The Influence of Electronic Service Quality, Sales Promotion, and Product Variation on Purchase Decisions and Satisfaction: a Study on BUKALAPAK

: Bukalapak, as one of the most popular online shopping applications in Indonesia, faces challenges in satisfying its customers. Despite offering a variety of products and services, there are significant complaints regarding the quality of service and the conducted promotions. This research aims to evaluate how product variation, sales promotions, and electronic service quality at Bukalapak influence purchase decisions and customer satisfaction, as well as the extent to which each of these factors impacts the decisions and satisfaction of Bukalapak's customers. The study employs a quantitative method, collecting data through a survey distributed via Google Form, and selecting samples using non-probability sampling with purposive sampling technique. Respondents included those who have shopped at Bukalapak while residing in Indonesia, including millennials, Generation X, and Generation Z. For data analysis, the research utilizes the Partial Least Squares Structural Equation Modeling (PLS-SEM) technique with the assistance of SmartPLS software. The results of this research indicate that respondents rate the electronic service quality, sales promotions, and product variation offered by Bukalapak, as well as the purchase decisions and customer satisfaction at Bukalapak, as Very Good. Electronic service quality, sales promotions, and product variation have a positive and significant impact on both purchase decisions and satisfaction.


INTRODUCTION
The use of the internet in Indonesia continues to increase every year, and this is not separate from the benefits of the internet that can assist in daily human activities.This phenomenon is supported by an article on indonesiabaik.id, which demonstrates the ongoing growth of internet usage in Indonesia from 2020 to 2023.The increased internet usage is utilized by companies to develop their businesses online through marketplaces.According to [13] online shopping habits have remained unchanged and are still frequently used even though active pandemic cases in Indonesia have decreased.The Indonesian society has numerous options for online shopping with the presence of several marketplaces.A survey from [4] indicates that Shopee, Tokopedia, and Lazada dominate the market by being the top three most visited.marketplace most visited.Behind this dominance, there are two other ecommerce platforms that stand out in the market, namely Blibli and Bukalapak.Bukalapak has a lower rating on the App Store compared to the other four e-commerce platforms; however, on Google Play, its rating is on par with Lazada and almost approaches the ratings of Tokopedia and Blibli.A more in-depth study on Bukalapak includes an analysis of its revenue.Every year, the third quarter consistently becomes the period where Bukalapak reaches its peak revenue, surpassing other quarters.However, after reaching the peak in the third quarter, the revenue experiences a decline in the fourth quarter.This downward trend continues into the first quarter of the following year, where the revenue further decreases compared to the fourth quarter of the previous year.Nevertheless, the second quarter always becomes a recovery momentum, with revenue showing an increase.Unfortunately, this positive momentum only lasts until the third quarter before experiencing another decline in the fourth quarter.This pattern illustrates the unique and recurring business dynamics of Bukalapak every year.Despite the decrease in Bukalapak visitors, the cyclical revenue trend remains consistent.Bukalapak has a lower rating on the App Store compared to the other four e-commerce platforms; however, on Google Play, its rating is on par with Lazada and almost approaches the ratings of Tokopedia and Blibli.In addition to providing ratings, Bukalapak's consumers also give reviews on the App Store and comments on Bukalapak's official Instagram posts regarding the quality of its services.According to [12], besides Sales Promotion, Product Variation can also influence consumer purchasing decisions because product variation provides more significant opportunities for consumers to find products that match their preferences.[1] state that the more diverse the product variations offered, the higher consumer satisfaction.This is because consumers can find all the products they need in one store, eliminating the need to search for products elsewhere.[10] states that purchase decisions are formed when consumer needs are met, with factors such as perceived service quality and the availability of product variations in e-commerce playing a role.Meanwhile, [5] suggests that marketers can stimulate purchase decisions through the development of effective advertising and sales promotion programs.and complete Product Variations offered, the higher the satisfaction of consumers, as consumers can find all the products they need in that store, eliminating the need to search for products in other stores.[10] states that purchase decisions are formed when consumer needs are met, with factors driving purchase decisions including the perceived quality of service by customers and the availability of product variations in e-commerce.Meanwhile, [5] suggests that marketers can stimulate purchase decisions through the development of effective advertising and sales promotion programs.The phenomena, facts, and issues mentioned above lead researchers to examine the influence of Electronic Service Quality, Product Variations, and Sales Promotion on Satisfaction through Purchase Decision.

