Sustainable Retail Strategies and Impulse Buying within Urban Commercial Ecosystems: Evidence from an Indonesian Shopping Mall with Hedonic Motivation Mediation and Positive Emotion Moderation

The study of consumer behavior in modern retail has evolved from a purely transactional perspective toward a broader understanding of retail systems as socio-economic infrastructures embedded within urban development processes. Marketing channels, competitive dynamics, spatial configurations, and experiential design collectively shape the behavioral environment in which consumption decisions occur. Retailers function as the final node in the distribution chain, mediating product–consumer interactions through physical outlets and increasingly integrated digital platforms. As modern retail expands alongside urbanization and income growth, it becomes both an economic driver and a structural determinant of consumption intensity. Within this broader urban-commercial ecosystem, impulse buying represents a behavioral mechanism through which retail strategies translate environmental stimuli into immediate purchasing outcomes.

The theoretical roots of impulse buying can be traced to mid-twentieth-century scholarship. Rook [9] connected early corporate investigations by Du Pont de Nemours and Co. (1945–1965) to subsequent academic inquiry, while Clover [1] introduced the foundational observation that consumers frequently deviate from planned shopping lists. Stern’s theory [2] of impulsive purchasing remains the most influential conceptual contribution, challenging classical assumptions of rational consumer behavior such as Maslow’s hierarchy of needs and Engel et al.’s [10] structured decision-making model. Stern [2] emphasized the potency of in-store stimuli in redirecting purchasing behavior away from deliberative planning toward spontaneous action. This shift reframed impulse buying as a patterned response to environmental cues rather than an isolated deviation from rationality.

Subsequent research identified two broad categories of determinants: individual predispositions and environmental triggers. At the psychological level, consumers may respond to salient stimuli without systematically evaluating marginal utility, especially when stimuli are cognitively accessible or emotionally charged [4]. Behavioral manifestations include unplanned decisions, limited reflection on consequences, emotional susceptibility, and responsiveness to promotional cues. From a retail systems perspective, these characteristics suggest that impulse buying is structured by the configuration of commercial environments rather than solely by internal consumer traits.

Among environmental stimuli, sales promotion constitutes one of the most visible and strategically deployed instruments. Retailers activate promotions across multiple touchpoints, including external advertising, in-store displays, digital media, loyalty programs, and price discounts. Shelf visuals, promotional signage, basket messages, and audiovisual installations are designed to heighten attention and accelerate decision-making. These tools convert consumer presence into transaction volume. However, although promotions may stimulate short-term revenue growth, their effectiveness in sustaining long-term retail performance depends on complementary structural elements. Overreliance on promotional intensity may generate volatility rather than stability within retail ecosystems.

Service quality provides a more durable foundation for retail sustainability. Kotler and Keller [11] conceptualized service quality as the totality of attributes that fulfill explicit and implicit customer needs, while Grönroos [12] distinguished between technical quality (what is delivered) and functional quality (how it is delivered). Parasuraman et al. [13] further demonstrated that gaps between expectations and perceived performance determine satisfaction and loyalty. In urban commercial systems, service quality contributes to trust formation, repeat visitation, and long-term relational stability. From a planning perspective, consistent service quality supports commercial resilience and reinforces the economic viability of retail centers beyond immediate promotional cycles.

Hedonic value has emerged as a central explanatory construct linking retail environments and impulse buying. Silvera et al. [14] argued that impulsive behavior frequently reflects the pursuit of hedonic goals rather than functional necessity. Shopping experiences characterized by enjoyment, excitement, and aesthetic engagement activate affective pathways that shorten decision times and increase spontaneous purchases [15, 16]. Clean, visually appealing, and atmospherically engaging retail spaces amplify these effects [17]. Park and Lennon [18] further demonstrated that hedonic value indirectly influences impulse buying through positive emotions. These findings suggest that retail spaces operate as experiential environments in which affective architecture shapes consumption behavior.

Within broader sustainability discourse, hedonic consumption occupies an ambiguous position. On one hand, experiential retail contributes to urban vibrancy and commercial vitality. On the other hand, intensified spontaneous consumption may challenge long-term responsible consumption objectives. Understanding the mechanisms that convert hedonic motivation into purchasing action is therefore essential for evaluating how retail strategies interact with sustainable development agendas.

