BSSS Journal of Management, Volume XVII, Issue-I

THE EFFECT OF AI-DRIVEN FEATURES ON CUSTOMER SATISFACTION AND REPEAT PURCHASE AMONG LAZADA USERS IN THAILAND

 

*Naw Htoo Aung, **Dr. Siriwan Kitcharoen, ***Dr. Bhumiphat Gilitwala

*Graduate School of Business and Advanced Technology Management, **Assistant Vice President for Educational Innovation and Graduate Studies ***Program Director MBA, Assumption University, Thailand

 

Abstract

Artificial intelligence (AI) has become a key driver in shaping today’s e-commerce platforms and customer experiences. The current study aims to analyze the effect of AI-driven features on customer repeat purchase among Lazada users in Thailand, and the mediating effect of customer satisfaction. A quantitative research method was used, where primary data were collected from Lazada users through an online questionnaire. A total of 409 Lazada users actively participated in the survey. The data were analyzed using SmartPLS. The findings showed that AI-driven personalization, customer control over personalization, and recommendation algorithm transparency have no statistically significant effect on customer repeat purchases, while customer satisfaction has a significant impact on customer repeat purchases at Lazada. Customer control over personalization and transparency in recommendations has a statistically significant effect on customer satisfaction; however, AI-driven personalization has no significant impact on customer satisfaction. Moreover, the results indicated that customer satisfaction served as a significant mediator between customer control over personalization and customer repeat purchase. Lazada customers are satisfied when they have the authority to control their personalized data, and the e-commerce recommendation algorithm is transparent and trustworthy. Satisfied customers become loyal customers. E-commerce sectors can implement the findings to improve customer satisfaction and retain customers.

 

Keywords: AI-Driven Personalization, Recommendation Algorithm Transparency, Customer Control over Personalization, Customer Satisfaction, Repeat Purchase, Lazada, Thailand


1.     Introduction

Artificial intelligence (AI) becomes an essential part of a modern e-commerce development due to AI transformation affecting customer engagement on digital platforms including customized product recommendations through an algorithmic system's visibility. Over the past few years, e-commerce such as Lazada in Thailand has implemented cutting-edge technology to meet consumer expectations and stay competitive with a growing market (Turatti, 2025).

The COVID-19 pandemic created Thailand's shift to online shopping platforms, allow more consumers to understand online shopping spaces. Due to the shifting landscape of online shopping, consumers began demanding a personalized and seamless shopping experience (Weiler & Gilitwala, 2023; Oh et al., 2023). AI is one critical tool in personalizing the digital shopping experience by allowing consumers to find suitable products effectively and with improved accuracy (Agoro et al., 2021; Ahmed et al., 2025).

1.1  Statement of the Problems

 

 

The deployment of AI technology is now an essential component of contemporary e-commerce platforms. As one of the premier online platforms in Thailand, Lazada's utilizes some AI functions to enhance the online shopping experience and enjoy a technological advantage in the online marketplace (Oh et al., 2025). Several studies regarding algorithms have examined AI functions in isolation from each other. Previous research regarding AI personalization described production improvements and customer purchase behavior (Ahmed et al., 2025; Agoro et al., 2021b). AI transparency studies have focused on how user trust develops based on their understanding of AI processes (Owen et al., 2023; Akbar et al., 2024). In addition, several studies have analyzed user modal control methods, which highlight that user control over AI-generated results improves user satisfaction (Bok, 2023; Geetha et al., 2024). AI tools assist consumers through shopping decision processes, although some consumers question the recommendations, algorithmic fairness, and use of their data collection (Bok, 2023; Owen et al., 2023; Akbar et al., 2024). Although AI is becoming increasingly common in e-commerce, Thai consumers have limited knowledge regarding AI functions. This study fills the gap of previous studies where the researchers implemented the single AI feature for each study; thus, the current study analyzes the effect of AI personalization, customer control over personalization, and recommendation transparency on customer satisfaction and repeat purchases at Lazada e-commerce in Thailand. The results will contribute to an in-depth understanding of how AI-driven features effect on e-commerce users in Thailand.

 

1.2  Research Questions

 

 

The following research questions support the purpose of the study:

1.      How does AI-driven personalization affect customer satisfaction among Lazada users in Thailand?

2.      How does recommendation algorithm transparency affect customer satisfaction?

3.      How does customer control over personalization affect customer satisfaction?

4.      To what extent does customer satisfaction affect repeat purchases in the context of AI-enabled e-commerce?

 

2.     Literature Review

2.1  Theories Related to The Research

 

E-commerce businesses must know what makes customers reorder. Theoretical frameworks that outline the ways individuals appraise their experience can better facilitate identifying customer satisfaction and long-term loyalty, or even patronage. Thus, these two models are helpful for our study, which explores the influence of AI-enabled features on customer satisfaction and intention to repeat purchases: Expectation Confirmation Theory (ECT) and the Technology Acceptance Model (TAM).

