The Role of Customer Satisfaction in Mediating the Effect of Service Claim and Brand Image on Customer Loyalty at PT Asuransi Sahabat Artha Proteksi in Jakarta

The purpose of this study was to determine the role of customer satisfaction in mediating the effect of Service Claim and Brand Image on Customer Loyalty at PT Asuransi Sahabat Artha Proteski in Jakarta. This research was conducted using the Probability Sampling method with the type of Simple Random Sampling. Researchers distributed questionnaires to 100 respondents who were customers of PT Asuransi Sahabat Artha Proteksi in Jakarta. Data analysis using SmartPLS Analysis version 3.3. The results of this study indicate that service claims have a positive and significant effect on customer loyalty. Brand image has a positive and significant influence on customer loyalty. Service claims have a positive and significant effect on customer satisfaction. Brand image has a positive and significant influence on customer satisfaction. Customer satisfaction has a positive and significant influence on customer loyalty. Customer satisfaction mediates the effect of service claims on customer loyalty. Customer satisfaction mediates the effect of brand image on customer loyalty.


I. Introduction
The service sector plays an important role in sustaining a country's economy. Economic growth is still an important goal in a country's economy, especially for developing countries like Indonesia (Magdalena and Suhatman, 2020). The more developed a country, the contribution of the service sector in gross domestic product will be even greater. One of the important service sectors is insurance services. As believers, we naturally don't need to worry too much about things that will happen in our lives because the Creator has a plan to ensure the safety of His people. However, as humans who were created with reason and mind, of course, we will consider all kinds of risks that can happen to us at any time, so it is only natural that we have protection/guarantee for our souls/wealth (Saparso, 2018).
A sense of security (peace of mind) in the future is a dream for everyone, whereas most people cannot predict the future. Therefore insurance is important because insurance protects a person or business entity against financial losses due to disasters or liability for emergencies, accidents, or other unexpected negative things. There are various types of insurance, some protect people and some protect business entities.
Public awareness of the importance of protection against various kinds of risks that can occur is one of the main reasons for the growth of insurance users lately. The insurance industry in Indonesia continues to increase in line with the development of public needs and market demand. The following table presents data on the number of insurance companies in Indonesia.  Companies in years 2017 -2019  Type  2017  2018  2019  Life Insurance Company  61  60  60  General insurance  79  79  79  Total  140  139 139 Sources: Otoritas Jasa Keuangan tahun 2020 One of the general insurance companies included in the data above is PT Asuransi Sahabat Artha Proteksi or commonly called Sahabat Insurance. PT Asuransi Sahabat Artha Proteksi is a general insurance company previously named PT. Bess Central Insurance and has been operating since 1997 with 16 branch offices throughout Indonesia.
Sahabat Insurance always puts forward the Positively Protected concept where the customer is the center of all management team activities. Every activity carried out is channeled positively to protect customer safety and satisfaction. On the other hand, the company also receives positive feedback so that it can be a constructive reference for a positive future for the company. The increasingly tight competition in the insurance sector forces insurers to be able to create competitive advantage in order to survive and develop. This makes PT Asuransi Sahabat Artha Proteksi continuously improve its performance in all lines including innovative products, services, networks, human resources and technology in order to attract new customers and retain existing customers.
Today's insurance industry is full of competition, and to win, companies must be able to keep their customers happy. Dissatisfied customers will leave the company and become customers of other companies that can provide better satisfaction, and in the end the company will be less profitable. If the service provided by insurance is not good, it will create a bad image for the company.
Measuring customer satisfaction is important so that companies can find out which attributes of a product make customers happy (Tecoalu, Wahyoedi & Kustiawan, 2021). For example, customers feel satisfied after receiving a claim from an insurance company because the payment process is fast, even faster than expected. Understanding customer perceptions helps insurance companies provide better service to customers, maintain customer satisfaction, and provide high loyalty to insurance companies.
In addition to service quality and satisfaction, brand image is one of the factors that can affect customer loyalty (Scorita and Nurmahdi, 2018). Society responds to the image or brand of a company differently, because the brand does not just come but must be achieved with an image building program. A positive brand image is able to gain or maintain customer loyalty (Wahyoedi, Winoto & Edyson, 2021).
The following table shows the gross premium income data for Sahabat Insurance for 2016-2020. Based on the background and data in table 3 above, there are several phenomena that are the reason for the researchers to conduct research, namely there is a very large gap between the national premium for general insurance in 2019 which is Rp. 89.52 Trillion, while the gross premium of Sahabat Insurance is only Rp. 359 billion or only 0.4% of the national gross premium. Then, the number of general insurance companies is quite a lot so that the competition in this industry is very tight. Furthermore, it can be seen that during 2016-2020 the company's premium income was unstable, not on the growth/increase graph. Even in 2020 there was a decrease of 18% when compared to premium income in 2019.
Based on the description of the background, the formulation of the problem in the study is as follows: 1. Is there any effect of service claim on customer loyalty? 2. Is there any influence of brand image on customer loyalty? 3. Is there any effect of service claim on customer satisfaction? 4. Is there any influence of brand image on customer satisfaction? 5. Is there an effect of satisfaction on customer loyalty? 6. Is there a role for customer satisfaction in mediating the effect of service claims on customer loyalty? 7. Is there a role for customer satisfaction in mediating the influence of brand image on customer loyalty?

