIs endorsement of a product or service is likely to shape consumer attitudes towards purchase intention and actual purchase positively [31,32].

Subjective norms refer to the social pressure on a consumer to conform to the preferences of significant others and social groups. Social media amplifies this pressure through witnessing positive online reactions towards products endorsed by SMIs [33]. Perceived behavioral control is a consumer's perceived ability to purchase a product. Factors related to perceived behavioral control include accessibility, purchase cost and the complexity of purchase. If a product endorsed by an SMI is immediately available to a consumer and comes at a low perceived cost, the consumer will have an easier time translating intentions into purchase [34].

3. Research Model and Hypotheses

We propose a research model grounded in SMI and consumer purchasing behavior studies with four constructs: SMIs, trust in SMIs, purchase intention, and actual purchase. Figure 1 illustrates the research model and the hypothesized relationships.

3.1. Research Model

Extant research supports a significant relationship between SMIs and consumer purchase intentions [14,35,36]. Studies demonstrate that SMIs can positively influence consumer attitudes towards products or services, potentially leading to a desire to emulate the influencer and engage in a positive behavioral outcome, such as a purchase [37]. Furthermore, research suggests that source-consumer similarity and post-quality perceptions can partially mediate the influence of SMIs on purchase intention [38]. The attractiveness and interactivity of SMIs have also been identified as factors that strengthen the connection between consumers and brands, ultimately influencing purchase intentions [39]. Also, SMIs can enhance brand recognition and purchase likelihood[18]. The “fear of missing out” (FOMO) has also been linked to consumers' increased purchase intention towards products endorsed by SMIs [40]. The literature indicates that individuals who regard influencers as credible, exhibit demographic or psychosocial similarities to them, assess their content as high-quality, and establish robust parasocial relationships are more likely to express intentions to consume the products they endorse [41,42]. In contrast, older demographic groups tend to demonstrate reduced intentions toward consumption [43]. Based on the foregoing, we propose the following hypothesis:

H1: 

Consumers exposed to SMIs content will demonstrate strong purchase intentions.

SMIs’ impact increases through their variety of content, such as reviews, tutorials, and endorsements, thus creating emotional bonds between them and their followers [44]. Several studies suggest that there is a relationship between SMIs and purchase behavior. For instance, it has been established that fashion influencers have a significant impact on customers’ buying decisions [45]. Also, social identification had a direct impact on young people’s online advertising engagement and purchasing behavior [34]. Finally, purchase intention has a positive impact on purchasing behavior. This suggests that if consumers have already demonstrated purchase intention, they are more likely to actualize their purchase intention into a purchase decision [17]. Therefore, we hypothesize that:

H2: 

Purchase intention will positively correlate with actual purchase completion rates among SMI-following consumers.

3.2. The Role of Trust

In the context of SMIs, trust can be defined as readers’ confidence in an influencer’s credibility, dependability, and authenticity. Consumers’ trust in an SMI affects how they respond or interact with an SMI’s recommendations and what they write [15]. For instance, a study found that trust in the SMI mediates the relationship between consumers and their buying intention. In turn, purchasing intentions and purchasing behavior are expressions of the influence of the SMIs’ trustworthiness and the strength of their recommendations [46] . SMIs create trust through their relevant material related to a product or service [47]. Having already established trust with their followers, these followers are more likely to listen to an influencer’s opinions [48].

Some authors argue that consumers value highly the knowledge and expertise of SMIs in certain areas, making them trusted sources of information [14]. Consumers can interact with SMIs using likes, comments, and sharing of content via social media [2]. The interaction between consumers and SMIs has a two-way direction and allows consumers to ask for advice, ask questions, and share experiences [2]. This allows consumers to gain SMI’s recommendations [44]. This can increase consumers' trust in brand recognition and stimulate product sales [49]. Kim and Kim [19] point out that it is crucial to build trust when conducting influencer marketing strategies. Informative SMI content, perceived SMI trustworthiness, and perceived SMI attractiveness can positively affect consumers’ trust, which can be conducive to brand awareness and purchase intentions [50]. Therefore, the following hypotheses are formed based on the literature:

H3: 

Content created by SMIs is likely to build strong levels of trust in products among their audience.

