The Influence of Service Quality, Business Capacity Development Programs, and Word of Mouth on Customer Satisfaction and Loyalty of Mekaar Permodalan Nasional Madani Customers in Bandung Branch

To improve service quality, PNM innovated by launching the Business Capacity Development (PKU) program. PNM also empowers existing customers to attract new prospective customers through Word of Mouth. There are two analysis methods used, namely outer model analysis with Convergent Validity, Discriminant Validity, Composite Reliability, Cronbach's Alpha and using inner model analysis with Hypothesis Test. Outcome from this study are Customer Satisfaction affecting Customer Loyalty. Service positively and insignificant affecting on Customer Loyalty. The PKU program not affecting Customer Loyalty. WOM positively and insignificant affecting on Customer Loyalty. Service quality affecting customer satisfaction. PKU program not affecting customer satisfaction. WOM has a significant positive effect on Customer Satisfaction

To improve service quality, PNM innovated by launching the Business Capacity Development (PKU) program. PNM also empowers existing customers to attract new prospective customers through Word of Mouth. There are two analysis methods used, namely outer model analysis with Convergent Validity, Discriminant Validity, Composite Reliability, Cronbach's Alpha and using inner model analysis with Hypothesis Test. Outcome from this study are Customer Satisfaction affecting Customer Loyalty. Service positively and insignificant affecting on Customer Loyalty. The PKU program not affecting Customer Loyalty. WOM positively and insignificant affecting on Customer Loyalty. Service quality affecting customer satisfaction. PKU program not affecting customer satisfaction. WOM has a significant positive effect on Customer Satisfaction

INTRODUCTION
Word of Mouth marketing strategy can make customers voluntarily provide testimonials of the advantage being a Mekaar customer and this strategy is considered quite effective for growing PNM customers, especially those in urban areas such as Medan, Semarang, Surabaya, Jakarta, and Bandung.
PT PNM Bandung Branch was established in February 2000. The coverage area includes 2 regencies/cities, namely Bandung City and Bandung Regency, and oversees 2 Regional Mekaar (Membina Ekonomi Keluarga Sejahtera), namely Bandung Region 2 and Bandung Region 3.
In 2021 the total disbursement of Mekaar financing (net lending) reached 870.4 billion with 244,173 customers. In 2022 the total disbursement reached 1.1 trillion with 298,418 customers. For more details can be seen in The data in Table 1 shows that in 2022 there was an increase in the amount of financing by 138% and the amount of distribution to customers by 122% compared to 2021. The same thing can also be seen in table 1.2 that the number of Business Capacity Development Trainings in 2022 also increased from 85 trainings in 2021 to 207 trainings in 2022 or an increase of 2,348%. The number of PKU training participants also experienced a very significant increase, namely 11,102 training participants compared to 2021 of 1,996 participants or an increase of 5,363%.  Based on the jobs, 48 respondents (48%) are grocery store workers, convection workers as many as 34 respondents (34%), farmers as many as 15 people (15%) and daily laborers as many as 3 people (3%). d. Length of Time as a Customer Based on the length of time being a customer of PT PNM Bandung Branch, 71 people (71%) are between 2-3 years, 20 people (20%) are between 3-4 years, 7 people (7%) are between 1-2 years, 2 people (2%) are more than 4 years and no respondent less than 1 year (0%). Descriptive data displays an overview of respondents answers contained in the questionnaire as well as respondents' responses. Researcher will describe in detail that are gathered in descriptive statistics. The submission of an empirical description using the average value, minimum value, and maximum value. This description will find what extent the respondents perceptions of the variables that are indicators in the study. The author will provide a more comprehensive description of the questionnaire answers received as follows: There are 6 indicators in Quality of service (QL) variable QL1-QL6, 0.16% strongly disagreed, 1.5% disagreed, 18% less agree, 54% agreed and 26.34% strongly agreed. Average result 4.05, which means interval interpretation is in "Agree" category. These means quality of service gathered from sample is in a good category, with the question from QL1 highest average value of 4.29, the question point is "The level of conformity between the quality of service provided by the company and your expectations as a customer is very high." There are 6 indicators in Business Development Program (PU) variable PU1-PU6, 2.5% strongly disagreed, 2.83% disagreed, 15.34% less agree, 52.33% agreed and 27% strongly agreed. Average result is 4.00, which means interval interpretation is in "Agree" category. These means the Business Development Program gathered from sample is in a good category, with the question from PU6 highest average value of 4.25 the question point is "Our service is very good in serving customer requests to upgrade."  (2023) There are 10 indicators in Word of Mouth Marketing (WOM) variable WOM1-WOM10, 0.4% strongly disagreed, 3% disagreed, 12.3% less agree, 59.3% agreed and 25% strongly agreed. Average result is 4.05, which means interval interpretation is in "Agree" category. These means Word of Mouth Marketing gathered from sample is in a good category, with the question from WOM7 highest average value of 4.28 the question point is "you will recommend your relatives to use our products / services." Percentage 0.17% 0,5% 9,5% 58,83% 31% 100% Source: Author's Analysis (2023) There are 6 indicators in Customer Satisfaction (KN) variable KN1-KN6, 0.17% strongly disagreed, 0.5% disagreed, 9.5% less agreed, 58.83% agreed and 31% strongly agreed. The average result in 4.2, which means interval interpretation is in "Agree" category. These means Customer Satisfaction gathered from sample is in a good category, with the question from KN5 highest average value of 4.29 the question point is "your initial perception of our service performance before using the products/services is very good." Source: Author's Analysis (2023) There are 10 indicators in Customer Loyalty (LN) variable LN1-LN10, 1.4% strongly disagreed, 2.2% disagreed, 8.6% less agree, 57% agreed and 30.8% strongly agreed. The average result is 4.14, which means interval interpretation is in "Agree" category. These means Customer Loyalty gathred from sample is in a good category, with the question from LN10 highest average value of 4.37 the question point is "you are very interested in referring the existence of the company."

