The Influence of Website Quality and Social Media Quality on Brand Image, with Brand Awareness and Electronic Word of Mouth Mediating (Case Study at Legend Coffee Yogyakarta)

The purpose of this study is to compare the quality of websites and social media to brand image, using brand awareness and E-WOM as mediation factors. The exogenous factors in this study are website and social media quality; the endogenous variables are brand image; and the mediating variables are brand awareness and E-WOM. The Legend Coffee coffee shop is the focus of this study, with a sample size of 200 respondents. The study's subjects were Legend Coffee consumers. The data analysis approach used in this work is Structural Equation Modeling (SEM) utilizing AMOS 24 analytic tools. Based on the results of this study, it shows that website quality has a positive and significant effect on brand awareness, social media quality has a positive and significant effect on brand awareness, website quality does not have a positive and significant effect on brand image, social media quality has a positive and significant effect on brand image, brand awareness has a positive and significant effect on E-WOM, E-WOM has a positive and significant effect on image brand, as well as brand awareness and E-WOM mediate website quality and social media quality to brand image.

users in Indonesia will reach 167 million, or around 60.4% of the population in Indonesia.We Are Social statistics reveals that the average Indonesian utilizes social media and the internet.As a result, today's firms must be more sophisticated and keep up with contemporary technical breakthroughs in order to compete with their competitors.
In the current digital era, coffee shops, which are one of the business sectors that are currently growing rapidly, must keep up with current technological developments.According to the Indonesian Coffee and Chocolate Entrepreneurs Association (APKCI), the number of coffee shops is always increasing every year.
Coffee shops, also known as cafés, must now comprehend and follow technological changes in order to market their business, therefore increasing brand awareness and improving the company's brand image.Although, many cafes now understand marketing strategies through digital marketing or many use modern marketing, many still do not understand how to determine digital marketing that suits their business, as well as understand the use of digital media and lack knowledge about how to plan, monitoring and maintaining effective use of digital media (Malesev & Cherry, 2021), and lack of access to cutting-edge technologies (Kergroach, 2020).Aside from that, cafés must recognize that having a presence on a website and social media is critical, but how they maintain the quality of both digital channels is even more crucial, as bad management of both digital media has a negative influence.Significant influence on customer views of the company's image, resulting in a decline in consumer purchase interest (Suryani, et al., 2020).
The problem indicated above has grown so common in every café that it has piqued the interest of many academics, some of whom have addressed the issue by investigating the impact of website quality (Wilson, et al., 2019) and social media quality on development.(Suryani, et al., 2020) discuss digital marketing and how to increase brand recognition and corporate image in the eyes of customers.Then, based on the practice carried out by (Renton & Richard, 2019) that in digital marketing strategies, building a solid brand and increasing brand equity is very important in doing business and securing growth in SMEs, there is a lack of studies regarding the relationship between the use of digital marketing tools and brand equity in SMEs.This impacts not just SMEs, but even café enterprises, requiring them to implement digital marketing tactics.
This research focuses on tactics that link theoretical inequalities by performing an empirical investigation on the importance of website and social media quality in enhancing brand image at Legend Coffee in Yogyakarta.Legend Coffee uses digital marketing using social media, namely TikTok, Facebook and Instagram, but on Legend Coffee's social media, this research only examines Legend Coffee's social media, specifically on Instagram, because Legend Coffee's Instagram social media followers have The largest number of followers is 10.1k, although now Legend Coffee has updated its Instagram account because the previous account was hacked by an irresponsible party, with the number of Legend Coffee Instagram followers now being 470 followers, social media can be used and used effectively to increase their brand image, therefore, efforts to identify factors and ways of working that can contribute to improving the brand image at Legend Coffee can be expected to expand their main interests, this can lead to additional job chances (Rahayu, & Day, 2017) and reduce poverty (Nursini, 2020).
The brand image will emerge in customers' brains in two ways.First, brand image can be derived directly from environmental explanations (such as websites and social media) of a certain brand.Second, exposure to websites and social media may indirectly increase brand image by improving consumer brand awareness, which means consumers will exchange information about businesses via websites or social media under the electronic word of mouth (E-WOM) framework, as a result, websites and social media may play an essential role in establishing brand image from two separate perspectives, directly and indirectly, by incorporating brand awareness and E-WOM (Suryani, et al., 2020).
Based on the discussion above, it became the researcher's interest to examine objects in cafes, and in this research we took one of the cafes that the majority of Jogja people are familiar with, namely, Legend Coffee.Legend Coffee itself was founded in 2012, with the concept of games and community café, the games available at Legend Coffee are such as, UNO, PS2, PS3, billiards, soccerboard, carom, dart games and so on, so this attracts the attention of consumers.Legend Coffee also offers several interesting programs, namely meetings, gatherings, there are even birthday and pre-wedding programs, apart from that, this café also has regular events such as gatherings, watching football matches, community workshops, and acoustic evenings every week (Tribunnewswiki.com, 2021).
At Legend Coffee, the majority of Jogja people already know about this café, which could be of interest to researchers in researching the marketing strategies carried out by Legend Coffee.Specifically, how the quality of the website and social media might enhance the Legend Coffee Café's brand image.This study is a straight duplication of earlier research, namely from (Suryani, et al., 2020).The study model is similar to prior research, but the distinction is in the item investigated, Legend Coffee in Yogyakarta, and the subjects, Legend Coffee consumers in Yogyakarta.