H4: 

Trust in SMIs correlates positively with consumer purchasing behavior, leading to a substantial increase in purchase intention.

H5: 

High trust in SMIs significantly enhances the probability of consumer purchasing behavior regarding products they endorse.

4. Materials and Methods

We employed a positivist quantitative research strategy to validate the proposed research model and its associated hypotheses. The quantitative approach is ideal for this study as it enables systematic collection and analysis of data to examine the relationships among SMIs, purchase intention, and behavior, emphasizing objectivity, measurability, and empirical validation [51].

4.1. Data Collection

We utilized a cross-sectional survey because it allowed us to gain a snapshot of the views of consumers on purchase intention and behavior. It also facilitated the identification of relationships between SMIs and purchase intention and behaviors without requiring longitudinal data [52]. Cross-sectional surveys are valuable in several disciplines due to their use, low cost, and ability to provide timely insights [53].

A principal element of surveys is the design of an appropriate questionnaire[54]. Accordingly, we designed a questionnaire from the literature on SMI, trust purchase intention, and behavior. Table 1 shows the operationalization and definitions of these constructs and their theoretical foundations. The survey instrument consisted of two parts. The first part was used to collect demographic data, including age, gender, education, employment status, and income. The second collected data on social media interactions between individuals and SMIs, trust purchase intentions and behaviors based on SMI recommendations. Respondents responded to the questions on a 7-point Likert scale. The 7-point scale provides finer granularity than a 5-point scale, allowing for better discrimination between neutral and slightly positive or negative responses [55].

We used a convenience sampling method, selecting participants based on their availability and willingness to participate. This approach was chosen for its practicality, cost-effectiveness, and efficient access to the target population. Although convenience sampling is time-saving and easy to implement, it carries the risk of sampling bias, which can potentially limit the generalizability of the results. Despite these limitations of convenience sampling, we considered it appropriate in this case because the study is exploratory [63] and allowed us to gather sufficient data to test the research model and hypotheses.

We distributed the survey link across various social media platforms, primarily Instagram and Facebook. The data collection process spanned a period of two months, specifically from March to May 2024, and resulted in a dataset comprising 232 responses. The participants in this study are residents of the Greater London area of the United Kingdom, providing a diverse representation of perspectives and experiences.

4.2. Data Analysis

We applied PLS-SEM, a variance-based statistical modelling technique, to validate the proposed research model and to test the hypotheses owing to the possibility of latent variables in the measurement model [64]. PLS-SEM's flexibility with non-normal data and small-medium sample sizes makes it ideal for this research. Additionally, it enables simultaneous hypothesis testing, enhancing the rigor of the analysis [65]. We opted for the choice of PLS-SEM because of its robustness in handling complex structural models with multiple latent variables and its more flexible requirement for sample size than covariance-based SEM. Furthermore, this technique facilitated the assessment of relationships among latent constructs and estimates of path coefficients to indicate the strength and direction of the relationships among constructs in our model [66,67]. The data were analyzed using Smart PLS (Version3), an advanced structural equation modeling software. This methodology enabled the exploration of intricate relationships within the dataset, allowing for a nuanced examination of latent constructs and their interrelations [68].

5. Results

The section presents the study findings. Firstly, descriptive analysis of the study sample is provided. Subsequently, the validity and reliability of the measurement model are evaluated. Finally, the structural model is examined, and the hypotheses of the relationships are tested.

