The Analysis of Internet Service Provider (ISP) Service Quality to Maintain Customer Loyalty in The B2B Market

The use of an internet network in an organization cannot be divorced from the numerous types of challenges that occur inside it, particularly those connected to network service quality., In addition, the high demand for service users has become a severe issue where people believe that the network needs to be optimally utilized. This research examined the connection between ISP service quality and B2B customer loyalty to business Internet services. This work employs Structural Equation Modeling with Partial Least Square methodology for SEM data analysis (PLS-SEM). The research revealed a favourable and statistically significant correlation between BP trust and BP loyalty. Because BP believes in the quality and service of ISP providers, they are typically loyal to BP. In addition, NQ and SP have a sizeable direct association with client loyalty. Moreover, this research showed that QSTS had a good and substantial effect on BP's level of trust. Meanwhile, CSTS, NQ, and SP have a favourable and significant impact on BP value.


INTRODUCTION
Industrial market segmentation is a significant marketing difficulty (Cortez et al., 2021). Crucial to a company's financial performance and growth are decisions about expanding existing product lines and capturing targeted markets. This pertains to the amount to which product expansion contributes to corporate profitability in both Business to Business (B2B) and Business to Customer B2C sectors and how companies should pursue expansion methods with the greatest influence on company earnings. Due to various significant differences between the B2B and B2C market structures, it is essential to understand the product and market preferences of B2B clients to adjust the offered management strategy solution. The B2B market comprises a small number of customers, is easier to identify, and is more accessible, resulting in a concentrated, small market (Liu et al., 2017).
Utilization of the Internet network in an organization cannot be divorced from the numerous types of difficulties that occur inside it, particularly those relating to the quality of network services. In addition, the high demand of service users has become a severe issue where people believe network utilization is suboptimal (Fatoni & Sandra, 2015). To retain clients successfully and efficiently, Internet Service Provider (ISP) must comprehend their customers' characteristics (Quach et al., 2016).
In a competitive business climate, service quality is a significant differentiator and a driver of service-based business (Lisa Zhao et al., 2013). Service quality has a significant impact on consumers. Good service is personable and responsive, making customers feel satisfied and at ease. Service quality is the consistent delivery of services that meet or exceed consumer expectations (Syauqi, 2019) ISPs can benefit from precise information regarding their B2B clients' evaluations of the quality of service they give. Such data can help managers design effective marketing plans to attain competitive advantage and long-term profitability. This study examined the relationship between ISP service quality and B2B customer loyalty to business Internet services.
This research employs a quantitative method based on the data analysis techniques of Structural Equation Modeling and Partial Least Square for SEM (PLS-SEM). Partial Least Square (PLS) is a method created to estimate the model path using latent variables and multiple indicators. The PLS method is the most suitable for forecasting, particularly for circumstances with formative indications (Mayasari et al., 2020).
By utilizing this PLS SEM method, ISPs can obtain precise information regarding the opinions of business partners on the quality of the company's services on the business internet market. This information can help business managers construct an Internet Service Provider (ISP) quality improvement plan to retain corporate internet clients.
The purpose of this research was to assess the effects of ISP service quality dimensions on customer value and business partner trust, as well as the effects of the value of business partners as intermediaries on the interaction between ISP service quality dimensions and business partner trust, Considering the consequences of ISP service quality aspects on consumer value and business partners' commintment in customer loyalty for business internet services, where ISPs provide internet services to business partners (B2B). Business To Business (B2B) refers to business conducted between individuals or organizations with common business interests, also known as inter-company business. (Muslim et al., 2019)

METHOD
The method used in this study is a quantitative method using the Structural Equation Modeling data analysis technique with the Partial Least Square approach for SEM (SEM-PLS). The independent variables consist of network quality (NQ), customer service and technical support (CSTS), quality of information and site support (QSIS), security and privacy (SP). The dependent variable consists of business partner value (BPV) and business partner trust (BPT). The customers in this study are business partners or ISP (Business Partner) resellers. The characteristics or demographics of the respondents that are used for this study described in the Table 1. This is modified according to the conceptual framework in Figure 1 : Based on the theory and framework presented in Figure 1, the following hypothesis is derived: H1a : "Network quality has a significant and positive influence on the value of B2B business partners." H1b : "Network quality has significant and positive influence on B2B business partners' commitment." H1c : "Network quality has a significant and positive influence on B2B business partner loyalty." H2a : "Customer service and technical support have a significant and positive influence on the trust of B2B business partners." H2b : "Customer service and technical support have significant and positive influence on the trust of B2B business partners" H2c : "Customer service and technical support have a significant and positive influence on the trust of B2B business partners" H3a : "The value of B2B business partners is positively and significantly affected by the quality of content and site assistance." H3b : "The quality of website content and customer service has a significant and positive influence on B2B business partners' commintment." H3c : "The quality of information and site support has a significant and positive influence on B2B business partners' loyalty." H4a : "Security and privacy have a significant and positive influence on the value of B2B business partners." H4b : "Security and privacy have significant and positive influence on the commintment of B2B business partners." H4c : "Security and privacy have a significant and positive influence on B2B business partners' loyalty." H5 : "B2B business partner loyalty is positively and significantly impacted by the value of business partners." H6 : "The value of business partners has a significant and positive influence on the trust of B2B business partners." H7 : "B2B business partner trust influences B2B business partner loyalty in a favorable and significant manner." Based on the hypothesis mentioned above, a survey was prepared and executed with ISP consumers as research samples. This study surveyed 46 business partners who utilize business internet services (B2B), the justification is because using the SEM PLS method can be used for predictive models using a small sample, even a minimum of 30 respondents (Sumarsono et al., 2021). This research implemented the method of purposive sampling, in which the researcher relied on his discretion to determine which individuals in the population would participate.
Respondents to this research are users of business Internet products from ISP companies. A questionnaire given to a sample of the target demographic served as the research survey's primary data collection method. Over one to two months, the ISP sends the respondents the web link to the online survey.