LITERATURE REVIEW
A. E -Service Quality [2] defined Service Quality as the result of a comparison between two key elements influencing service quality, namely the expected service and the perceived service.[3] defines Sales Promotion as one of the seven aspects of the promotional mix, with the other six being advertising, personal selling, direct marketing, public relations, corporate image, and exhibitions.Sales Promotion is a short-term marketing strategy designed to achieve specific objectives.It differs from advertising in its goal, as it aims to create a sense of urgency to prompt immediate action rather than building a brand over an extended period.[8] state that Product Variation is one crucial aspect of marketing, involving the development, arrangement, and promotion of various types of products to meet customer needs and desires.The more diverse and complete the product variations offered, the higher the customer satisfaction.In this context, customers no longer need to search for products in other stores because they can find all the products they need in that store.[10] defines a purchase decision as a series of steps that begins with consumers realizing their needs, searching for alternatives, evaluating choices, and ultimately selecting a product or service that can meet their specific needs.This decision is significantly influenced by internal factors such as beliefs, attitudes, knowledge, personality, perception, lifestyle, roles, and social status of consumers, as well as external factors like culture, group membership, and social class.[6], Kepuasan pelanggan adalah indikator sejauh mana perusahaan telah memenuhi serangkaian persyaratan pelanggan.Ini mencerminkan sejauh mana perusahaan berhasil memenuhi harapan dan kebutuhan pelanggan serta memberikan pengalaman yang memuaskan.Tantangan terbesar dalam belanja online adalah menyediakan dan mempertahankan kepuasan pelanggan.

F. Hypothesis
HI : E-Service Quality has a significant positive effect on Purchase Decision H2 : Product Variation has a significant positive effect on Purchase Decision H3 : Product Variation has a significant positive effect on Purchase Decision H4 :Purchase Decision has a significant positive effect on Satisfaction H5 : E-Service Quality has a significant positive effect on Satisfaction H6 : Sales Promotion has a significant positive effect on Satisfaction H7 : Product Variation has a significant positive effect on Satisfaction

METHODOLOGY
The approach to theory development used in this research is deductive, which involves reasoning using a sample that is part of the population, so the theory starts with something general, and then specific hypotheses are derived and subsequently tested [11] According to [11], the type of investigation in this research is causal because it seeks to find out whether one variable causes another variable to change, making it a cause-and-effect study.The researcher's involvement in this study is termed minimal interference, as described by [11], meaning the researcher's intervention is minimal.This is because the research uses a questionnaire, where the researcher actively asks participants to provide information in the form of responses to statements made.The operationalization of variables involves breaking down the variables in the study into smaller indicators, making it easier to collect the necessary data for the research and facilitating its measurement.In this study, data is measured using an ordinal scale.An ordinal scale allows ranking of responses without a fixed numerical difference between each answer.The population in this research consists of consumers using the services of Bukalapak and belonging to the millennial, generation X, or generation Z in major cities in Indonesia, particularly provincial capitals.The population size for this research cannot be determined due to the broad geographical coverage and the researcher's limitations in obtaining data on Bukalapak users.This research employs a non-probability sampling technique.
According to [7], non-probability sampling is a method that does not provide an opportunity for every member of the population to become a sample.The data collection process in this study is conducted through a survey.Cooper define a survey as a method of data collection used to obtain information over a predetermined period and systematically implemented.The essence of a survey is to obtain comparable data from the selected sample, allowing for the identification of similarities and differences.This research uses Confirmatory Factor Analysis (CFA) to analyze the measurement capacity (loading factor) of each indicator transformed into items in the questionnaire.The study utilizes multivariate analysis techniques.As explained by [7] multivariate analysis is a method that allows researchers to study more than two variables.In this study, the chosen multivariate analysis technique is Structural Equation Modeling (SEM).In SEM (Structural Equation Modeling), there are two approaches, namely Covariance Based Matrix Structural Equation Modeling (CB-SEM) and Variance Based Matrix Structural Equation Modeling (VB-SEM), in which the Partial Least Square (PLS) analysis technique is employed.CB-SEM is more oriented towards theory confirmation, while VB-SEM focuses more on predicting relationships between variables and theory development [7] RESULT AND DISCUSSION In this study, primary data was obtained through a questionnaire filled out by respondents, namely consumers who have made purchases on Bukalapak while in Indonesia between February and August 2023.The total number of respondents meeting the criteria was 300 respondents.Respondent characteristics include gender, age, highest education level, occupation, domicile, income, and the frequency of purchasing on Bukalapak.