Expectation Confirmation Theory (ECT) specifies satisfaction because of comparing expectations to outcomes. Intuitively speaking, AI-enabled features (personalization, automation, and transparency) can succeed (satisfaction) when the system performs according to expectations or exceeds the user's expectations. Faster, relevant, and well-timed interactions can make customers feel satisfied and therefore want to return shopping. With ECT, satisfaction outcomes can be contextualized in set standards (i.e., expectations), which makes for an especially good fit in the e-commerce environment that relies heavily on expectations of speed, convenience, and personalization (Ahmed et al., 2025; Hardcastle et al., 2025). More specifically, if an AI tool is designed to do something, like uncover hidden patterns in customer data, users expect that it delivers exactly what it is designed to do. The outcomes are critical for maintaining user satisfaction and therefore fostering loyalty (Ahmed et al., 2025; Hardcastle et al., 2025).

Technology Acceptance Model (TAM) describes a relationship that exists between users and technology. TAM has two components, such as perceived usefulness and perceived ease of use. The two components define whether a user will adopt and continue to use a respective technology. In the context of AI-enabled personalization, if customers perceive the AI tool to be useful and feel they can easily interact with it, they will have a positive experience and are likely to remain a loyal customer. Therefore, meaningful AI, combined with usability, was a significant factor of overall experience and intention to repurchase (Alkudah & Almomani, 2024; Turatti, 2025). ECT defines the importance of meeting and exceeding expectations, while TAM highlights usefulness and ease of use. When combined, both theories help explain the preference for returning to AI-enabled experiences and platforms that showcase trust, speed, and customer satisfaction.

 

2.2  Repeat Purchases

 

 

Repeat purchase behavior indicates that a consumer is willing to purchase from that same platform again, based on prior experience that was favorable. In e-commerce platforms like Lazada, consumers are likely to return because they trust that platform, they are familiar with the interface, and they see a good perceived value in their purchases. Previous studies reported that customer satisfaction is a significant predictor of repeat purchase behavior. If customers' needs are met and/or exceeded, they are likely to return for future purchases (Tufahati et al., 2021; Jiradilok et al., 2013).

AI-driven platforms facilitate repeat purchases by recommending relevant products at appropriate times. The impact of AI-powered personalization improves satisfaction and has a significant effect on customers' willingness to repeat purchase behavior. The relevance increases customers perceived efficiency and ease of use the platform (Ahmed et al., 2025). Personalization minimizes search time and tailor’s users’ experience, which creates loyalty (Hardcastle et al., 2025). Consumers expect their personalized experience are ethical and reliable. Transparency and control offer a reduction in uncertainty and build their confidence over time (Owen et al, 2023; Akbar et al, 2024). Demonstrating transparency into the algorithms used can enhance users’ confidence. When consumers know how their recommendations are generated and, importantly, feel the personalization system is fair, their trust in the platform becomes much stronger. In turn, their trust relates positively to their willingness to repeat purchase behavior.

2.3 Literature review related to the relationship between each variable

 

This section discusses the relationships between independent variables; understanding how these variables interact is essential for identifying which AI features most effectively enhance customer satisfaction and drive repeat purchases on e-commerce platforms.

2.3.1 AI-Driven Personalization  Þ Customer Satisfaction

 

AI-driven personalization increases customer satisfaction through product recommendations, promotional offerings and outstanding navigation flows by utilizing an individual user’s interactions, preferences, and past purchases. When users are provided with suggestions that mirror their interests and individual preferences, they identify with the personalized approach and feel validated and recognized. This feeling can help decrease decision fatigue and improve browsing convenience, leading to contentment. (Ahmed et al., 2025) highlight that personalization done accurately and respectfully increases customers' enjoyment, perceived convenience, and loyalty. (Hardcastle et al., 2025) also illustrate that personalization helps develop emotional engagement and emotional comfort, promoting contentment as well as a longer-term preference for using the platform.

Owen et al. (2023) highlighted that personalization is considered ethical and transparent to avoid relationship concerns regarding manipulation and privacy. Personalization can positively influence contentment with relationships; however, it is highly relevant in terms of how it is respectful and transparent. On a platform like Lazada, making personalization ethical without undue burden is effective in simplifying shopping by improving the user experience, while strengthening the user-platform relationship through enhanced trust resulting from a positive user experience.

