The rapid development of technology and information has caused various impacts on human activities. The effect is felt on multiple dimensions of life such as economic, political, social, and cultural. This development is motivated by human attitudes toward meeting their needs. In addition, this rapid development will have positive and negative impacts depending on how to respond to it. The emergence of interconnected networks, commonly called the internet, is a consequence of rapid technological advances. Along with the times, the internet was initially only used for military purposes, but now the internet can be used by all levels of society. The practicality and sophistication offered by the internet are added values ​​that trigger people to access the internet more often. The internet provides various kinds of information that people need. It is predicted that internet users will increase over time. As evidenced by statistical data, predictions of internet users in Indonesia in 2017-2023 are increasing.

The internet also plays a role in the field of electronic commerce or electronic commerce (e-commerce) into a new era in commerce. The e-commerce business sector in Indonesia is predicted to grow along with companies that are starting to change and develop their business in digitalization. This creates a situation where the public or consumers find shopping more accessible by using gadgets from home. In this situation, sellers must also follow technological developments related to selling their wares. They inevitably have to follow results that occur by choosing one of the e-commerce as a means of selling their wares. Likewise, consumers must be intelligent and wise in choosing e-commerce as a place to buy. There are many e-commerce in Indonesia, all of which can be the choice of traders (sellers) and buyers (buyers) in buying and selling transactions. The e-commerce names include Shopee, Tokopedia, Bukalapak, Lazada, Blibli, Orami, Bhinneka, Ralali, JD.ID, and Sociolla. The result can be seen from the average number of e-commerce visitors per month in the fourth quarter of 2020, which is presented in Table 1.

Based on research results from the iPrice website, it was stated that the number of Shopee visitors in the fourth quarter of 2020 was first in the marketplace, with the highest number of website visitors in Indonesia.

Table 1: Data on E-commerce Visitors in Indonesia

Source: Iprice, 10 February 2021

The average monthly visit to Shopee last quarter of last year was recorded at 129.3 million visitors. iPrice data informs that this marketplace received 71.5 million visits in the first quarter of 2020. The second place is Tokopedia, which in the fourth quarter of 2020 had an average number of visitors per month of 114.67 million. The third place is Bukalapak, with an average monthly number of 38.58 million visitors. Successively from big to small then are Bukalapak, Lazada, Blibli, Orami, Bhinneka, Ralali, JD.ID, and Sociolla. Because of Shopee's position as the market leader in this e-commerce business, Shopee was chosen as the object of this research. This is also supported by the existence of Shopee, which has grown and offers various needs of the community. Various types of products are offered by Shopee, including beauty products, data packages, food, fashion, household appliances, and electronic equipment payment of electricity bills can be done through the Shopee application. In addition, Shopee also offers a payment system through a joint account that can be mediated by credit card, instant debit, shopeepay or shopeepay later. The payment system using a joint account is claimed to be safer because Shopee is the third party for transactions between sellers and buyers [1].

Of course, by buying and selling e-commerce through Shopee, sellers and buyers do not meet in person. Buyers only see display images and descriptions of goods. In this case, of course, there can be a risk that is detrimental to the buyer; the goods sent by the store do not match what the buyer sees. In this case, many things cause buyers to decide whether or not to buy goods. These factors include experience, service, rating, other customer reviews and others. Therefore, in addition to selling various types of products, Shopee also provides a review column facility (online customer review) and an online customer rating column that provides space for consumers to provide reviews and comments regarding the quality of products and services to sellers. This review and rating column also serves to assist customers in determining and assigning goods that meet customer expectations. The existence of a review column and a rating column is beneficial for other potential consumers to know the quality of the product and the quality of service provided by the seller honestly.

The review and rating columns function to minimize buyers' risks. There are several risks encountered by online customers described by Turben et al. The first risk is the incompatibility of the original item with the image on the online display. This is common because the display image has been engineered in such a way to attract consumer demand. The second risk is the damage to the product received. Product damage can occur due to shipping or the goods sent have experienced a defect from the manufacturer. The third risk is the seller's error during data collection and product packaging, where there is often a mismatch of goods, sizes, colours and types. The fourth risk is a delay in delivery which usually occurs due to the negligence of the seller or courier who is not observant in data collection and package grouping. The last risk is the emergence of fraud or fraudulent actions. Constraints encountered from online shopping activities today include potential consumers being faced with various risks, one of which is not having access to a direct assessment of whether the goods to be purchased following the qualifications and expectations expected by customers. This research is based on previous research, as shown in the following Table 2:

Based on the table above, it can be seen that many researchers conducted research on various variables with different results. For this reason, researchers will discuss how to improve Purchase Decisions in the Shopee marketplace.

