Price Dominance on Loyalty and Repurchase of Bottled Water

( Dominansi Harga terhadap Loyalitas dan Pembelian Ulang Air Minum Dalam Kemasan )

Joice Putri Ulina Panjaitan^*1), Minto Waluyo²⁾

¹⁾Program Studi Teknik Industri, Universitas Pembangunan Nasional “Veteran” Jawa Timur, Indonesia

²⁾ Program Studi Teknik Industri, Universitas Pembangunan Nasional “Veteran” Jawa Timur, Indonesia

*Email Penulis Korespondensi: 22032010137@student.upnjatim.ac.id

Abstract . The high demand for bottled water has triggered intense competition within the industry. This study aims to analyze the effects of brand image, brand awareness, price, and promotions on purchase decisions, brand loyalty, and repurchase intention for BW X in Surabaya. Data were collected from 180 respondents through an online questionnaire and analyzed using CB-SEM via AMOS 26. The results indicate that brand image and price have a significant positive effect on purchase decisions, whereas brand awareness and promotions do not. Regarding brand loyalty, price and brand awareness exert a significant positive influence, while promotions and brand image are insignificant. For repurchase intention, only price proves to have a significant positive impact. Furthermore, purchase decisions significantly enhance brand loyalty, which in turn drives repurchase intention. These findings emphasize the importance of competitive pricing and the need for BW X to create better promotional innovations to foster future brand loyalty .

Keywords - B ottled Water ; Brand Loyalty ; Purchase Decisions ; Repurchase Intention

I. INTRODUCTION

The need for clean water positions Bottled Water (Air Minum Dalam Kemasan or AMDK) as an essential commodity for modern society. Based on the 2023 National Socioeconomic Survey (Susenas) data, 40.64% of Indonesians consume bottled water daily, and this figure surges drastically to 96.01% in urban areas like Surabaya 1 2. This high demand has triggered massive industry growth while creating fierce market competition, as evidenced by the thousands of bottled water products registered with the Indonesian Food and Drug Authority / BPOM 3. In this dynamic market landscape, producers are required to precisely understand the drivers of consumer behavior to maintain market share and win the competition. Therefore, continuous evaluation of market dynamics is imperative, as the massive proliferation of domestic brands intensifies the struggle for consumer attention and long-term retention 3, 1.

As a new player launched by a leading Fast-Moving Consumer Goods (FMCG) company, BW X faces an existential challenge in penetrating the dominance of established brands. Building brand awareness from scratch is a crucial initial foundation, as consumers often rely on familiar brands to reduce perceived risks and fulfill broader social or emotional values in daily consumption products 4. However, amidst these introductory efforts, BW X was confronted with a reputation crisis due to a viral negative issue on social media platforms regarding doubts about its water quality and source clarity. This issue directly threatens the brand image of BW X, which theoretically heavily influences the level of consumer trust, a fundamental basis for consumers in evaluating an AMDK brand before making a purchase decision 5.

To systematically evaluate the complex dynamics among a tested brand image, low brand awareness, competitive pricing, and promotional offers, this study draws upon two foundational frameworks: Customer-Based Brand Equity (CBBE) and the Theory of Planned Behavior (TPB). According to the CBBE framework, brand awareness (salience) and brand image (meaning) act as cognitive building blocks that shape the initial purchase decision and eventually culminate in brand loyalty (resonance) 6 7. Once this loyalty is established, TPB provides a mechanism to understand how this profound commitment (acting as a positive attitude), combined with pragmatic evaluations of price and promotion (acting as perceived behavioral control), dictates the ultimate behavioral outcome: repurchase intention. Furthermore, empirical evidence supports that a well-perceived brand image has a significant impact on continuously driving consumers' repurchase intentions through loyalty 8.

Various prior literatures have examined the relationships between these marketing variables and consumer behavior, yet they show inconsistent results. There is an academic debate where s study found that brand image has a dominant effect on purchase decisions in the bottled water industry 9 while other studies emphasized the price and promotion factors more, stating that brand image is insignificant 10 11. Conversely, other research indicates that brand awareness, brand image, price, and promotion simultaneously exert a significant influence on purchase decisions 12. The novelty of this study lies in its unique context and comprehensive approach: it explicitly evaluates consumer behavior toward a new brand navigating a viral reputation crisis alongside aggressive promotional strategies, translating these complex dynamics into predictive mathematical model. Therefore, the purpose of this study is to investigate the influence of brand image, brand awareness, price, and promotion on purchase decisions, brand loyalty, and repurchase intentions related to BW X products. It also aims to analyze the impact of purchase decisions on brand loyalty and how brand loyalty affects repurchase intentions. By examining these relationships, the study seeks to provide a comprehensive understanding of the key factors driving consumer behavior toward BW X in Surabaya. Additionally, the research intends to develop a mathematical model that accurately represents the interactions between these variables. This model will serve as a valuable tool for predicting and enhancing customer loyalty and repurchase behavior.

