ANTECEDENTS OF E-LOYALTY AND ITS IMPACT TO ONLINE REPURCHASE INTENTION

Research aims to examined further affect from application design, customer services, fulfillment, responsiveness, compensation, contact, e-service quality, eWOM, e-recovery on e-loyalty and its impact to online repurchase intention. Data analysis method used PLS-SEM through total of 164 respondents Sayurbox application's users. The finding showed from nine antecedents, likewise from e-loyalty mediating variable, eight variables had significant affect to online repurchase intention and not included contact variable which appears not significant. The strongest affect came from responsiveness and application design. Meanwhile, the direct affect on e-loyalty revealed to be the strongest by e-recovery . customers compensation, contact, quality, e-WOM, e-recovery, pada e-loyalty online repurchase intention. Metode analisis


INTRODUCTION
In 2020, WHO announced that there is an corona virus  outbreak. This trigger to various new policies that Indonesian government need to take to precoutious relates to widen impact from the corona virus in changing people's act to carry out various activities from home, amongst them is shopping for food daily needs through online (e-grocery). With the existence of an e-grocery, people would get various kinds easily when obtaining and fulfilling the food ingredients needed during these current pandemic.
With this kindly situation, the start-ups development, especially e-grocery in Indonesia is grow up day by day. Sayurbox is an e-grocery platform which serves to sell fruits and vegetables online with a farm-to-table concept. During these pandemic, Sayurbox business has continued to increase, even 5 times higher than before the pandemic occured (Novika, 2020). With these increase, Sayurbox must be able to fill their consumer needs in order to increase eloyalty and online repurchase intention.
Online repurchase intention is a consumer's desire to repeat visits in the near future (Kotler & Keller, 2018). Repurchase act is often associated with loyalty. However, the two are different, repurchase behavior involves re-purchasing the same particular brand over and over again, while loyalty reflects as an psychological commitment to a particular brand (Tjiptono, 2012in Hidayah & Aniarwati, 2018. Online repurchase intention could be defined as a purchase application. In the process of collecting data for this research object, the primary data used through a questionnaire with a Likert scale of 1-5 which was distributed by online via google form. AD.1 Sayurbox application displays a visually attractive design.

AD.2
The information I need to transact on the Sayurbox Application is quite complete. AD.3 Sayurbox application informs you of discounts or free shipping. AD.4 The Sayurbox application is easy to use in transactions.
Customer Service (CS)

CS.1
Customer Service on the Sayurbox Application handles complaints quickly.
CS.2 Customer Service on the Sayurbox Application handles product returns with well. CS.3 Service reliable by customer service.

CS.4
Customer Service on the Sayurbox App offers the ability to talk to customers directly if there is a problem occurs.

CS.5
The Sayurbox application provides a telephone number to contact the company.
The product was delivered acoording to the time that agreed by Sayurbox. FL.2 The product ordered arrived in good condition. FL.3 Sayurbox application sends the goods ordered by the customer. FL.4 The offers made to customers is real. FL.5 Sayurbox application is available for delivery within the desired timeframe.

Responsiveness (RS)
RS.1 Sayurbox Apps will provide an adequate alternative for product returns.

RS.2
Sayurbox will provide information if the transaction is not successful.

RS.3
Sayurbox will address potential problems quickly. RS.4 Sayurbox will offer a meaningful warranty.

Compensation (CP)
CP.1 Sayurbox will provide compensation if the goods ordered not delivered on time.
CP.2 Sayurbox will provide compensation for problems that occur in the Application.
CP.3 Sayurbox will pick up the product If the customer wants to return from home or office.

Contact (CT)
CT.1 Sayurbox provides a telephone number that is easy to contact. CT.2 Sayurbox has available online service staff.
CT.3 Sayurbox provides contact phone numbers, whatsapp/chatbot, email that can be contacted.
When customer have problems with order on Sayurbox app, customer service department would be please to help.

ESQ.2
The overall quality of the e-grocery service on the Sayurbox application is very good. ESQ.3 E-grocery service at Sayurbox application should be provided as promised.

