The Influence of Augmented Reality on Purchase Intentions through Consumers of Madame Gie Products

: Beauty products are needed by every individual, with various types of products offered, including quality, price, brand, and packaging. Therefore, the Purchase intentions of beauty products are very diverse. Madame Gie is one of the beauty product brands that has plunged into the use of Augmented Reality to make it easier for consumers to choose product variances. This study aims to determine and analyze the effect of Augmented Reality experiences on the purchase intention of Madame Gie products in Bandung, with Consumers' Control as an intervening variable. This study uses a quantitative approach with a questionnaire through Google Form of 385 samples and analyzed by the Smart-PLS statistical program based on this research, the AR experience positively influences consumer Purchase Intention toward Madame Gie products. Consumer’s Control also plays a crucial role in the relationship between AR experience and Purchase Intention. Therefore, companies need to consider the importance of giving control to consumers in AR technology usage to maximize its impact on consumer purchase intention and enhance the performance of their beauty product marketing.

involved in creating and delivering value.This active involvement can increase consumers' control over product and service experiences.Greater consumer control naturally implies a wider, more complex, adaptive, and open decision-making process (Whang, 2021).
In an advanced preliminary study, it was found that 13 respondents, or 43.4 percent, said that shopping at Madame Gie's online store does not save consumers time.Additionally, 16 respondents, or 53.5 percent, stated that shopping at the online store is not easy to understand.This poses a challenge for companies to make online stores easier for everyone to use, especially for potential consumers who are visiting Madame Gie's online store for the first time.According to Whang (2021), Augmented Reality Experience significantly influences Purchase Intention.The presence of Augmented Reality technology allows potential consumers to experience products virtually through Augmented Reality, making it easier for consumers and consequently increasing purchasing interest.This is related to consumers' control over understanding and perceiving purchases through detailed visual information about products in Augmented Reality technology (Whang, 2021).Madame Gie is one of the pioneers of AR Virtual Try-On in Indonesia.The Virtual Try-On feature is a response to consumer problems during the pandemic.Large-scale social distancing policies by the government have made it difficult for consumers to try out products, especially cosmetics, which generally have testers available for consumers due to varying skin profiles (Whang, 2021).
Previous research by Watson (2018) has stated various positive benefits for companies after implementing Augmented Reality technology in beauty products, as indicated by increased sales conversion and repurchase rates, suggesting that companies should implement it.Based on the above description, this research is aimed at analyzing the correlation or relationship between consumer experiences in using Augmented Reality technology.The novelty of this research is the focus on beauty products with the Madame Gie brand as one of the local brands and its correlation with the user experience of Augmented Reality.This will be conducted through a survey of consumers and potential consumers in the Bandung area.

LITERATURE REVIEW Augmented Reality
Augmented Reality is a technology that allows consumers to engage impersonally with virtual representations with their own body as a substitute for real-world try-outs, thus easily providing size and fit guidance (Judistira, 2022).According to Hilty (2020), Augmented Reality (AR) can be a solution for consumers to reduce direct contact through virtually displayed information (Ghiffari, 2022).Augmented Reality utilizes two characteristics or indicators as follows (Hilty, 2020): 1. Vividness Vividness is defined as the 'technology's ability to generate rich sensorial-mediated environments.Clearer product displays tend to influence consumers' cognitive processing according to Keller and Block.Research by Safitri (2022) reveals that Augmented Reality can be well-received by cosmetic product consumers, thus influencing purchase intention (Safitri, 2022).

Interactivity
Interactivity is defined as technology operating through telecommunication channels, such as telephones, to provide person-to-person or machine-to-machine interactions that mimic interpersonal communication.Perceived interactivity is a concept still evolving.According to Yohanes' research (2023), Interactivity influences purchase intention.Furthermore, according to research by Monalisa (2021), interactivity together with advertising influences purchase intention.

Purchase Intention
According to Kotler as cited in Alghifari (2021), purchase intention is a mindset that emerges after stimulation from a product they see, followed by a desire to purchase and possess it.The indicators of purchase intention according to Ferdinand as cited in Alghifari (2021) are as follows: 1. Transactional interest, which is a person's tendency to purchase a product.2. Referential interest, which is a person's tendency to refer the product to others.3. Preferential interest, which indicates a person's behavior with a primary preference for the product.This preference can be changed if something happens to their preferred product.4. Exploratory interest, which indicates a person's behavior of constantly seeking information about the desired product and searching for other information that supports the positive attributes of the product.