THEORETICAL REVIEW Theoretical Foundation
To serve as a foundation for study, a fundamental theoretical notion and technique must be established.The following fundamental theories are applied in: 1.Schematic Theory.
The conceptual foundation for this study is based on schematic theory (Bartlett & Burt, 1933).Schematic theory outlines how customers receive information via learning {(Kocyigit, O., & Ringle, 2011), (Hawkins & Mothersbaugh, 2020), (Halkias, 2015)}.It emphasizes that learning happens through external stimulation and customer communication with their surroundings.At this time, it is reasonable to infer that external stimuli from the digital world originate via two digital channels: websites and social media.
Businesses utilize websites and social media platforms as strategic communication tools to communicate not only the company's brand profile but also the products or services supplied to the target market { (Liao, et al., 2006), (Galati, et al., 2016), (Suryani, et al., 2021a)}.Customers can obtain stimulation regarding products, services, and brands via websites operated by firms and social media platforms (Suryani, et al., 2021), furthermore, both websites and social media platforms, such as Facebook and Instagram, may be considered as social networks, as they allow users to collect information about brands, communicate with one other, and varied brand experiences, including their impressions of other brands, with brand social linkages using E-WOM (Chu & Kim, 2011), so, in this point, there are two paths discussed in schematic theory by offering vital insights into how today's websites and social media might increase the learning process and contribute to boosting brand image in our social context (Suryani, et al., 2021).

Brand Image
Brand image is a consumer's view of a brand as reflected in brand associations stored in their memory (Keller & Swaminathan, 2020).Building a solid brand image necessitates marketing applications that instill powerful, pleasant, and distinctive connections with the brand in the mind.The benefit of the brand itself is that it has its own worth and meaning that can be conveyed to customers through product or service qualities (Keller & Swaminathan, 2020).According to (He, et al., 2013) stated that brand image, a brand is an intangible asset that a company can provide unique value and commitment to consumers by forming a different identification.
As is known, a brand is a symbol of brand attributes, name, package, reputation, and complex advertising style.Then the brand image itself is the overall response of consumers to a brand, which can provide results from customer conclusions from significant information about the brand, so a clear brand image can enable consumers to be able to identify and test products or services, as well as reduce intellectual risks on purchases, confirm with what needs are met and enjoy differentiation and satisfaction.

Electronic Word of Mouth
The internet has created new communication channels that empower both providers and customers, allowing for the exchange of information and opinions from company to consumer as well as consumer to consumer.Electronic wordof-mouth (eWOM) refers to good and negative words made by potential, present, or past consumers about a product or company, which are accessible to many people and institutions via the internet (Hennig-Thurau, et al., 2004).In the explanation in schematic theory (Bartlett & Burt, 1933), E-WOM can occur when customers discuss and express their ideas about a product or company with their social circle via the website or social media they use.Therefore, it is considered that E-WOM plays a vital function in marketing websites and social media to boost brand image.(Suryani, et al., 2021).4. Brand Awareness (Hutter, et al., 2013) stated that brand awareness is the degree to which a brand is seen to have a strong presence in the thoughts of customers.(Sitorus, et al., 2022) said that brand awareness is customers' capacity to recognize or recall a brand based on a certain product category.Brand awareness can also be used by companies as information and a means of providing a deeper understanding of the company's product brand to consumers and consumers.customers to buy a brand that is already known so that consumers and customers will feel safe and avoid various risks from using the company's product brand.Apart from that, brand awareness includes brand recognition and recall performance.Brand recognition refers to the consumer's capacity to corroborate past exposure.Brand recall is the consumer's capacity to remember a brand from memory when given a product category, a need satisfied by the category, or a purchase or usage circumstance as a cue (Keller & Swaminathan, 2020).