Service Claims
Service is any action or activity that can be offered by one party to another, which is essentially intangible and does not result in the ownership of anything. Its production may or may not be linked to a single physical product (Kotler and Keller, 2016).

Claim
Claims in insurance terms are defined as requests or demands for payment of benefits in accordance with the provisions stipulated by the insurance policy (Puspitasari, 2015).

Brand Image
According to Setiadi (2015) brand image is a representation of the overall perception of the brand and is formed from information and past experiences of the brand.

Customer Satisfaction
Customer satisfaction is a post-purchase evaluation, where the perception of the performance of the selected product/service alternative meets or exceeds expectations before purchase (Lovelock and Jochen, 2011).

Customer Loyalty
The definition of loyalty according to Lovelock and Jochen, (2011) is the loyalty of a customer to use a company's product for a long time.

Conceptual framework
From the data that has been described above, therefore we researchers try to create a framework of thought as follows:

Figure 1. Research Concept Model
Hypothesis H1: Service Claim has an effect on customer loyalty. H2: Brand image has an effect on customer loyalty. H3: Service Claim has an effect on customer satisfaction. H4: Brand image has an effect on customer satisfaction. H5: Customer satisfaction has an effect on customer loyalty. H6: Customer satisfaction mediates the effect of service claims on customer loyalty H7: Customer satisfaction mediates the effect of brand image on customer loyalty

III. Research Methods
The subjects in this study were policy holders at PT Asuransi Sahabat Artha Proteksi in Jakarta. The reason for choosing this location is because the location is considered to have many respondents who are customers of PT Asuransi Sahabat Artha Proteksi. While the object of research is a variable that can be measured and researched by the author. The objects in this study are service claims, brand image, customer satisfaction and customer loyalty.
The population in this study are customers of PT Asuransi Sahabat Artha Proteksi in Jakarta and the number is unknown. Therefore, the authors conducted research based on the sample. The number of samples will be taken as many as 100 respondents. The sample includes customers of PT Asuransi Sahabat Artha Proteksi in Jakarta. Determination of this sample size is based on the approach of Haier et al., (2014).
The sampling method used in this study is to use the Probability Sampling technique. This study used a simple random sampling technique. In this research, there are 3 types of variables studied which consist of dependent variable, independent variable, and intervening variable.
In this study the independent variables used are service claim (X1) and brand image (X2) and the intervening variable is customer satisfaction (Z) and the dependent variable in this study is customer loyalty (Y).
In this study, primary data was sourced from questionnaires distributed directly to customers of PT Asuransi Sahabat Artha Proteksi in Jakarta. Data collection techniques will be carried out through questionnaires submitted to each selected respondent. The scale used in this study is a 5-point Likert scale for each statement submitted to the respondent.
The data analysis technique in this study uses Partial Least Square (PLS) with SmartPLS software version 3.3 PLS is a Structural Equation Modeling (SEM) equation model with an approach based on variance or component-based structural equation modeling. This study has a complex model and a limited number of samples, so the data analysis uses SmartPLS software. PLS-SEM analysis consists of two sub-models, namely the measurement model or the outer model and the structural model or the inner model.