Literature Review

Definition of Online Customer Review: Online Customer Reviews (OCRs) are a form of word of mouth in online sales. The article is that prospective buyers get product information from consumers who have felt the product's benefits. Therefore, customers will find it easier to obtain information about the advantages and disadvantages of the goods they want. In addition, consumers are easier to compare with similar products sold by other sellers because of the rapid use of digital marketing so that it can provide benefits for other consumers, namely, consumers do not need to visit different sellers directly, and consumers no longer need to visit the store directly, do not need a long time to get information about the goods they want to buy.

Online customer reviews work the same as word of mouth, but there are two differences between online customer reviews and word of mouth. First, the influence of traditional word of mouth has boundaries between local social networks. The impact of online customer reviews can extend beyond the local community, as consumers across the country can access customer reviews online via the Internet. Online customer reviews are relevant when paired with electronic word of mouth. Second, traditional word of mouth is not a supplier decision variable. Online customer reviews can be decided by the seller, when and whether the seller will provide them to consumers on their website. Shopee sellers can give consumers the option to choose to provide product reviews. Some studies claim that what drives consumers to look for customer reviews online is conflicting information, lack of information, increased awareness, loyalty behaviour, product features, and discount offers on products. Shopping sites enhance consumers' ability to make better purchasing decisions [5].

Table 2: GAP Research Online Customer Review and Online Customer Rating

Source: Various journal articles

Figure 1: Relationship between Variables

Online Customer Rating

Rating or often considered as customer views on a specific scale in the popular rating system on Shopee, namely by assigning a star image. The more stars are given, the better the seller's rating. Generally, the rating is a vehicle for conveying the feedback given by the seller. This rating is another opinion given by many people or as an average rating of buyers and buyer reviews on the characteristics of different goods or products or those provided by the seller's services. The rating can be interpreted as an assessment of the preferences of users of a product for their experience, referring to the emotional state they experience when interacting with dream products in a high-profile environment and being a representation of consumer opinions using an exclusive scale. The impact of customer reviews on reviews before deciding to buy something depends on how often customers review a product. Product reviews by consumers reflect consumer satisfaction in using the product and how online sellers serve consumers. The number of stars can be related to the quality of an item sold online.

Definition of Purchase Decision

The buying decision process is a psychological process for consumers or buyers, starting from the phase of paying attention to goods or services, which, if impressed, will enter the phase of interest (interest) to study further the characteristics of the product or service. If the intensity of interest is vigorous, it continues until the phase of wanting/interested (desire) because the goods or products are in sync with their needs. Then buy (action to buy) the goods or services offered. The purchase decision is one of the steps in the buying process prior to post-purchase behaviour. Consumers are faced with several other choices when entering the last purchase decision term. As a result, consumers will act to decide to buy products following the choices that have been determined. Below are some definitions of purchasing decisions based on experts. 

Buchari Alma in suggests that purchasing decisions are as follows: "Purchase decisions are consumer decisions that are influenced by the economy, finance, technology, politics, culture, product, price, location, promotion, physique, people and processes. The behaviour of consumers is formed to conclude in the form of responses to which products to buy. Based on Kotler and Armstrong define purchasing decisions as follows: consumer behaviour is the study of how individuals, groups and organizations choose, buy, use and dispose of goods, services, new ideas or experiences to meet their needs and wants. This means that purchasing decisions are part of consumer attitudes, namely the study of how individuals, groups and organizations determine, buy, use and how goods, services, ideas or experiences satisfy their needs and desires.

Trust

According to Crosby [6], consumer trust is the belief that the supplier of a product or service can accept the mandate to behave in a way that satisfies the long-term interests of consumers. Trust is defined as the fundamental originating training and retention relationship between customers and online sellers. Gefen concluded that if the online seller's information is clear, accurate and complete, customers will believe that this online store is not dealing with mere opportunities, which will increase customer confidence in the service. Trust refers to Robbins' theory, including:

• Integrity: Truth- fullness

• Competence: Insight and technical skills as well as personal standards

• Consistency: Reliable, predictable and well considered at handling situations

• Loyalty: Willingness to protect and maintain good relationships

• Openness: Willingness to share opinions using freely

Conseptual Framework

Based on the variables described above, namely online customer reviews, online customer ratings, purchasing decisions and e-trust. The framework of thinking related to the relationship between research variables can be visualized as shown in the image below.