Website Quality
Web quality is a very undefined concept for the most part, existing scientific research discusses the meaning of web quality aspects descriptively without describing its main dimensions to provide a tested scale for measuring it (Aladwani & Palvia, 2002).In practice, (Hsu, et al., 2014) verified that certain shopping sites obtain and keep more consumers than others, due to greater website design elements, a user-friendly experience, and a dependable system, resulting in Websites Quality play a vital role in differentiating brands and are an important factor in determining success in a business (Shin, et al., 2013).
Consider the importance of high electronic service quality in online company success through website quality (Noorshella, et al., 2015).According to (Janita & Miranda, 2013) stated in their research conducted on companies serving the business-to-business (B2B) sector, the quality of a website is connected with privacy limitations, efficiency, added value, and information usefulness, meanwhile, research according to (Jeon & Jeong, 2017) the quality of a website's function, personalization, and reputation all contribute to its overall quality.Despite inconsistent findings in previous research regarding dimensions of website quality, this study uses a simple construct that takes into account several key factors of website quality, such as ease of use of the interface, ease of finding products or services, well-organized information, and system availability (Tsao, et al., 2016).

Social Media Quality
According to (Michaelidou, et al. , 2011) Social media offers a new means for people to engage with their social circles, as well as for businesses to communicate and promote their target market products, services, or brands.A lot of prior studies have indicated that social media plays a critical part in corporate success nowadays.Marketing communications using social media may improve brand equity.Brands that are active on social media and succeed in building strong interactions with customers tend to have higher brand equity (Schivinski & Dabrowski, 2016).According to (Panigyrakis, et al., 2020) show that the usage of social media can boost client loyalty to companies.Through continuous and personalized interactions with customers, social media can build stronger relationships between brands and customers, which in turn increases brand affinity.
The significance of social media in business is critical in today's digital age.Social media can be a powerful tool for establishing a brand, raising brand recognition, and engaging with potential customers.Additionally, it is important to realize that social media user trends and preferences are constantly changing (Tajvidi & Karami, 2021).According to (Teo, et al., 2019) a more straightforward social media quality construct, consisting of content clarity, content quality, and content relevance (Helal, et al., 2018)  Source: (Suryani, et al., 2020) Figure 1.Thinking Framework Hyphothesis H1: High-quality websites increase brand recognition.H2: Social media quality positively impacts brand awareness.H3: High website quality improves brand image significantly.H4: Social media quality positively impacts brand image.H5: Brand awareness has a favorable and considerable impact on E-WOM.H6: E-WOM has a favorable and considerable impact on brand image.H7: Brand awareness and E-WOM mitigate the impact of website quality on brand image.H8: Brand awareness and E-WOM balance the impact of social media quality on brand image.

METHODOLOGY
The research object is the problem being studied.according to (Sekaran & Bouige, 2017) states that the object of research is anything that may differentiate or cause changes in value, and this value can occur at different periods for the same object or person, or at the same time for many objects or individuals.The focus of this study is Legend Coffee in Yogyakarta City.
The research subject is where the variables are attached.The subject is one of the parts or members in the sample.(Sekaran & Bouige, 2017) state that research subjects are parties who are used as sources of information or data sources for research.The subjects of this research are customers at Legend Coffee in Yogyakarta City.
This research relied on original data.According to (Sekaran & Bouige, 2017)., primary data refers to information received directly from the first person by researchers on variables of interest for specific study aims.Methods that can be used by researchers to search for primary data are observation, interviews and questionnaires (Sekaran & Bouige, 2017).
In this quantitative study, non-probability sampling techniques are utilized, which is a kind of sampling where elements in the population do not have the same opportunity.Each element or member of the population is picked to be the sample for this methodology (Sekaran & Bouige, 2017).The purposive sampling approach was utilized in this study, which is a sampling strategy that selects purposeful or subjectively aimed samples.This purposive sample is carried out when the researcher understands that the information needed can be obtained from a particular target group who has been able to provide information and they have met the criteria determined by the researcher (Ferdinand, 2016).
The following are the exact criteria used to choose respondents for this study: 1. Respondents with a minimum age of 17 years, the age limit was taken because at that age, they are considered adults and can make wise decisions.

