A Study on the Impact Dimensions of Customer Experience on Cross?border E?commerce Platforms from the Perspective of E?Commerce Politeness: The Case of Apparel

for evaluating platform politeness. Agrawal (2023) shows that product risk and opaque return/exchange policies undermine consumer trust in online shopping. Wojciechowski et al. (2023) point directly to sizing mismatch issues causing high return rates; when users suffer returns due to inaccurate information, the platform’s “impoliteness” experience emerges. From a cross-cultural perspective, differences in apparel consumption raise higher demands on platform politeness. Eldemerdash & Abou-Ghali (2023) note that fashion preferences are shaped by cultural and social variables, implying that cross-border platforms must adapt to diverse cultural expectations. Kim et al. (2009), from a technology acceptance view, confirm that perceived ease of use and usefulness affect user adoption. Rossolov et al. (2021) find in emerging markets that apparel is a priority good and consumers are highly price-sensitive, meaning price-related communication experiences directly influence overall evaluations of platform politeness. In summary, the particularities of apparel transactions make them highly dependent on information accuracy, after-sales protection, and cross-cultural adaptation—elements that are central Experience Dimensions of E-Commerce Politeness. The empirical analysis in this study will further reveal which specific touchpoints on DHgate are most critical for polite or impolite experiences. 2.3 Current Status and Development of E-Commerce Politeness The theoretical roots of E-Commerce Politeness trace back to human–computer interaction studies on social software. Whitworth (2005) proposed the concept of “politeness computing,” arguing that politeness is a key social requirement in human–computer interaction: impolite software drives users away, while polite software attracts them. Whitworth (2009) later defined politeness as a core requirement of social software, essentially about “giving users choice,” manifested as respect, transparency, and personalization. Whitworth & Ahmad (2013) argued that social technologies must consider both social and technical needs. These early studies laid the theoretical foundation for understanding user interaction quality on e-commerce platforms. In recent years, scholars have applied politeness frameworks to e-commerce research. Chen & Huang (2025) examine platform service quality from a politeness perspective and construct a model of how politeness affects user value co-creation. Chen & Lu (2025) extend this perspective to logistics platforms, arguing that platform E-Commerce Politeness is key to enhancing user trust and satisfaction and shaping brand image. Chen & Ye (2025), focusing on food delivery platforms, identify politeness-related issues in interface design, platform governance, and after-sales service. Collectively, these studies indicate that E-Commerce Politeness is an integrative concept affecting user experience across interaction touchpoints. From the negative side, Huang & Liu (2025) find that user complaints often stem from perceived moral violations by service agents, which reflect users’ perceptions of “platform impoliteness.” Widiana & Sri (2024) analyze sellers’ verbal behaviors in live-stream commerce and show that P-ISSN 2695 2416 appropriate verbal strategies positively affect user experience. Calista & Kuntjara (2023) further point out that expressions of politeness are culturally dependent; users from different cultural backgrounds have varying expectations of politeness, which is particularly relevant for cross-border platforms like DHgate. Overall, existing research has revealed the connotations and impacts of E-Commerce Politeness from multiple angles, but few studies focus on user politeness experience for specific categories on cross-border B2B platforms. This study targets DHgate’s apparel category to identify the core Experience Dimensions of E-Commerce Politeness that influence user experience in that context. 3. Research Methodology 3.1 Critical Incident Technique The Critical Incident Technique , formally proposed by Flanagan (1954), is a method for identifying effective or ineffective behaviors by collecting and analyzing user-described specific incidents. Since its inception, CIT has been widely applied across fields: in service research, Gremler (2004) reviewed 141 studies and confirmed CIT’s value for exploring service issues; in e-commerce service quality assessment, Ingaldi & Brožová (2020) used CIT to analyze key satisfaction and dissatisfaction incidents on e-commerce sites; in mobile app UX, Salo & Frank (2017) employed CIT to reveal links between contextual features and user responses; in police crisis intervention, Watson et al. (2010) and Compton et al. (2014) applied CIT to evaluate officer training, and Peterson & Densley (2018) and Watson et al. (2017) further validated its applicability for practice evaluation. In summary, CIT has been extensively validated as an effective tool for identifying key factors that influence user experience. Therefore, this study adopts the Critical Incident Technique to investigate DHgate apparel users’ E-Commerce Politeness experience, providing a solid methodological basis. 