 

2.3.2 Recommendation Algorithm Transparency Þ Customer Satisfaction

 

Transparent recommendation algorithm improves customer satisfaction by making the reasons why AI suggests an offering more visual and understandable. When customers understand where the recommendation came from, they can view the platform in a way that contains fair analysis, offers reliability, and provides a user perspective (Owen et al., 2023). Akbar et al. (2024) also stated that transparent user systems can increase user’s satisfaction by increasing the appropriateness and reliability of AI-generated suggestions. Alkudah and Almomani (2024) mentioned the drawback of having overly technical and complicated explanation instructions, as it can confuse customers with low digital literacy. Thus, transparency needs to be designed responsibly, having just enough transparency for the neediest of customers to be satisfied but not disrupting their connotative equilibrium

2.3.3 Customer Control over Personalization Þ  Customer Satisfaction

 

Customer satisfaction can be correlated to whether users are able to take control of their personalization settings, such as data preferences, recognition of product types, and suggestions. Control features provide users with the ability to shape their shopping experience, offering a sense of freedom and autonomy, which can effectively reduce user frustration with suggestions that may seem either irrelevant or invasive. Alkudah and Almomani (2024) stated that digital control empowers users and can improve users’ perceptions of fairness and trust in the system. Turatti (2025) echoes the sentiment by indicating that when customers feel a sense of control in their digital ecosystem, they may become more emotionally committed to the process, which can lead them to return. Turatti (2025) suggests that excessive personalization options may confuse users or deter them from using any personalization, thereby minimizing the benefits. It highlights that control systems should be designed to be simple, intuitive, user-centered, and easy to use. For e-commerce platforms, a sensible approach is to offer users meaningful control, ensuring improvements in user satisfaction by reflecting a sense of participation and personalization in a relevant and respectful manner.

 

2.3.4 Customer Satisfaction Þ Repeat Purchase

 

Customer satisfaction is a key factor in growing repeat purchase behavior in e-commerce. A user with an expectation of a satisfactory shopping experience from a platform is likely to return to make purchases. Satisfied users develop emotional and psychological commitments that lead to loyalty and contribute to the development of habitual buying behavior over time. Hardcastle et al. (2025) claim that emotional engagement resulting from satisfaction often translates into trust and familiarity, which fosters repeated interactions. Tufahati et al. (2021) caution that satisfied users are also more likely to elicit positive word-of-mouth behaviors, leading to repeat purchases and new users to the platform. Furthermore, Jiradilok et al. (2013) found that satisfaction was one of the strongest indicators of repurchase intention, particularly when users experienced ease, responsiveness, and value in Thailand. Maintaining customer satisfaction for e-commerce platforms is essential for an individual's retention on the platform and the long-term growth of the market through repeat purchase and referrals.

2.     Research Methods and Materials

 

2.1  Conceptual framework

Repeat Purchases (RP)

Recommendation Algorithm Transparency (RAT)

 

AI-driven Personalization (AIP)

 

Customer Control over Personalization (CCP)

 

Customer Satisfaction

(CS)

H1

H2

H3

H4

 

 

 

 

 


Figure (1) Conceptual framework

The following hypotheses are developed based on the conceptual framework:

H1: AI-driven personalization (AIP) positively and significantly affects the customer satisfaction.

H2: Recommendation algorithm transparency (RAT) positively and significantly affects the customer the satisfaction.

H3: Customer control over personalization (CCP) positively and significantly affects the customer satisfaction.

H4: Customer satisfaction positively and significantly affects the repeat purchases of Lazada users in Thailand.

3.2 Research Design

 

In this study, a cross-sectional design was used as data collection occurred at a single time point. The data collection proceeded on across-sectional basis for the inclusion of investigating the naturally occurring, which may have influenced the three AI-driven features: personalization, transparency of the algorithm, and control by customer (McCarthy et al., 2022), which imposed customer satisfaction and their consequent behavior in repeat purchasing. The goal of the research was not to manipulate the research participants but to measure their perceptions of AI systems in the Lazada environment as it existed in real-time.

The research instrument was a self-administered online questionnaire that contained structured items that were measured on a five-point Likert scale to ensure that access to the study was meaningful, to reach as many potential respondents as possible, and to normalize options provided to respondents. The final data file was run through the SEM technique for analysis. The design was aligned to the study without influencing the outcomes, suggesting it was appropriate to explore consumer experiences in AI-enabled e-commerce contexts.

 

3.3 Sampling Plan

3.3.1 Target Population

 

The target population for this study encompasses Lazada users in Thailand who have bought at least one product on Lazada (online) within the last three months before data collection and who have experienced the AI features of Lazada. Research conducted by SEMrush (2025) states that Lazada Thailand receives around 37.6 million site visitors per month, which indicates a large and active customer group online. Despite no public number for unique users, this statistic strongly supports the viability of collecting data using an online survey and provides assurance there are enough unique users in the target population to apply a statistical analysis approach like Structural Equation Modeling (SEM).