Consumers who have experience purchasing several products at
Legend Coffee via the website.3. Consumers who have ever read reviews on social media or the website for Legend Coffee.In this case, by establishing specific criteria, it can be assumed that consumers with this experience can assess the performance of the Café Legend Coffee service in promoting its products or services to targeted consumers via websites and social media, with a sample size of up to 200 respondents in this study.The sample size is determined using the measuring approach described by (Hair, et al., 2009), which stipulates that the sample size should be 100 or greater, with a minimum sample size of 5 to 10 times the number of indicators calculated.The number of indicators in this research is 17.If you multiply the number of indicators by ten, the number of respondents required in this research is 17 x 10 = 170, which may be calculated as the number of respondents, namely a minimum of 170.
Data collecting strategies are the most strategic stage of research, as the primary goal is to get data.The data gathering method for this study is to distribute questionnaires.According to (Sekaran & Bouige, 2017), a questionnaire is a written question that will record answers from respondents with clearly defined alternatives.The questionnaire used was a Google Form distributed to respondents who had experience purchasing several products at Legend Coffee located in Yogyakarta.
The Likert scale is a sort of scale that is often used in research to assess a subject's level of agreement or disagreement with the statement or question presented.Each scale option is given a numerical weight, which can be used to analyze data or describe the level of perception and attitudes of respondents towards the social phenomenon being studied (Sekaran & Bouige, 2017).

Instrument Quality Test
Instrument quality tests are tests used to test the quality of data from questionnaires.Test the instrument's quality by doing a validity and reliability test.When data is collected, it is hoped that the research results would be valid and reliable, therefore validity and reliability are fundamental requirements for obtaining valid and reliable research results, but this is not implied by using a valid and reliable instrument.Its validity and reliability are tested, the research results (data) automatically become valid and reliable (Sugiyono, 2017).(Sekaran & Bouige, 2017) say that good research tools should have validity and reliability as crucial provisions.Research findings can be scientifically justified, therefore information on the validity and reliability of a measuring tool must be communicated.

Validity Test
According to (Sekaran & Bouige, 2017) validity testing is a questionnaire measurement technique that will be used to examine whether it is valid or invalid.This validity test measures each question item in the questionnaire that the researcher will distribute to find out whether the questions are valid or invalid.Validity testing in this research uses Confirmatory Factor Analysis (CFA), which is measured by Convergent Validity, which can be seen from significant factor loadings with estimates that must be equal to 0.50 or more and ideally must be 0.70 (Ghozali, 2018).

Reliability Test
Reliability testing is a test that demonstrates the extent to which a variable is measured without causing bias (error).This test is needed to measure consistency using research instruments (Sekaran & Bouige, 2017).The results obtained will then be compared with other questions.The criteria used in reliability testing in this research are Construct Reliability (CR), where a variable is said to be trustworthy if the CR value is > 0.70.However, reliability of 0.60 -0.70 is still acceptable given that the validity of the indicators in the model is adequate (Ghozali, 2018).

Data Analysis and Hypothesis Test
In this research, different test tools are employed to test the hypothesis, which are as follows: 1. Descriptive Analysis Descriptive analysis is statistics used to evaluate data by describing or illustrating the acquired data.Because there is no desire to make conclusions that apply to the broader public or generalizations, descriptive statistics offer facts through tables, graphs, pie charts, pictograms, calculation modes, median, mean (central tendency measurement), deciles, percentiles, calculating data spread using average and standard deviation, and computing percentages.In descriptive statistics, you may also use correlation analysis to determine the strength of the association between variables, generate predictions with regression analysis, and compare two or more averages without having to assess their significance (Sugiyono, 2017).

Structural Equation Modeling (SEM) Analysis
This study's data was analyzed using Structural Equation Modeling (SEM) software and AMOS.Structural Equation Modeling (SEM) is a collection of statistical approaches that allow evaluating a succession of very "complex" connections concurrently (Ferdinand, 2016).