RESULTS AND DISCUSSION
Data processing occurs following the research stage of data collection, which occurs after respondents complete the survey web link and submit their responses. Using the PLS-SEM (Partial Least Square) method, data processing is done with the help of the SmartPLS program. The reliability and validity of the Cronbach Alpha and Barlett Tests were employed to examine the questionnaire's reliability and validity (Muslim et al., 2019). SmartPLS 3.0 statistical software, specifically made to estimate structural equations based on variance, is the instrument employed. Figure 2 displays the computation outcomes based on the framework ( Figure 2).

Test Measurement Model (Outer Model)
The Outer Model Test is a test to see if the research tool can produce valid and dependable data, which are essential for excellent research (Law & Fong, 2020). In this research, the outer model, which consists of convergent validity, discriminant validity, reliability, AVE value, and commonality, is evaluated using SmartPLS 3.0 from the research results to examine the connection between constructs (latent variables) and indicators.
A convergent validity measuring model describes the connection between indicators and latent variables. The processing results at this point take the form of each indicator's loading factor value, which is declared valid if it is > 0,70. The Business Partner Loyalty (BPL.3) indicator is not included in the model since it has a loading of less than 0,70 and is not significant based on the outer loading ( Figure 2) of the model. The BPL.3 indication is then removed, and the model is once more calculated. BPL 3 statement regarding customers experiencing great difficulties when they have to move to other internet service providers. Figure 3 displays the SmartPLS graph's output.   The BPL.3 indication is not included in the model since it has a loading of less than 0,70 and is not essential, according to Outer Loading  Table 1. The model is also re-estimated when the BPL 3 indication is taken out. Table 1 shows the outcomes of the Outer Loading re-estimation (second model modification). Every indicator's loading factor value is changed, and the modification is deemed legitimate if it is a value > 0,70. When utilizing a number greater than 0,70, each factor for each indicator is legitimate.

Discriminant Validity
The degree to which a construct is wholly distinct from other constructs is known as discriminant validity (a construct is unique). The Heretroit-Monotrait Ratio (HTMT ratio) value is the most accurate measurement available right now, however. A construct has excellent discriminant validity if the Heretroit-Monotrait Ratio (HTMT ratio) value < 0,90 between constructs. Alternatively, to put it another way, it is determined that a construct is discriminantly valid if its HTMT value (correlation value) against itself is higher than that of other constructs. Table 3 displays the outcomes of utilizing SmartPLS 3.0 to process Discriminant Validity. As can be observed, the HTMT value of a construct has a more significant correlation with itself than other constructions, denoted in bold in Table 3. The CSTS construct has a correlation value of 0,861, which is greater than the value of HTMT to NQ (0,736); QSIS (0,746); SP (0,803); BPT (0,754); BPV (0,703) and BPL (0,798). For other constructs, the same holds. This demonstrates that the latent construct outperforms other blocks' ability to anticipate their block's indications.

Reliability
A reliability test examines a measurement's consistency.
Composite reliability and Cronbach's alpha were used to measure the build reliability test. Based on Table 4 below, the results of the Construct Reliability and Validity test conducted using the SmartPLS 3.0 program show that composite reliability and Cronbach's alpha are all suitable constructs, with values > 0,70, indicating that all construct indicators are reliable or pass the reliability test. Based on the results of Table 4 above, it can be said that all construct indicators are trustworthy or pass the reliability test because the composite reliability and Cronbach's Alpha for all constructs are good, i.e., > 0.70.

Structural Model Test (Inner Model)
By examining the R-square value, which is the model's goodness-of-fit test, the structural model is put to the test. Examine the connections between exogenous and endogenous factors and the connections between R-Square, F-Square, Direct Effect, Indirect Effect, and Total Effect.

R-Square
R-Square measures the percentage of variation in the value of the endogenous (affected) variable that can be accounted for by the explanatory variable (exogenous). This can be used to determine whether a model is sound or flawed. Criteria ; -If the value of R 2 = 0,75 → The model is substantial (strong) -If the value of R 2 = 0,50 → The model is moderate -If the value of R 2 = 0,5 → The model is weak (bad) The variables Business Partner Loyalty (BPL), Business Partner Trust (BPT), and Business Partner Value (BPV) have a value of R square > 0,75 based on Table 5 below, where the outcomes of these values suggest that the model is considerable (strong).

