A Structural Equation Model of Customer Retention in the Supermarket Industry in Uganda ()
1. Introduction
In Uganda’s competitive supermarket industry, retaining customers has become a significant challenge as businesses face high costs associated with new customer acquisition, service quality complaints, regulatory changes, and economic constraints, converging to indicate a bleak future in the supermarket sector. Uganda’s economic constraints stem from structural dependence on low-value industries and sectors, relatively weak infrastructure, institutional inefficiencies, and limitations in human capital. Addressing these bottlenecks requires comprehensive reforms that promote industrialisation, skills development, infrastructure investment, good governance, and economic diversification. This turbulent business environment in Uganda is characterised by high uncertainty, rapid change, and instability driven by economic, political, technological, and environmental factors. Success in such a context depends on resilience, innovation, and adaptability to validated models.
The supermarket landscape in Uganda is characterised by frequent openings, closures, and relocations as supermarkets struggle to achieve sustainable operations. For instance, big supermarkets such as Uchumi and Nakumatt are notable examples. This makes customers grapple with finding dependable sources for their grocery needs. Customer attrition remains a pressing issue, particularly in the consumer market (Singh et al., 2023). However, research exploring the underlying causes of this phenomenon, especially in developing economies, remains scarce (Milan et al., 2015). Scholars have emphasised the importance of conducting studies in diverse cultural contexts, undertaking comparative analyses, and utilising robust statistical methods to enhance understanding of consumer behaviour and develop actionable strategies to reduce customer churn, as customer retention has even become more critical in the digital economy (Saturi et al., 2025). This study aims to address these gaps by employing a structural equation modeling approach to examine the determinants of customer retention in Uganda’s supermarket industry, and thereby propose a Supermarket Customer Retention (SCR) model.
2. Theoretical Perspectives of Customer Retention
Relationship development and maintenance in marketing literature featured in the 1970s gained more academic attention in the early 1990s and sustainably remained the same (Möller & Halinen, 2010). The debate on relationship marketing, its problems, and direction has been shaped by several international academic engagements, such as seminars, colloquia, and conferences that have taken place worldwide. For example, the American Marketing Association seminar in Berlin in 1996, the Monash University, Emory University, and Dublin conferences in 1993 and 1997, respectively. In addition to disseminating the conference outputs, special issues of journals such as the Journal of the Academy of Marketing Science (1995), the European Journal of Marketing (1996), the Asia-Australia Marketing Journal (1996), and the Journal of Marketing Management (1997) were launched, and several academic articles on relationship marketing were published.
According to Kotler and Keller (2008), relationship marketing establishes enduring, mutually beneficial connections with important stakeholders to win and keep their business. Relationship marketing is characterised by having a sale as the beginning of the relationship, aiming at a continuous, harmonious relationship, delivering value, and considering the customer as the centre of business success, fostering a long term relationship by retaining the customer. Any corporate organisation’s ability to retain customers is critical to its growth (Donovan & Rossiter, 2002). It provides stability and strength, much like the base of a tall building. Businesses that put their present customers first benefit from their continued business and provide opportunities for organic growth through happy customers and goodwill.
This study is anchored on Stimulus-Organizational-Response (SOR) Theory. The theory proclaims that stimuli from the environment affect an individual’s emotions and perception, causing reactions that influence the individual’s behaviour according to the direction of the stimuli (Zhang et al., 2022a). SOR explains customer retention by showing how external factors such as service quality, digital design, pricing, and communication shape customers’ internal cognitive and emotional states, including trust, satisfaction, and perceived value. These internal states then drive retention behaviors such as repeat purchases, loyalty, and positive word-of-mouth (Hussain et al., 2023). In essence, SOR highlights that customer retention is not only about what businesses offer, but how customers interpret and emotionally respond to those experiences. However, its limitations in complexity, measurement, and environmental adaptation mean it should be supplemented with broader behavioural and contextual models, especially in dynamic, developing markets like Uganda.
3. Literature Review
Customer retention is the company’s capability to keep its customers over a long time (Singh et al., 2023; Ranaweera & Prabhu, 2003a; Trivellas & Dargenidou, 2009; Pranindyasari, 2025), and it is a critical factor for business success across industries in any economy (Danesh et al., 2012) as it ensures a stable market for the business. The literature presents customer retention as synonymous with future behaviour intentions (Danesh et al., 2012), behavioural intention (Cronin et al., 2000), repurchase, customer loyalty, and long term customer relationship (Chen & Wang, 2009).