3.3.2 Sampling Unit

 

The sampling unit for this study is an individual Lazada user that meets the screening criteria set in this study. Although there were multiple respondents per Lazada user, each respondent is treated as a single independent unit of analysis, as the responses reflect their own experience and perceptions of AI features of the Lazada platform. Each of the respondents' data was completed individually through a self-administered online questionnaire; as such, each user's answers were unique, confidential, and uninfluenced by other participant responses.

 

3.3.3 Sample Size and Sampling Procedure

 

 

This study collected responses from 409 valid participants, which exceeds the minimum sample size required for conducting statistical analysis using structural equation modeling.

The study employed an online convenience sampling method, inviting participants who met the defined screening criteria for this study. Respondents were permitted to participate if they had made at least one purchase from Lazada within the last three months and had interacted with the AI features on Lazada in some way, such as a consumer of personalized recommendations or algorithm-generated suggestions. The questionnaire was administered online through the researcher distributing the questionnaire using commonly accessible online platforms such as Facebook groups and through LINE messaging apps, as these are platforms frequently used by Lazada shoppers in Thailand, thereby allowing the researcher to conveniently obtain responses from a convenient but relevant subset of Lazada users.

The screening questions were incorporated at the beginning of the questionnaire so that only qualified respondents could complete the full survey. The participation component of the research was completely voluntary and anonymous, ensuring the ethical standards in relation to data collection and participation remained intact.

 

4. Results and Discussion

4.1 Reliability Testing

 

The construct reliability and validity are measured by Cronbach’s alpha and composite reliability. The reliability results in Table (1) show that all constructs achieved Cronbach’s alpha and composite reliability (CR) values above the recommended threshold (0.70). AI-driven Personalization (AIP) had the highest alpha (.922) and CR (.942), indicating excellent internal consistency. Customer Control over Personalization (CCP) also performed strongly (α = .846, CR = .891). The lowest reliability values were for Customer Satisfaction (CS) (α = .786, CR = .854), but these are still acceptable. The Average Variance Extracted (AVE) values for all constructs were greater than 0.50, ranging from .540 (CS) to .766 (AIP), which confirms convergent validity. These results indicate that the constructs explain more than half of the variance in their items and can be considered reliable and valid for further testing.

Cronbach's alpha 𝛼

Composite reliability

(rho_a)

Composite reliability

(rho_c)

Average variance extracted

(AVE)

AIP

0.922

0.924

0.942

0.766

CCP

0.846

0.846

0.891

0.621

CR

0.816

0.819

0.872

0.576

CS

0.786

0.788

0.854

0.540

RAT

0.808

0.810

0.867

0.566

Table (1). Construct reliability and validity for measurement model

(Source: survey data 2025)

 

 

 

4.2 Descriptive Analysis

Gender of the respondent

Out of 409 respondents, 242 were female, making up 59.17% of the total sample. Male respondents accounted for 153, or 37.41%, while 14 participants (3.42%) preferred not to disclose their gender. These results show that female participants form the majority, which may reflect the strong presence of women in online shopping. At the same time, the significant number of male respondents ensures that the findings reflect opinions across both genders. Consumers also expect their personalized experience to be ethical and reliable. Demonstrating transparency into algorithms used can enhance users’ confidence of how it works. When consumers know how their recommendations are generated, and, importantly, feel the personalization system is fair, their trust becomes much stronger in the platform. In turn, their trust relates positively to their willingness to repeat purchase behavior. Transparency and control offer a reduction in uncertainty and build their confidence over time (Owen et al, 2023; Akbar et al, 2024).

Figure (2) Gender

(Source: survey data 2025)

 

Age of the respondents

The age distribution shows that most respondents are young adults. The largest group is those aged 26–30 years (167 respondents, 40.83%), followed by 31–35 years (104 respondents, 25.43%) and 20–25 years (102 respondents, 24.94%). A smaller group of respondents were aged 18–20 years (16 respondents, 3.91%) and 36 years or above (20 respondents, 4.89%). Together, these results confirm that most participants are within the prime working and shopping ages, which makes them highly relevant for online retail studies.

Figure (3) Gender

(Source: survey data 2025)

Current Occupation

The respondents also come from diverse occupational backgrounds. Students make up the largest portion, with 172 participants (42.05%), followed by employees with 128 participants (31.30%). Freelancers represent 73 respondents (17.85%), while business owners account for 36 respondents (8.80%). This mix indicates that Lazada is widely used across different occupational groups, from young students to working professionals and entrepreneurs. The high number of students suggests that younger users are particularly active in online shopping.