Measurement Model (Outer Model)
This model expalin how every indicator ties to its latent variable. The test carried out on the outer model by using Confimatory Factor Analysis (CFA) technique.
CFA of the construct useful look at validity of every indicator and testing reliability of the construct. Validity criteria are based on discriminant validity and convergent validity. What is meant by convergent validity is a measure of a measure of a construct that must correlated, while what is meant by discriminant validity is a measure of a measure of a different construct that cannot be strongly correlated. Indicators meet convergent validity when loading factor > 0.7, but for research in early stages, a loading 0.5 to 0.6 good enough, and is indicated by an Average Variance Extracted (AVE) value > 0.5. Average Variance Extracted (AVE) is a latent variable that describe half even more variance in its indicators. Construct reliability is measured by Composite Reliability and Cronbach's Alpha. Composite Reliability is used to prove the accuracy, consistency of the instrument in measuring constructs. Cronbach Alpha and Composite Reliability reliable if the value > 0.7 (Ghozali, 2014).   All Outer Loading value above 0.7 means all indicators of the Service Quality (QL) are valid or fulfill convergent validity.

Business Capacity Development Program Variable (PU)
The results of Validation Test are shown below:     The correlation score of the construct is higher than the correlation score with other constructs. For example, the KN1 indicator (indicator of the Customer Satisfaction variable) has an Outer Loading value of 0.856 which is higher than the Outer Loading values in other constructs, namely 0.593 -0.610 -0.599 -0.721. The table also shows that the variable KN1-KN6 indicator Outer Loading value that is higher than the Outer Loading values in other constructs. Likewise, the LN1-LN10 indicator of the Customer Loyalty variable, the PU1-PU6 indicator of the Business Development Program variable, the QL1-QL6 indicator of the Service Quality variable, the WOM1-WOM10 indicator of the Word of Mouth Marketing variable, have a higher Outer Loading value than the Outer Loading value in other constructs. So all latent constructs shows nice discriminant validity since they can predict indicators in their blocks better than indicators in another blocks.

B. Discriminant Validity
For further discriminant validity test, let us look at the AVE (Average Variance Extracted) value. The AVE value is nice if it is greater than 0.5 (Imam Ghozali, 2014). Every variable has Cronbach's alpha score above 0.6, lowest value 0.840 from the Service Quality variable (QL), and the highest value 0.920 from the Customer Loyalty variable (LN). it means it have fulfill the score of Cronbach's alpha. These criteria have met the reliability criteria and are reliable.

D. Structural Model Testing (Inner Model)
Inner Model or also called Inner Relation is explanation of relationship between latent variables. There are two types of testing, R2 test and hypothesis test.

Evaluating the R² Value
Furthermore, testing predictive relevance or Q-Square for structural models to find out how great the observation of its parameters. It is known that if the Q-Square value> 0, it can be said that the model has predictive relevance. To calculate the Q-Square value, you can use: Q 2 = 1 -(1 -R1 2 ) (1 -R2 2 )..... (1-Rp 2 ) where R , R1 2 2 2 , ... Rp 2 is the R-square of the dependent variable. Q 2 = 1-(1-0.670)(1-0.402) Q 2 = 1-(0.330)(0.598) Q 2 = 1 -0.1973 Q 2 = 0.8026 From above calculations, the value of Q 2 or predictive relevance in this study is 0.8026. From these results it is known that this research model have predictive relevance because Q 2 is greater than 0 and can be said to be good because it is close to the value of 1.

Hypothesis Test
Bootstrap resampling method is used to test hypothesis between constructs. Determine the value of the t table by looking at the vertical degree of freedom value and the value of the meaning limit based on the hypothesis. DF = n -k *n = total respondents DF = 100 -5 *k = number of research variables DF = 95 *DF = Degree of Freedom In statistic table, t-table value with a value of 95 is 1.985 with significance level (α) 5% (0.05). The decision-making method is: -If P-Values above 0.05 or t count below t table, Ho accepted Ha rejected.
If P-Values below 0.05 or t count above t table, Ho rejected Ha accepted.