3.2 Research Design This study applies CIT to DHgate’s apparel category by analyzing user reviews to identify critical incidents that affect E-Commerce Politeness. Data were drawn from publicly accessible DHgate user reviews, yielding 490 valid comments: 251 satisfaction incidents and 239 dissatisfaction incidents. Three coders independently classified incidents according to a coding framework; inter-rater reliability was tested, and core Experience Dimensions affecting users’ E-Commerce Politeness were identified based on the classification results. Based on CIT and the research design, this paper constructs the research logic framework for E-Commerce Politeness experience (see Figure 1). The framework presents the study’s core ideas and content modules. P-ISSN 2695 2416 Figure 1 Research logic framework for E-Commerce Politeness in cross-border apparel 4. Data Analysis 4.1 Data Collection The dataset originates from DHgate’s apparel category; all data are publicly accessible user reviews on the platform. To ensure ethical compliance, all collected data were anonymized to remove any personally identifiable information. A purposive sampling strategy was used to ensure both market representativeness and content depth. The selection criteria and procedure were: 1. Top‐selling items: Items were selected by sales ranking to capture products with large user bases and abundant reviews, ensuring market‐level representativeness. 2. Time window: Reviews were collected from February to April 2025, a period of platform stability, to ensure timeliness and consistency. 3. Per‐item sampling: For each selected product, 10–15 reviews were purposively chosen, balancing satisfied (5‐star) and dissatisfied (1‐star) comments and prioritizing detailed reviews that include specific usage scenarios or descriptions to ensure informational depth. Following this procedure, a qualitative dataset of 490 user reviews was constructed. Due to its focused source (single platform and vertical category), continuous time span, sales‐based representativeness, and high information density, the dataset is suitable for exploratory thematic analysis and provides a solid empirical basis for understanding consumer experience in this market. 4.2 Classification Principles This study, after further data analysis and consolidation, divided the collected satisfied and dissatisfied apparel incidents into five categories, as shown in the table below: P-ISSN 2695 2416 Table 1 · Event Classification Names and Definitions 4.3 Reliability and Validity Testing Reliability testing of the CIT is a key step to ensure the objectivity and dependability of classification results. It comprises two indicators—individual classification consistency and inter-coder consistency—both essential for securing the reliability and validity of critical incident classifications. 4.3.1 Individual Classification Consistency Individual classification consistency refers to the degree of agreement between two classifications made by the same coder on the same set of incidents at different times. When the individual consistency of two or more coders exceeds 0.8, it indicates good stability and agreement (Flanagan, 1954). In this study, three coders performed the classifications: the first is a university lecturer specializing in e-commerce, the second is a practitioner in the cross-border apparel e-commerce industry, and the third is an experienced cross-border e-commerce buyer. Individual consistency was assessed for each coder (see Tables 2 and 3). All three coders achieved individual consistency greater than 0.8, indicating satisfactory stability and consistency of the classification results. Table 2 · Individual Classification Consistency of Coders Event Satisfied events Dissatisfied events Number of mutual individual consistencies Individual classification consistency Number of mutual individual consistencies Individual classification consistency Coder 1 227 0.904 217 0.908 Coder 2 226 0.90 219 0.916 Coder 3 211 0.841 209 0.874 Satisfied Event Category Definition Dissatisfied Event Category Definition Product Quality (satisfied/unsatisfied) Overall performance in functionality, comfort, material, durability, workmanship (authenticity), and reliability (failure rate). Customer Service (satisfied/unsatisfied) Seller‐provided consultation, Q&A, and complaint‐handling services. Logistics Service (satisfied/unsatisfied) System capability for delivery timeliness, packaging protection, delivery accuracy, and tracking transparency. Description Consistency (satisfied/unsatisfied) Match between the actual item and seller’s description in color, quantity, specifications, and other details. Cost‐performance Ratio Perceived value for money; whether customers consider the product worth buying. Claims Mechanism Seller’s compensation rules and procedures for damage, loss, or delay. P-ISSN 2695 2416 4.3.2 Inter-coder Consistency Inter-coder consistency refers to the degree of agreement among different coders when classifying the same set of incidents. It is important for reflecting the clarity and objectivity of the coding framework. To ensure rigor and accuracy, this study conducted two rounds of coding with a 14-day interval; the results from both rounds were used to assess inter-coder consistency (see Table 3). Table 3 · Number of Mutual Consistencies Among Coders Event Satisfied events Dissatisfied events Mutual Consistencies Coder 1 Coder 2 Coder 3 Coder1 Coder 2 Coder 3 Coder 1 227 -- -- 217 -- -- Coder 2 218 226 -- 203 219 -- Coder 3 202 203 211 197 190 209 Holsti’s (1969) reliability analysis was employed to assess the degree of agreement among multiple coders; the calculation formulas are presented in Equations (1) and (2). R= (N×A) 1+[(N−1)×A] (1) A= 2M12 n1+n2+ 2M23 n2+n3+ 2M13 n1+n3 N (2) Where: R = Reliability N = Number of coders A = Average mutual consistency M = Number of identical classifications among coders n = Number of samples each coder classified Table 4 · Classification Reliability Classification Average mutual consistency (A) Reliability (R) Satisfied Events 0.938 0.980 Dissatisfied Events 0.915 0.983 The results indicate that the reliability for both satisfied and dissatisfied incidents exceeds 0.8, meeting scholarly requirements; this demonstrates that the classification results are reliable and provide a solid data foundation for subsequent studies of user experience on cross-border e-commerce platforms based on e-commerce politeness. 