Test Measurement Model or Outer Model 3.1 Validity test
There are several stages of testing to be carried out, namely convergent validity, average variance extracted (AVE), and discriminant validity. a. Convergent Validity Convergent validity can be met when each variable has an AVE value above 0.5, with the loading value for each item also has a value of more than 0.5 (Ghozali and Latan, 2015). b. Average Variance Extracted (AVE) For good requirements, if the AVE of each question item is greater than 0.5 (Ghozali and Latan, 2015).

c. Discriminant Validity
This validity test explains whether the two variables are quite different from each other.

Reliability Test
In addition to the validity test, model measurement is also carried out to test the reliability of a construct. Reliability test was conducted to prove the accuracy, consistency, and accuracy of the instrument in measuring the construct. In PLS-SEM using the SmartPLS 3.3 program, measuring the reliability of a construct with reflexive indicators can be done in two ways, namely Cronbach's Alpha and Composite Reliability. The construct is declared reliable if the composite reliability value and Cronbach's Alpha is above 0.7 (Ghozali and Latan, 2015).

Structural Model or Inner Model
The structural model or inner model shows the relationship or the strength of estimates between latent variables or constructs based on substantive theory. a. Q-Square b. R-Square c. F-Square d. Goodness of Fit (GoF) Test e. Estimate For Path Coefficients

Direct and Indirect Effect Test
Tests using path analysis using SEM PLS. Path analysis shows the direct or indirect effect of the independent variable on the dependent variable with the mediating variable.
To see the direct influence between exogenous and endogenous variables, the Path Coefficients range from +1 to -1, the closer to +1 the relationship between the two construct variables is getting stronger, while the closer to -1 the relationship between the two variables is negative.
Meanwhile, to see the indirect effect, in the Specific Indirect Effect, if it has a p-value > 0.05, it indicates that the indirect effect of the variable (mediation variable) has an insignificant impact (Ghozali, 2021: 200).

Validity test
Based on the construct model in Figure 2, the results of the validity test are depicted in the table below:  (2021) From the data in table 4, convergent validity is seen from the loading factor value for each indicator. According to Hair et al (2017), the loading factor parameter with a rule of thumbs > 0.7 is considered to meet the convergent validity requirements. The research data above shows that the loading factor has a value above 0.7. So it can be said that the research data is said to be valid.  (2021) It can be seen that the cross loading value greater than 0.7 can be said to be valid, and the research data can be said to be valid.  (2021) From the data in table 6, the discriminant validity test is also carried out by looking at the discriminant validity value. The value of discriminant validity in this study is seen from the value of the root parameter of Average Variance Extracted (AVE) and the correlation of latent variables with the rule of thumb of the root of Average Variance Extracted (AVE) > correlation of latent variables. And the second is seen from the cross loading parameter value with a rule of thumbs > 0.7, then by fulfilling these conditions it will be declared that the research is valid.  (2021) The results of this study obtained the Average Variance Extracted (AVE) value of all variables above 0.5 which can be seen in table 7. This means that the research conducted has met the requirements of the convergent validity test. This is in accordance with the opinion of Ghozali and Latan (2015) Convergent validity of a construct with reflective indicators evaluated with the Average Variance Extracted (AVE) value should be equal to 0.5 or more.

Reliabilty Test
Testing the reliability of the instrument using the coefficient value of composite reliability and Cronbach's Alpha. An instrument is declared reliable if the Cronbach's Alpha value> 0.7 then it can be concluded that it is reliable. The reliability test in this study used Composite Realibilty and Cronbach's Alpha, showing the following results:

a. Composite Reliability
The reliability test was carried out by looking at the composite reliability value. The results of composite reliability are said to be reliable if they have a value above 0.7. The following is the composite reliability value in table 8.  (2021) From the research above, it was found that the composite reliability value above, it was concluded that the variables carried out by the study were said to be reliable with composite reliability above 0.7, so they were considered to have good reliability (Ghozali and Latan, 2015). From table 8, it can be seen that the composite reliability value is above 0.7, so all variables in this study are said to be reliable.

b. Cronbach's Alpha
The reliability test is also strengthened with Cronbach's Alpha with the recommended value above 0.7 table 9 shows the results of Cronbach's alpha.  Table 9 shows that the value of Cronbach's Alpha is that all constructs are above 0.7 which means that it meets the recommended value. The results of all reliability tests through composite reliability and Cronbach's alpha show that all indicators are reliable.