Online Customer Review dan E-Trust

Online customer reviews can be interpreted as an electronic form of word of mouth. Electronic word of mouth is a statement made by potential customers, current product consumers, and past consumers about a product, service, or business that has been made available to anyone, users and institutions via the Internet. It is easily accessible with these advances, and users are smarter in making transactions. Most consumers or buyers who buy and sell site users often consult previous customer reviews to increase their trust in the targeted product. In previous research, Fikri et al. stated that customer reviews positively affect customer reliability. Based on this description, the following hypotheses were obtained:

• H1: Online customer reviews have a positive effect on trust in the Shopee buying and selling site

Online Customer Rating and Trust

Consumers are faced with several choices when entering the last purchase decision term. As a result, consumers will decide to buy products according to the choices that have been determined. Based on Buchari, Alma suggests that purchasing decisions are as follows: "Purchase decisions are consumer decisions that are determined by the financial economy, technology, politics, culture, product, price, location, promotion. The online customer rating is not much different from an online customer review, where an online customer rating is an assessment using symbols or numbers to determine how good or bad a product is. Previous customers usually give ratings, and ratings are also given according to the actual and emotional state of the product the customer. In a previous study, Sarmis et al. stated that Rating positively influences customer e-trust. Rating is also a reference for customers or customers before making a transaction for a product on the shopee buying and selling site. Based on this description, the following hypotheses were obtained:

• H2: Online customer rating has a positive effect on trust in the Shopee buying and selling site

Online Customer Review and Purchase Decision

Customer purchasing decisions are often associated with online customer reviews from previous customers. The reason is that most customers use reviews to make transaction decisions. Consumers are faced with several choices when entering the last purchase decision term. As a result, consumers will decide to buy products according to the choices that have been determined. According to Meidiyansayh suggests that purchasing decisions are as follows: "Purchase decisions are consumer decisions that are determined by the financial economy, technology, politics, culture, products, prices, locations, promotions, from physical evidence, people and process. What to buy." Based on previous research, HT Hariyanto, L Trisunarno et al. stated that online customer reviews affect the customer's physical evidence in making decisions to buy or make transactions. Based on this description, the following hypotheses were obtained:

• H3: Online customer reviews have a positive effect on purchasing decisions on the Shopee buying and selling site

Online Customer Rating and Purchase Decision

Rating is a user opinion given to the seller to show the quality of the products and services owned by the seller. The more stars are given, the more it shows a good rating. In this digitalization period, there are many developments, especially in the online buying and selling system, for example, the shopee buying and selling site, where shopee provides a rating feature for customers to give a rating or assessment of the product purchased freely and the rating becomes a reference for other customers to decide to purchase the product. In a previous study, Wahyudi, Rinuastuti, Sarmo et al. [7] stated that online customer ratings positively influence other customers' reparation decisions. Based on this description, the following hypotheses were obtained:

• H4: Online customer rating positively affects purchasing decisions on the Shopee buying and selling site

Trust and Purchase Decision

Trust is the beginning of the relationship creation and maintenance between customers and online sellers. The customer's trust will be realised when the seller provides accurate information following reality. This trust is the basis for customers to purchase the desired product. Gefen concluded that if the information provided by online sellers is clear, correct and complete, customers will believe that this online store is not only dealing with opportunities but will increase customer confidence in the online services provided by the seller. In previous research, Yuniati et al. stated that the development of the internet has an impact on business, where sellers are required to be able to provide information that attracts customer trust in order to increase sales of a product. Based on this description, the following hypotheses were obtained:

• H5: Trust positively affects purchasing decisions on the Shopee buying and selling site

Population and Sample 

Population: A population is a group of people, animals, plants or objects with specific characteristics to be studied. At the same time, Sugiyono (in Kamal, 2017) argues that the population is a generalization area consisting of objects/subjects with unique qualities and characteristics determined by the researcher. To be studied, and therefore conclusions will be drawn. It can be interpreted that the population is not only people but also objects and other natural objects. The population of this study consisted of Shopee users aged 18 to 55 years.