Danesh et al. (2012) assures businesses that customer retention ensures business stability and fosters profitability and organic growth through consumer satisfaction and goodwill. Companies prioritising customer retention benefit from long-term profitability, reduced marketing costs, enhanced brand loyalty, and business survival and growth.
While extensive research has been conducted on the concept of customer retention, significant literature gaps remain in studies focusing on least-developed countries (LDCs) (Alkitbi et al., 2020). These economies face unique economic, cultural, and infrastructural challenges influencing customer retention strategies. Understanding these dynamics is crucial for developing effective retention strategies tailored to these contexts (Jashari-Mani, 2024). A systematic review of customer engagement research shows that most studies stem from developed economies, resulting in a knowledge gap regarding customer retention in LDCs (Hollebeek et al., 2022). Due to the variations in market structures, consumer behaviour, and economic conditions, findings from developed markets may not be directly applicable or generalisable to LDCs. This study seeks to address this research gap by offering insights from the developing economy perspective within the context of Uganda.
Research in the developing economy markets has established that service quality dimensions, such as tangibility, reliability, and responsiveness, indirectly influence customer retention through customer satisfaction. However, there is limited empirical evidence on how these relationships manifest in LDCs, where socio-economic conditions and service delivery infrastructure differ significantly. Many studies Orel and Kara (2014), Venetis and Ghauri (2004), and Ranaweera & Prabhu (2003b) suggest that customer satisfaction mediates the link between service quality and customer retention, but these findings need validation from the LDC perspective (Kheng et al., 2010).
Technological innovation has been widely recognised as a driver of customer retention in developed markets (Paul et al., 2024). Digital platforms, data analytics, and personalised marketing enhance customer experiences, leading to increased loyalty. However, research on how technological advancements influence retention in LDCs remains scarce (Tamirat & Zewdie, 2024). Limited internet penetration, inadequate digital infrastructure, and low technology adoption rates pose challenges that could alter the effectiveness of tech-driven retention strategies (Li et al., 2023). Future studies should explore how businesses in LDCs can leverage cost-effective digital solutions to improve customer retention (Gil-Gomez et al., 2020).
While Customer Relationship Management (CRM) strategies have been extensively studied in developed economies, their applicability in LDCs remains underexplored. Effective CRM mechanisms, including loyalty programmes, relationship marketing, and personalised customer interactions, have been shown to enhance customer retention in well-established markets (Ferrer-Estévez & Chalmeta, 2023). However, cultural variations and resource constraints in LDCs may require different CRM approaches (Opoku, Agyapong, & Alhassan, 2025). A CRM literature review suggests that most frameworks are designed for structured economies with advanced data management systems, limiting their applicability to LDCs’ business environments (Chikweche & Fletcher, 2013).
Cultural factors greatly influence consumer behaviour, yet studies on customer retention often fail to incorporate cultural dimensions (Alsaleh et al., 2019). Research has shown that shopping habits, brand perceptions, and service expectations vary based on national culture. A literature review on culture and consumer behaviour found that while cultural influences are significant, many studies do not integrate cultural dimensions into their methodologies (Zhang et al., 2022b). A cross-cultural investigation is necessary to understand the role of culture in shaping customer retention strategies in LDCs.
Given the scarcity of customer retention studies in LDCs, comparative research could provide valuable insights (Lim & Rasul, 2022). For instance, customer retention models developed in developed economies could be tested in LDCs to assess their relevance and statistical significance (Lovemore et al., 2023). A study in Bangladesh on customer retention strategies highlighted the need for comparative research to refine conceptual models and test hypotheses across different service industries (Leverin & Liljander, 2006).