II. MetHod

This study employed a quantitative approach using a survey method through the distribution of a 5-point Likert scale questionnaire (ranging from 1 = Strongly Disagree to 5 = Strongly Agree). The sample size determination technique referred to the Maximum Likelihood Estimation (MLE) guidelines, which require a representative sample of 100 to 200 observational units to maintain estimation stability and statistical power 13. The respondents were selected using a purposive sampling technique, with the specific criteria for inclusion were individuals residing in Surabaya who have purchased BW X at least once.

In this study, brand image is measured through brand benefit and competence, brand association, and reputation 14,15. The operational indicators for brand awareness include brand recall, brand salience during consumption, brand recognition, and likeability 16 17. Price is assessed based on affordable price, competitive price, price-quality alignment, and price-benefit alignment 18. The promotion variable is evaluated through media, frequency, quality of message, promotional reach, attractive gift offers, and discount amount 19 20 21. Purchase decision is measured using indicators such as product attachment, recommendations to others, brand choice, and distributor choice 22 23. Loyalty is assessed through satisfaction, brand linkage, prioritizing the brand, and recommending the brand to others 24 25. Lastly, repurchase intention is measured by continuous consumption, sharing experiences, trust, reliability, and functionality 26 27.

The collected data were analyzed using the Covariance-Based Structural Equation Modeling (CB-SEM) approach with the assistance of AMOS software. This approach was selected due to its advantage in simultaneously testing unnomplex theoretical models 28. Before evaluating the structural relationships, the data underwent classical assumption tests. Multivariate normality was assessed using the Critical Ratio (C.R.) of multivariate kurtosis, while multivariate outliers were detected using the Mahalanobis distance () criterion. Following the assumption tests, the data were evaluated through a measurement model (Confirmatory Factor Analysis) to ensure the validity and reliability of the instruments. Convergent validity is met if the loading factor () value is significant and the Critical Ratio (CR) , while discriminant validity was tested by comparing the correlation values between constructs. Furthermore, inter-item correlations were examined to detect potential multicollinearity, ensuring correlation coefficients remained below 0.8, while internal consistency was confirmed when the construct reliability value met the minimum threshold of 0.70 13.

Once the instruments were declared valid and reliable, the testing proceeded to the structural model to evaluate the causality among variables. The feasibility of the theoretical model was evaluated using comprehensive Goodness of Fit criteria. In instances where the initial structural model failed to meet these stringent fit criteria, model modifications were executed by referring to the Modification Indices (MI), provided that the Standardized Residual values were maintained within the acceptable boundary of -2.58 to 2.58. Ultimately, once the modified model achieved an ideal fit, hypothesis testing was conducted by analyzing the direction, magnitude, and statistical significance of the standardized path coefficients within the simultaneous equation model.

According to 13 model is considered fit with the empirical data if it meets the following criteria, as presented in table 1.

Table 1. Cut-Off Value of GoF Index 13

III. Results and Discussion

The questionnaire was distributed throughout the Surabaya area using Google Forms, targeting respondents who met the criteria of having purchased BW X products at least once. The table 2 presents the analysis of the respondents' characteristics.

Table 2.Respondent’s Highest Level of Education

From table 2, respondents’ highest level of education shows that the majority, 40%, have completed high school or an equivalent. Close behind, 39% of respondents hold a diploma ranging from Diploma I to Diploma III. Additionally, 17% have attained a Diploma IV or Bachelor's degree, while only a small portion, 2%, have completed elementary or junior high school. Similarly, 2% of respondents hold a Master's or Doctorate degree. This distribution indicates that most respondents have education levels ranging from secondary to undergraduate, reflecting a relatively well-educated sample.

In addition, table 3 presents the occupations of the respondents. The data in this table shows a variety of jobs held by respondents. Thus information provides a more comprehensive picture of socio-economic backgrounds of the respondents participating in this study.

Table 3.Respondent’s Employment Status

From table 3, the employment status of the respondents reveals that the largest group, 41%, are college students, followed by 26% who are civil servants. Freelancers make up 18% of the respondents, while unemployed individuals account for 4%. A smaller portion, 3%, are students, and 8% fall into other employment categories. Overall, this distribution shows a diverse range of employment backgrounds, with a strong representation from the student and civil servant populations.