Sampling Technique

The sample is part of the number and characteristics possessed by the population, Sugiyono [8]. This study uses a purposive sampling technique. Information sources are determined using specific objectives and considerations. Thus the determination of respondents is based on the goals set. The sampling criteria are as follows:

• Respondents are male and female

• 18-55 years old

• Is a user of the Shopee buying and selling site

According to Amirullah, the sample is 10 times or more than the number of indicators. This research has 12 indicators, 20.14 x 12 = 245 respondents. Roscoe also added that the sample should range from 30-500, so the researcher decided to take a sample of 245 respondents.

Type and Data Source

Primary Data: Primary data is obtained directly by researcher Sugiyono. The data in this study is information obtained directly from the source. In the form of a questionnaire filled out directly by the respondent, the respondent is 18-55 years old and is a Shopee buying and selling system user.

Secondary Data

While secondary data means that the origin is not personally conveying data to data collectors Sugiyono. In this study, the data used is data from relevant journals, books and websites.

Data Collecting Method

The questionnaire is a method of collecting information by giving respondents a set of questions or written statements to answer in [8]. Opinions on the questionnaire are based on the Likert scale of the questions given to the informants. According to Sugiyono, in [8]. The Likert scale is used to look at the behaviour, comments, and thoughts of people or organizations in social areas; these social areas are approved explicitly by researchers, after which they are said to be research variables. The Likert scale includes:

• For the answer "STS" strongly disagree given a value = 

• For the answer "TS" disagree given a value = 2

• For the answer "N" neutral is given a value = 3

• For the answer "S" agree to be given a value = 4

• For the answer "SS" strongly agree given a value = 5

The ordinal scale above is stated in the Vulnerable Scale (RS) as follows:

The above sum is as follows:

RS= (5−1) x 5 = 0.8

The standardization of the five categories are:

1.00 - 1.08            :               Very low

1.81 - 2.60            :               Low

2.61 - 3.40            :               Enough

3.41 - 4.20            :               Height

4.21 - 5.00            :               Very high

Research Type

Qualitative Analysis: Qualitative analysis is an analysis in the form of an explanation of the questionnaire distributed to respondents. As a first step, researchers must analyze the results or respondents' responses to the measured variables. Based on these responses, conclusions were obtained based on respondents' responses.

Quantitative Analysis

Based on Sugiyono, quantitative research is a research method based on the philosophy of positivism, used to examine specific populations or samples, data collection using research instruments, quantitative or statistical data analysis, using the aim to test predetermined hypotheses.

Data Analysis Method

The data analysis technique is a systematic method used to present and analyze information. In line with the expected example in this research dispute, the data analysis tool used is SEM (Structural Equation Modeling), which is operated using the Smart PLS program. SEM is a model used to carry out or test relatively tricky research. Partial Least Square (PLS) is a type of SEM analysis with a component base using formative constructs. PLS-SEM aims to spread hypotheses or construct hypotheses using prediction orientation [9]. PLS can be used to describe the relationship between variables or claim to use prediction-oriented techniques. Moreover, PLS is very helpful for anticipating the dependent variable by including many independent variables. The PLS-SEM test consists of 2 submodels: the measurement model (Measurement example/Outer model) and the structural example (Structural model/Inner example). The data analysis technique is a systematic method used to present and analyze information. In line with the expected example in this research dispute, the data analysis tool used is SEM (Structural Equation Modeling), which is operated using the Smart PLS program. SEM is a model used to carry out or test relatively tricky research. Partial Least Square (PLS) is a type of SEM analysis with a component base using formative constructs.

Research Object Desription

The data in this study were taken using a questionnaire method distributed online using the Google Form media to Shopee user respondents throughout Indonesia. The questionnaire distribution period started on November 10, 2021, until December 10, 2021, with a population of Shopee users throughout Indonesia. The method used is purposive sampling. The number of respondents who filled out the questionnaire was 245 people. It can be seen that from 245 respondents, 73 people (29.8%) were male, and 172 people (70.2%) were female. Most of the respondents were 18-25 years as many as 217 people (88.57%). While the ages of 26-33 years were 12 people (4.9%), 34-41 years were 8 people (3.27%), 42-49 years were 6 people (2.45%), and 50-55 years were as many as 2 people (0.82%). Meanwhile, in terms of occupation, 43 people (17.55%) are employees, 11 people (4.49%) are entrepreneurs, 10 people (4.08%) are householders, and 170 people (69.39%) are college students, and 11 people are students. (4.49%) have other jobs. Based on 245 respondents who use Shopee, as many as 238 people (97.14%), while the remaining 7 people (2.86%) do not use Shopee. Therefore, in the follow-up analysis using SEM analysis, the data used were 238 people, which were taken from respondents who used Shopee.