Despite theoretical alignment, empirical findings concerning the relationships among sales promotion, service quality, hedonic motivation, positive emotion, and impulse buying remain inconsistent. Some studies confirm significant promotional effects [5, 6], whereas others report non-significant outcomes [7]. Service quality demonstrates positive associations in physical retail contexts [8] but weaker relationships in certain digital commerce settings [19]. Mixed findings also characterize the links between promotional strategies and hedonic motivation, as well as between service quality and hedonic motivation [6, 20]. These inconsistencies indicate that contextual variables—such as store format, product category, demographic composition, and stimulus intensity—moderate observed effects.

Positive emotion plays a central role within the psychological architecture of impulse buying. Rook and Gardner [21] argued that joy and excitement can bypass cognitive deliberation, stimulating immediate purchasing behavior. Subsequent research confirmed that positive affect enhances receptivity to stimuli and accelerates the transformation of intention into action [22]. Watson and Tellegen [23] distinguished between positive and negative affect, highlighting their asymmetrical behavioral consequences, while Muruganantham and Bhakat [24] emphasized the salience of positive emotional states in impulsive consumption. Baron et al. [25] identified satisfaction, comfort, and enthusiasm as affective states easily activated by store atmospherics and persuasive messaging. Earlier evidence further indicated that impulsive buyers exhibit stronger emotional responses than non-impulsive consumers [26] and that retail environments reshape perceptions of quality and value through affective channels [15].

Financial capacity and payment infrastructure constitute enabling conditions that facilitate the translation of affective impulses into actual transactions. Hoch and Loewenstein [27] demonstrated that immediate stimuli combined with financial slack increase the likelihood of repetitive and higher-volume purchases. Bayley and Nancarrow [28] conceptualized perceived resource availability as flexible access to funds, including non-cash instruments. The expansion of digital payment systems has reduced transaction friction, making spontaneous purchases easier to execute. In emerging urban economies, savings ownership and non-cash payment adoption reflect structural shifts in financial accessibility. These developments reshape the operational context in which impulse buying occurs and integrate consumer behavior into broader financial and technological infrastructures.

Collectively, the literature outlines an integrated relational framework. Sales promotion activates attention and arousal; service quality structures experiential reliability; hedonic motivation heightens affective sensitivity; positive emotion converts stimulus perception into behavioral action; and financial infrastructure enables execution. However, inconsistencies across empirical studies underscore the need for models that explicitly test mediation and moderation effects. The interaction between functional retail strategies and affective processes remains insufficiently specified, particularly within rapidly transforming urban retail environments.

Kupang provides a relevant empirical context. In 2023, the city recorded 126 modern retail outlets dominated by national chains [29]. Lippo Plaza Kupang functions as a central commercial node integrating diverse retail formats, promotional intensity, and increasing cashless payment adoption. Such conditions create a retail ecosystem conducive to impulse buying while simultaneously raising questions regarding consumption patterns and long-term commercial sustainability. Empirical research within this local context remains limited.

This study therefore tests an integrated structural model examining the influence of sales promotion and service quality on impulse buying, incorporating hedonic shopping motivation as a mediator and positive emotion as a moderator (Figure 1). By incorporating contemporary indicators such as savings ownership and non-cash payment instruments, the study reflects evolving consumption practices within digitally mediated urban retail systems. In doing so, it contributes to a more nuanced understanding of how retail strategy, emotional processes, and financial accessibility interact within expanding commercial infrastructures.

Figure 1. Conceptual framework

4.1 Respondent characteristics

The respondents in this study comprised 145 individuals with diverse demographic and socioeconomic backgrounds. Based on age, the majority were between 21 and 50 years old, dominated by those aged 21–30 (33.10%) and 41–50 (30.35%), while smaller proportions were in the 31–40 (26.20%) and 51–60 (10.35%) age groups. In terms of occupation, most respondents were civil servants, military personnel, police officers, or employees of state-owned enterprises (59.31%), followed by entrepreneurs (22.76%) and private employees (17.93%). Educational attainment was largely at the undergraduate level (53.79%), with smaller shares at the senior high/vocational (20.69%), postgraduate (17.93%), and diploma (7.59%) levels (see Table 1).