Inner Model Test
The inner model test is used to show whether there is a relationship or power estimate between latent variables or constructs. The inner model analysis uses 2 ways, with R-square and Q-square

R-square test
The R-square test was used to test the value of the independent variable on the dependent variable. Evaluation of the predictive ability of the model was carried out using Rsquare, providing criteria for R-square values of 0.67 (strong), 0.33 (moderate) and 0.19 (weak). (Jogianto and Abdilah, 2009) Table 10 shows the R-square value. The table shows the R-square value of Customer Satisfaction (M) = 0.700 = 70.0%, this means that the service claim and brand image variables together affect the customer satisfaction variable by 70% and the remaining 30% is influenced by other variables that are not researched. Meanwhile, the R-square value of customer loyalty is 0.728 = 72.8%, this means that the service claim, brand image and customer satisfaction variables affect the customer loyalty variable by 72.8% and the remaining 27.2% is influenced by other variables not examined. .

Q-Square Test
After seeing the R-square value, the inner model analysis can be done by calculating the Q-square value. The Q-square value can be searched by performing calculations using the formula, as follows: The results of the calculation of the Q-square formula obtained a value of 0.918 or 91.8%. Thus, the research data explains as much as 91.8% and the remaining 8.2% are other factors that are not related to the research conducted.

Hypothesis test
Hypothesis testing using path analysis (path analysis) using SEM PLS. Path analysis shows the direct or indirect effect of the independent variable on the dependent variable with the mediating variable.
The bootstrapping method can be used for various things, one of which is to determine the value of t-statistics as is done in the Partial Least Square SEM model. The bootstrapping method or resampling up to 5000 times will be able to calculate the Standard Deviation value so that it can then calculate the t-statistical value by dividing the regression coefficient by the Standard Deviation. Significance testing is carried out to determine the significance of direct and indirect effects, where the T-statistics requirement must be greater than the T-value. T-value used is 1.96 Sources: SmartPLS (2021)   (2021) Hypothesis measurement can make comparisons between t-table and t-statistic values. The probability value, the p-value with 5% alpha is <0.05, with a t-table value of 5% alpha is 1.96 If the t-statistic value is higher than the t-table value, it means that the hypothesis is accepted, and vice versa if the t-statistic value is lower than the t-table, meaning that the hypothesis is rejected. This study tested four existing hypotheses, as follows:

a. First Hypothesis
The first hypothesis test is to test the influence of service claims on customer loyalty of PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Service claim has no effect on customer loyalty H1: Service claim has a positive and significant effect on customer loyalty.
Based on table 11, the service claim variable has a positive and significant effect on the customer loyalty variable with a t-statistic of 2,267 > 1.96, with a significance value of 0.024 <0.05. Thus H0 is rejected and H1 is accepted. So the first hypothesis is accepted, meaning that service claims have a positive and significant influence on purchasing decisions. This means that if the service claim provided by the company is good, then this will lead to customer loyalty to PT Asuransi Sahabat Artha Proteksi

b. Second Hypothesis
The second hypothesis test tested the influence of brand image on customer loyalty of PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Brand image has no influence on customer loyalty. H2: Brand image has a positive and significant influence on customer loyalty Based on table 11, the brand image variable has an effect on the customer loyalty variable with a t-statistic of 3.093 > 1.96, with a significance value of 0.002 <0.05. Thus H0 is rejected and H2 is accepted. So the second hypothesis is accepted, meaning that brand image has a positive and significant influence on customer loyalty. This means that if the company's brand image is good, it will lead to customer loyalty to PT Asuransi Sahabat Artha Proteksi

c. Third Hypothesis
The third hypothesis test tested the influence of service claims on customer satisfaction at PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Service claim has no effect on customer satisfaction H3: Service claim has a positive and significant effect on customer satisfaction.
Based on table 11, the service claim variable has a positive and significant effect on the customer satisfaction variable with a t-statistic of 4.942 > 1.96, with a significance value of 0.000 <0.05. Thus H0 is rejected and H3 is accepted. So the third hypothesis is accepted, meaning that service claims have a positive and significant effect on customer satisfaction. This means that if service claims are easier to do, then this will lead to increased customer satisfaction at PT Asuransi Sahabat Artha Proteksi.