5.1. Descriptive Statistics

A summary of the study sample and descriptive statistics is depicted in Table 2. The average age of the sample is 34, with a gender distribution of 48% male and 52% female. A sizable portion of the sample (34%) is aged between 28 and 37, while those in the 18-27 age group account for 31%. Participants aged 38-47 represent 24%, with smaller percentages from the 48-57 (9%) and 58-67 (2%) age groups. Most participants are well-educated, with 52% holding a university degree, 29% pursuing postgraduate studies, 11% with further education qualifications (A-levels or BTECs), 5% having completed secondary education, and 3% with only primary education. The survey also indicates considerable social media engagement among participants, who, on average, use four social media platforms and follow approximately nine SMIs. Half of the sample are full-time employees, and part-time workers make up 15%. Students comprise almost a quarter of the sample (24%). The remaining respondents are either unemployed (2.6%) or self-employed (7.8%). Income levels range from £10,000 to above £60,000. Over 60% of respondents earn below £20,000, while a smaller proportion (4.3%) earn above £60,000.

5.2. Measurement Model Validity Assessment

Convergent and discriminant validity analyses were conducted to evaluate the validity of the measurement model. Convergent validity assesses how well the indicators measure the underlying constructs [69]. The average variance extracted (AVE) values for all the indicators pertaining to each construct were computed. Subsequently, composite reliability (CR) and Cronbach’s alpha coefficients were examined to evaluate internal consistency and reliability of the measurement model. Table 3 summarizes the measurement model assessment results. The results show that the study constructs have a high level of reliability. The Cronbach’s alpha of all the constructs falls within the acceptable range from 0.746 to 0.899. The internal consistency reliability was satisfactory since all the CR values were above 0.7 [70].

The AVE values for all constructs exceed the recommended threshold value of 0.5, ranging from 0.692 to 0.868. This implies that each construct accounts for a significant proportion of the variance in its respective indicators, thereby demonstrating strong convergent validity [71]. Established criteria require that a latent variable should explain at least 50% of the variance in its associated indicators to ensure adequate measurement reliability. The AVE values confirm that all constructs meet this condition, with each explaining over half of the variance. Therefore, the results offer robust evidence for the convergent validity of the constructs, affirming their reliability and accuracy in evaluating the theoretical dimensions [70].

The Heterotrait-Monotrait (HTMT) ratio of correlations is a robust measure of discriminant validity to ensure that the study constructs in the measurement model are distinct. As shown in Table 4, our analysis shows strong evidence of discriminant validity, as all HTMT values fall below the threshold of 0.85 [72]. We observed ratios ranging from a minimum of 0.114 (between trust and SMIs) to a maximum of 0.725 (between purchase intention and actual purchase), further corroborating the discriminant validity of the construct.

5.3. Measurement Model Reliability Assessment

Internal consistency reliability is fundamental for robust PLS-SEM analyses. Evaluating the measurement model through an estimation of indicator reliability is essential to establish the validity and reliability of the constructs under investigation before structural model analysis. Indicator reliability, computed through outer loadings, is the proportion of variance explained by its corresponding latent construct. Following conventional thresholds suggested by Hair et al [70]. Loadings over 0.708 are considered acceptable, as they indicate that the constructs account for at least 0.708 of the indicator’s variance. As presented in Table 5, most indicators had loadings exceeding this threshold. However, a few indicators had loadings from 0.4 to 0.7, although not optimal, and were retained. This is consistent with recommendations by Hair et al [70], who maintain that keeping such indicators can improve model fit and the robustness of subsequent structural analysis, provided the constructs have adequate internal consistency and reliability.

5.4. Structural Model Assessment

The structural model analysis outcomes, which include standardized path coefficients, significance thresholds, and fit indices, are presented in Table 6. These findings elucidate both the predictive validity and explanatory power of the hypothesized relationships within the model.

Bootstrapping analysis provided mixed evidence regarding the proposed pathways. Hypothesis H1, which posits a positive influence of SMIs on consumers’ purchase intentions, received support (β = 0.722, t = 8.434, p < 0.05). This finding aligns with meta-analytic evidence on digital persuasion [10]. Conversely, Hypothesis H2, suggesting that SMIs positively affect actual purchase behaviors, lacked support, indicated by a non-significant path coefficient (β = 0.112; t = 1.233; p > 0.05).