Regarding income, Table 1 shows that the largest group earned between 2,000,001–3,500,000 (41.38%) and 3,500,001–5,000,000 (31.03), with 17.93% earning above 5,000,000 and 9.66% earning 2,000,000 or less. Almost all respondents (96.55%) owned bank savings, while only 3.45% used cooperatives for savings. In terms of non-cash payment facilities, the majority used ATMs (82.76%) and mobile banking (17.24%), with no respondents reporting credit card use. Finally, the shopping frequency per month was relatively high, with 55.17% visiting Lippo Plaza Kupang once or less, 40% visiting two to four times, and 4.83% visiting more than five times per month. This profile indicates that the respondents represent active, educated, and financially independent consumers with a strong reliance on banking services and a growing adoption of digital payment systems.

Table 1. Respondent characteristics

4.2 Descriptive analysis

A descriptive analysis was employed to illustrate respondents’ perceptions of each research variable. Responses were statistically described using interval calculations and index values derived from the scale’s range. The index values served as the basis for interpreting the mean scores of each indicator. The categories were classified as follows: very low (1.00–1.80), low (1.81–2.60), moderate (2.61–3.40), high (3.41–4.20), and very high (4.21–5.00). Accordingly, the interpretation of variables was based on the mean value of each indicator, providing a systematic overview of the respondents’ perceptions.

Table 2. Descriptive statistics of variables

The descriptive analysis (Table 2) indicates that all variables were rated very highly by the respondents. Sales promotion (X1) scored 4.50, supported by high values for membership cards, discounts, cashback, and discount information, confirming effective promotional strategies. The service quality (X2) averaged 4.47, reflecting services that meet or exceed customer expectations. Hedonic shopping motivation (Z) scored 4.47, suggesting that consumers are emotionally driven by pleasure, entertainment, and social interaction. Positive emotion (M) recorded 4.49, showing that consumers experience high enjoyment and enthusiasm while shopping. Impulse buying (Y) obtained the highest score at 4.53, demonstrating that unplanned purchases are prevalent. Overall, the findings highlight the strong interplay of promotion, service, hedonic motives, and positive emotions in fostering impulse buying, positioning Lippo Plaza Kupang to maintain consumer appeal and retail competitiveness in the future.

4.3 Validity and reliability test

The measurement model estimated using the Maximum Likelihood method in AMOS, as summarized in Table 3, demonstrates an overall good model fit. Except chi-square, all indices, including RMSEA, CMIN/DF, TLI, GFI, AGFI, CFI, and NFI, fell within the recommended thresholds, indicating that the proposed endogenous constructs met the goodness-of-fit criteria. Eight indices confirmed model adequacy, and no indicator elimination was required for the confirmatory factor analysis of exogenous variables.

Table 3. Validity and reliability test results

Construct validity was further confirmed through Table 3, which shows that all indicators across exogenous variables achieved critical ratio values greater than twice the standard error and probabilities of less than 0.05. This finding validates that all observed indicators are significant measures for their respective constructs. The factor loadings presented in Table 3 reinforce these results, as all values exceed 0.50, and most are greater than 0.90. This indicates that the indicators for sales promotion (X1) and service quality (X2) operate as coherent measures within their latent variables.

4.4 Measurement model evaluation

The measurement model demonstrated acceptable levels of fit across both exogenous and endogenous constructs (Table 4).

Table 4. Measurement model results

For the exogenous constructs (Promotion and Service Quality), the model yielded a good fit (χ² = 38.891, df = 26, p = 0.079, RMSEA = 0.050, CMIN/DF = 1.496, GFI = 0.945, AGFI = 0.902, TLI = 0.979, CFI = 0.985, NFI = 0.956) (Figure 2).