Outer Model Testing (Measurement Model)

This research model will be analyzed using the Partial Least Square (PLS) method and assisted by SmartPLS 3.0 software.

Table 3: First Iteration Loading Factor Value 

Source: Data Processed (2022)

Convergent Validity

Convergent Validity can be determined using the Loading Factor value. The definition of a loading factor is a value that shows the correlation between the value of a question item with a construct indicator that measures the construct, and the validity criterion is the loading factor value greater than 0.7. based on the results of data processing using SmartPLS 3.0, the loading factor value is obtained as follows:

Based on the data processing results using smartPLS 3.0, the loading factor value is obtained, as shown in Table 4.6. The results show that most of the loading factor values are above 0.7, which means the indicator of the variable is declared valid. Exceptions for indicators X1.1, X1.4, X2.7 and Z1 have a value less than 0.70, meaning these indicators need to be eliminated or removed from the model. The loading factor value after the four indicators are eliminated as follows:

The table shows that the loading factor value less than 0.7 is on the X1.2 and X1.3 indicators, so both indicators need to be eliminated or deleted. The loading factor value after the two indicators are eliminated as follows:

Based on Table 6, it is shown that if the loading factor value, which is less than 0.7, is on the X1.6 indicator only, then the X1.6 indicator needs to be eliminated or deleted. The loading factor value after the X1.6 indicator is eliminated as follows:

Based on the test results, the table shows that the loading factor value is more than 0.7. This means that the research indicators meet the validity criteria.

Table 4: Second Iteration Loading Factor Value 

Source: Data Processed (2022)

Table 5: Third Iteration Loading Factor Value

Source: Data Processed (2022)

Table 6: Fourth Iteration Loading Factor

Source: Data Processed (2022)

Discriminant Validity

The benchmark for discriminant validity is through the cross-loading value. The value of cross loading shows a correlation between the construct and its indicators and indicators of constructs from other blocks. Discriminant Validity can be determined in a model if the correlation value between constructs is higher than the correlation between other constructs. After processing the data using smart PLS 3.0. The results of cross-loading are as follows:

Table 7: Cross Loading Value

Source: Data Processed (2022)

Table 8: Value of Average Variance Extracted (AVE) and Square Root of AVE

Source: Data Processed (2022)

Based on the results of data processing using smartPLS 3.0, it can be seen in the table that the cross-loading value shows that the correlation value between constructs is higher than the correlation between other constructs. So all constructs can be said to have good discriminant validity.

The subsequent measurement compares the AVE root value with the correlation between constructs. The criterion is that the AVE root value is higher than the correlation between constructs. A measurement model has a better discriminant validity value if the square root value of the AVE in each construct is greater than the correlation between the two constructs in the model. The requirement is the AVE value >0.50. The AVE value and its square root in each construct can be shown in the table below:

The table shows that the AVE value of all constructs is more significant than 0.50, with the smallest value being 0.625 for the E-Trust variable and the most significant being 0.754 for the OCReview variable (X1). This value has met the minimum standard of 0.50. Next is to compare the value of the square root with the correlation between constructs in the model. In this study, the results of the correlation between constructs and the square root value of AVE are shown in the Fornell-Larcker Criterion table as follows: The each construct is greater than the correlation value, so the construct in this research model can be said to have good discriminant validity.

Table 9: Fornell-Larcker Criterion

Source: Data Processed (2022)

Table 10: Composite Reliability Value

Source: Data Processed (2022)

Table 11: R-Square Value

Source: Data Processed (2022)

Table 12: Hasil Path Coefficients

Source: Data Processed (2022)

Composite Reliability

In addition, to the value of convergent and discriminant validity, the measurement of the Outer model can also be done through the composite reliability value. The reliable construct criteria have a composite reliability value of more than 0.7. The Smart PLS output results for the composite reliability value can be shown in the following table:10

Based on the results of the composite reliability calculation shows that the value is above 0.7. It can be interpreted that the value has good reliability and meets the minimum standards.