Hypothesis H3, which asserted that SMIS significantly bolster consumer trust, was robustly supported (β = 0.819, t = 23.986, p < 0.05). However, Hypothesis H4, which estimates that perceived trust in SMIs positively impacts consumers’ purchase intentions, was not supported (β = 0.141; t = 0.849; p > 0.05). Intriguingly, Hypothesis H5, which suggested that SMI recommendations drive actual purchases mediated by trust, was also unsupported (β = -0.971; t = 7.108; p < 0.05).

6. Discussion

6.1. Contributions

The results reveal a complex interplay between SMIs, trust, and consumer purchasing decisions, challenging several assumptions in the extant literature. Additionally, this research makes two contributions that address critical gaps in the current body of knowledge regarding the behavioral impacts of SMIs. This is particularly relevant in light of the inconsistencies identified in prior studies [10,16].

First, our confirmation of a significant intention-behaviour gap within SMI contexts warrants attention. Despite strong positive correlations between SMIs and purchase intentions, as supported by earlier research [13,37], these intentions account for only a marginal portion of the variance in actual purchasing behavior. The expressed intentions account for only 2.1% of the variance in actual purchasing behavior (R² = 0.021). This discrepancy challenges a key assumption of the TPB, which suggests that intention is a precursor to actual behavior [28]. It reinforces criticisms against reliance on intention-based measures in influencer studies [16]. Notably, this disparity between intention and action may be explained by contextual variables such as price sensitivity [1], product availability, and promotional activities.

Secondly, the study's most theoretically significant finding pertains to the paradoxical role of trust within influencer marketing dynamics. While prior research has consistently viewed trust as a positive mediator in digital persuasion [1,18], our analysis reveals a counterintuitive negative association between trust and purchase intention (β = -0.971, p < 0.05). This finding contradicts the prevailing consensus in marketing literature and suggests that the relationship between trust and consumer decision-making follows more complex mechanisms than previously imagined. Drawing on persuasion knowledge theory [73], we suggest that this inverse relationship emerges when consumers perceive influencer endorsements as overly commercialised, triggering scepticism that outweighs initial trust. This interpretation is consistent with emerging work on “overtrust” effects in digital contexts [19] and requires a reconsideration of how trust operates within influencer marketing.

The interpretation of our findings should be situated within the broader context of cultural and platform-specific dynamics that influence consumer reactions to SMIs. Our study focused on a Western, English-speaking demographic, yet existing literature highlights significant cross-cultural variances in audience engagement with influencer content. For example, research indicates that collectivist societies may display markedly different trust behaviors in influencer endorsements, owing to heightened group conformity [19]. This suggests that the negative trust effects we identified could either be mitigated or intensified in varying cultural contexts.

Moreover, platform architecture and behavioral norms significantly mediate influencer efficacy in ways our research could not comprehensively address. The inherently visual curation of Instagram's platform ecology may heighten authenticity concerns relative to the perceived spontaneity associated with TikTok content [74]. Consequently, our findings, primarily derived from Facebook and Instagram users, may reflect distinct trust dynamics that would emerge on short-form video platforms or within niche community spaces, where digital relationships are cultivated differently. Also, the intention-behavior gap we observed is likely to vary considerably across platforms, influenced by interface design elements such as integrated shopping features and the friction involved in external purchasing pathways.

6.2. Implications

The implications of these findings are significant for theory and practice. From a theoretical perspective, they necessitate a reevaluation of the dynamics of trust in the influencer context. This suggests a need to move beyond positive-linear models and adopt a more nuanced framework that incorporates varying conditions and the possibility of contradictory effects. Furthermore, these results highlight the critical need for incorporating behavioral metrics in influencer marketing research as opposed to solely relying on intention proxies, which may lead to overestimations of campaign efficacy.