ESQ.4
The application provides booking confirmation information via email or my phone. ORP.2 I intend to extend the e-grocery in the Sayurbox App rather than discontinue its use.
ORP.3 I intend to continue doing e-grocery in the Sayurbox App rather than using any alternative means.
The research population was all people who have made transactions using the Sayurbox application in areas that include Sayurbox services (Jakarta, Bandung, and Surabaya). The sample selection method used in this research is non-probability sampling with purposive sampling technique based on the author's considerations. Because the number of customer data cannot be ascertained, so this research used the Lemeshow formula. Through the calculation of the Lemeshow proportional formula, a sample from 96 respondents was obtained. However, based on the minimal sample approach for the SEM-PLS method, the minimum sample size required is 160 respondents (Kock & Hadaya, 2018), so based on consideration, the sample that meets the requirements and kindly used in this research amounted to 164 respondents.
This research used descriptive statistical analysis and inferential statistical analysis. Inferential analysis is a statistical analysis method which processes data from samples then would be represnted and generalized to the entire population, in order to draw adequate conclusions and approach the population, a sampling technique and an adequate number of samples are needed (Hair et al., 2014). Hypothesis and correlation between variables would be analyzed through SEM-PLS modeling which produces a structural model and measurement model.

RESULTS AND DISCUSSION
The outer model test is first test in SEM-PLS modeling which is intended to discover the validity and reliability of the data arranged in 4 parts of the test, namely indicator reliability, construct reliability, construct validity and discriminant validity. Based on the reliability indicator, 33 of the 44 reflective indicators used in the research survey were obtained, so that 11 indicators who did not meet the requirements were removed from the model and a second test need to conducted. In the second test, it was found that 33 indicators from the research variables in the framework had an outer loading value above 0.708, except for the CP.1 indicator of the compensation variable related to the process of giving compensation to customers if there was a problem in the delivery of orders that is not on time, where this value as the required limit (Hair et al., 2014).