 

Figure (4) Gender

(Source: survey data 2025)

Purchasing Frequency

Purchasing frequency results show how often respondents shop on Lazada. The highest proportion, 169 respondents (41.30%), reported shopping twice per month. Another 136 respondents (33.25%) shop once per month, while 104 respondents (25.43%) reported purchasing three times or more per month. These results suggest that most participants are frequent and engaged shoppers, with the majority making at least one to two purchases each month. This confirms that the respondents are active online consumers who can reliably evaluate Lazada’s AI-driven features and their influence on satisfaction and repeat purchase.

Figure (5) Gender

(Source: survey data 2025)

 

4.3 Outer Loading Model

 

The outer loading model presents the factor loadings of the structural equation modeling. The minimum acceptable threshold of factor loading score is 0.7. If the item scores less than 0.7, it needs to be removed from the model. According to the results shown in table (2), the factor loadings of AI-driven personalization (AIP1 to AIP5); customer control over personalization (CCP1 to CCP5); customer repeat purchase (CR1 to CR5); customer satisfaction (CS1 to CS5), and recommendation algorithm transparency (RAT1 to RAT5) are more than 0.7 which means that the items are reliable.

 

AIP

CCP

CR

CS

RAT

AIP1

0.926

 

 

 

 

AIP2

0.796

 

 

 

 

AIP3

0.924

 

 

 

 

AIP4

0.794

 

 

 

 

AIP5

0.924

 

 

 

 

CCP1

 

0.722

 

 

 

CCP2

 

0.848

 

 

 

CCP3

 

0.848

 

 

 

CCP4

 

0.767

 

 

 

CCP5

 

0.746

 

 

 

CR1

 

 

0.758

 

 

CR2

 

 

0.808

 

 

CR3

 

 

0.773

 

 

CR4

 

 

0.730

 

 

CR5

 

 

0.723

 

 

CS1

 

 

 

0.720

 

CS2

 

 

 

0.778

 

CS3

 

 

 

0.755

 

CS4

 

 

 

0.711

 

CS5

 

 

 

0.706

 

RAT1

 

 

 

 

0.743

RAT2

 

 

 

 

0.783

RAT3

 

 

 

 

0.732

RAT4

 

 

 

 

0.764

RAT5

 

 

 

 

0.737

Table (2). SEM results outliner

(Source: survey data 2025)

4.4 Path coefficient

 

The table (3) shows the direct effect of AI-driven personalization, customer control over personalization, and recommendation algorithm transparency on customer satisfaction and repeat purchase. The results indicate that AI-driven personalization has a weak no statistically significant effect on repeat purchase (path coefficient 0.109, p-value 0.275). Moreover, AI-driven personalization has no statistically significant effect on customer satisfaction (path coefficient 0.09, p-value 0.244). Customer control over personalization has no statistically significant effect on customer repeat purchase (path coefficient 0.068, p-value 0.286), although it has statistically significant direct effect on customer satisfaction (path coefficient 0.306, p-value < 0.01). Customer satisfaction also has direct significant effect on customer repeat purchase (path coefficient 0.703, p-value <0.01). Recommendation Algorithm Transparency has no statistically significant effect on customer repeat purchase (path coefficient -0.027, p-value 0.811), although it has direct significant effect on customer satisfaction (path coefficient 0.40, p-value < 0.01).

The results indicate that customer satisfaction is the critical factor of repeat purchase. When customer satisfy with customer control over personalization and recommendation algorithm transparency, they repurchase products.

 

Path Coefficient

P values

AIP Þ CRP

0.109

0.275

AIP Þ CS

0.090

0.244

CCP Þ CRP

0.068

0.286

CCP Þ CS

0.306

0.000

CS  Þ CRP

0.703

0.000

RAT Þ CRP

-0.027

0.811

RAT Þ CS

0.400

0.000

Table (3) - Direct Effect

(Source: survey data 2025)

Indirect effect (mediating effect of CS)

Table (4) shows the mediating effect of customer satisfaction. The results indicate that the AI-driven personalization has no statistically significant indirect effect on repeat purchase through customer satisfaction (path coefficient 0.0634, p-value 0.2331), which means that customer satisfaction does not perform as a significant mediator between the relationship of AI-driven personalization and repeat purchase. However, the indirect effects of customer control over personalization and recommendation algorithm transparency on repeat purchase are statistically significant through customer satisfaction (path coefficient 0.2154 and 0.2815, respectively, p-value < 0.01), which means that customer satisfaction has mediating effect between the relationships of customer control over personalization, recommendation algorithm transparency, and customer repeat purchase.