In contrast, the endogenous construct model (Hedonic Shopping, Positive Emotion, and Impulse Buying) achieved only a partial fit (χ² = 163.429, df = 84, p = 0.000, RMSEA = 0.081, CMIN/DF = 1.946, GFI = 0.866, AGFI = 0.809, TLI = 0.925, CFI = 0.940, NFI = 0.886), although the indices remained within an acceptable threshold for continued structural testing (Figure 3).

Figure 2. Exogenous variable measurement model with CB-SEM

Figure 3. Endogenous variable measurement model with CB-SEM

Table 5 shows that all factor loadings were significant (CR > 2, p < 0.001), confirming convergent validity. For exogenous constructs, Promotion exhibited factor loadings ranging from 0.910 to 0.976 (M ≈ 0.93), while Service Quality ranged from 0.888 to 0.996 (M ≈ 0.91). For endogenous constructs, Hedonic Shopping showed factor loadings between 0.774 and 0.973 (M ≈ 0.86), Positive Emotion ranged from 0.779 to 0.977 (M ≈ 0.84), and Impulse Buying ranged from 0.854 to 0.990 (M ≈ 0.89). These results indicate that the observed items strongly represent their respective latent variables.

Table 5. Measurement model results after modification

Discriminant validity was also confirmed. The correlation between Promotion and Service Quality (0.805) remained below the 0.90 threshold, with the square root of the AVE values (Promotion = 0.912; Service Quality = 0.881) exceeding the inter-construct correlations. Similarly, for the endogenous constructs, the square roots of the AVE values were greater than their corresponding inter-construct correlations, supporting discriminant validity.

The revised measurement model (Figure 4 and Table 5) demonstrated satisfactory convergent validity, with all standardized loadings above 0.50, critical ratio values exceeding twice the standard error, and significance at p < 0.05. The model assumptions of normality and the absence of outliers were also met. The revised overall model achieved an acceptable fit, allowing further assessment of construct reliability and validity. The composite reliability values for Sales Promotion (0.952), Service Quality (0.946), Hedonic Shopping Motivation (0.946), Positive Emotion (0.927), and Impulse Buying (0.939) exceeded the recommended threshold of 0.70, indicating very high reliability. Likewise, the AVE values (0.832, 0.776, 0.746, 0.808, and 0.721) were above the minimum requirement of 0.50, confirming that each construct was well represented by its indicators. Collectively, these results support the reliability and convergent validity of the constructs in the model.

Figure 4. Measurement model after modification

4.5 Model diagnostics

Data analysis using CB-SEM with AMOS confirmed that the measurement model met the required assumptions (Table 6). No univariate or multivariate outliers were detected, with Z-scores below ± 3 and the maximum Mahalanobis distance (44.181) below the chi-square threshold. Normality was satisfied at both the univariate and multivariate levels, as the skewness and kurtosis values remained within ±2.58. Multicollinearity was not an issue, with exogenous construct correlations below 0.90, while the determinant value, although small, was acceptable under SEM assumptions.

The modified overall model achieved an adequate level of fit, with a GFI of 0.781, which is considered marginal but supported by other fit indices. Convergent validity was established as all standardized loadings exceeded 0.50, critical ratios were greater than twice the standard error, and p-values were significant at < 0.05. These results confirm that all constructs are reliable and valid, providing a sound basis for further testing of the structural model.

Table 6. Measurement model diagnostics

4.6 Results of CB-SEM with AMOS

Data analysis was conducted using CB-SEM in AMOS. After tabulating the questionnaire data and ensuring the absence of univariate and multivariate outliers, normality, and multicollinearity issues, the model was estimated using Maximum Likelihood. The initial model showed a marginal fit, with only three indices approaching the recommended cutoff values, while the chi-square, RMSEA, GFI, and AGFI indicated a poor fit (Figure 5). The model includes exogenous constructs X1 and X2 and latent constructs Z and M, which mediate their effects on the endogenous construct Y. Each latent variable is measured through multiple observed indicators (e.g., X1 by x1.1–x1.3, X2 by x2.1–x2.4, Y by y1–y5). The diagram also depicts the standardized regression weights between constructs, with several significant paths (e.g., from Z → Y = 0.36, M → Y = 0.25, X2 → Z = 0.83), while other paths appear weaker or non-significant.