1. Respondents Profile
2. Assumption Tests

Prior to evaluating the structural model, data screening was conducted to ensure the dataset met the required assumptions for CB-SEM analysis, specifically regarding multivariate outliers and normality.

The detection of multivariate outliers was performed using the Mahalanobis distance () criterion. Given that the model consists of 30 indicators, the Chi-square threshold at a significance level of p < 0.001 and df = 30 is 59.703. The results of the normality assessment, including the skewness and kurtosis coefficients for each indicator as well as the multivariate critical ratio, are summarized in Table 4.

Table 4. Assesment of Univariate and Multivariate Normality

Based on the table 4, five observations (numbers 45, 126, 169, 44, and 62) were identified as exceeding this threshold, with the highest value being 67.478. However, these observations were retained in the analysis as they represent genuine variations in consumer behavior toward the brand, and their inclusion ensures the representativeness of the sample.

1. Multivariate Outlier Test
2. Normality Test

The normality of the data was assessed through univariate and multivariate normality tests. The results of the normality assessment, including the skewness and kurtosis coefficients for each indicator as well as the multivariate critical ratio, are summarized in Table 5.

Table 5. Assessment of Normality

Based on Table 5, the multivariate Kurtosis critical ratio (c.r.) was 16.956, which is above the ±2.58 threshold, indicating that the assumption of multivariate normality was not strictly met. However, univariate normality was maintained, as all items exhibited skewness values between -1.430 and -0.572 and kurtosis values between -0.495 and 1.788. According to 29, data distributions with skewness and kurtosis coefficients within the range of ±2.0 are considered acceptable and do not severely deviate from the normal distribution, particularly in large-scale assessments. Since all indicators in this study fell within these specified limits, the data were deemed suitable for further analysis using Maximum Likelihood (ML) estimation, which is known to be robust to moderate non-normality when univariate distributions are within acceptable boundaries.

As a fundamental step in Covariance-Based Structural Equation Modeling (CB-SEM), the conceptual model was visualized through a path diagram using AMOS. This diagram precisely maps the causal relationships between four exogenous variables (Brand Image, Brand Awareness, Price, and Promotion) and three sequential endogenous variables (Purchase Decision, Brand Loyalty, and Repurchase Intention) representing BW X consumer behavior. This visualization serves as the mathematical foundation for estimating regression coefficients and evaluating the overall goodness of fit. The path diagram is presented as follows, as shown in figure 1.

Figure 1. Path diagram Purchase Decision, Brand Loyalty, and Repurchase Intention BW X in Surabaya

To test the causal relationships, the instrument's feasibility was evaluated through a measurement model using CB-SEM via AMOS 26 software. Shown in figure 2, displays the measurement model used in this study.

Figure 2. Measurement Model of Purchase decision, Brand loyalty, dan Repurchase intention BW X in Surabaya

The feasibility of the instrument and the Goodness of Fit of the measurement model were rigorously evaluated ahead of casual testing. The initial measurement model demonstrated a good fit in several key criteria, specifically CMIN/DF (1.328 2.00) and RMSEA (0.043 0.08). Although the Chi-Square was significant (509.801) and certain comparative indices were marginal (GFI = 0.847, AGFI = 0.815, TLI = 0.929, CFI = 0.937), the model remains statistically viable. This confirming excellent internal consistency and qualifying the data for subsequent structural refinement. Validity and significance tests of the research instrument were conducted, as presented in table 4.

Table 6. Validity and Significancy Test of indicators

The convergent validity test results from the table 4 showed that all indicators utilized possessed a Critical Ratio () value greater than its with a significance level 1.697. Based on a significance level of 0.05 and degrees of freedom (df) = 30 (total number of indicators). This proves that each item validly represents its respective construct. Construct Reliability (CR) was calculated using the formula (1).

............................................................................(1)

In this formula, Std. Loading represents the standardized factor loadings of the indicators, while denotes the error variances associated with each indicator. The results of the construct reliability calculations are presented in table 5.

Table 7. Construct Reliability for Measurement Model

Table 5 presents the construct reliability for the measurement model, showing the estimated standardized regression weights and error values for each indicator within their respective constructs. Each construct, labeled from X1 to Y3, consists of multiple indicators with varying regression weights, reflecting their strength in representing the underlying construct. The construct reliability values range from 0.772 to 0.928, indicating good internal consistency across all constructs. Higher regression weights suggest stronger indicators, while the error values represent the measurement uncertainty for each item. Overall, the table confirms that the measurement instrument demonstrates satisfactory reliability, making it suitable for further analysis.

The structural model tested in this study is shown in figure 3. It depicts the relationships between the latent variables (X1 to X4 as exogenous variables and Y1 to Y3 as endogenous variables) along with their observed indicators. The model's goodness of fit indices are also presented, indicating an acceptable fit between the hypothesized model and the observed data.