Figure 1. Conceptual Framework
The framework of thinking in the above diagram aims to understand the influence of Augmented Reality on Purchase Intention.Consumers' Control mediates the process and behavior of consumers, which can help reduce uncertainty that may arise from transaction processes or behaviors, thus resulting in higher satisfaction.
1. Influence of Augmented Reality on Purchase Intention According to Steuer as cited in Whang (2021), high-quality product presentation helps reduce risk by creating a sense of non-mediation, reducing barriers between consumers and products.When customers feel as if the product is physically present with them, negative impressions can decrease because the AR experience provides a richer product presentation with a high level of telepresence (Sustaningrum, 2023)

RESEARCH METHODOLOGY
This research employs a quantitative method through a survey with respondents criteria located in the Bandung area, who have used Augmented Reality in the last 6 months, and have purchased Madame Gie from an online store in the last 6 months.This research utilizes the formula from Bernoulli as follows.

Figure 2. Result of Normality Test
Based on the results of the normality test shown in the above diagram, it indicates that the data is not normally distributed, as indicated by the data distribution shown in the points on the diagonal axis (Pratama, 2021).Therefore, data analysis in this study is conducted using the Structural Equation Modeling (SEM) method to test the model.Partial Least Squares Structural Equation Modeling (PLS-SEM) is a non-parametric data analysis method that does not require the assumption of data distribution.PLS-SEM can be used with data that is not normally distributed because the PLS algorithm transforms non-normal data through the central limit theorem (Marliana, 2020).
Partial Least Squares Structural Equation Modeling (PLS-SEM) is a combination of interdependence and dependence techniques.The SEM method is based on the analysis of total variance and includes both the Measurement Model and the Outer Model.The steps in applying this method involve using a measurement model called Confirmatory Composite Analysis to identify the contribution of each measured variable to its construct and evaluate the reliability and validity of the model (Joseph, 2019).Outer Model: The measurement model defines latent constructs, which are hypothesized but unobservable concepts represented by observable or measurable variables.According to Joseph (2019), the Outer Model is evaluated through Convergent Validity, Discriminant Validity, and Internal Consistency.Convergent Validity: Convergent validity is a metric of the overall measurement model that measures the extent to which indicators of a construct converge, thereby explaining item variance.An indicator is considered to have good convergent validity if the loading is >0.70, and an acceptable Average Variance Extracted (AVE) is >0.50 or higher (Joseph, 2019).Discriminant Validity: Discriminant validity is a metric that indicates the extent to which a construct differs from other constructs.Cross Loadings is the first approach when assessing discriminant validity, with values >0.70 (Joseph, 2019).Internal Consistency Reliability: Internal consistency states that all items or indicators should measure the same construct and thus be highly correlated.Internal consistency can be measured using Cronbach's alpha and composite reliability.Composite reliability values of 0.60 to 0.70 are acceptable in exploratory research (Joseph, 2019).Based on the results of testing the Avarage Variance Extracted (AVE) value, it shows that the Augmented Reality (X) variable obtained an AVE value of 0.749.Then the Purchase Intention (Y) variable obtained an AVE value of 0.764.Next, the Consumer's Control (Z) variable as an intervening variable shows an AVE value of 0.783.Thus, all variables in the research are above the value of 0.50 so they are declared valid and reliable for use in measurement.The process of testing discriminant validity involves comparing the results of cross-loading values against certain criteria.According to Husnawati (2019), if the loading values for a construct are higher than those for other constructs, it indicates good discriminant validity.The data presented indicates that the loading value exceeds those of other variables.Therefore, the constructs in this research model are considered valid and suitable for measuring variables, allowing for further testing and analysis to proceed.Variables in the model can be considered valid if they meet the criteria of Composite Reliability with values greater than 0.6.This testing can be supported by Cronbach's Alpha values, where if the obtained value is greater than 0.7, the variable can be considered to meet the criteria or be reliable.Based on the data provided, the variable Augmented Reality (X) Vividness obtained a Composite Reliability value of 0. The R 2 value close to 1 indicates that almost all the necessary information to describe the independent variable is possessed by the independent variable (Ghozali, 2021).Based on the data in the table of R 2 calculation results above, it shows that in the Behavioral Control construct model, it obtains a value of 0.674, meaning that the Augmented Reality variable is able to explain the Consumer's Control variable in the Behavioral Control indicator by 67.