The model fit indices showed that the initial model did not meet the recommended thresholds. The Chi-Square value is 1069.458 with p = 0.000, indicating a poor fit. The RMSEA value of 0.136 is above the acceptable cutoff (< 0.08), suggesting a model misfit. Other indices, such as GFI = 0.652, AGFI = 0.583, TLI = 0.684, CFI = 0.716, and NFI = 0.650, all fall below the conventional cut-off of 0.90. The CMIN/DF ratio was 3.650, which is higher than the ideal threshold (< 3). Collectively, these results indicate that the initial model requires re-specification and modification to achieve a better overall fit for the data.

After evaluating the initial model (Figure 5), several modifications were made to improve the overall fit of the structural equation model. The modified model (Figure 4) incorporates additional paths and covariance relations between the error terms to address areas of misfit identified in the modification indices. This refinement resulted in a more parsimonious and theoretically consistent measurement.

The modified model demonstrated substantial improvements in the goodness-of-fit indices. The Chi-Square value decreased to 518.896 with a ratio of CMIN/DF = 1.834, which is within the acceptable threshold (< 3). The RMSEA = 0.076 is below the 0.08 cutoff, indicating an adequate fit. The incremental fit indices also improved significantly, with GFI = 0.788, AGFI = 0.737, TLI = 0.901, CFI = 0.914, and NFI = 0.830, many of which exceed the recommended 0.90 standard or approach it closely. These improvements suggest that the modified model provides a more accurate representation of the underlying theoretical framework than the initial model.

The structural paths remained consistent with theoretical expectations, with notable significant effects observed from exogenous constructs (X1, X2) toward mediating constructs (Z, M) and subsequently to the endogenous construct (Y). The inclusion of correlated error terms also reflects the recognition of shared variance among the observed indicators, enhancing model reliability without distorting theoretical integrity.

The modified structural model demonstrated significant relationships among the constructs (Table 7). Sales Promotion had a strong and positive effect on Impulse Buying (β = 1.86, p < 0.001) and also significantly influenced Hedonic Shopping Motivation (β = 0.48, p < 0.05). Service Quality showed the highest impact on Impulse Buying (β = 5.40, p < 0.001) and also contributed significantly to Hedonic Shopping Motivation (β = 1.49, p < 0.001). In turn, Hedonic Shopping Motivation had a direct positive effect on Impulse Buying (β = 3.30, p < 0.001), confirming its role as a mediator between exogenous and endogenous constructs (Table 7).

Table 7. Causality test result of the overall model fit after modification

Moderation testing further revealed that Positive Emotion strengthened the effects of both Sales Promotion and Service Quality on Impulse Buying. Although the coefficients were relatively small (β = 0.09 and β = 0.05, p < 0.05), their significance demonstrates that emotional factors play a reinforcing role in consumers’ impulse-buying decisions. This suggests that while service quality and hedonic motives are the primary drivers, the presence of positive emotions amplifies consumer responsiveness to sales promotions and service experiences.

In terms of indirect and total effects, both Sales Promotion and Service Quality were found to influence Impulse Buying indirectly through Hedonic Shopping Motivation (Table 8). The total effects indicated that Service Quality exerted the greatest overall impact on Impulse Buying, followed by Hedonic Shopping Motivation and Sales Promotion. This hierarchy highlights the dominant role of service quality in shaping consumer behavior, with hedonic motives acting as the key psychological mechanism through which these effects are transmitted.

Table 8. Standardized direct, indirect, and total effects (modified model)

The model’s explanatory power was substantial. The squared multiple correlation (R²) for Hedonic Shopping Motivation was 0.986, meaning that Sales Promotion and Service Quality together explained 98.6% of its variance, leaving only 1.4% unexplained by other factors. For Impulse Buying, the R² was 0.751, indicating that 75.1% of the variance was explained by Sales Promotion, Service Quality, Hedonic Shopping Motivation, and Positive Emotion as a moderator. The remaining 24.9% of the variance was attributable to other unmeasured factors.