Inner Model Testing (Structural Model)

After testing the outer model, it is then testing the inner model (structural model). The evaluation of the inner model can be done by looking at the R-Square value on the dependent construct and the T-statistic value based on the path coefficient test. The higher the R-Square value, the better the research model. The path coefficient value shows the significance level in hypothesis testing.

Determination Test or Analysis of Variance (R2)

A determination test or analysis of variance (R2) is a test conducted to determine the effect of the independent variable on the dependent variable. The value of the coefficient of determination can be shown in the table:

The R-square value in Table 4 shows that the OCRview (X1) and OCRating (X2) variables can explain the E-Trust construct variable by 55.2%, and other constructs explain the remaining 44.8% outside those studied in this study. While OCRview (X1) and OCRating (X2) were able to explain the Purchase Decision variable of 58.8%, the remaining 41.2% was explained by other constructs outside those studied in this study.

Hypothesis test

Hypothesis testing is carried out based on the inner (structural) model, which includes R-Square, parameter coefficients, and T-statistics. The values ​​that need to be considered are the significance values ​​between constructs, T-statistics, and p-Values. Testing is done with the help of SmartPLS 3.0 software through Bootstrapping testing. The criteria used in this study were T-statistic > 1.96 with a significance level of p-value 0.05 (5%) and a positive beta coefficient. The results of the research model and hypothesis testing can be shown in the figure:

Figure 2: The Result of Research Model 

The Effect of Online Customer Reviews on Trust

The direct test results on the Online Customer Review variable with the Shopee object showed positive and significant results on the Trust variable. Based on the beta coefficient value of 0.222 and T-statistic of 3.155>1.96 (significance level 5%). Based on these measurement values, both meet the minimum standard, which means that there is a positive influence on the Online Customer Review variable on the Trust variable. Based on the results of respondents' responses, the X1.8 indicator ranks at the top with the highest average value of 4,038, which shows that online reviews on Shopee that provide information about the products they want to buy correctly have a significant impact on consumers. These results follow research conducted by which shows the Online Customer Review variable on trust.

Influence of Online Customer Rating on Trust

The direct test results on the Online Customer Routine variable with the Shopee object showed positive and significant results on the Trust variable. Based on the beta coefficient value of 0.588 and T-statistic of 8.340>1.96 (significance level 5%). Based on these measurement values, both meet the minimum standard, which means that there is a positive influence on the Online Customer Rating variable on the Trust variable. Based on the respondents' responses, the X2.6 indicator ranks at the top with the highest average value of 4,441, indicating that consumers feel comfortable and safe when the intended seller has a high rating. These results follow research conducted by, which shows that the Online Customer Rating variable positively affects trust.

The Effect of Online Customer Reviews on Purchase Decisions

The direct test results on the Online Customer Review variable with the Shopee object showed positive and significant results on the Purchase Decision variable. Based on the beta coefficient value of 0.152 and T-statistic of 2.546>1.96 (significance level 5%). Based on the measurement values, both meet the minimum standard, which means that there is a positive influence on the Online Customer Review variable on purchasing decisions. These results follow research conducted by which shows that the Online Customer Review variable positively affects purchasing decisions.

Influence of Online Customer Rating on Purchase Decision

The direct test results on the Online Customer Rating variable with the Shopee object show positive and significant results for the Purchase Decision variable. Based on the beta coefficient value of 0.274 and T-statistic of 3.330>1.96 (significance level 5%). Based on the measurement values, both meet the minimum standard, which means that there is a positive influence on the Online Customer Rating variable on the Purchase Decision variable. These results follow research conducted by [10] which shows that the Online Customer Rating variable positively influences purchasing decisions.

The Effect of Trust on Purchase Decisions

The direct test results on the Trust variable with the Shopee object showed positive and significant results on the Purchase Decision variable. Based on the beta coefficient value of 0.437 and T-statistic of 5.803>1.96 (significance level 5%). Based on the measurement value, both of them meet the minimum standard, which means that there is a positive influence of the Trust variable on the Purchasing Decision variable. These results follow research conducted by [11] which shows that the Trust variable positively affects the Purchasing Decision variable.