 

Path coefficient

P values

AIPÞCSÞCRP

0.0634

0.2331

CCPÞCS ÞCRP

0.2154

0.0002

RATÞCSÞCRP

0.2815

0.0010

Table (4) - Indirect effect (Mediating effect of customer satisfaction

(Source: survey data 2025)

Total effect (direct effect + indirect effect)

The total effect presents the overall effect of AI-driven personalization, customer control over personalization, and recommendation algorithm transparency on customer repeat purchases. Table (5) presents the total effects, showing that AI-driven personalization has no statistically significant total effect on customer repeat purchases (path coefficient 0.172, p-value 0.111). Customer control over personalization has a statistically significant total effect on customer repeat purchase (path coefficient 0.2832, p-value < 0.01). However, the transparency of the recommendation algorithm has no statistically significant total effect on customer repeat purchases (path coefficient 0.2541, p-value 0.0793).

Table (5) Total effect (Direct effect + Indirect effect)

 

Direct effect

Indirect effect

Total Effect

P-value

AIPÞCRP

0.109

AIPÞCSÞCRP

0.0634

0.1720

0.1111

CCPÞCRP

0.068

CCPÞCSÞCRP

0.2154

0.2832

0.0018

RATÞCRP

-0.027

RATÞCSÞCRP

0.2815

0.2541

0.0793

AIPÞCS

0.090

-

-

0.090

0.2442

CCPÞCS

0.306

-

-

0.306

0.0001

RSTÞCS

0.400

-

-

0.400

0.0003

CSÞCRP

0.703

-

-

0.7033

0.0000

(Source: survey data 2025)

4.4 Hypotheses Results and Conclusion of the Study

 

β 0.400(P-value 0.0003)

β 0.306 (P-value 0.0001)

β 0.09 (P-value 0.244)

β 0.703 (P-value 0.000)

Repeat Purchases (RP)

Recommendation Algorithm Transparency (RAT)

 

AI-driven Personalization (AIP)

 

 

Customer Control over Personalization (CCP)

 

Customer Satisfaction (CS)

 

 

 

 

 

 


Figure (6)) Hypotheses Results of path coefficient and P-value

The results indicate that consumer satisfaction is the strongest driver of repeat purchases in Lazada. The path of satisfaction and repeat purchase is significant (β = 0.703, p < .001), so if a consumer is satisfied with the purchase experience, then they are likely to buy again. When people are satisfied with the buying experience in e-commence platform, they develop trust in the system, recognize the system as trustworthy, and are confident that future purchases are smooth and rewarding. Satisfaction reduces uncertainty about the experience and switching intentions because it is easier to buy again within comfortable system that is familiar than search for a new one. Consequently, customers satisfaction builds loyalty by converting a one-time customer into a repeat purchaser.

The results also indicated the primary factors influenced on customer satisfaction were recommendation transparency and user control. Recommendation Algorithm Transparency was the primary influence on satisfaction, having the significant effect size (β = 0.400, p < .001). When AI system explains why it is recommending a product, customers see it as fair. This fairness builds trust, and trust supports confidence to buy. The second strongest influence was customer control over personalization (β = 0.306, p < .001). Customers who can change their settings feel more in control and less worried about privacy. This also builds trust, may lead them to switch from old retailers, and finally gives comfort while improving satisfaction. However, AI based personalization was not statistically significant (β = 0.090, p = .244). This shows that in current times personalization has become an essential expectation of online shopping and does not create any additional satisfaction unless the process includes transparency and a sense of control.

The findings reveal that AI-driven personalization has no statistically significant effect on customer satisfaction or repeat purchases among Lazada users in Thailand. However, Ahmed et al. (2025) and Hardcastle et al. (2025) found that personalization can enhance enjoyment, reduce decision fatigue, and lead to emotionally engaged users. Moreover, Owen et al. (2023) noted that personalization builds trust when deployed ethically and transparently. In the case of Lazada, users likely see personalization as an essential feature of browsing at an e-commerce site, but not the creative and innovate feature. For Lazada users, the value of personalization depends more on whether it is transparent and gives them control than on personalization alone.

The findings demonstrate strong positive effect of recommendation algorithm transparency on customer satisfaction of Lazada users in Thailand. Owen et al. (2023) have suggested that trust builds beliefs of transparency that minimize feelings of manipulation, while Akbar et al. (2024) report that satisfied customers are more likely to accept that the appropriateness of the recommendation. Transparent recommendation system is perceived as a source of credibility also as an element of fairness enhances customer satisfaction.

The study found that customer control over personalization has a significant and positive impact on customer satisfaction. Alkudah and Almomani (2024) emphasized that control over user's digital interactions fosters a sense of fairness and trust in the system. Turatti (2025) similarly found that control also enhanced the emotional commitment from users, as it made them feel more invested and valued in the experience.