Figure 3. Structural Model of Purchase decision, Brand loyalty, dan Repurchase intention BW X in Surabaya

The initial evaluation of the structural model indicated that the empirical data did not perfectly fit the hypothesized model. The initial Goodness of Fit tests yielded a significant Chi-Square value of 721.690 and marginal comparative fit indices (GFI = 0.784, AGFI = 0.743, TLI = 0.817 and CFI = 0.835). Although the CMIN/DF and RMSEA were within acceptable limits the overall results necessitated model refinement.

An examination of the Modification Indices (M.I.) revealed significant redundancies among several error terms, most notably a high index value (33.590) between error variable. To resolve these misspecifications and improve model fit without violating theoretical assumptions, a model modification was executed by correlating the error terms exhibiting the M.I. values. This targeted modification effectively reduced the Chi-square value, subsequently elevating the comparative indices and further minimizing the RMSEA.

This modified model is the result of refining the initial model to improve the fit between the model and the empirical data. Figure 4 illustrates the structural relationships among purchase decision, brand loyalty, and repurchase intention for BW X in Surabaya.

Figure 4. Modification Model of Purchase decision, Brand loyalty, dan Repurchase intention BW X in Surabaya

The initial evaluation of the structural model indicated that the empirical data did not fully fit the conceptual model. necessitating model modification. Based on the Modification Indices (M.I.) guidelines, the modification focused on correlating error terms that exhibited high redundancy without violating theoretical assumptions. Following the modification, the final model demonstrated a highly significant and ideal Goodness of Fit. The probability value improved to 0.359 (> 0.05) accompanied by a Chi-Square value of 363.277 (df = 359). Other primary fit indicators were perfectly met: CMIN/DF stood at 1.012 ( 2.00) and RMSEA reached a remarkably minimal error rate of 0.008 ( 0.08). Although GFI (0.889) and AGFI (0.856) remained at marginal levels, major comparative indices such as TLI (0.997) and CFI (0.998) exceeded the excellence threshold ( 0.95). To provide further insight, the table 6 presents the Standardized Regression Weight for the modification model.

Table 8. Standardized Regression Weight for Modification Model

According to the Standardized Regression Weight table for the modification model presented in table 6, following the confirmation of the model's fit, the evaluation proceeds to assess the validity and significance of the causal relationships within this final structural model. The estimation results, as presented in the structural regression weights table, utilize the Critical Ratio (C.R.) and Probability (P) values to determine the significance of each specific path. A structural path is considered statistically valid and significant if the absolute C.R. value exceeds 1.697 which ( ). Based on the output, several pathways demonstrate robust statistical significance, notably those exhibiting C.R. values of 2.269 (p = 0.023), 2.313 (p = 0.021), 2.226 (p = 0.026), 2.355 (p = 0.019), and 2.677 (p = 0.007). Additionally. one path shows a C.R. of 1.818 (p = 0.069). Conversely, paths possessing C.R. values below the acceptable threshold above the 1.697 limit, fail to meet the significance criteria. The relationships defined by these particular paths are therefore deemed statistically insignificant. The distribution of these valid and invalid parameters, alongside their Estimated Standardized Regression Weights (ESRW), serves as the definitive groundwork for concluding the simultaneous equations and the subsequent hypothesis testing.

1. Path Diagram
2. Measurement Model
3. Structural Model
4. Modification Model
5. Structural Equations (Simultaneous Equations)

Following the attainment of an optimal structural model fit, the simultaneous equations were formulated to mathematically represent the causal relationships among the latent variables. These equations utilize the standardized regression weights to illustrate the magnitude and direction of the effects from the exogenous variables (Brand Image, Brand Awareness, Price, Promotion) to the endogenous variables (Purchase Decision, Brand Loyalty, Repurchase Intention). The formulated structural equations for the BW X brand are:

= f () + f () + f () + f ()

= 0.415 – 0.173 + 0.372 + 0.150

The equation indicates that Purchase Decision is positively and predominantly influenced by Brand Image () with a coefficient of 0.415. followed closely by Price () at 0.372. Conversely, Brand Awareness () displays a negative coefficient of -0.173. while Promotion () contributes weakly but positively with a coefficient of 0.150.