Predictive Relevance (Q 2 )
Advanced measurement in structural models utilizes Predictive Relevance or Q 2 , aimed at determining the magnitude of the values obtained in the model along with parameter estimation.The criteria for the strength and weakness of the model are determined based on the results, where a strong model has a value of 0.35 or higher, a moderate model has a value of 0.15 or higher, and a weak model has a value of 0.20 or higher.The calculation of predictive relevance uses the equation Q (Darwin, 2020).Based on this equation, the calculation result of Q2 is shown to be 0.98, indicating that the model in the study falls into the strong category because it is greater than the value of 0.35.0.92=1-(1-0.674)(1-0.697)(1-0.834).The GoF value is obtained by taking the square root based on the average communalities index, then multiplied by the mean value of R 2 within the range of 0 to 1, with interpretations categorized into three groups including small category with GoF value of 0.1 or more, medium category with GoF value of 0.25 or more, and large category with GoF value of 0.36 or more.Thus, the GoF equation used is GoF=√(Com x R 2 ) where Com is the mean Communality, and R 2 is the mean of R 2 (Maryani, 2020).Based on the data above, in the testing of Goodness of Fit SEM used for evaluating empirical model based on research results and whether it is aligned to be accepted according to existing indexes.Therefore, based on the data above, the GoF value obtained is 0.75 or greater than 0.35, indicating that the GoF of the research model falls into the large category.Based on the conclusion that the three research variables show that the experience of using Augmented Reality by consumers and potential consumers of Madame Gie beauty products can influence Purchase Intention both directly and through Consumer's Control as an intervening variable, the following practical recommendations can be given: a. Madame Gie brand is advised to make efforts to improve the quality of the final product results through Augmented Reality experiments, add product variants that can be tried by consumers through this feature to increase Augmented Reality usage, which can impact Purchase Intention.b.Madame Gie brand can develop Augmented Reality as a step to improve its relationship with consumers and potential consumers, such as improving the quality of display and implementing various types of products in one try-on feature on Augmented Reality to enhance the perception of ease as part of Consumer's Control, which can impact Purchase Intention of consumers and potential consumers.c.Another recommendation for Madame Gie brand is to add campaigns focusing on the use of Augmented Reality, such as organizing pre-launch events for new products to be tried out early through this feature to attract consumer interest in new products, increase consumer usage of the feature, and hopefully drive increased Purchase Intention of consumers and potential consumers of the Madame Gie brand.d.Recommendations for other companies in the beauty product sales sector can also adapt Augmented Reality technology with try-on features to increase Purchase Intention and Consumer's Control for beauty products through knowledge of detailed product information and perceptions that can be directly applied to consumers and potential consumers through e-commerce platforms.

Hypothesis Testing Result (Bootstrapping)
Consumer's control is defined as the self-regulation of consumers and their ability to organize, guide, regulate, and direct behavior towards positive consequences.According to McMillan et al., as cited in Whang (2021), the indicators of consumer's control include the following (Whang, 2021): 1. Behavioral Control Behavioral control refers to consumers' perceptions of what they can do to influence a given situation.According to research by Peña-García (2020), Behavioral Control influences consumers' Purchase Intention.2. Cognitive Control Cognitive control refers to the prediction and understanding of the next steps in performing tasks.Consumers evaluate the information provided and integrate it to predict and interpret the next steps in specific situations.Based on previous research by Zhan (2022), Cognitive Control has a positive and significant influence on Purchase Intention.