Hypothesis Test
In this study, hypothesis testing is performed using SEM and AMOS software.

Instrument Quality Test 1. Validity Test
The instrumental validity test determines if the instrument being studied fits the validity requirements.In this study, there were 17 lists of statements representing each variable, with a total of 200 respondents utilizing SEM using AMOS 24 software.
Table 1 shows the results of verifying the instrument's quality using the CFA validity test using AMOS 24.
Tabel The test findings are considered legitimate if the factor loading (standardized loading) value exceeds 0.5 (Ghozali, 2018).The results of this test reveal that the factor loading value for each variable exceeds 0.5.Based on these findings, we may infer that all indicators are deemed legitimate and can be utilized in this study.

Reliability Test
A reliability test determines the dependability of a measuring device.In this study, reliability testing is performed using CR (Construct Reliability), which has the following criteria: if the CR value is more than 0.7, the variable is considered reliable.To measure dependability, use the following formula: Test findings are considered credible if their build reliability value exceeds 0.7 (Ghozali, 2018).This test's findings demonstrate that the CR values for the five research variables are all more than 0.7.Based on these findings, we may infer that the complete research instrument is dependable and can be used in this study.'

Hypothesis Test Results
In line with the paradigm proposed in this study, namely using an SEM data analysis tool which is operated using AMOS 24 software.The steps for SEM analysis according to (Ghozali, 2018) are as follows: 1. Develop a Model Based on Theory The first step is to create a model based on the ideas presented in chapter III.This study model includes two exogenous factors (website and social media quality), one endogenous variable (brand image), and two mediating variables (brand awareness and E-WOM).

Arrange a Path Diagram
The next step is to arrange the model as a route diagram.This route diagram is used to depict direct relationships, which are denoted by straight lines, whilst dotted lines are used to represent links through variables.In SEM, measuring the link between variables is referred to as model structure.In the second stage, the model indicated in the flow diagram is transformed into a structural equation.

Figure 2. Structural Equation
Figure 2 shows that the path diagram, both structural and measurement model equations were converted.As seen in the image above, there are five variables, namely website quality which has four statements, social media quality which has five statements, brand awareness which has three statements, E-WOM which has five statements, and brand image.There are three statements.

Matrix and Model Estimation
The matrices used are correlation and covariance and model estimation using maximum likelihood (ML) estimation.The maximum likelihood estimate can be met with the following assumptions: a. Sample Size Table 3 indicates that the multivariate and univariate normality tests are normally distributed, with critical ratio (c.r.) values for kurtosis and skewness ranging from -2.58 to +2.58.Multivariate data also meets normal assumptions, with a value of 0.556 falling within the range of ± 2.58.

Identification of Ouliers
The AMOS Mahalanobis Distance output identifies multivariate outliers.The criteria are at the p < 0.001 level, with 17 statement items, resulting in a degree of freedom value of 17.To get the Chi-square table in Excel, navigate to the Insert-Function-CHIINV sub-menu and enter the probability and number of measured variables as shown below:   The output comments for the model show that the df value is 163.The model's positive df value indicates over-identification, allowing for further data analysis.
5. Have Goodness of Fit Criteria The primary purpose of SEM is to determine how well the hypothesized model matches the data sample.Goodness of fit findings are shown in the following data: The results in Table 6 indicate that the research model is a good fit.Significancy This study's probability is considered unsuitable as the significant probability (0.000) falls below the cut-off-value standard of ≥ 0.05.The Root Mean Square Error of Approximation (RMSEA) is an index that accounts for chi-square values in big samples.The study model's RMSEA score of 0.062 is within the required range of ≤ 0.08, indicating its fit.The Goodness of Fit Index (GFI) indicates an overall model fit of 0.874 based on the squared residual value.The score is close to the suggested standard of > 0.90, suggesting that the research model is moderately fit.
The Adjusted Goodness of Fit Index (AGFI) adjusts the GFI based on the suggested degree of freedom vs the entire model.The AGFI for this model is 0.837.The number is close to the acceptable threshold of > 0.90,suggesting a marginal fit of the study model.CMIN/DF is a parsimonious suitability indicator that assesses the model's goodness of fit based on the number of estimated coefficients required to attain suitability.The CMIN/DF result in this study was 1.771, which is below the acceptable value of 2.00, indicating that the research model is fit.
The Tucker Lewis Index (TLI) is an appropriateness metric that is less affected by sample size.The TLI score in this study was 0.911, above the suggested value of ≥ 0.90, indicating that the research model is fit.
The Comparative Fit Index (CFI) is mostly independent of sample size and model complexity.This study's CFI result of 0.924 is within the suggested DISCUSSION 1. Impact of Website Quality on Brand Awareness.According to the study results in table 4.14, the estimated coefficient value is 0.244, the C.R. is 2.662, and the P value is 0.008.Based on these findings, hypothesis 1 is validated.The value P = 0.008 < α 0.05 fits both the conditions and the C.R. value.2.662 > 1.96 also passes the standards, implying that Website Quality has been proved to have a positive and significant effect on Brand Awareness.