= ff () + f() + f () + f () + f ()

= 0.292 (0.415 – 0.173 + 0.372 + 0.150) -0.055 + 0.547 + 0.311 - 0.167

= 0.121 - 0.051 + 0.109 + 0.044 - 0.055 + 0.547 + 0.311 - 0.167

= 0.066 + 0.496 + 0.420 – 0.123

This equation shows that Brand Loyalty is highly influenced by Brand Awareness () with the largest coefficient of 0.547, followed by Price () at 0.311. In contrast, Brand Image () and Promotion () exhibit negative structural paths with coefficient of -0.055 and -0.167, diminishing their roles in building long-term loyalty.

1. Equation 1: Purchase Decision
2. Equation 2: Brand Loyalty
3. Equation 3: Repurchase Intention

= fff () + f() + f () + f () + f ()

= 0.737 (0.292 (0.415 – 0.173 + 0.372 + 0.150) -0.055 + 0.547 + 0.311 - 0.167) – 0.092 – 0.166 + 0.325 + 0.100

= 0.737 (0.066 - 0.598 + 0.420 – 0.123) – 0.092 – 0.166 + 0.325 + 0.100

= 0.049 - 0.366 + 0.309 – 0.091 – 0.092 – 0.166 + 0.325 + 0.100

= – 0.043 + 0.200 + 0.634 + 0.009

The final equation reveals that Repurchase Intention is Price () emerges as the most dominant and robust predictor, contributing a massive coefficient of 0.634, followed by Brand Awareness () at 0.200. Brand Image () has a small negative effect with a coefficient of -0.043, while Promotion () shows a very minor positive influence with a coefficient of 0.009.

Prior to detailing the structural relationships, a summary of the hypothesis testing results based on the modification model is presented in Table 7. The evaluation compares the Critical Ratio (C.R.) against the ( = 1.697). The table below systematically presents the detailed outcomes of the hypothesis testing, including the critical ratios and their comparison against the established significance threshold.

Table 9. Hypothesises Results

the modification structural estimation reveals that Brand Image has a positive and significant effect on Purchase Decision. Similarly, Price serves as a strong and significant driver. Thus, H1 and H3 are supported. Conversely. Brand Awareness and Promotion do not yield significant impacts, meaning H2 and H4 are rejected. This indicates that for consumers of BW X in Surabaya, a positive brand reputation and competitive pricing align better with their immediate purchasing parameters than promotional exposure or basic brand recognition.

In shaping consumer loyalty, Brand Awareness inversely acts as a highly significant positive predictor, supported by favorable Price perceptions. Furthermore, the prior Purchase Decision significantly reinforces Brand Loyalty. Consequently, H6, H7, and H13 are supported. In contrast, Brand Image and Promotion fail to significantly influence loyalty, rejecting H5 and H8. This proves that long-term loyalty for BW X is built upon deep-rooted brand familiarity (awareness), sustained price satisfaction, and the success of the initial buying experience, rather than superficial promotional cuts.

Testing the direct effects on Repurchase Intention shows that Price is the sole external factor exerting a significant direct influence, thereby supporting H11. The other external variables Brand Image, Brand Awareness, and Promotion, do not reach significance, rejecting H9, H10, and H12. Most importantly, the structural model successfully validates a crucial progression: Brand Loyalty acts as the strongest and most significant predictor of Repurchase Intention, firmly supporting H14. This highlights that cultivating profound brand loyalty is the most vital mechanism to guarantee continuous consumption and positive post-purchase behavior for BW X.

1. Hypothese s Testing
2. Discussion

The structural estimation results provide a comprehensive map of consumer behavior for BW X in Surabaya. While several hypotheses were supported, the emergence of non-significant results in certain pathways offers a critical point for analytical discussion, particularly considering the brand's current status as a new player facing a reputation crisis.

The findings reveal that Brand Image (H1) and Price (H3) are the primary drivers of purchase decisions, whereas Brand Awareness (H2) is not. Viewed through the CBBE framework, this confirms that for BW X brand meaning (image) is far more critical than mere brand salience (awareness). Consumers in Surabaya do not purchase simply because they "know" the brand; rather, they evaluate the competence of the parent FMCG company behind BW X as a guarantee of quality. This suggests that a positive image serves as a vital defense mechanism against perceived risks. Furthermore, the insignificance of Promotion (H4) indicates that in a saturated urban AMDK market, aggressive discount strategies or advertisements lose their efficacy if not accompanied by strong brand trust. This supports the findings of 10, which state that pragmatic factors such as price and image are more decisive in the bottled water industry than momentary promotional stimuli.