For practitioners, our findings offer strategic recommendations that can inform decision-making and practice. The observed weak correlation between consumer intention and actual behavior implies that brands should enhance conventional engagement metrics, such as likes and shares, by incorporating direct behavioral tracking methods. This approach would enable a more accurate assessment of campaign effectiveness. Secondly, the trust paradox indicates that criteria for influencer selection should prioritize perceived authenticity instead of only focusing on audience size or reported trust metrics. The findings underscore the critical role of contextual factors, suggesting that influencer marketing initiatives be strategically synchronized with pricing strategies and distribution channels. This alignment is essential for optimizing the conversion of consumer intentions into actual purchasing behaviors.

6.3. Limitations and Future Research Directions

The methodological limitations of this study require careful interpretation of its findings. The use of convenience sampling, while pragmatically justified for exploratory research [75], introduces potential selection bias that may compromise the generalizability of our results. The sample’s demographic skew, predominantly comprising young, low-income participants, further limits the applicability of our findings to broader consumer segments. Given the well-documented influence of age and socioeconomic status on responses to SMIs [34], the limited demographic scope of our study raises critical questions regarding the generalizability of our findings to older or higher-income cohorts. This limitation is symptomatic of broader challenges in influencer marketing research, characterized by prevalent reliance on convenience sampling and the predominance of homogenous participant groups. Moreover, although our findings offer valuable insights into consumer segments, they do not adequately illustrate effects at the population level.

The cross-sectional design also imposes constraints on causal inference. While our model identifies significant associations between constructs, longitudinal or experimental designs would better establish changes over time and rule out alternative explanations [52]. Future studies should employ stratified sampling techniques to ensure adequate representation across age groups, income levels, and cultural contexts [52]. Such approaches would address the current overreliance on convenience samples that plagues much of the literature [16].

Our findings highlight the need for more nuanced investigations into the intention-behavior gap in influencer marketing. While we identify a weak translation of purchase intentions into actual behavior, the underlying mechanisms remain underexplored. Subsequent studies should examine contextual moderators such as price sensitivity [1], product availability, and competitive promotions, which may disrupt intention-behavior consistency. Additionally, mediating factors such as perceived authenticity [18] and social validation warrant further research to explain why even trusted SMIs often fail to drive conversions.

The unexpected negative relationship between trust and intention demands particular attention in future work. This finding contradicts prevailing assumptions in the literature [15] and suggests that trust’s role in digital persuasion is more complex than previously theorized. Experimental designs could systematically test boundary conditions, such as varying levels of perceived commercialization, to identify when and why trust backfires. Comparative studies across influencer types (e.g., micro-influencers vs. celebrities) and cultural contexts [76] would further clarify whether this phenomenon is platform-specific or generalizable.

7. Conclusions

We examined SMIs, purchase intention, purchase behavior, and the role of consumer trust. This study makes two contributions. First, the findings affirm the capacity of SMIs to impact consumers' purchase intentions. However, we observed a substantial intention-behavior gap, as our model explained only 2.1% of the variance in actual purchasing behavior. This empirical evidence of weak behavioral translation questions the theoretical assumption of intention-behavior alignment within digital contexts. Second, we discover a negative relationship between trust and intention, which contradicts most of the literature. This paradox implies that highly polished and overtly commercial influencer content may activate persuasion knowledge, converting trust into skepticism. Our findings substantiate growing concerns regarding authenticity deficits in influencer marketing [1], while challenging the universal applicability of trust-based persuasion through SMIs. These contributions call for a paradigm shift from the uncritical adoption of influencer marketing strategies to evidence-based frameworks that recognize: (1) the fragile nature of digital trust, (2) the mediating influence of platform-specific contextual factors, and (3) the essential distinction between engagement, purchase intention and purchase behavior.

These findings suggest that brands should prioritize authenticity in influencer collaborations, align partnerships with pricing and distribution, and focus on sales metrics rather than just engagement. Future research should explore trust’s negative effects, cross-cultural differences in influencer effectiveness, and use longitudinal studies to assess behavioral outcomes over time.