2.
The Impact of Website Quality on Brand Awareness According to the study results in table 4.14, the estimated coefficient value is 0.244, the C.R. is 2.662, and the P value is 0.008.Based on these findings, hypothesis 1 is validated.The value P = 0.008 < α 0.05 fits both the conditions and the C.R. value.2.662 > 1.96 also passes the standards, implying that Website Quality has been proved to have a positive and significant effect on Brand Awareness.

3.
The Effect Impact of Website Quality on Brand Image.According to the study results in table 4.13, the estimated coefficient value is 0.027, the C.R. is 0.259, and the P value is 0.795.Based on these findings, H3 is not supported.The value P = 0.795 < α 0.05 falls short of meeting the standards and C.R. value.5.211 > 1.96 likewise does not match the criterion, implying that Website Quality has no positive or substantial effect on Brand Image.4.
The Impact of Social Media Quality on Brand Image.According to the study results in table 4.13, the estimated coefficient value is 0.440, the C.R. is 3.419, and the P-value is 0.000.Based on these findings, H4 has been supported.The value P = 0.000 < α 0.05 fits both the conditions and the C.R. value.3.419 > 1.96 also passes the standards, implying that Social Media Quality has been proved to have a positive and significant impact on Brand Image.

5.
The Impact of Brand Awareness on E-WOM.According to the study results in table 4.13, the estimated coefficient value is 0.541, with a C.R. of 4.298 and a P value of = 0.000.Based on these findings, H5 is supported.The value P = 0.000 < α 0.05 fits both the conditions and the C.R. value.4,298 > 1.96 also fits the criterion, implying that Brand Awareness has been proved to have a positive and significant effect on E-WOM.6.
The Impact of E-WOM on Brand Image.According to the study results in table 4.13, the estimated coefficient value is 0.694, the C.R. is 5.211, and the P value is 0.000.Based on these findings, H6 is supported.The value P = 0.000 < α 0.05 fits both the conditions and the C.R. value.5.211 > 1.96 also passes the standards, implying that E-WOM has been proved to have a positive and significant impact on Brand Image.This study examines the impact of indirect effects between variables by comparing the value of standardized direct effects to total effects.7.
The impact of website quality on brand image using Brand Awareness and E-WOM as mediating factors.According to the research findings, the impact of Website Quality on Brand Image is mediated by Brand Awareness and E-WOM, can be seen from the direct effects value and total effects value, this refers to (Suryani, et al., 2021a) with the

Figure 3 .
Figure 3. Mahalanobis Distance Limit Values was established to assist businesses in improving the quality of their social media content.B.Research ModelThe following model depicts the relationship between the variables "The Influence of Website Quality and Social Media Quality on Brand Image, with Brand Awareness and E-WOM as Mediating Variables."This connection model may be illustrated in the image below: table obtained is 40.79,so all data on the Mahalanobis Distance that is greater than 40.79 are multivariate outliers.The following is the Mahalanobis d-squared data: Table 4. Outliers Test Results

Table 4
shows no values larger than 40.79 for Mahalanobis Distance, indicating no outliers in the processed data.4.Identify the Structural ModelThe estimation findings might reveal whether or not there is an identification difficulty.SEM analysis can be performed only if the model identification findings indicate that the model is over-identified.This identification is accomplished by examining the df value of the model constructed.

Table 6 .
Assess Goodness of Fit Source: Data processed by AMOS24 (2023)