According to Solomon (2020) an individual who uses their credibility and online presence to affect many facets of their target audience's lives is known as a social media influencer. By convincing people to buy or utilize particular goods and services, they have a big influence on how consumers behave. Through their content and interactions with their audience, social media influencers also affect routines, attitudes, and behaviors, such as food habits, lifestyle decisions, and communication styles.

Becker et al. (2022) research, which applies planned behaviour theory, demonstrates that customer opinion is positively impacted by the credibility of social entrepreneurship messages (knowledge and trustConsidering the significance of the idea of planned behavior in marketing, opinions about the

organic cosmetics sector are greatly impacted by the legitimacy, skill, and attractiveness of celebrities (Murwaningtyas et al., 2020).

H1: The credibility of social media influencers significantly and favourably impacts attitudes towards Wardah on TikTok Indonesia

Smith and Gallicano (2015) define SMU as an assortment of online activities. According to Pop et al., (2020), claim that social networking sites significantly influences individuals' attitudes and opinions when they are making purchasing decisions. According to studies by Nekmahmud et al. in 2022, SMU has an impact on SN and PBC.

H2: Social media usage significantly and favourably impacts attitudes towards Wardah on TikTok Indonesia

H3: Subjective norms of Wardah on TikTok Indonesia are greatly and favorably impacted by social media use.

H4: Perceived behavioral control of Wardah on TikTok Indonesia are greatly and favorably impacted by social media use.

When consumers choose a relatively lifestyle class on social media, social media marketing has been considered the best alternative for marketers and advertisers (Lee et al., 2018). Social media marketing significantly and favorably affects subjective norms since people use social media to gather information, make decisions about what to buy, and revise those decisions. This makes it a great way for firms to create online marketing strategies (Sun & Wang, 2020). According to research by Hassan et al. (2021) and I. Khan (2022) which demonstrate that social media marketing can affect consumer attitudes.

H5: Social media marketing significantly and favourably impacts attitudes towards Wardah on TikTok Indonesia

H6: Social media marketing significantly and favourably impacts subjective norms towards Wardah on TikTok Indonesia

The fundamental concept of planned behavior is behavioral intention, which indicates the degree to which an individual plans to carry out specific actions based on their behavior (Ajzen, 1991). According to earlier research by Nekmahmud et al. (2022), the theory of planned behavior variable has a substantial impact on an individual's motivation to make an online purchase.

H7: Attitude significantly and favourably impacts online purchase intention towards Wardah on TikTok Indonesia

H8: Subjective norms significantly and favorably impacts online purchase intention towards Wardah on Tiktok Indonesia

H9: Perceived behavioral control significantly and favourably impacts online purchase intention towards Wardah on TikTok Indonesia

Purchase intention is the likelihood that a customer will buy a good or service in the near future (Ajzen & Fishbein, 1980). According to research by Zhao et al. (2024) attitude mediates the association between purchase intention and influencer traits such attractiveness, credibility, and interactivity. Through

attitude, Celep & Çorumlu, (2022) discovered a positive relationship between social media engagement and internet sales. Social media material can influence customers' desire to buy in more subjective ways, per research by Mubdir et al. (2024). Through perceived behavioral control, social media use favorably increases online purchase intentions, according to research by Sharma et al. (2022). According to studies by Divsalar et al. (2023), social media marketing that is mediated by behavior can affect consumers' desire in making online purchases. Zhang et al. (2023) demonstrated that purchase interest is significantly impacted by subjective norms that are altered by social media marketing.

H10: Credibility of Social Media Influencer positively influences Online Purchase Intention through Attitude towards Wardah on TikTok Indonesia

H11: Using social media on TikTok Indonesia positively influences online purchase intention by influencing attitudes towards the Wardah brand on TikTok Indonesia

H12: Using social media on TikTok Indonesia positively influences online purchase intention by influencing subjective norms towards Wardah on TikTok Indonesia

H13: Using social media on TikTok Indonesia positively influences online purchase intention by influencing Perceived Behavioural Control towards Wardah on TikTok Indonesia

H14: Through attitudes toward Wardah on TikTok Indonesia, Social media marketing has a beneficial effect on online purchase intention.

H15: Through subjective norms toward Wardah on TikTok Indonesia, Social media marketing has a beneficial effect on online purchase intention.

The study's research framework is depicted in Figure 1.

The participants in this study are TikTok users from Indonesia who are aware of or have seen content from Wardah brand. The participants in this study are TikTok users from Indonesia who are aware of or have seen content from products under the Wardah brand. Because of the purposive sampling strategy

employed in this study, non-probability sampling, not every member of the population has an equal chance of being included in the sample. The respondents must be active TikTok users in Indonesia (17–50 years old), familiar with the Wardah brand, have seen Wardah influencer content within the last 12 months, and be interested in purchasing Wardah products online.

A questionnaire survey that was distributed online through social media to participants who fit the study's respondent requirements was utilized to gather data because of the quantitative technique. To ensure that respondents who fill out the questionnaire match the requirements, there are 12 screening questions to ensure that they are in accordance with the research. To learn more about respondents' perceptions of marketing on Tiktok social media, including influencer credibility, Social media's effects on attitudes, perceived behavioral control, and subjective norms, 32 statements were given to them. The purpose of these messages was to sway their decision to buy Wardah products online.