There is a Significant Effect of Service Quality (QL) on Customer Loyalty (LN)
t statistics 0.426 lower than the t table 1.985, P-Values 0.671 higher than α = 0.05, means Ha is rejected, so there is no effect of the Service Quality (QL) on Customer Loyalty (LN). The coefficient value positive (0.047), means Service Quality (QL) variable affecting Customer Loyalty (LN) variable by 4.7%. H2 "Service Quality (QL) has a significant effect on Customer Loyalty (LN)" is rejected.

There is a Significant Effect of the Business Development Program (PU) on Customer Loyalty (LN)
t statistics 0.612 which is lower than the t table 1.985, P-Values 0.541 higher than α = 0.05, means that Ha is rejected, so there is no effect of the Business Development Program (PU) on Customer Loyalty (LN). The coefficient value negative (-0.064), means Business Development Program (PU)negatively affecting Customer Loyalty (LN) variable by -6.4%. H3 "Business Development Program (PU) has a significant effect on Customer Loyalty (LN)" is rejected.

There is a Significant Influence of Word of Mouth Marketing (WOM) on Customer Loyalty (LN)
t statistics 1.637 which is lower than the t table 1.985, P-Values = 0.104 higher than α = 0.05, meaning that Ha is rejected, so there is no effect of the Word of Mouth Marketing (WOM) on Customer Loyalty (LN). The coefficient value positive (0.247), means Word of Mouth Marketing (WOM) variable affecting Customer Loyalty (LN) variable by 24.7%. H4 "Word of Mouth Marketing (WOM) has a significant effect on Customer Loyalty (LN)" is rejected.

There is a Significant Effect of Service Quality (QL) on Customer Satisfaction (KN)
t statistics 3.488 which is higher than the t table 1.985, and the P-Values = 0.001 lower than α = 0.05, means that Ha is accepted, so there is no effect of the Service Quality (QL) on Customer Satisfaction (KN). The coefficient value positive (0.284), means Service Quality (QL) variable affecting Customer Satisfaction variable by 28.4%. H5 "Service Quality (QL) has a significant effect on Customer Satisfaction (KN)" is accepted.

There is a Significant Effect of the Business Development Program (PU) on Customer Satisfaction (KN)
t statistics value 1.470 lower than the t table 1.985, and the P-Values = 0.144 higher α = 0.05, means that Ha is rejected, so there is no effect of the Business Development Program (PU) on Customer Satisfaction (KN). The coefficient value positive (0.144), means Work Development Program (PU) variable affecting Customer Satisfaction variable by 14.4%. H6 "Business Development Program (PU) has a significant effect on Customer Satisfaction (KN)" is rejected.

There is a Significant Influence of Word of Mouth Marketing (WOM) on Customer Satisfaction (KN)
t statistics 5.492 is higher than the t table 1.985, and the P-Values value = 0.000 lower α = 0.05, means that Ha is accepted, so there is an influence of the Word of Mouth Marketing (WOM) variable on Customer Satisfaction (KN). The coefficient value is positive (0.489), meaning that the Word of Mouth Marketing (WOM) variable affecting Customer Satisfaction variable by 48.9%. H7 "Word of Mouth Marketing (WOM) has a significant effect on Customer Satisfaction (KN)" is accepted.

Service Quality (QL) Significantly Affecting on Customer Loyalty (LN) Through Customer Satisfaction (KN)
t statistics 2.639 higher than the t table 1.985, and the P-Value 0.009 is greater than α = 0.05, means that Ha is accepted, so there is an significant influence of the Service Quality (QL) on Customer Loyalty (LN) through Customer Satisfaction (KN). The coefficient value is positive (0.124), means Customer Satisfaction (KN) affecting or increases the influences of the Service Quality (QL) variable on the Customer Loyalty (LN) variable by 12.4%. H8 "Service Quality (QL) significantly affecting Customer Loyalty (LN) through Customer Satisfaction (KN)" is accepted.

Business Development Program (PU) Significally Affecting on Customer Loyalty (LN) Through Customer Satisfaction (KN)
t statistics 1.342 lower than the t table = 1.985, and the P-Values = 0.182 higher than α = 0.05, means that Ha is rejected, so there is no significant effect of the Business Development Program (PU) on Customer Loyalty (LN) through Customer Satisfaction (KN). The coefficient value positive (0.063), meaning that the Customer Satisfaction (KN) variable affecting or increases the influence of the Business Development Program (PU) variable on the Customer Loyalty (LN) variable by 6.3%. H9 "Business Development Program (PU) has a significant effect on Customer Loyalty (LN) through Customer Satisfaction (KN)" is rejected.

Word of Mouth Marketing (WOM) Significantly Affecting on Customer Loyalty (LN) Through Customer Satisfaction (KN)
t statistics 2.579 higher than the t table 1.985, and the P-Values = 0.011 lower than α = 0.05, means Ha is accepted, so there is an affect of Word of Mouth Marketing (WOM) significantly on Customer Loyalty (LN) through Customer Satisfaction (KN). The coefficient value positive (0.212), meaning that the