Addressing the research gaps in customer retention within LDCs is essential for academic and practical advancements. Future studies should focus on empirical research that considers the unique socio-economic and cultural contexts of LDCs. By examining service quality, technological innovation, CRM effectiveness, and cultural influences, researchers can develop tailored customer retention strategies that promote business sustainability and economic development in these regions. As customer retention continues to be a foundation for business growth, further research will provide actionable insights for companies operating in LDCs. Relational norms turn customer relationships from simple transactions into lasting partnerships based on trust, mutual respect, and shared goals (Teo et al., 2025). This relational base is vital for maintaining customer loyalty and achieving long-term profitability. Service quality plays a vital role in customer retention as it determines customer satisfaction, loyalty, and long-term engagement with a brand (Venetis & Ghauri, 2004; Ferrer-Estévez & Chalmeta, 2023). When service consistently meets or exceeds expectations, customers are more likely to remain loyal, repurchase, and recommend the service to others (Leverin & Liljander, 2006). High service quality also reduces customer defection, enhances trust, and strengthens brand reputation. In competitive markets where products are similar, typical in the Ugandan case, superior service quality provides a key differentiator that helps organisations maintain lasting customer relationships and improve profitability through repeat business and positive word-of-mouth. This study attempts to close these research gaps that have come about due to the LDC business environment as well as the socioeconomic conditions.
4. Methodology
A cross-sectional quantitative survey research design was adopted. A five-point Likert scale questionnaire with scale items modified from various sources was administered to supermarket customers, based on a random sampling technique. The data was cleaned, and structural equation modelling (SEM) was used as it can consider all the variables and specified parameters simultaneously. Some of the relevant methodological actions taken are presented in the following sections.
4.1. Data Cleaning
Data cleaning was considered necessary as impure data could influence results in an undesirable manner, which could distort accurate findings. One of the data-cleaning actions was identifying the outliers in the data. Tukey’s method was adopted, where a box plot was used to display the data distribution on a five-point number summary consisting of minimum, the first quartile (QI), the median (QII), the third quartile (QIII), and maximum (Mazarei et al., 2024). In this case, outliers are cases plotted as individual points beyond the whiskers below the minimum or above the maximum, which extend to 1.5 times the interquartile range (IQR) from the first quartile (QI) and the last quartile (QIII).
Based on this criterion, the cases that lie far from the minimum and maximum boundaries have been identified as outliers. They were found crowding on the items specified in Table 1. As you can see, Alpha (α) after the data cleaning indicates lower unexaggerated values. The corrective measure to clean the data was to exclude the cases from the data.
Table 1. Items that indicate outliers in the dataset.
S# |
Item |
Item label in the file |
Outlier cases |
1 |
RN 1 |
My supermarket is supportive |
401, 391, 389, 388 |
|
RN 12 |
My supermarket is flexible when my/our choice changes |
414, 425, 416, 402 |
2 |
SC 6 |
The behaviour of my supermarket instils confidence in me |
390, 414, 397, 384 |
|
SC 16 |
I am uncertain about the results if I change to another supermarket |
400, 397, 381, 371 |
3 |
CT 1 |
My supermarket is reliable as it keeps its promises |
416, 414, 400, 397 |
|
CT 8 |
I believe my supermarket staff have a high level of integrity |
407, 390, 366, 311 |
4 |
CS 2 |
The supermarket staff treat me fairly |
330, 389, 365, 328 |
|
CS 3 |
The supermarket staff pay genuine attention to my interests and needs |
414, 389, 384, 383 |
|
CS 4 |
My supermarket employees respond to my needs as I expect |
414, 402, 391, 367 |
|
CS 5 |
My supermarket products are easily accessible |
289 |
|
CS 11 |
If I had to purchase again, I would feel differently about buying from this supermarket |
414, 387, 379, 374 |
|
CS 12 |
My choice to purchase from this supermarket is always a wise one |
416, 384, 360, 344 |
|
CS 13 |
If I feel good regarding my decision to buy from this supermarket |
369, 416, 389, 365 |
5 |
CR 1 |
I intend to be a client of my supermarket in the future |
414, 406, 400, 391 |
|
CR 6 |
If I had to choose, I would choose my current supermarket again |
341, 414, 407, 335 |
|
CR 11 |
I will continue to purchase from this supermarket because I value it |
407, 360, 384, 335 |
Primary data (2025).