Overall, the revised model demonstrated strong validity, reliability, and robustness. Service Quality emerged as the most influential factor in driving Impulse Buying, both directly and indirectly, followed by Hedonic Shopping Motivation and Sales Promotion. Although weaker in magnitude, positive Emotion contributed as a meaningful moderator, reinforcing the pathways from exogenous to endogenous constructs. These findings support the robustness of the structural model and provide empirical evidence that both functional (service quality) and emotional-hedonic factors jointly shape impulse-buying behavior.

This study examined the structural relationships among sales promotion, service quality, hedonic shopping motivation, positive emotion, and impulse buying within the context of an emerging urban retail center. Using Lippo Plaza Kupang as the empirical setting, the findings confirm that both sales promotion and service quality exert significant direct effects on impulsive purchasing behavior. Promotional instruments such as discounts, loyalty cards, and cash-back programs increase the immediacy of decision-making, while the SERVQUAL dimensions—reliability, responsiveness, assurance, empathy, and tangibles—create an experiential environment that enhances affective readiness to purchase.

Beyond these direct influences, the study demonstrates the importance of indirect psychological mechanisms. Sales promotion and service quality both strengthen hedonic shopping motivation, which in turn significantly increases impulse buying. Hedonic motivation therefore operates as a mediating pathway translating retail strategy into behavioral outcomes. Positive emotion further moderates these relationships, indicating that consumers in elevated affective states are more responsive to promotional cues and service encounters. However, the moderating strength varies across relational pathways, particularly in the link between service quality and impulse buying, suggesting heterogeneity in consumer responsiveness within evolving urban retail environments.

From a broader perspective, these findings illuminate how retail systems shape consumption dynamics within expanding commercial infrastructures. Promotional intensity accelerates purchasing cycles, while service quality stabilizes experiential trust and reinforces repeat visitation. Together, these mechanisms contribute to retail competitiveness and economic vitality in emerging urban centers. Shopping malls such as Lippo Plaza Kupang serve not only as transactional spaces but also as economic anchors that stimulate local commercial activity and employment. In this respect, impulse buying supports short-term revenue growth and contributes to the financial sustainability of retail operations.

At the same time, the amplification of spontaneous consumption raises important considerations within the discourse of sustainable development. While commercial vitality is essential for urban economic resilience, increasing reliance on impulse-driven purchasing may influence broader consumption patterns. The integration of digital payment systems and expanded financial accessibility further lowers transactional barriers, facilitating immediate spending decisions. Consequently, retail planning strategies must balance competitiveness with long-term considerations related to consumption intensity and financial behavior. Aligning retail growth with responsible consumption principles remains a strategic challenge for urban commercial development.

Practically, the results suggest that retail managers and commercial planners should consider both structural and experiential dimensions of retail environments. Innovative but measured promotional strategies, consistent service delivery, and atmospherically engaging retail spaces can enhance consumer engagement while maintaining operational stability. Rather than focusing solely on transactional acceleration, retail strategies may benefit from integrating experiential quality with broader objectives of customer retention and relational sustainability.

Several limitations warrant acknowledgment. First, the study was conducted within a single shopping center in Kupang, which limits the generalizability of the findings to other retail formats or geographic regions. Second, although the sampling design was theoretically justified, the focus on working women represents a specific consumer segment and may not capture broader demographic diversity. Third, the relatively modest sample size and restricted set of variables—sales promotion, service quality, hedonic motivation, and positive emotion—do not encompass all structural, social, or cultural determinants of impulse buying behavior. Additional contextual factors such as spatial layout, income stratification, lifestyle segmentation, and macroeconomic conditions may further influence purchasing dynamics.

Future research should therefore extend analysis across multiple retail centers and urban regions to enhance external validity. Longitudinal designs could also provide insight into how impulse buying behavior evolves over time within changing economic and technological conditions. Incorporating additional variables related to financial literacy, consumption awareness, or sustainable consumption attitudes may offer a more comprehensive understanding of how retail strategies interact with broader development objectives. Such investigations would contribute to a deeper evaluation of the relationship between retail competitiveness, consumer behavior, and long-term urban sustainability.