4.3.3 Validity Analysis In reliability and validity testing, validity refers to the extent to which a measurement instrument accurately captures the intended research construct. This study ensures the scientific rigor and applicability of the coding framework through three types of validity checks. 4.3.3.1 Content Validity Content validity assesses whether the measurement items sufficiently cover the core domains of the research objective. As Haynes et al. (1995) note, content validity concerns the representativeness of measurement items for the target behavioral domain and should be verified through systematic literature review and expert judgment. The coding dimensions in this study P-ISSN 2695 2416 were derived from literature on cross-border e-commerce platforms, apparel products, and E-Commerce Politeness, and they encompass key dimensions such as Product Quality, Customer Service, Logistics Service, Description Consistency, Cost-performance Ratio, and Claims Mechanism—thereby ensuring comprehensive coverage of the core domains of users’ E-Commerce Politeness experience. 4.3.3.2 Expert Validity Expert validity examines whether the content and structure of the measurement instrument are endorsed by domain experts. Cronbach & Meehl (1955) emphasize that when no single criterion measure is available, multiple lines of evidence—including expert judgment—are necessary to support an instrument’s interpretability. In this study, scholars in cross-border e-commerce and platform operations personnel were invited to review the coding dimensions. Based on their feedback, definitions and boundaries of the dimensions were adjusted to ensure the coding framework aligns with both industry practice and academic understanding. 4.3.3.3 Face Validity Face validity refers to whether the instrument appears to measure what it claims to measure. Thomas et al. (1992) argue that instruments should “look valid” to participants and users; otherwise, participant cooperation and the acceptability of results may be affected. The coding dimensions in this study were refined through repeated team discussions to ensure that names and definitions are intuitive and closely correspond to issues reflected in user reviews, thereby providing logical coherence. In summary, this study demonstrates satisfactory content validity, expert validity, and face validity, ensuring the effectiveness of the coding results. 4.4 Classification Results From the five categories into which satisfied and dissatisfied incidents were each divided, key incident for each category were extracted as examples; see Table 5 for illustrative critical incidents. Table 5 · Examples of Critical Incidents Category Satisfied Incident (example) Dissatisfied Incident (example) Product Quality 'Quality exceeded expectations; from the back label to the details everything was perfect; will repurchase!' 'Worst quality I've ever received! Hoodie material is cheap. Design is okay but fabric is terrible. Do not buy!' Customer Service 'Excellent! Seller communicated smoothly. I selected the wrong size by mistake; the seller patiently helped adjust it—very considerate.' 'Terrible customer service; the seller ignored me and lied. Warning everyone: don't waste your money!' Logistics Service 'Would give 1000 stars if I could! Fastest delivery ever; tracking was precise.' 'Don't buy this jacket! It showed as delivered but I didn't receive it; contacting the courier and appealing to the seller was useless.' P-ISSN 2695 2416 Category Satisfied Incident (example) Dissatisfied Incident (example) Description Consistency 'Love this T-shirt! Size exactly matches the description even though I usually use US sizing. Highly recommend!!' 'Poor product quality and random sizing. The knit I ordered is very different from the advertisement; the seller refuses to communicate.' Cost-performance Ratio (Satisfied) / Claims Mechanism (Dissatisfied) 'T-shirt offers excellent value for money; great price and surprisingly good quality! Recommended!' 'Item never arrived; seller refused refund and closed the return request—no refund. Big scam! Kept stalling about restocking but never fulfilled; had to dispute with the bank. Do not buy!' Preliminary statistics were performed on the classified data to understand user feedback across different categories, as shown in Tables 6 and 7: Table 6 · Summary of Satisfied Event Categories Category Coder 1 Coder 2 Coder 3 Average Rank Product Quality 182 178 177 179.00 1 Customer Service 18 21 18 19.00 3 Logistics Service 33 32 33 32.67 2 Description Consistency 11 12 10 11.00 4 Cost‐performance Ratio 7 8 13 9.33 5 From Table 6, among the 251 satisfied apparel incidents divided into five categories, the highest mean is Product Quality, reaching 179 (approximately 71%), which is far greater than the combined averages of the other four