This study’s finding confirmed that customers' satisfaction was a powerful tool on Lazada users repeat purchase behavior in Thailand. Hardcastle et al. (2025) highlighted those positive experiences the user has creates trust and familiarity, which encourages users to engage repeatedly. Jiradilok et al. (2013) demonstrated that satisfaction had the most decisive influence on repurchase intention. Customer satisfaction was the most important factor to ensure long-term customer loyalty. Lazada must ensure that their users' experience is satisfactory through good quality service, transparent communication, and providing reliable experiences.

 

4.5 Recommendations

Lazada should prioritize customer satisfaction as a key factor in building repeat purchase behavior. Lazada should develop strategies that enhance customers’ shopping experience through AI features to improve satisfaction and, consequently, repeat purchases. Lazada needs to improve its personalization by making the system transparent. Moreover, Lazada should develop simple and user-friendly control functions that allow customers to adjust recommendation levels and decide what personal data can be used. Giving users this sense of autonomy will strengthen their feeling of being respected and valued, which directly contributes to satisfaction. Lazada should also ensure that the overall service experience is reliable and efficient. Fast shipping, accurate delivery, and responsive customer service are important for building positive emotions toward the platform. When customers enjoy a smooth and trustworthy shopping journey, they are more likely to recommend the platform to others and continue purchasing repeatedly. By focusing on transparency, user control, and reliable service, Lazada can strengthen satisfaction, increase loyalty, and secure long-term growth through repeat purchases.

4.6 Further Study Recommendations

 

The current study targeted only Lazada to find the impact of AI features on customer satisfaction and repeat purchase. Further studies should do in other e-commerce platforms as Shopee, and SHEIN. This study implemented the quantitative research method, and used three AI features (personalization, transparency, customer control over personalization) to study how AI features on customer satisfaction and repeat purchase. Further researchers should use other AI features as Chatbot, voice assistants, and visual search tools. Moreover, future research should apply different research methods as qualitative and mixed-method to get in-depth understanding on AI-features of e-commerce platform. Future studies should collect data from different e-commerce users and nations to understand the different consumer perception on e-commerce.

 

5. Conclusions

This study shows that customer satisfaction is the most powerful predictor of repeat purchase for Lazada consumers in Thailand. While AI-driven personalization did not establish any significant direct effect, recommendation transparency and customer control were both found to explain customer satisfaction significantly in this research. Satisfied customers are more likely to repurchase, be loyal, and recommend others to buy products through Lazada, thus making their purchasing experience and have positive retention effects. Consequently, the results demonstrate that satisfaction is a major mediator of AI recommendations on customer retention. Overall, the current research provides insights into how e-commerce AI-driven features could be better designed and managed. By focusing on customer satisfaction, Lazada and similar businesses can reap the dividends of sustainable growth through repeated purchases and long-term loyal customers.

 

 

References

 

1.     Abinesh & Dulloo, 2024, The impact of AI-Driven personalization on customer satisfaction in E-Commerce: balancing technology, transparency, and control. (2024). In Journal of Computational Analysis and Applications (Vol. 33, Issue 2, pp. 649–655)

2.     Agoro, H., Templer, S., Maddog, M., & University of Ibadan. (2021b). AI-Powered Personalization: Enhancing user experience through customized shopping recommendations. https://openurl.ebsco.com/EPDB%3Agcd%3A12%3A28051788/detailv2?sid=ebsco%3Aplink%3Ascholar&id=ebsco%3Agcd%3A181584469&crl=c&link_origin=scholar.google.co.th

3.     Ahmed, S. M. M., Owais, M., Raza, M., Nadeem, Q., & Ahmed, B. (2025). The impact of AI-Driven personalization on consumer engagement and brand loyalty. Qlantic Journal of Social Sciences., 6(1), 311–323. https://doi.org/10.55737/qjss.v-iv.24313

4.     Akbar, M. U., Ibrahim, N., Nabil, S. J., Iqbal, K. A., & Islam, A. (2024). The influence of artificial intelligence on consumer trust in E-Commerce: Opportunities and ethical challenges. European Journal of Theoretical and Applied Sciences, 2(6), 250–259. https://doi.org/10.59324/ejtas.2024.2(6).20

5.     Alkudah, N. M., & Almomani, T. O. (2024). The integration of artificial intelligence techniques in E-Commerce: enhancing online shopping experience and personalization. Global Journal of Economic and Business, 14(6), 640–658. https://doi.org/10.31559/gjeb2024.14.6.8

6.     Bok, S. K. (2023). Enhancing User Experience in E-commerce through Personalization Algorithms. In Haaga-Helia University of Applied Sciences, Haaga-Helia University of Applied Sciences (pp. 1–61) [Thesis]. https://www.theseus.fi/bitstream/handle/10024/815645/Bok_Sun%20Khi.pdf?sequence=2&isAllowed=y