Figure 1. Outer Research Framework Model
In the construct reliability test, it was found that all variables had cronbach's alpha values above 0.7 and composite reliability values between 0.7 to 0.95 as required (Hair et al., 2014). In the construct validity test, all the research variables have an AVE value > 0.50 as required (Hair et al., 2014). Further validity test is discriminant validity with the Fornell-Lacker Criterion. The outcomes showed that the value of each construct is greater than the correlation between one construct to another in the model.  Based on these four parameters of the reliability and validity test results on the outer model as above, namely indicator reliability (outer loading), construct reliability, construct validity and discriminant validity, a general conclusion that can be conveyed from this measurement model is all indicators are reliable and valid to measure the construct. each specifically and deserves to be continued in the next analysis stage, namely the inner model test.
In the inner model analysis, one-tailed hypothesis test was carried out by the re-sample or bootstrapping method. Before examine the hypothesis, this research looks at the model quality parameters used in the inner model through VIF, R 2 , f-Square, Q 2 , Q 2 -Predict. Subsequently, the significance test is to determine whether the hypothesis can be supported and show the path analysis through the specific indirect affects test results. At the end, an importance-performance analysis is added based on IPMA data that uses the total affect value on the target construct to provide manager input in setting priorities (Hair et al., 2014).
In the multicolinearity test, it was found that all variables had a value of less than 3 or then what required (Hair et al., 2014), so that the inner VIF value was ideal and there were no multicollinearity problems. In the coefficient of determination test, R 2 value for the e-loyalty variable was 0.690, erecovery was 0.492, e-service quality was 0.421 and online repurchase intention was 0.455, so it can be said that the variables in this research framework can predict accurately moderate to dependent variable in this research framework (Hair et al., 2014). In the f-square test, it was found that from those independent variables, e-recovery and e-loyalty have a significant affect with a value above 0.15, so it is said that e-recovery and eloyalty have a substantial affect on online repurchase intention and necessary to consider in increasing the repeat purchases by customers on the Sayurbox application. In the Q 2 test, it was found that the online repurchase intention, e-service quality, and erecovery variables have moderate relevance prediction abilities with Q 2 values of 0.335, 0.349, and 0.355, while the variable with the highest Q 2 value is the e-loyalty variable of 0.535. Then, in the Q 2 -Predict test, it is known that the online repurchase intention variable has a Q 2 -Predict value that is greater than the Q 2 value and is close to large predictive relevance, so that the research framework that uses three independent variables and three mediating variables is said to be sufficient to be applied to research. next about customer loyalty in the Sayurbox application in increasing repeat purchases. An overview from the hypothesis test from this research framework can be seen in the  According to the significance test, it tells that from this ten hypothesis in the research framework which have been tested, nine hypothesis revealed to have significant results with positive coefficient values based on the direction on the hypothesis proposed. However, for H6 with the path of the influence from the contact to e-recovery, the results were not significant so the hypothesis did not supported.
Application design is proven to have a positive and significant affect on e-service quality with a T-statistic value of 4.781 > 1.645 (t-table) and a standardized coefficient of 0.315. This result has confirmed the prior research by Rita et al. (2019), and Blut (2016) who defined that online services for online shopping with diverse populations. The findings which mentioned above can provide managerial implications in providing services for Sayurbox customers, when it is necessary to prioritize an application design process which attractive and ease to use by customers. This is important because customers who expect the ease and aesthetics of the beauty of the application will attract and loyal in using the Sayurbox application in doing e-grocery thereby will increase repeating purchases. Therefore, the Apps design should be emphasize on its advantage with aesthetic design which reflects on a strong and associative image of the brand, so as to attract customers to visits (Blut, 2016;Díaz & Koutra, 2013).
Customer services is proven have a significant positive affect to e-service quality with a T-statistic value of 2.666 > 1.645 and a standardized coefficient of 0.176. This result is confirmed with prior research from Blut (2016). The findings above could provide managerial implications that served good service quality for customers, companies need to be more attentive to the capacity and ability of customer services in handling policy returns during and after sales (Blut, 2016). In online business, customers sometimes carry out the entire buying process themselves without help of customer service (McLean & Wilson, 2016). Some of online businesses opened online customer service which allows customers to request further detailed information about the product they wish to purchase. Those Companies usually use web-based synchronous media such as live chat facilities, online help desks, and social networking sites (Turel & Connelly, 2013). According to Blut (2016), customer service may contribute to the quality of eservices. Therefore, it is necessary to prioritize customer service processes that are able to provide fast, precise, friendly, and informative service to customers therefore the customers can be loyal in doing e-grocery at Sayurbox application so as to increase the repeat purchases.
Fulfillment is proven have a significant positive affect to e-service quality with a Tstatistic value of 3.604 > 1.645 through a standardized coefficient of 0.283. This finding is in line with prior research by Rita et al. (2019), and Blut (2016). The result above can provide managerial implications in serving guidelines to customers Sayurbox which needs to prioritize the process fulfillment or product fulfillment properly, on time, and based on customer orders. This is important because customers need to make sure the transactions they make and fulfilled properly and according to their expectations (Ataburo et al., 2017), so that customers would feel fast in transactions in the Sayurbox application and create customer loyalty which later expected to increase repeat purchases.
Responsiveness is proven have a significant positive affect to e-recovery with a T-statistic value of 4.444 > 1.645 and a standardized coefficient of 0.376. This finding is in line with the research by Shafiee & Bazargan (2018), Ataburo et al. (2017), and (Bernardo et al., 2013) on online services. The result above can provide managerial implications in serving recovery services for Sayurbox customers, it is necessary to prioritize the fast of response to complaints and problems which faced by customers, thus customers feel valued and will not disappointed in doing transaction at the Sayurbox application. This will increase customer confidence in sustainable purchases at Sayurbox application (Shafiee & Bazargan, 2018). This is important because customers need to ensure that the aftersales process will not disappoint and troublesome to customers and it is hoped that customers will feel fast in transacting at the Sayurbox application and create customer loyalty which is later expected to increase repeat purchases.