categories and is about five times the mean of Logistics Service (32.67), the second-ranked category. This indicates that DHgate performs well in product quality for apparel products. Table 7 · Summary of Dissatisfied Event Categories Category Coder 1 Coder 2 Coder 3 Average Rank Product Quality 139 138 133 136.67 1 Customer Service 14 21 19 18.00 4 Logistics Service 37 35 36 36.00 2 Description Consistency 31 28 34 31.00 3 Claims Mechanism 18 17 17 17.33 5 From Table 7, among the 239 dissatisfied apparel incidents divided into five categories, dissatisfaction with Product Quality has an average of 136.67 (approximately 56%), ranking first. The second and third highest averages are Logistics Service (36) and Description Consistency (31), with similar mean counts. These results suggest that issues such as poor product quality, long delivery times, lost packages, seller shipping errors, and false advertising on the DHgate platform have produced an extremely poor level of e-commerce politeness, severely undermining customers’ experiential perceptions. P-ISSN 2695 2416 5. Conclusions and Recommendations 5.1 Conclusions The study results indicate that product quality is the most critical dimension affecting users’ perceptions of e-commerce politeness. In satisfied incidents, product-quality-related events account for as much as 71%, far exceeding other dimensions; in dissatisfied incidents, product quality also ranks first, accounting for 56%. This finding suggests that when a product’s functionality, materials, and workmanship meet or exceed user expectations, users perceive the platform as “polite”; conversely, when product quality is poor, users experience pronounced “impoliteness.” As the final deliverable in user–platform interactions, product quality directly determines users’ overall evaluations of the platform experience. Logistics service is the second most influential dimension on user experience, ranking second in both satisfied and dissatisfied incidents. Experiences related to delivery timeliness, tracking accuracy, and delivery completeness form an important part of users’ service perceptions. When users encounter delivery delays, lost packages, or opaque tracking information, this information vacuum directly triggers feelings of “impoliteness.” Notably, description consistency ranks third among dissatisfied incidents, indicating that issues such as incorrect sizing and mismatches between images and descriptions are also key causes of negative experiences. When the actual item deviates from the seller’s description, users’ trust in the platform is undermined, producing a sense of being “deceived” and perceived impoliteness. In summary, the e-commerce politeness experience for DHgate’s apparel category exhibits a multidimensional structure, with product quality, logistics service, and description consistency being the three most critical dimensions. Platform and seller performance in these three areas directly determines users’ overall feelings during interactions. When problems occur in these areas, users encounter what this study defines as “platform impoliteness.” 5.2 Recommendations Based on the above findings, this study offers the following recommendations for DHgate, its sellers, and consumers. 5.2.1 Recommendations for the Platform The platform should prioritize improvements in the three dimensions most important to users: product quality, logistics service, and description consistency. For product quality, establish a dynamic rating mechanism to issue warnings or limit traffic for sellers with high complaint rates, and preferentially allocate traffic to high-quality sellers. For logistics service, coordinate with logistics providers to clarify timeliness standards for each stage and enhance the real-time transparency of tracking information. For description consistency, strengthen spot checks of product page descriptions, constrain sellers who engage in exaggerated or false advertising, and optimize after-sales claims procedures to shorten response times and prevent users from falling into a “no channel for complaints” impolite experience. 5.2.2 Recommendations for Sellers Sellers should strictly control product quality and ensure that shipped items match samples; provide truthful and accurate product descriptions to reduce expectation gaps caused by information asymmetry; and respond promptly and patiently to customer inquiries and complaints, avoiding ignoring or shirking responsibilities. Sellers should also pay attention to user communication regarding logistics: proactively inform users of progress when delays or damages P-ISSN 2695 2416 occur due to logistics, avoid leaving users in an information vacuum, and, within their capabilities, choose more reliable logistics channels to improve the delivery experience. 5.2.3 Recommendations for Consumers Consumers should carefully read product descriptions and size charts before ordering, consult other users’ comments on product quality and sizing accuracy, and, when necessary, confirm key information with sellers via platform customer service. If quality issues or description inconsistencies arise, consumers should promptly communicate through platform channels and retain evidence, reasonably using the platform’s claims mechanism to protect their rights. At the same time, consumers should understand that cross-border e-commerce logistics have inherent uncertainties, set expectations accordingly, and share honest feedback through reviews when they have a positive experience to help other users.