In alignment with the CBBE hierarchy, loyalty or brand resonance represents the pinnacle of the consumer experience. The significance of Purchase Decision (H13) and Price (H7) toward loyalty underscores that post-purchase satisfaction and consistent economic value are fundamental prerequisites for loyalty. Interestingly, Brand Awareness (H6), which did not influence the initial purchase, emerged as a significant predictor of loyalty. This indicates that brand awareness plays a "long-game" role; once a consumer tries the product, familiarity with the brand's presence becomes a psychological anchor that prevents brand switching. Conversely, the failure of Brand Image (H5) and Promotion (H8) to directly influence loyalty proves that loyalty for BW X in Surabaya is "performance-based." While image may attract consumers to try the product (H1), only product quality and price stability can keep them loyal. Promotion is shown to be merely transactional and incapable of fostering long-term emotional commitment 24.

Through the lens of the Theory of Planned Behavior (TPB), repurchase intention is understood as a behavioral intention governed by attitudes and perceived behavioral control. The finding that Brand Loyalty (H14) is the strongest predictor of repurchase intention confirms that loyalty has been deeply internalized as a positive attitude. Loyal consumers automatically intend to consume BW X again without needing to re-evaluate the brand at every purchase occasion. However, the direct significance of Price (H11) on repurchase intention indicates the presence of perceived behavioral control. Even when consumers feel loyal, financial control (price rationality) remains a determinant of whether that intention will be realized. Meanwhile, the insignificance of Brand Image (H9) and Brand Awareness (H10) toward repurchase intention reinforces the role of Loyalty as a full mediator. Elements of brand equity must first undergo the process of loyalty formation before they can trigger long-term behavioral intentions 8.

1. The Drivers of Purchase Decision
2. Transformation of Experience to Brand Loyalty
3. Formation of Repurchase Intention
4. Novelty and practical Implication

The novelty of this research lies in uncovering consumer behavior patterns toward a new brand facing a reputation crisis. This study proves that brand image serves as the "entry ticket" to the market, while brand awareness and price stability act as the "retention tools." For BW X management, these results suggest a strategic shift: rather than allocating excessive budgets to mass promotions (which proved insignificant for loyalty), it is more effective to focus on stabilizing brand trust through quality communication and maintaining price competitiveness at the retail level to secure long-term repurchase cycles.

IV. Conclusions

Based on the structural model testing of BW X consumer behavior in Surabaya, this study draws several key conclusions regarding the dynamics of consumer decision-making. Initially, during the purchase decision stage, consumers are significantly influenced by Brand Image and Price. They prioritize the product's reputation and the alignment of price with the benefits received, whereas mere Brand Awareness or exposure to Promotion is insufficient to trigger an initial purchase. As consumers transition towards building Brand Loyalty, the determining factors shift. Brand Awareness and Price stability emerge as the primary drivers, supported by the positive experience from their previous purchase decision. Conversely, promotional strategies relying heavily on discounts indicate a negative impact on the formation of long-term loyalty, confirming the occurrence of spurious loyalty where consumers easily switch brands once the promotional period ends. Ultimately, in achieving Repurchase Intention, Price stands as the only external stimulus with a significant direct effect. However, the most crucial finding of this structural modeling is that Brand Loyalty acts as the absolute and dominant mediator. Consumers exhibit high repurchase intentions only when they have passed through a satisfactory purchase decision phase and have established genuine loyalty to the BW X brand.

References

[2]BPS, “Distribusi Persentase Rumah Tangga Menurut Kabupaten/Kota dan Sumber Air Minum di Provinsi Jawa Timur, 2023,” Badan Pusat Statistik. Accessed: Nov. 14, 2025. [Online]. Available: https://jatim.bps.go.id/id/statistics-table/3/YzBaMlduSlFVbTVrUnpWeU9YRTJka0pVTTFkU1FUMDkjMyMzNTAw/distribusi-persentase-rumah-tangga-menurut-kabupaten-kota-dan-sumber-air-minum-di-provinsi-jawa-timur.html?year=2023

[3]Joniwarta, W. Priatna, A. R. Hamdani, and A. D. Alexander, “Implementasi Fuzzy Logic Pada Sistem Kontrol pH Air Mineral Berbasis IOT,” Indones. J. Comput. Sci., vol. 12, no. 4, pp. 2064–2077, 2023, doi: 10.33022/ijcs.v12i4.3356.

[4]M. Nuril Ihsan and B. Sutedjo, “Pengaruh brand image dan brand trust terhadap kepuasan konsumen dan dampaknya terhadap minat beli ulang,” Forum Ekon., vol. 24, no. 1, pp. 170–176, 2022, doi: 10.30872/jfor.v24i1.10586.

[5]M. Dwi Setiawan, “Pengaruh Kualitas Harga Dan Promosi Produk Terhadap Keputusan Pembelian Air Mineral Dalam Kemasan (Amdk) Merek Aqua,” J. Cafe., vol. 3, no. 2, pp. 29–37, 2022, doi: 10.51742/akuntansi.v3i2.613.