Direct Effect (Direct Effect)
Analysis of the direct effects of exogenous (internal) and endogenous (internal) variables on each other is known as the "DIRECT EFFECT" method (endogenous -If the P-Values value < 0,05, then significant.
-If the P-Values value > 0,05, then it is insignificant.

Hypothesis Test
By examining the t-statistics and probability values, hypotheses are tested. Evaluation is carried out by examining the importance of the connection between exogenous and endogenous components. If the statistical t count value is more than or equal to the t-table value, the test employs a t-test. The connection between endogenous and exogenous elements is then found to have a significant impact, and vice versa (Sumarsono et al., 2021) The statistical value used for alpha 5% in t-statistical hypothesis testing is 1,96. Therefore, the hypothesis is rejected if the t-statistic < 1,96 and accepted if it is > 1,96. a. Testing Hypothesis 1a: "The value of B2B customers and business partners is positively and significantly impacted by network quality." The findings of testing hypothesis 1a show a relationship between the value of business partners (BPV) and the Internet network quality (NQ) variable, with a total effect value of 0.346 and a t-statistic value of 2,827. Given that this result is higher than the t-table value of 1,96, either Ha is accepted or the effect is both positive and significant. b. Test Hypothesis 1b (Network quality has a positive and significant effect on the trust of B2B business partners) The findings of testing hypothesis 1b demonstrate a relationship between Internet network quality (NQ) and B2B business partner trust (BPT), with a total effect value of -0,004 and a t-statistic value of 0,028. Given that this number is smaller than the ttable cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. c. Testing a Hypothesis 1c (Network quality has a positive and significant effect on the loyalty of B2B business partners) The findings of testing hypothesis 1c, there is a relationship between the Internet network quality (NQ) variable and the loyalty of B2B business partners (BPL), with a total effect value of 0.567 and a tstatistic value of 3,776. Given that this result is higher than the t- The testing 2c show that the association between the Customer Service and Technical Support (CSTS) variables impacts Customer Loyalty (BPL), which has a total effect value of 0,132 and a t-statistic value of 0,903. Given that this number is smaller than the t-table cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. g. Testing Hypothesis 3a (Quality of information and site support has a positive and significant effect on the value of B2B business partners) The findings of testing hypothesis 3a show that the Customer Value (BPV), which has a total effect value of 0,089 and a t-statistic value of 1,066, is influenced by the connection between the quality of information and site support (QSIS) variables. Given that this number is smaller than the t-table cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. h. Hypothesis Testing 3b (The quality of information and site support has a positive and significant effect on the trust of B2B business partners) The results of testing hypothesis 3a, there is a correlation between customer trust (BPT) and the quality of information and site support (QSIS) variables, with a total effect value of 0,479 and a t-statistic value of 3,678. Given that this result is higher than the t-table value of 1,96, either Ha is accepted or the effect is both positive and significant. i. Testing a hypothesis in 3c (The quality of information and site support has a positive and significant effect on the loyalty of B2B business partners) The findings of hypothesis testing 3c show a relationship between customer loyalty (BPL) and the quality of information and site support (QSIS) variables, with a total effect value of -0,087 and a t-statistic value of 0,646. Given that this number is smaller than the t-table cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. j. Testing Hypothesis 4a (Security and privacy have a positive and significant effect on the value of B2B business partners) The findings of testing hypothesis 4a show a correlation between security and privacy (S & P) variables and business partner value (BPV), with a total effect value of 0,240 and a t-statistic value of 3,019. Given that this result is higher than the t- (BPT), with a total effect value of 0,221 and a t-statistic value of 1,903 respectively. Given that this number is smaller than the ttable cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. l. Test of Hypothesis 4c (Security and privacy have a positive and significant effect on the loyalty of B2B business partners) The findings of hypothesis testing 4c show a correlation between security and privacy (S&P) variables and business partner loyalty (BPL), with a total effect value of 0,390 and a t-statistic value of 2,965. Given that this result is higher than the t- The findings of testing hypothesis number 5 show a relationship between the variable business partner value (BPV) and business partner loyalty (BPL), and this association has a total effect value of 0,049 and a tstatistic value of 0,147. Given that this number is smaller than the t-table cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. n. Testing Hypotheses No. 6 (Customer value has a positive effect on B2B business partner trust) According to the results of testing hypothesis number six, there is a relationship between the variable Business Partner Value (BPV) and Business Partner Trust (BPT), with a total effect value of 0,247 and a t-statistic value of 0,636. Given that this number is smaller than the t-table cutoff of 1,96, it may be said that H0 is either rejected or that the effect is negative and minor. o. Testing theory no. 7 (Customer trust has a positive effect on B2C customer loyalty) According to the findings of testing hypothesis number seven, there is a relationship between business partner loyalty and trust, with a total effect value of 0,318 and a t-statistic value of 2,454. Given that this result is higher than the t-table value of 1,96, either Ha is accepted or the effect is both positive and significant.