Compensation is proven have a significant positive affect to e-recovery with a T-statistic value of 3.803 > 1.645 and a standardized coefficient of 0.298. This result is in line with Shafiee & Bazargan (2018), Ataburo et al. (2017), and (Bernardo et al., 2013). The findings above can provide managerial implications which served recovery services for Sayurbox customers, compensation is needed for problems faced by customers. Compensation, guarantee or assurance are needed for customers so that trust is formed in transactions (Ataburo et al., 2017). Customers feel safe and believe that they will get the best when doing e-grocery at Sayurbox application. Compensation can be in the form of refunds, exchange of damaged goods, loyalty points or also could be provide as e-voucher if the product did not match what was ordered. This is important because customers need to ensure the aftersales process will not disappoint and not troublesome for customers if problems occur and it is hoped that customers will feel fast in transacting at Sayurbox application and formed customer loyalty which is later expected to increase repeat purchases.
Contact is not proven to have an affect on e-recovery, because the T-statistic value is 1.356 < 1.645. This result is in line with the research by Shafiee & Bazargan (2018), Ataburo et al. (2017), and (Bernardo et al., 2013) who urges the opposite. This explains that the availability level of telephone or online assistance at the Sayurbox application did not have an impact to the erecovery process and still needs to be considered, especially on the availability of telephone numbers in the help feature of the Sayurbox application. This findings can also provide managerial implications which served in recovery services for Sayurbox customers, it is necessary to increase telephone or online contact information which is effective to customers when finding something to order so that customers will feel that Sayurbox application provides complete information compared to others tho customers have characteristics and profiles which differents. Through this contact line or those online customer service, it is hoped that all the issue that customer faced would be solved specifically (Shafiee & Bazargan, 2018).
E-services quality is proven have a significant positive affect to e-loyalty with a T-statistic value of 4.315 > 1.645 and a standardized coefficient of 0.260. This result is in line with prio research by Shafiee & Bazargan (2018), and Ataburo et al. (2017). These research findings can provide managerial implications which served online services for Sayurbox customers, it is necessary to have good quality and satisfying e-services for customers, such as providing an attractive and simply to use application display, having a customer service department, reliable security, and fulfilling all requirements. customer orders properly and on time (Shafiee & Bazargan, 2018). Thus customers would feel satisfied in transacting at Sayurbox application and creating customer loyalty which leads to increase repeat purchases. e-WOM proved to have a significant positive affect on e-loyalty with a T-statistic value of 4.171 > 1.645 and a standardized coefficient of 0.282. This result is in line with prior research by Perera et al. (2019), and (Khan & Azam, 2016) on the impact of e-loyalty on social media perspective. However, what is interesting from the respondent's profile is that there are as many as 63% of respondents who do not follow Sayurbox's social media accounts. This could be seen that most of the Sayurbox customers were not affected by UGC (user generated content) but through FGC (firm generated content) (Baker et al., 2016). That interpreted above can provide managerial implications which served online services for Sayurbox customers, it is necessary to have a good e-WOM that is conveyed well to others through official social media from Sayurbox, therefore it can impact to other people in transacting at the Sayurbox application and become more confident about its quality. Thus, customers will feel satisfied in transacting at Sayurbox application and formed customer loyalty which leads to increase repeating purchases. E-recovery proved to have a significant positive affect on e-loyalty with a T-statistic value of 9.881 > 1.645 and a standardized coefficient of 0.447. This result is in line with prior research by Shafiee & Bazargan (2018), and Ataburo et al. (2017). These findings above can provide managerial implications in serving recovery services for Sayurbox customers, it is necessary to have adequate e-recovery services and it is hoped that consumers will reduce disappointment from consumers. E-recovery can be in the form of a fast response recovery service, then compensation for problems faced by customers, as well as the availability of calls or online contacts that are easily accessible by customers in case of problems or obstacles (Shafiee & Bazargan, 2018). This can make customers feel safe and believe that they will get the best treatments when doing e-grocery at the Sayurbox application. It is important because customers need to ensure that the aftersales process will not go down and troublesome for customers if problems occur so as customers will feel quick in transacting at the Sayurbox application and formed customer loyalty that leads to increase repeat purchases.
E-loyalty is proven to have a significant positive affect to online repurchase intention with a T-statistic value of 17,257 > 1,645 and a standardized coefficient of 0.674 which is the largest value among all paths in the research framework. This result is in line with prio research by Shafiee & Bazargan (2018). The findings above can provide managerial implications which providing services so that customers are loyal to the Sayurbox application, it is necessary to have e-loyalty in the form of providing good quality services, honest and interesting content information, as well as the existence of recovery services for aftersales that are match to expectations. This could make customers feel safe and believe that they will get the best when doing e-grocery at the Sayurbox application. This is important because customers need to ensure that the transaction process and aftersales will succeed and will not burden the customers if problems occur and it is expected that customers will feel fast in transacting at the Sayurbox application and create customer loyalty (Kotler & Keller, 2018), which leads to increase repeat purchases.
In the Importance Performance Map Analysis (IPMA), it can be seen that in the upper right quadrant there is an e-loyalty variable, thus this variable has shown good performance and considered important by consumers in this case are Sayurbox customers who have made transactions in the application. Elicited from this, it is advised for Sayurbox managers to pay attention to the factors which affect customer loyalty in terms of service quality and recovery services for the obstacles experienced by respondents. On these research, the most important thing is to aware in improving customer loyalty, namely by improving recovery services for customers who experience aftersales problems when transacting at the Sayurbox application, especially in the responsiveness process. Sayurbox managers are advised to allocate resources and budget in an effort to competence and improve the quality of the recovery service. Things that can be done for example by recruiting officers to handle complaint or forming a department who can solved the problem faced by customers quickly, clearly, and can communicate by providing solutions to the problems that they report. The company need to create the SOP, and a SLA for each category of problems faced by customers, so that department resolution can provide answers quickly and according to company regulations. The management of human resources needs to followed this programme by increasing communication training to improve skills in service delivery at the Sayurbox application. Furthermore, it is necessary to improve the performance of applications which are more sophisticated, attractive, and easy to use by customers, add features needed by customers such as the refund and return feature or the complaint feature, then the customers do not need to look for Line contacts or contact information to be able to convey their problems.