[6]M. P. Putra and S. Purwanto, “The Influence of Brand Image and Brand Awareness on Brand Loyalty in Packaged Drinking Water Ades in Surabaya City,” J. Ekon. dan Bisnis Digit., vol. 2, no. 4, pp. 1137–1146, 2023, doi: 10.55927/ministal.v2i4.5899.

[7]H. D. Putra, D. Himawati, D. Suzana, and D. M. Oktavilantika, “The Influence of Social Media Marketing, e-WOM, Brand Awareness, and Brand Image on Brand Loyalty of Herbs Product,” Almana J. Manaj. dan Bisnis, vol. 7, no. 1, pp. 77–83, 2023, doi: 10.36555/almana.v7i1.2105.

[8]R. P. Aggarini and S. K. Prastiwi, “The Influence of Green Marketing, Lifestyle, And Brand Image on Repurchase Intention,” Glob. Rev. Econ. Bus. Strateg., vol. 1, no. 1, pp. 57–70, 2025, doi: 10.34013/jk.v8i1.1410.

[9]L. Fransisca, S. Thomas, Andi, and R. Putra, “Analysis Of The Effect Of Price, Advertising, And Brand Image On Purchasing Decisions Of Aqua In Pekanbaru City,” Lucr. J. Bisnis Terap., vol. 4, no. 2, pp. 222–234, 2024, doi: 10.33369/bicemba.2.2024.62.

[10]R. S, A. R. Mus, H. Damis, and A. Mahmud, “The Intervening of Purchase Decision Between Marketing Mix, Consumer Behaviour and Brand Image on Costumer Statisfaction,” J. Manaj. Bisnis, vol. 8, no. 2, pp. 382–396, 2021, doi: 10.33096/jmb.v8i2.925.

[11]F. Herman, R. Indra, Kurniawati, M. Christiawan, and M. Aras, “Factors Influencing Brand Image and Purchase Intention in Indonesia’s Furniture Distribution Channels,” J. Distrib. Sci., vol. 22, no. 7, pp. 33–42, 2024, doi: 10.15722/jds.22.07.2024.07.33.

[12]D. Octavia, R. Arifin, and M. K. A. B. Suharto, “Pengaruh Brand Awareness, Brand Image, Kualitas Produk, Harga dan Promosi Terhadap Keputusan Pembelian Air Minum dalam Kemasan (AMDK) Merek AQUA (Studi Kasus pada Masyarakat di Kecamatan Pandaan Jawa Timur),” e-Jurnal Ris. Manaj., vol. 11, no. 23, pp. 10–27, 2022, [Online]. Available: https://jim.unisma.ac.id/index.php/jrm/article/view/30096/22761

[13]M. Waluyo and M. Rachman, Mudah Cepat Tepat dalam Aplikasi Sructural Equation Modeling, 2nd ed. Malang: PT. Literasi Nusantara Abadi Group, 2024.

[14]A. Liyono, “Pengaruh Brand Image, Electronic Word Of Mouth (E-WOM) dan Harga Terhadap Keputusan Pembelian Produk Air Minum Galon Crystalline Pada PT. Pancaran Kasih Abadi,” J. Ekon. Manaj. Dan Bisnis, vol. 3, no. 1, pp. 73–91, 2022, doi: 10.32815/jubis.v3i1.1089.

[15]K. Pandiangan, M. Masiyono, and Y. Dwi Atmogo, “Faktor-Faktor Yang Mempengaruhi Brand Equity: Brand Trust, Brand Image, Perceived Quality, & Brand Loyalty,” J. Ilmu Manaj. Terap., vol. 2, no. 4, pp. 471–484, 2021, doi: 10.31933/jimt.v2i4.459.

[16]N. Arianty and A. Andira, “Pengaruh Brand Image dan Brand Awareness Terhadap Keputusan Pembelian,” Maneggio J. Ilm. Magister Manaj., vol. 4, no. 1, pp. 39–50, 2021, doi: 10.30596/maneggio.v4i1.6766.

[17]C. Clarissa and I. Bernarto, “The Influence of Brand Ambassador, Brand Awareness, Brand Image and Prices on Purchase Decisions on Online Marketplace,” Bus. Entrep. Rev., vol. 22, no. 2, pp. 273–288, 2022, doi: 10.25105/ber.v22i2.14966.

[18]D. Darmawan and S. Arifin, “The Studi Empiris tentang Perilaku Konsumen Produk Air Minum dalam Kemasan Ditinjau dari Variabel Harga dan Kesadaran Merek,” Ideas J. Pendidikan, Sos. dan Budaya, vol. 7, no. 3, p. 179, 2021, doi: 10.32884/ideas.v7i3.434.