This study investigates a particular population or sample using a quantitative approach grounded on the paradigm of positivism. With the aim of descriptive and causal research by providing an overview of the phenomenon and also testing the causal relationship between variable. Because this study used a purposive sampling technique in conjunction with non-probability sampling, not every member of the population has a comparable opportunity of being included in the sample. Positive correlations have been shown between influencer credibility, social media usage, social media marketing, attitude, subjective norms, perceived behavioral control, and desire to make an online purchase, according to the research findings, which were examined using the SmartPLS program and SEM-PLS.

32 construct elements demonstrate exceptional consistency, according to the reliability study results using SmartPLS, values ranging from 0.70 to 0.955 for Cronbach's alpha. If this result surpasses the 0.70 criterion, it can be considered dependable. The results of the composite reliability analysis are used to gauge the concept's internal consistency; a value of 0.70 or greater is considered to be of high quality. Therefore, as seen in table 1, the value of this construct indicates good reliability for this investigation.

To ensure that each questionnaire has accuracy with the research, the validity test is carried out on each variable with factor loading analysis and also AVE, resulting in table 2.

All six variables have factor loading values more than 0.70, which suggests that the indicator accurately captures the idea being examined, according to the findings of the AVE and factor loading tests. Thus, it can be said that every variable in the research satisfies the validity test. AVE testing has criteria that must be more than 0.50 to indicate that the construct effectively represents the concept being measured so that it is considered to have convergent validity that. The AVE test results show that the AVE value has good convergent validity because it is above 0.50.

A paradigm for evaluating discriminant validity, the Fornell Larcker Criterion, highlights that a concept must be more effective at explaining the variation of its own indicators than the variance of other constructs. The variance of its own indicators must be more adequately explained by the root of the AVE value than the variation of indicators from other structures, according to Hair et al. (2002).

It is possible to conclude that every construct in this model exceeds the requirements for discriminant validity because Table 3's data demonstrates that each construct's root AVE is higher than the correlation between it and other constructs. In other words, it is easier to explain the variance of indicators from one model construct than it is to explain the variance of indicators from other constructs.

The inner model, as defined by Hair et al. (2022), describes the relationship between the model's latent variables. This model is designed to test theories on the causal relationship between latent variables and assess the extent to which latent variables may explain other variables. R-square, F-square, and Q-square are used for inner model testing.

As can be seen in the above table, R-square indicates the extent to which the independent variable mode may account for changes in the dependent variable. Table 4 shows that Attitude's R-square is 0.463, meaning that the marketing, usage, and trustworthiness of social media influencers account for 46.3% of its explanation, leaving the other 53.7% unaccounted for. With an R-square of 0.259, perceived behavioural control is explained by social media use in 25.9% of cases, with the remaining 74.1% going unstudied. Social media usage and marketing account for 28.8% of the R-squared value of subjective norms (0.288), with the remaining 71.2% going unanalyzed. 49.9% of the R-squared 0.499 for Online Purchase Intention is explained by SMIC, SMU, SMM, Attitude, SN, and PBC; the remaining 50.1% is not taken into consideration.

The F-square statistic is used to assess the independent variable's effect size on the dependent variable. This measure illustrates the extent to which the independent variable accounts for the variance in the dependent variable. Table 5 displays the F-squared test used in this investigation.

Based on the F-square test, the independent variables make various contributions to the dependent variable. Some relationships show significant contributions, such as Attitude to Online Purchase Intention (medium influence, 0.266), Perceived Behavioral Control to Online Purchase Intention (small, 0.073), Social Media Influencers Credibility to Attitude (small, 0.082), Social Media Marketing on Attitude (small, 0.039) and Subjective Norms (small, 0.014), Social Media Usage on Attitude (medium, 0.160), Perceived Behavioral Control (large, 0.349), and Subjective Norms (medium, 0.190), and Subjective Norms on Online Purchase Intention (small, 0.026).

An study called Q-square is used to evaluate a model's predictive significance. The Q-square measures how well the model predicts the value of the undiscovered dependent variable using the provided independent variables. A blindfolding approach is used to carry out the Q-square calculating process. The model has predictive relevance if Q^2>0, which means it can accurately forecast the dependent variable.

Table 6 shows that the four variables have a Q-square value> 0 or can be said to have predictive relevance. This means that this model can be used to measure the suitability of model predictions in predicting the values of the original data.

The accuracy with which the model fits the data is shown by a statistic known as goodness of fit. Indicators like SRMR, Chi-square, and NFI are used in goodness of fit testing to assess the model's adequacy. A low Chi-square, a low SRMR score, and an NFI near 1 all suggest that the model fits the data well.

The saturated model fits better than the estimated model, according to table 7 goodness of fit test findings. The estimated model does not fit the data, as indicated by the SRMR values of 0.092 in the estimated model and 0.048 in the saturated model. The saturated model is more appropriate, as indicated by the chi-square values of 1255.861 from the saturated model and 1357.031 from the estimated model. Both the estimated model's and the saturated model's NFIs are near 1, indicating that the saturated model better fits the data.

According to Hardani et al. (2020), hypotheses are predictions regarding the relationship between variables in study that are based on pertinent theories but have not been validated by the available empirical data. This study used a one- tailed test with a partial hypothesis test (T test), which has a value of t_tabel 1.65 with a 95% confidence level and a 5% significant level.

The findings of evaluating 15 study hypotheses about the connections between online purchase intention, SN, PBC, attitude, SMU, SMM, and the credibility of social media influencers are shown in Table 8. Based on Table 8, fifteen assumptions demonstrate that each one has a noteworthy and advantageous impact. This is evidenced by p-value of no more than 0.50 and t- statistic of more than 1.65.