4.2. Reliability Assessment of the Research Instrument
Reliability is a fundamental criterion in evaluating the quality of research instruments, particularly in ensuring the consistency of measurement outcomes across items intended to assess the same construct. One of the most widely employed statistical indicators for internal consistency is Cronbach’s Alpha (α), which measures the degree of inter-item correlation within a scale. The value of Cronbach’s Alpha reflects how closely related a set of items is as a group, thereby indicating the reliability of the scale (Hair, Black, Babin, & Anderson, 2010a).
Interpretation of Cronbach’s Alpha values often depends on the research context, including the complexity of the construct, the research purpose, disciplinary standards, sample characteristics, and the testing environment. As a general guideline, values of α ≥ 0.90 are considered excellent, while those below α < 0.60 are typically deemed unacceptable (Hair et al., 2010a). However, exceedingly high values (i.e., α > 0.90) may suggest redundancy among items, implying potential duplication or overlap due to excessive inter-item similarity. For structural equation modelling (SEM), an optimal range of reliability is typically considered to be between 0.80 and 0.89 (Hair, Black, Babin, Anderson, & Tatham, 2010b), balancing both reliability and item uniqueness.
In this study, reliability analysis was conducted to assess the internal consistency of the data collection instrument. The results, presented in Table 2, demonstrate that the instrument met the established thresholds for acceptable reliability with an Alpha (α) above 0.7.
Table 2. Preliminary reliability analysis.
Number |
Variable |
Number |
Alpha (α) |
1 |
RN |
16 |
0.880 |
2 |
SC |
21 |
0.915 |
3 |
CT |
10 |
0.872 |
4 |
CS |
15 |
0.893 |
5 |
SQ |
17 |
0.905 |
6 |
CR |
11 |
0.890 |
Overall |
|
90 |
0.973 |
4.3. Demographic Profile of Respondents
The respondents’ characteristics are insightful and are presented in Table 3. Personal data was collected on gender, age, marital status, employment status, length of time shopping in a supermarket, highest education level, income range, and region of residence in the country.
Table 3. Demographic profile of respondents.
Variable |
Characteristics |
Frequency |
Percentage |
Gender |
Male |
190 |
41.1 |
|
Female |
201 |
51.6 |
|
Other |
001 |
00.3 |
Age |
Less than 25 Years |
151 |
38.2 |
|
26 - 30 Years |
99 |
25.1 |
|
31 - 35 Years |
58 |
14.7 |
|
36 - 40 Years |
36 |
09.1 |
|
41 - 45 Years |
30 |
07.6 |
|
46 - 50 Years |
20 |
05.1 |
|
51 Years and above |
01 |
00.3 |
Marital Status |
Single |
185 |
46.8 |
|
Married, have child/ren |
182 |
46.1 |
|
Married, no child/ren |
24 |
06.1 |
|
Widow/widower |
04 |
01.0 |
Employment |
Civil Service |
99 |
25.1 |
|
Private Employment |
162 |
41.0 |
|
Not Employed |
134 |
33.9 |
Being Customer |
≤ or = 6 Months |
98 |
24.8 |
|
7 - 12 Months |
54 |
13.7 |
|
13 - 18 Months |
23 |
05.8 |
|
19 - 24 Months |
23 |
05.8 |
|
<24 Months |
197 |
49.9 |
Education Level |
Bachelors |
243 |
61.5 |
|
Masters |
55 |
13.9 |
|
Others |
97 |
24.6 |
Residence |
Central |
209 |
52.9 |
|
Western |
46 |
11.6 |
|
Eastern |
53 |
13.4 |
|
Northern |
87 |
22.0 |
Income |
Less than UGX 500,000 |
185 |
46.8 |
|
UGX500,001 - 1,000,000 |
86 |
21.8 |
|
UGX1,000,001 - 1,500,000 |
48 |
12.2 |
|
UGX1,500,001 - 2,000,000 |
19 |
4.8 |
|
UGX2,000,001 - 2,500,000 |
32 |
8.1 |
|
UGX2,500,001 - 3,000,000 |
23 |
5.8 |
|
More than UGX3,000,000 |
02 |
0.5 |
Data were collected from a total of 395 participants. The demographic profile of the respondents provides valuable insights into the characteristics of supermarket shoppers in Uganda. In terms of gender distribution, female respondents constituted the majority, accounting for 51.6% of the sample. This suggests a relatively higher participation of women in supermarket shopping activities in Uganda.