Fourth Hypothesis
The fourth hypothesis test is to test the influence of brand image on customer satisfaction of PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Brand Image has no influence on Customer Satisfaction. H4: Brand Image has a positive and significant effect on customer satisfaction.
Based on table 11, the brand image variable has an influence on the customer satisfaction variable with a t-statistic of 3.746 > 1.96, with a significance value of 0.000 <0.05. Thus H0 is rejected and H4 is accepted. So the fourth hypothesis is accepted, meaning that brand image has a positive and significant influence on customer satisfaction. This means that the better the brand image of a company, it will further increase customer satisfaction.

Fifth Hypothesis
The fifth hypothesis test tested the influence of customer satisfaction on customer loyalty at PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Customer satisfaction has no effect on customer loyalty. H5: Customer satisfaction has a positive and significant influence on customer loyalty.
Based on table 11, the customer satisfaction variable has an influence on the customer loyalty variable with a t-statistic of 3.444 > 1.96, with a significance value of 0.001 <0.05. Thus H0 is rejected and H5 is accepted. So the fifth hypothesis is accepted, meaning that customer satisfaction has a positive and significant influence on customer loyalty. This means that if customer satisfaction increases, then it will increase customer loyalty.  (2021) From the results of the significance test for the mediating effect above, the hypothesis can be explained as follows:

a) Sixth Hypothesis
The sixth hypothesis test tested the role of customer satisfaction in mediating the effect of service claims on customer loyalty at PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: Customer satisfaction is not able to mediate the effect of service claims on customer loyalty. H6: customer satisfaction is able to mediate the effect of service claims on customer loyalty.
Based on table 12, the customer satisfaction variable was able to mediate the effect of service claim on the customer loyalty variable with a t-statistic of 2.722 > 1.96, with a significance value of 0.007 <0.05. Thus H0 is rejected and H6 is accepted. So the sixth hypothesis is accepted, meaning that customer satisfaction mediates the effect of service claims on customer loyalty. This means that if the service claims provided are getting better, it will increase customer satisfaction, then this will further increase customer loyalty of PT Asuransi Sahabat Artha Proteksi.

b) Seventh Hypothesis
The seventh hypothesis test tested the role of customer satisfaction in mediating the influence of brand image on customer loyalty at PT Asuransi Sahabat Artha Proteksi, with the following hypothesis: H0: customer satisfaction is not able to mediate the effect of brand image on customer loyalty. H6: customer satisfaction is able to mediate the effect of brand image on customer loyalty.
Based on table 12, the customer satisfaction variable was able to mediate the effect of brand image on the customer loyalty variable with a t-statistic of 2.603 > 1.96, with a significance value of 0.010 <0.05. Thus H0 is rejected and H7 is accepted. So the seventh hypothesis is accepted, meaning that customer satisfaction mediates the effect of brand image on customer loyalty. This means that the better the brand image of a company, the higher the level of customer satisfaction, then this will have an impact on customer loyalty of PT Asuransi Sahabat Artha Proteksi.

V. Conclusion
Based on the research conducted, the conclusions obtained are as follows: a. Service claims have an influence on customer loyalty. So it can be said, an easy service claim will affect customer loyalty at PT Asuransi Sahabat Artha Proteksi. b. Brand image has an influence on customer loyalty. So it can be said, a good brand image of a company will affect customer loyalty. c. Service claims have an influence on customer satisfaction. So it can be said, the service claim felt by customers will determine the level of customer satisfaction. d. Brand image has an influence on customer satisfaction. So it can be said, a good brand image in the minds of customers will affect the level of customer satisfaction. e. Customer satisfaction has an influence on customer loyalty. So it can be said, if a customer is satisfied with a company, it will increase their loyalty to the company. f. Customer satisfaction is able to mediate the effect of service claims on loyalty. So it can be said, if the service claims provided are getting better, it will increase customer papacy which in turn will increase customer loyalty. g. Customer satisfaction is able to mediate the influence of brand image on customer loyalty.
So it can be said, if the company's brand image is getting better, it will increase customer satisfaction and customer loyalty.