[19]E. M. Dilasari, G. Yosita, and V. F. Sanjaya, “Pengaruh Cita Rasa Dan Promosi Melalui Media Sosial Terhadap Keputusan Pembelian Konsumen Kopi Janji Jiwa Bandar Lampung,” REVENUE J. Manaj. Bisnis Islam, vol. 3, no. 1, pp. 25–42, 2022, doi: 10.24042/revenue.v3i1.10448.

[20]M. G. Maulana and W. Gunadi, “Pengaruh Promosi Dan Kualitas Pelayanan Terhadap Keputusan Pembelian Online Di Tokopedia Wilayah Dki Jakarta,” JIMEN J. Inov. Mhs. Manaj., vol. 3, no. 2, pp. 183–197, 2023, [Online]. Available: https://jom.unsurya.ac.id/index.php/jimen/article/view/119/339

[21]A. Nurdiansah and Widyastuti, “Pengaruh Price Discount Terhadap Customer Satisfaction Dan Repurchase Intention (Studi Pada Pengguna Shopee Food),” J. Ilm. Bid. Sos. Ekon. Budaya, Teknol. Dan Pendidik., vol. 1, no. 8, pp. 1495–1514, 2022, doi: 10.54443/sibatik.v1i8.198.

[22]A. S. Romadon, M. A. Pramusinto, and S. N. Kamelia, “Pengaruh Kualitas Produk, Citra Merek, Dan Harga Terhadap Keputusan Pembelian Melalui Shopee,” Solusi, vol. 21, no. 3, pp. 259–269, 2023, doi: 10.26623/slsi.v21i3.7080.

[23]R. Simangunsong and Morina, “Pengaruh Promosi, Harga Dan Kepercayaan Terhadap Keputusan Pembelian Di E-Commerce Shopee Pada Mahasiswa Universitas Hkbp Nommensen Medan,” J. Ilmu Sos. Dan Polit., vol. 1, no. 2, pp. 131–149, 2021, doi: 10.51622/jispol.v1i2.413.

[24]A. A. Annisawati and A. Situmeang, “Pengaruh Harga Dan Brand Experience Terhadap Brand Loyalty Pada Produk Teh Celup Walini Pt Perkebunan Nusantara (Ptpn) Viii Bandung,” J. Bisnis dan Pemasar., vol. 12, no. 1, pp. 2087–3077, 2022, [Online]. Available: https://ejurnal.ulbi.ac.id/index.php/promark/article/view/2131/951

[25]L. Q. Taslim, Faizatun, and I. H. Hatta, “Pengaruh Brand Image Dan Brand Attachment Terhadap Brand Loyalty Dengan Variabel Intervening Keputusan Pembelian (Studi Pada Pelanggan Apotek Sb),” COSTING J. Econ. Bus. Account., vol. 7, no. 6, pp. 6408–6420, 2024, doi: 10.31539/costing.v7i6.13183.

[26]L. Tataningtyas and E. Tjahjaningsih, “Faktor Yang Mempengaruhi E-Trust Serta Dampaknya Pada Repurchase Intention,” E-Bisnis J. Ilm. Ekon. dan Bisnis, vol. 15, no. 1, pp. 55–67, 2022, doi: 10.51903/e-bisnis.v15i1.616.

[27]M. F. Yusuf, R. Badi’ah, and A. F. Dedyansyah, “Meningkatkan Repurchase Intention melalui Keamanan dan Kepuasan Konsumen: Studi Kasus Marketplace di Indonesia,” JEBDEKER J. Ekon. Manaj. Akunt. Bisnis Digit. Ekon. Kreat. Entrep., vol. 4, no. 2, pp. 295–304, 2024, doi: 10.56456/jebdeker.v4i2.266.

[28]Z. Putlely, Y. A. Lesnussa, A. Z. Wattimena, and M. Y. Matdoan, “Structural Equation Modeling (SEM) untuk Mengukur Pengaruh Pelayanan, Harga, dan Keselamatan terhadap Tingkat Kepuasan Pengguna Jasa Angkutan Umum Selama Pandemi Covid-19 di Kota Ambon,” Indones. J. Appl. Stat., vol. 4, no. 1, p. 1, 2021, doi: 10.13057/ijas.v4i1.45784.

[29]S. Demir, “Comparison of Normality Tests in Terms of Sample Sizes under Different Skewness and Kurtosis Coefficients,” Int. J. Assess. Tools Educ., vol. 9, no. 2, pp. 397–409, 2022, doi: 10.21449/ijate.1101295.