19 Operational
Definition of Variables: a) Independent Variable (X) The independent variable is a variable that can influence the emergence of the dependent variable.The independent variable in this research is Augmented Reality with the dimensions of Vividness and Interactivity.b) Dependent Variable (Y) The dependent variable is a variable that is influenced by the presence of independent variables.The dependent variable in this research is Purchase Intention.c) Mediating Variable (Z) The mediating variable is a variable that can influence the relationship between the independent variable and the dependent variable, creating an indirect and unobserved relationship.The mediating variable in this research is Consumers' Control with the dimensions of Behavioral Control and Cognitive Control.The questionnaire in this research uses a Likert scale with 5 response criteria: Strongly Agree 5 points, Agree 4 points, Neutral 3 points, Disagree 2 points, and Strongly Disagree 1 point.As for the data analysis technique, it utilizes descriptive statistics and Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis with the assistance of Smart-PLS software, consisting of validity and reliability testing, as well as inner model analysis including Coefficient of Determination (R^2 Value), Stone-Geisser's (Q^2), Goodness of Fit test, and hypothesis testing using t-tests based on p-value criteria.
.47191/ijcsrr/V7-i2-38, Impact Factor: 7.943 IJCSRR @ 2024 www.ijcsrr.org1203 * Corresponding Author: Debora Dwi Rahayu Nugroho Volume 07 Issue 02 February 2024 Available at: www.ijcsrr.orgPage No. 1199-1211 Inner Model: The structural model tests the relationships expressed in a series of equations that can estimate a series of separate but interdependent multiple regression equations (Joseph, 2019).Coefficient of Determination (R^2 Value): The Coefficient of Determination or R^2 value ranges from 0 to 1, with higher levels indicating higher prediction accuracy.R^2 values of 0.20 are considered high, while in marketing-focused research, R^2 values of 0.75, 0.50, or 0.25 for endogenous latent variables are used.These values are described as substantial, moderate, or weak, respectively (Joseph, 2019).Stone-Geisser's (Q^2): Stone-Geisser's (Q^2) greater than zero for reflective endogenous latent variables indicates the predictive relevance of the path model for specific dependents in the construct.The Q^2 value can be calculated using two different approaches: cross-validated redundancy for the structural path model estimation and cross-validated communality to estimate construct scores for endogenous target constructs to predict missing data points (Joseph, 2019).In this way, this research utilizes SmartPLS software to examine the relationships between variables and predict relationships between constructs, confirming theories and explaining the existence of relationships between latent variables.Goodness of Fit Test: Goodness of Fit represents a model used to represent the covariance matrix of the indicators used.Therefore, the value of GoF is considered good if there is a difference between the observed covariance matrix and the estimated covariance matrix.If the GoF value produces a good result, then a research model is considered suitable and appropriate for use in a study.A model in research is said to have a good fit if it has a GoF value of 0.38, is said to have a marginal fit if it has a value of 0.25, and is said to have a poor fit if it has a value of 0.1 (Ghozali, 2021).Hypothesis Testing: Hypothesis are defined as temporary assumptions supported by empirical data in a research study and derived from the theoretical basis underlying the formation of the conceptual model of the study.To test hypotheses through PLS, the bootstrap resampling method is used by examining the t-statistic (Ulfa, 2021 largest cumulative number: 385 x 5 = 1925 and the smallest cumulative number: 385 x 1 = 385 b.Range value = largest percentage levelsmallest percentage level or total answer scale = 100 percent -20 percent /5 = 16 percent, so that based on these results we get an interval of 16 percent with the following assessment criteria.

Figure 5 .
Figure 5. Evaluation of the Structural Model (Inner Model) 4 percent.Then, in the Cognitive Control construct model, it shows a result of 0.697, indicating that the Augmented Reality variable is able to explain the Consumer's Control variable in the Cognitive Control indicator by 69.7 percent, and Purchase Intention by 0.834.

Figure 6
Figure 6.Hypothesis Testing Result

2581-8341 Volume 07 Issue 02 February 2024 DOI: 10.47191/ijcsrr/V7-i2-38, Impact Factor: 7.943 IJCSRR @ 2024 www.ijcsrr.org 1205
(Fazriansyah, 2022)urement of indicators aimed at determining the correlation between indicator scores and Loading Factor using the value criteria stated in Loading Factor on each indicator, which if it is greater than 0.50 then it can be declared valid(Fazriansyah, 2022).Based on the data above in the Outer Loading value table, it shows that all items in each indicator and variable depict good validity and can be continued in further testing.