CONCLUSIONS
Based upon the empirical tests that have been carried out, it is finalized that: 1) Application design, customer service, and fulfillment were validated to have a positive and significant affect on e-service quality; 2) Responsiveness and compensations were certified to have a significant positive affect on e-recovery, while contact did not proven; 3) E-service quality, e-WOM, and e-recovery are certified to have a significant positive affect on e-loyalty; 4) E-loyalty is proven have a significant affect on online repurchase intention. Through this research outcomes, it is hoped that: 1) Sayurbox management needs to maintain their quality of application, variety of product quality and maintain communication with customers through Sayurbox's social media accounts. In addition, the availability of couriers in the morning should be considered, Sayurbox need to cooperate with third-party logistics in the delivery process because most customers choose the delivery time in early morning; 2) The company provides a loyalty program by means of loyalty points that can be obtained if the customer transaction is successful and the points earned can be used as a means of payment on the Sayurbox application or exchanged for prizes such as e-vouchers or free shipping. In addition, the management needs to monitor loyalty by conducting periodic and ongoing surveys regarding the services provided so that management can focus on maximizing in services according to customer expectations and providing good service quality, providing information on interacctive social media, as well as fast and effective recovery services. clear; 3) The company forms a department that handles complaints handling or resolution departments by implementing SOPs and SLAs to respond to the problems faced by customers quickly, clearly, and can communicate by providing solutions to the problems that they reported, therefore customers feel satisfied and trust to do transact in the Sayurbox application; 4) By the research limitation, it is recommended that further research need to examine particularly relates to the affect of the refund and return feature, delivery time and courier selection, as well as access to the most important contacts for customers, so that they can find more precise results on customer loyalty which leads to increase the repeat purchases by customers.