Concerning age distribution, the largest proportion of respondents (38.2%) was aged below 25 years, while the smallest proportion was observed among those aged 51 years and above. This age distribution aligns with national demographic trends, as over 70% of Uganda’s population is under the age of 30 (Uganda Bureau of Statistics, 2024), and is thus appropriately reflected in the study sample.
Marital status data indicate that a significant proportion of supermarket shoppers were single, representing 46.8% of respondents. Regarding employment, 41.0% of respondents reported working in the private sector, compared with 25.1% in the public sector, suggesting a higher prevalence of private-sector employment among supermarket patrons. Customer loyalty appears relatively strong, with 49.9% of respondents having shopped at the same supermarket for more than two years. In terms of educational attainment, a majority (61.1%) of respondents reported holding at least a bachelor’s degree, indicating a relatively well-educated consumer base.
Geographically, the largest share of respondents (52.9%) resided in Central Uganda, which includes the capital city, Kampala. Finally, income data show that 46.8% of respondents fall within the lowest income bracket, highlighting the significant representation of lower-income individuals among supermarket customers.
4.4. Assessment of the Model Fit
Numerous scholars in structural equation modelling (SEM) have proposed a range of model assessment thresholds to guide the evaluation of model fit. These thresholds, however, are not universally fixed and may vary depending on several contextual and methodological factors. Key determinants include the nature and objectives of the study, the characteristics of the respondent population, sample size, the novelty of the measurement instruments, and the structural complexity of the model. As such, researchers are advised to interpret and apply these thresholds with flexibility, taking into account the specific conditions of their study (Hair et al., 2010a; Schumacker & Lomax, 2010; Little et al., 2007). The threshold for various fit indices is presented in Table 4.
Table 4. Threshold for various fit indices for a model assessment.
Measure |
Name of measure |
Description |
Threshold |
χ2 |
Normed Chi-Square |
Assessment of the overall fit between the sample and covariance matrices |
P. Value > 0.05 |
RMSEA |
Root Mean Square Error of Approximation |
Values closer to Zero represents a good fit |
RMSEA < 0.08 |
CFI |
Comparative Fit Index |
Comparison of fit between target model and independent/null model |
CFI ≥ 0.90 |
NNFI Or TLI |
(Non) Normed Fit Index Or Tucker Lewis Index |
1) Indicates the model of interest relative to the null model 2) NNFI is preferable for smaller samples 3) Sometimes, NNFI is called TLI |
NNFI ≥ 0.90 NFI ≥ 0.90 |
AGFI Or GFI PNFI |
(Adjusted) Goodness of Fit Parsimonious fit |
The proportion of variance accounted for Relative fit indices adjusted to the other fit indices |
AGFI ≥ 0.90 GFI ≥ 0.90 PNFI ≥ 0.50 |
There is an ongoing debate in the literature regarding the appropriate number and types of indices necessary to determine model fit in structural equation modelling. While no universal consensus exists, Hair et al. (2010b) recommend using three to four fit indices to sufficiently assess model adequacy. Similarly, Meyers et al. (2006) argue that three indices are adequate, provided that each represents a different category of fit assessment criteria. Guided by these recommendations, the present study evaluated both the measurement and structural models using fit indices drawn from four distinct categories: absolute fit indices (χ2, GFI), incremental or relative fit indices (TLI, NFI), parsimonious fit indices (PNFI), and noncentrality-based indices (RMSEA, CFI).
Confirmatory Factor Analysis (CFA) was conducted for each construct included in the study. The CFA results indicated that all constructs met the recommended thresholds for model fit, thereby validating their one-dimensionality according to theoretical expectations. Specifically, Relational Norms (RN), Switching Cost (SC), Customer Trust (CT), Customer Satisfaction (CS), and Customer Retention were all confirmed to be unidimensional constructs. In contrast, the Service Quality (SQ) construct exhibited multidimensionality, aligning with established theoretical frameworks. The analysis supported the theoretical position of five constituent dimensions of service quality: tangibility, responsiveness, empathy, reliability, and assurance. The various CFA fit models, along with their corresponding fit indices, are presented in Table 5.
Table 5. CFA results of the constructs.
Variable |
CMINDF ≤ 5 |
P. Value > 0.05 |
PNFI ≥ 0.50 |
GFI ≥ 0.90 |
NFI ≥ 0.90 |
TLI ≥ 0.90 |
CFI ≥ 0.90 |
RMSEA < 0.08 |
RN |
3.30 |
0.000 |
0.6333 |
0.967 |
0.943 |
0.937 |
0.959 |
0.077 |
SC |
3.10 |
0.000 |
0.5830 |
0.977 |
0.972 |
0.968 |
0.981 |
0.730 |
CT |
3.30 |
0.000 |
0.5721 |
0.983 |
0.964 |
0.949 |
0.974 |
0.077 |
CS |
3.40 |
0.000 |
0.6203 |
0.951 |
0.928 |
0.930 |
0.948 |
0.078 |
SQ |
3.07 |
0.000 |
0.6910 |
0.922 |
0.890 |
0.920 |
0.920 |
0.073 |
CR |
3.62 |
0.000 |
0.6040 |
0.966 |
0.950 |
0.945 |
0.963 |
0.081 |
Having run CFA on all the constructs successfully, a measurement model was extracted and tested for fitness using the same criteria. The Measurement model is presented in Figure 1, and the model fit is presented in Table 6.
Table 6. Measurement model.
GoF Category |
Fit Index |
Threshold |
Result |
Achievement |
Absolute fit |
Relative χ2 CMINDF |
≤5 |
2.011 |
Achieved |
|
Chi-Sq/df |
P-Value ≥ 0.05 |
0.000 |
Not achieved |
|
GFI |
≥0.90 |
0.874 |
Not achieved |
Relative fit |
NFI |
≥0.90 |
0.875 |
Not achieved |
|
TLI |
≥0.90 |
0.905 |
Achieved |
Parsimonious fit |
PNFI |
≥0.50 |
0.755 |
Achieved |
Centrality-based indices |
RMSEA |
≤0.08 |
0.051 |
Achieved |
|
CFI |
≥0.90 |
0.915 |
Achieved |
Figure 1. Measurement model.
Following a measurement model fit, a structural equation model was extracted. The structural model was assessed using the same chosen fit indices. The results turned out to indicate a possible customer retention model in the Ugandan context (Table 7, Figure 2).
Table 7. Structural model.
GoF Category |
Fit Index |
Threshold |
Result |
Achievement |
Absolute fit |
Relative χ2 CMINDF |
≤5 |
2.055 |
Achieved |
|
Chi-Sq/df |
P-Value ≥ 0.05 |
0.000 |
Not achieved |
|
GFI |
≥0.90 |
0.880 |
Not achieved |
Relative fit |
NFI |
≥0.90 |
0.875 |
Not achieved |
|
TLI |
≥0.90 |
0.906 |
Achieved |
Parsimonious fit |
PNFI |
≥0.50 |
0.758 |
Achieved |
Centrality-based indices |
RMSEA |
≤0.08 |
0.052 |
Achieved |
|
CFI |
≥0.90 |
0.916 |
Achieved |
Figure 2. Structural model.
4.5. The Supermarket Customer Retention Model
Following the structural model assessed using four fit indices, i.e., absolute, relative, parsimonious, and centrality-based indices, as recommended in Hair et al., (2010b), and Meyers et al., (2006), the study now proposes the Supermarket Customer Retention (SCR) Model as presented in Figure 3.
Figure 3. Supermarket customer retention model.
5. Discussion
In Structural Equation Modelling (SEM), the relative importance of variables is assessed through the achievement of the fitting of the models, as the relative importance of the variable is integrated and measured using at least three sets of fit indexes (Hair et al., 2010b). In addition, factor loadings from the measurement model show the strength of indicators in representing their latent constructs. In this case, referring to the measurement model (Figure 1), factor loadings of Service quality (SQ), Customer Satisfaction (CS), and relational norms (RN) all range from 0.6 to 0.9, signifying the strength of the contribution of the latent construct. Referring to the other latent constructs, i.e., customer trust (CT), and Switching Costs (SC), the factor loadings are still within an acceptable range, i.e., above 0.5 to 0.7 are relatively low. This therefore means for the contexts of the Ugandan supermarket environment, SQ, CS, and RN are the variables the management could invest in as a priority to retain customers.
The result of a fitting model, as assessed using multiple fit indices, provides rich ground for discussion. Despite socio-cultural and economic differences, there are many similarities between the determinants identified in this study and those identified decades ago and current findings. For instance, Ranaweera and Prabhu (2003a, 2003b), Danesh et al. (2012), and Hollebeek et al., (2022), where there is a consistent voice that to retain customers, regardless of industry, customer satisfaction, trust, and management of switching costs are critical. These studies come from different socioeconomic and cultural environments. This convergence of customer behaviour reflects the truth and provides a basis for recommendation.
SOR theory is therefore found to be an appropriate model that the supermarkets can adopt to induce favourable customer behaviour. The organism presents behaviour detection features to exhibit planned behaviour response, i.e., retention (Pranindyasari, 2025). It is critical to develop and keep customer data to trace the emerging customer behaviour to develop customer retention strategies (Saturi et al., 2025). The results equally agree with earlier studies (Chen et al., 2023) as well as recent studies (Liu & Zhao, 2025), where exchange and communal norms are critical in developing a beneficial mutual relationship. Service quality experience is a strong repurchase intention factor in the retail business sector (Teo et al., 2025). The customer retention model that this study has confirmed is hence strongly supported.
6. Conclusion
The study confirms that customer retention in the supermarket sector is significantly influenced by service quality, customer satisfaction, customer trust, and switching costs. These factors consistently appear across different socio-cultural and economic contexts, emphasising their universality. The findings, therefore, highlight that Ugandan supermarkets, despite operating in a distinct local environment, cannot overlook these fundamental drivers if they aim to develop loyal and steady customer bases.
The convergence of findings with both earlier and recent studies shows that supermarkets must regard retention as a strategic outcome rooted in both relational and transactional components. Service quality and satisfaction form the foundation of positive experiences, while trust and perceived switching costs act as barriers against customer defection. This aligns with the Stimulus-Organism-Response (SOR) theory, which proposes that attractive environments and responsive customer engagement promote retention behaviour. Gathering and analysing customer data to track behavioural changes is thus essential for supermarket managers in Uganda.
The validated customer retention model offers strong guidance for practice and policy. For practitioners, supermarket managers should focus on investing in staff training, customer care, and service process improvements, while utilising loyalty programmes to increase switching costs. Policymakers, on the other hand, can support customer protection frameworks that foster trust, enhance service standards, and promote fair competition. Collectively, these measures will enable the Ugandan supermarket sector to nurture long-term customer relationships, boost competitiveness, and support sustainable business growth.
Low customer loyalty in Uganda’s supermarket industry is mainly caused by poor service quality, which damages customer satisfaction and trust, resulting in frequent complaints and higher switching rates. Strengthening relationship management practices that build trust and increase satisfaction helps reduce customers’ likelihood of leaving. Improving service reliability, responsiveness, and empathy further decreases dissatisfaction and encourages repeat business. Additionally, introducing switching barriers such as loyalty programs and membership benefits raises the cost of switching and discourages customer turnover. Overall, these measures improve customer retention and address the sector’s ongoing loyalty issues.
7. Recommendations
First, the supermarket managers could improve service quality through training the staff, streamlining service processes, and maintaining a welcoming store ambience to build customer satisfaction, trust, and an appropriate experience. Secondly, supermarkets may introduce and strengthen loyalty programmes such as reward cards, discounts, and special offers to increase customer commitment and reduce the likelihood of switching to competitors. In other words, invest in increasing customer switching costs. Third, supermarket management should adopt a systematic data collection and analyse such customer data to understand shopping patterns and use this information to design strategies that encourage long-term customer retention.
8. Future Research Direction
This study aimed to establish a customer retention model in the supermarket sector in Uganda. The limitation in this study is that it is general, based on a cross-sectional research design. The study did not make a critical analysis of the determinants of customer retention in relation to the respondent characteristics. Future research that endeavours to establish consumer category perspectives would make customer retention better understood in the supermarket sector. For instance, how gender, age categories, marital status, employment status, and other consumer attributes influence customer retention would be insightful. It would be necessary to examine specific other market categories of groceries, garments, electronics, and others to assess if consumer behaviour would vary. Comparative study using the same data collection instruments in selected least developed economies and developed economies would confirm and build a strong theoretical foundation for customer retention in the modern business environment.