7.     Geetha, M., Rhetheyu, R., Department of Information Technology, Puducherry Technological University, S, R., Umapathy, K., & Department of Information Technology, Puducherry Technological University. (2024). A survey on E-Commerce recommendation systems using artificial intelligence and current trends for personalization to improve customer experience. In International Journal of Engineering Research & Technology (IJERT) (Journal-Article V13IS030252). http://www.ijert.org

8.     Hardcastle, K., Vorster, L., & Brown, D. M. (2025). Understanding customer responses to AI-Driven personalized journeys: impacts on the customer experience. Journal of Advertising, 176–195. https://doi.org/10.1080/00913367.2025.2460985

9.     Jiradilok, T., Malisuwan, S., Madan, N., & Sivaraks, J. (2013). The impact of customer satisfaction on online purchasing: a case study analysis in Thailand. Journal of Economics Business and Management, 5–11. https://doi.org/10.7763/joebm.2014.v2.89

10.  Kafey, A. (2025). The role of artificial intelligence in enhancing customer experience in e-commerce among young consumers (18–35). In Haaga-Helia University of Applied Sciences, Haaga-Helia University of Applied Sciences (pp. 1–40) [Thesis].

11.  Liu, C., & Zhang, Z. (2024). Can artificial intelligence (AI)-driven personalization influence customer experiences? - a quantitative study on TikTok integration with artificial intelligence. In Master’s Thesis (Vol. 15) [Thesis; PDF]. Uppsala University. https://uu.diva-portal.org/smash/get/diva2:1874165/FULLTEXT01.pdf

12.  LIU, X. (2020). Research on the impact of online shopping experiences on customer’ repeated purchase intension. In an independent study submitted in partial fulfillment of the requirements for the degree of business administration.

13.  Nuralam, I. P., Yudiono, N., Fahmi, M. R. A., Yuliaji, E. S., & Hidayat, T. (2024). Perceived ease of use, perceived usefulness, and customer satisfaction as driving factors on repurchase intention: the perspective of the e-commerce market in Indonesia. Cogent Business & Management, 11(1). https://doi.org/10.1080/23311975.2024.2413376

14.  Oguntibeju, O. O., Adonis, M., & Alade, J. (2024). Systematic Review of Real-Time Analytics and Artificial Intelligence Frameworks for Financial Fraud Detection. IJARCCE, 13(9). https://doi.org/10.17148/ijarcce.2024.13920

15.  Oh, Z. J., Odebunmi, A. T., Tu, G. Y., Teoh, Y. E., Thoe, W. M., Toh, Y. J., Toh, Y. X., Ali, A. J., & Kee, D. M. H. (2025). The influence of artificial intelligence on customer retention in the Asian E-Commerce market. Asia Pacific Journal of Management and Education, 8(1), 170–186. https://doi.org/10.32535/apjme.v8i1.3865

16.  Owen, A., Moore, J., IU International University of Applied Sciences, & Antony Owen. (2023). Ethical Considerations in AI-Driven E-Commerce Solutions: Balancing Personalization andprivacy.https://www.researchgate.net/publication/389250000_Ethical_Considerations_in_AI-Driven_E-Commerce_Solutions_Balancing_Personalization_and_Privacy.

17.  Rita, P., Oliveira, T., Nova Information Management School (NOVA IMS), Universidade Nova de Lisboa, & ISCTE Business School (IBS), Instituto Universitário de Lisboa. (2019). The impact of e-service quality and customer satisfaction on customer behavior in online shopping. InHeliyon (p.e02690) [Journal-article]. https://doi.org/10.1016/j.heliyon.2019.e02690

18.  Tufahati, N., Safa’atul Barkah, C., Tresna, P. W., Chan, A., & Business Administration Study Program, Faculty of Social and Political Sciences, Padjadjaran University, Indonesia. (2021). The impact of customer satisfaction on repurchase intention. In Journal of Business & Applied Management: Vol. Vol.14 (Issue No. 2, pp. 177–186) [Journal-article]. http://journal.ubm.ac.id/ (Original work published 2021)

19.  Turatti, R. C. (2025). The impact of AI-Power marketing automation in E-Commerce. International Seven Journal of Multidisciplinary, 4(2), 266–272. https://doi.org/10.56238/isevmjv4n2-009

20.  Weiler, A. S., & Gilitwala, B. (2023). Why Bangkokians use online food delivery services after COVID-19 restrictions have been lifted. Rajagiri Management Journal, 18(2), 151–166. https://doi.org/10.1108/ramj-08-2023-0244