Based on the research results on Shopee involving 245 respondents, conclusions can be used to answer the research. The question is, "What is the influence of Online Customer Reviews, Online Customer Ratings on purchasing decisions mediated by Trust?". Based on these questions, 5 research hypotheses have been formulated, namely: 

• Online Customer Reviews have a positive effect on trust
• Online Customer Ratings have a positive effect on trust
• Online Customer Reviews have a positive effect on purchasing decisions
• Online Customer Ratings have a positive effect on purchase decisions
• Trust has a positive effect on purchasing decisions

Suggestions

The company can use this research as an evaluation material to improve trust in Shopee, which will ultimately improve purchasing decisions. Evaluation to increase trust can be done by paying attention to the review and rating variables. The study's results indicate that online customer reviews positively and significantly affect trust and purchase decisions. Therefore, sellers at Shopee need to prioritize quality in products and services. A little lack can disappoint consumers and will vent dissatisfaction through reviews. Based on respondents' answers, reviews on Shopee are trustworthy and reliable and provide complete information about the product you want to buy, provide accurate information about the product you want to buy, and provide accurate information about the product you want to buy. That is, consumers have used reviews to determine whether they want to buy or not. If there is a bad review, it will reduce consumer confidence in the product. In this case, Shopee needs to increase trust in the company by increasing the credibility of the sellers there.

The online customer rating variable also positively and significantly affects e-trust and purchasing decisions. Giving an online rating can increase the effectiveness of online shopping. Rating ratings are straightforward to understand. Ratings make it easier for consumers to judge a product, consumers can achieve shopping goals efficiently, and consumers feel comfortable and safe when sellers have ratings (ratings) tall. A higher rating indicates a good seller reputation. To anticipate low ratings, sellers must pay attention to service and anticipate bad reviews. It is necessary to make a strategy so that consumers do not give low ratings, such as giving low prices, giving discounts and providing compensation if there is a discrepancy in the goods ordered so that consumers do not give low ratings. Shopee must have the ability to secure transactions, provide the best service for its customers, provide benefits for its customers, provide good faith to provide satisfaction to its customers and meet what is expected by customers. Shopee must always maintain its reputation so that Shopee's existence is increasingly recognized by other parties related to it. 

Shopee is not only used by consumers but also by suppliers and distributors of delivery services in order to able to increase consumer confidence to improve purchasing decisions.

1. Bayu, D.J. and M.A. Ridhoi. “10 e-commerce dengan pengunjung terbesar pada kuartal IV 2020.” Databoks Katadata, 2021.

2. Liu, Yi and Xinlin Tang. “The effects of online trust-building mechanisms on trust and repurchase intentions: An empirical study on eBay.” Information Technology & People, 2018, https://doi.org/10.1108/ITP-10-2016-02 42.

3. Zhu, L. et al. “How online reviews affect purchase intention: A new model based on the stimulus-organism-response (S-O-R) framework.” Aslib Journal of Information Management, 2020.

4. Yang, Y. et al. “The value of reputation in electronic marketplaces: A moderating role of customer experience.” Journal of Research in Interactive Marketing, 2019, https://doi.org/10.1108/JRIM-11-2018-0151.

5. Devedi, Sujatha et al. “A study on parameters of online reviews content that influence consumers’ buying behaviour: An Indian perspective.” 2017.

6. Alsenani, Y.S. et al. “ProTrust: A probabilistic trust framework for volunteer cloud computing.” IEEE Access, vol. 8, 2020, pp. 135059–135074.

7. Wahyudi, T. “Pengaruh online customer review dan online customer rating terhadap e-trust konsumen remaja Kota Mataram pada pembelian produk fashion Shopee online shop.” Jurnal Riset Manajemen, vol. 19, no. 1, 2019, pp. 1–12.

8. Kamal, M.B. “Pengaruh kepemimpinan dan pengawasan terhadap disiplin kerja karyawan pada PT. Perkebunan Nusantara III (Persero).” Jurnal Ilmiah Manajemen dan Bisnis, vol. 1, no. 1, 2017, pp. 1–17.

9. Ghozali, I. and H. Latan. Partial least squares: Konsep, teknik dan aplikasi menggunakan program SmartPLS 3.0 untuk penelitian empiris. 2nd ed., Semarang: Badan Penerbit Universitas Diponegoro, 2015.

10. Serra-Cantallops, A. et al. “The impact of positive emotional experiences on eWOM generation and loyalty.”

11. Jayadi and Mawardi. "Pengaruh perceived benefits, perceived risks dan customer trust terhadap customer loyalty." 2019, pp. 353–357.