A. Measurement (Outer) Model
In PLS-SEM, the "outer model" is equivalent to the measurement model, which is used to test validity and reliability.This stage employs convergent validity with loading factors and Average Variance Extracted (AVE), discriminant validity through crossloading, Fornell-Larcker, and HTMT, as well as testing reliability using Cronbach's Alpha and Composite Reliability.2) NFI (Normed Fit Index) serves as an indicator of model acceptance.A value of NFI approaching 1 indicates a more optimal model fit.(3) RMS_theta measures how much correlation exists between the outer model residuals.For the model to be considered appropriate, this measure should approach zero.(4) Exact Model Fit is a criterion that evaluates how close the implicit correlation matrix is to the observed correlation matrix.The model is considered fit if the difference between these two matrices is very small.The results of this study indicate that E-service Quality, Sales Promotion, and Product Variation are factors that can enhance Purchase Decision and, ultimately, increase Satisfaction when shopping online.The importance of Satisfaction is supported by), stating that providing the best service to customers has many benefits for the company.Satisfied customers will return and provide positive word-of-mouth promotion, serving as a free promotional tool and boosting the company's confidence.Customer satisfaction is only felt when the purchased product meets their expectations.This indicates the occurrence of a transaction, where the transactions that take place are one of the outcomes of the Purchase Decision made by customers.

CONCLUSION AND RECOMMENDATION
E-Service Quality has a positive and significant influence on Purchase Decision, with a path coefficient of 0.350.The generated Tvalue is >1.65, specifically 5.798.The significance P-value obtained is 0.000.The f square value is 0.174, indicating that the influence of E-Service Quality on Purchase Decision falls into the moderate category.This suggests that the better the electronic service quality provided by Bukalapak, the more consumers are inclined to make a purchase decision.Sales Promotion has a positive and significant influence on Purchase Decision, with a path coefficient of 0.284.The generated T-value is >1.65, specifically 3.171.The significance P-value obtained is 0.001.The f square value is 0.115, indicating that the influence of Sales Promotion on Purchase Decision falls into the moderate category.This implies that the more attractive the sales promotions offered by Bukalapak, the more consumers are inclined to make a purchase decision.Product Variation has a positive and significant influence on Purchase Decision, with a path coefficient of 0.306.The generated T-value is >1.65, specifically 5.121.The significance P-value obtained is 0.000.The f square value is 0.135, indicating that the influence of Product Variation on Purchase Decision falls into the moderate category.This suggests that the more diverse the product variations offered by Bukalapak, the more consumers are inclined to make a purchase decision.Based on the discussion of the research results, there are several recommendations divided into practical and theoretical suggestions, as follows: Bukalapak must continue to improve the quality of electronic services, including website design, fulfillment, customer service, and security/privacy.This improvement can be achieved by using evaluations and feedback from consumers.Bukalapak can enhance the effectiveness of sales promotions by aligning them more closely with the needs and desires of consumers.The organization of contests, sweepstakes, and coupons can be designed more creatively to increase attractiveness.Bukalapak is advised to continually expand and enrich the variety of products offered.This includes improvements in brand variation, product assortment, product sizes, and product quality.Bukalapak can enhance purchase decisions by providing more comprehensive and accurate product information, as well as facilitating the purchase process and payment methods.

Table 2 . Reability Testing Cronbach's Alpha Composite Reliability Keterangan E-Service Quality
Square For the first criterion, which is R-Square, it evaluates how well each endogenous latent variable explains the dependent variable.There are three categories in R-Square: (1) 0.75 indicates a model with strong predictions; (2) 0.5 indicates a model with moderate predictions, and (3) 0.25 indicates a model with weak predictions.
b. Model Fit In evaluating model fit in Smart-PLS, there are several important criteria to consider: (1) SRMR (Standardized Root Mean Square Residual).The model is considered to have a good fit if the SRMR value is less than 0.10 or 0.08.(

Table 4 . Model Fit
Square is useful for measuring how well the path model can predict the values of the original data.Based on [reference], a Q-Square value greater than 0 indicates that exogenous construct variables have predictive relevance for endogenous construct variables.Therefore, the model can be reused with the same measurement conditions and assumptions.Hypothesis he one-tailed hypothesis test with a significance level of 5% has a critical Z-value of 1.65.The f-square value represents the magnitude of the influence generated by the hypothesis test, and its significance is divided into three categories: (1) 0.02 indicates a small or weak effect; (2) 0.15 indicates a moderate or moderate effect, and (3) 0.35 indicates a large or strong effect. ISSN: