What drives Chinese users’ satisfaction in EV charging? Uncovering and prioritizing service design factors through a hybrid decision approach

1 Introduction

In recent years, the electrification of transportation has emerged as a pivotal strategy for achieving carbon neutrality and enhancing energy efficiency (Xiong et al., 2022; Peng et al., 2023a). Electric vehicles (EVs), as a form of green mobility, play a crucial role in fostering sustainable transportation development and attaining carbon neutrality (Bie et al., 2023). Nevertheless, the industrialization of EVs in China continues to encounter significant challenges, particularly regarding the quality of charging facilities, which has not met user expectations (Ren and Sun, 2025). This shortfall represents a major constraint on the industry’s growth. In recent years, the number of electric vehicles in operation has increased significantly, resulting in an average of nearly 100 million daily charging requests. However, the quality of service at charging stations has not met user expectations (Pu et al., 2023).

In 2020, the EV charging facilities sector was identified as one of the seven new infrastructure domains. Currently, the Chinese market presents significant investment opportunities for the development and enhancement of the charging station industry (Wu et al., 2023). Surveys indicate that by the end of 2024, China had installed 1.797 million public charging piles, resulting in an average of 3.413 charging piles per vehicle. Consequently, the overall ratio of vehicles to charging piles was approximately 2.5:1 (Zhang et al., 2026).

User experience and satisfaction at public charging stations significantly influence the number of monthly active users. Consequently, it is essential to investigate strategies for improving the usage experience of EV users, ensuring both convenience during use and comfort while waiting (Habbal and Alrifaie, 2024).

Currently, public charging stations remain primarily focused on technical aspects. Research in this area predominantly addresses site planning, enhancements in charging system efficiency, and optimization of charging routes, among other topics (Ray et al., 2023; Faustino et al., 2023; Rana et al., 2023; Zeng et al., 2024; Cao et al., 2025; Guobin et al., 2026). Song et al. (2024) delineated the fundamental driving path of EVs into three segments to meet the application requirements of reinforcement learning in the context of charging path planning for EVs. Huang and Kockelman (2020) employed genetic algorithms to identify optimal locations and design features for charging stations, aiming to maximize profits while enabling passengers to monitor congestion levels at these stations and adjust their route choices accordingly. Pan et al., 2024 introduced an intelligent charging strategy for peak hours to address the path planning challenge while minimizing charging costs.

The design and research of EV charging stations continue to be a prominent topic. When optimizing the design of these stations, it is essential to consider the varied needs of users and to coordinate the different facilities within the station. This approach aims to enhance both the quality of charging services and the overall user experience. To comprehensively address the diverse demands of users in the design and service of charging stations (Mamkhezri, 2025; Lin et al., 2021) this paper introduces the AHP-DEMATEL composite model. This model identifies core user demands, thereby providing a foundation for the optimized design of charging stations and improving the user charging experience.

2 Literature review and hypotheses development

The primary process of product and service innovation entails the collection and analysis of user demands. In contrast to traditional engineering products, the design of charging stations incorporates interrelated and dynamically evolving hardware, service, and software requirements over time (Pratiwi et al., 2024). Consequently, employing a multi-attribute decision-making method can facilitate the development of a demand decision-making model for charging stations. This approach allows for the scientific ranking of each function based on its significance, thereby offering designers a more intuitive reference framework (Deng et al., 2026; Liu et al., 2024; Wang et al., 2024; Peng et al., 2023b).

AHP was introduced by Professor Saaty in the early 1970s as an effective approach for addressing complex problems through the resolution of hierarchical subproblems. It is widely utilized for ranking user requirement weights (Liu et al., 2023b; Chen et al., 2024). Xu et al. (2025) employed AHP to develop a demand hierarchy model for sports service robots, subsequently applying weighted calculations to quantify and rank user demands, thereby enhancing the user experience. Liu et al. (2023a) created a restaurant chair design evaluation system focused on user needs by utilizing AHP and fuzzy AHP. However, AHP is inherently subjective and has certain limitations when applied in isolation, which may compromise the accuracy of user demand weights (Du and Zhu, 2025; Wang and Qian, 2025). DEMATEL was initially proposed by scholars at the Battelle Institute in Geneva in 1971 and is primarily used to identify key elements within complex systems. Essentially, it conceptualizes complex systems as directed graphs with weights and employs relevant mathematical theories to address intricate problems (Zhang et al., 2025). Both AHP and DEMATEL possess distinct advantages and disadvantages when utilized independently for requirements analysis, as illustrated in Table 1. Consequently, employing the AHP-DEMATEL combined model to establish the comprehensive influence degree can mitigate the limitations associated with using either method in isolation. This approach allows for an objective assessment of the contributions of influencing factors to the design of charging stations, thus offering a more effective methodology for their design (Zhou et al., 2018).

TABLE 1. Advantages and disadvantages of AHP and DEMATEL methods.

3 Methodology and data collection procedure

3.1 User demand research

This study systematically examined users’ charging needs, behavioral patterns, and pain points across various scenarios through in-depth interviews and structured questionnaires, thereby establishing a foundation for developing a user role model. The investigation was conducted from three dimensions. The first dimension encompasses the basic information of users, focusing on the characteristics of the target demographic. The second dimension addresses vehicle usage, gathering data on the purchase motivations of target users, their travel scenarios, and the mileage required to meet their needs. The third dimension pertains to charging practices, which involve collecting information on how target users typically search for charging stations and the challenges they currently face during the charging process.

The questionnaires were distributed online to EV users and offline at charging stations and EV 4 S stores. A total of 653 valid responses were collected. By summarizing and integrating the charging behavior characteristics of users, it was determined that user satisfaction is relatively low, suggesting significant opportunities for optimizing and enhancing the charging station experience. The findings from the questionnaire survey will serve as initial points for improvement, guiding the design of the service process in subsequent stages.

3.2 Record and organize the user’s charging behavior process

This stage comprises two steps. Initially, an expert interview study was conducted (Bradley et al., 2021), involving 100 participants: 20 researchers specializing in charging stations, 20 university professors, 20 regional managers from State Grid, 20 EV owners, and 20 senior designers.

Based on expert recommendations and an analysis of the charging behavior of EVs, the charging service process can be summarized as follows: (1) EV owners search for available charging stations; (2) Upon locating a usable charger, the owner parks the EV in the designated space; (3) The owner retrieves the charger and initiates the charging process via the charging service platform; (4) The owner waits at the service station while charging occurs; (5) Once the vehicle is fully charged, the owner unplugs the charger and departs from the charging station.

Accordingly, the design team performed user behavior observations at various charging stations, focusing on peak electricity consumption periods from 8:00 to 11:00 and from 18:00 to 23:00. This approach allowed for a comprehensive examination of user charging behavior, encompassing spatial navigation as users sought charging stations, the operational process of charging, the specific locations utilized for charging, the steps involved in using the charging equipment, and the number of charging piles accessed by users during the 4–5-h peak periods. The observers employed a structured behavior observation record form to sequentially mark and time-code each of the aforementioned key behavior nodes, utilizing fixed-point video recording as the foundational basis for verification.

3.3 User requirement analysis based on user journey mapping

Based on the research data, user characteristics were visually represented to generate the user role model necessary for this study (Li et al., 2023; Han et al., 2026). The target demographic comprises non-operational fast-charging EV users aged 23 to 44. Typical users were selected for non-participatory observation through a purposeful sampling method. Key behavioral touchpoints during the users’ interactions were documented, and the behavioral data were systematically organized, as illustrated in Figure 1.

User experience maps (Chhabra et al., 2023) have emerged from the field of service design, highlighting the transformations of users across various scenarios and temporal dimensions. These maps facilitate the analysis of the usage process at charging stations and users’ interaction behaviors throughout the engagement. By identifying user pain points, they enable the extraction of potential design optimization opportunities. To enhance the connection between users and the product, the team employed custom analysis cards to systematically organize and evaluate the critical information provided by users. In relation to the charging process, they explicitly defined user touchpoints, behavioral expectations, and goal orientations at each stage, thereby identifying users’ latent needs, as illustrated in Figure 2. Upon reviewing negative user feedback, several pain points were identified: the charging station lacks non-electric service facilities, operational status prompts, and other essential functions. Following this analysis, key optimization directions were distilled, including the enhancement of service-oriented supporting facilities and the improvement of the lighting within the charging station’s wayfinding system.

3.4 Construct a hierarchical structure model for charging stations

Based on the findings from the user experience map, expert evaluations were integrated to extract the core dimensions of user needs, which are summarized into five key aspects: recognizability, usability, trustworthiness, reliability, and perceptibility of information. The design of charging stations must incorporate a clear identification system, prioritize user-friendly features, and facilitate easy navigation. Furthermore, the interactive functions of charging stations should align with users’ charging behaviours to promote rational usage. To enhance user experience and alleviate charging-related anxiety, charging stations should offer service facilities such as rest areas, entertainment options, and dining services. Additionally, the safety of users during the charging process must be ensured, thereby eliminating operational risks. Charging stations should also facilitate the efficient absorption and utilization of information by users. A hierarchical analysis model for charging stations has been constructed based on the interactions among users, products, and the environment (Kwok and Tang, 2023), as illustrated in Figure 3. The criterion layer (Qi and Kiesu-Kim, 2024) is denoted by B ₁, B₂, B₃, and B₄, while the sub-criterion layers are represented as B_i1, B_i2, B_in (i = 1,2). This structure is detailed in Table 2.

TABLE 2. Sub criteria layer definition.

4 Results

4.1 Quantitative analysis of subjective weights

The design team established an evaluation group comprising five experts and fifteen target users, utilizing the nine-level importance scale method (Ma et al., 2025). They performed pairwise comparisons of the constructed index systems at each level, developed corresponding judgment matrices, and calculated the Analytic Hierarchy Process (AHP) weight values W ¹ for each level of indicators using the eigenvector method. Additionally, they conducted consistency tests on the judgment results by calculating the consistency ratio (CR) to ensure the reliability of the assessment outcomes.

To ensure the validity and reliability of the research data, this study employed the outlier elimination method to preprocess the questionnaire data and remove extreme values. The questionnaire results were analysed and discussed through multiple iterations of the Delphi method until the expert group’s opinions achieved stable convergence, thereby constructing a comprehensive judgment matrix (Sokolov et al., 2025). Following consistency testing, the consistency ratios of all judgment matrices fell below the acceptable threshold of 0.1, and the weights for each index were derived, as presented in Table 3.

TABLE 3. AHP weight values and optimal ranking of each indicator.

4.2 The influence relationship among user demands

To investigate the correlation characteristics among user demand indicators for charging stations, the DEMATEL method was employed to adjust the initial weight W ¹ (Du and Li, 2021; Sonar et al., 2025). The 528 valid samples utilized for the DEMATEL analysis were exclusively obtained from the user database of the charging operation platform. All participants were required to possess a minimum of 6 months of experience with electric vehicles and to have utilized at least two distinct types of public charging stations. The 0–3 scale method was employed, where 0 indicates no impact, 1 signifies low impact, 2 represents medium impact, and 3 denotes high impact. This approach quantitatively assesses the 18 × 18 direct impact relationship matrix established based on 18 user demand indicators. Let X = (a_ij) _*n*×*n*, where a_ij denotes the direct influence degree of a_i on a_j. If i = j, then a_ij = 0. The calculation formulas of the direct influence matrix M and the comprehensive influence matrix T are shown in Equations 1, 2, where I is the identity matrix.

M = 1 max 1 ≤ i ≤ n ∑ j = 1 n a ij X (1)

T = M I − M − 1 (2)

By analyzing the D-R-M-N index set within the user demand dimension, the degree of contribution of the interactions among the indicators to the comprehensive weight coefficient is precisely represented. Specifically, utilizing the comprehensive influence matrix T, the sums of each row and each column are computed to derive the influence degree D_i and the affected degree R_i. Subsequently, the centrality N_i and the cause degree M_i are calculated (Fatima et al., 2025), as illustrated in Table 4.

TABLE 4. Comprehensive impact relationship between charging station demand index.

The effect D_i denotes the synthesized influence values across all indicators in relation to other indicators, and its calculation formula is shown in Equation 3:

D i = ∑ i = 1 n t i j , i = 1 , 2 ⋯ n (3)

The degree of influence Ri represents the comprehensive impact value of each column of indicators on other indicators, and the calculation formula is shown in Equation 4:

R i = ∑ j = 1 n t i j , j = 1 , 2 ⋯ n (4)

The centrality N _i denotes the influence among indicators, and its calculation formula is shown in Equation 5. A causality greater than 0 signifies that this indicator exerts an impact on other indicators. Conversely, a causality less than 0 indicates that this indicator is influenced by other indicators.

N i = D i − R i (5)

The cause degree M_i reflects the status and role of this indicator, and its calculation formula is shown in Equation 6. When the centrality is larger, it indicates that the importance of this indicator is stronger.

M i = D i + R i (6)

4.3 Calculate the comprehensive weight value

The integration of AHP and DEMATEL mitigates the limitations associated with relying on a single method, thereby yielding more comprehensive and precise weight results (Wu and Liao, 2021; Wang and Hsueh, 2013), as shown in Equation 7. This approach effectively quantifies and assesses the relative influence of each indicator within the decision-making process. The detailed results are presented in Table 5.

W = M i × W 1 ∑ j = 1 n M i × W 1 , i = 1 , 2 ⋯ n (7)

TABLE 5. The comprehensive rights value of various requirements indicators.

4.4 Calculate the comprehensive weight value

By comparing and analyzing the comprehensive weight values with those obtained through the single hierarchy analysis method, it is evident that the ranking results of the comprehensive weights exhibit greater accuracy and reliability. For example, although B ₁₁ (consistency) ranks relatively low when assessed using the AHP alone, it achieves a high ranking in M_i and exerts a significant influence on other metrics. Consequently, under conditions of controllable costs and feasible technology, it should be prioritized for development. The AHP weights for B ₃₃ (functional facilities) and B ₄₅ (occupancy enforcement) are relatively high; however, they do not adequately account for the interdependencies among the indicators. Two of these indicators are result-oriented factors, and their performance is readily affected by preceding indicators. To enhance this function, it is essential to improve the fundamental indicators and achieve an overall enhancement through indicator optimization. Furthermore, requiring charging stations to install parking space locks or provide functional supporting facilities necessitates the coordinated optimization of a multi-dimensional indicator system, which will inevitably increase R&D investment costs. Therefore, the priority order should be adjusted moderately in descending order to facilitate the optimal allocation of resources.

The B ₃₁ (visual aids) AHP possesses a relatively high weight; however, it serves primarily as an auxiliary function and does not significantly enhance user satisfaction. Consequently, a comprehensive assessment and trade-off analysis should be conducted during the development process, taking into account the overall context. Although the weights of B ₁₂ (information hierarchy) and B ₁₃ (product attributes) in the AHP are not substantial, they are fundamental causal factors that exert a considerable influence on other indicators, warranting priority consideration. Notably, the combined weight values of B ₂₃, B ₃₂, B ₄₁, and B ₅₁ are relatively high and should also be prioritized. An analysis of the M_i and N_i values reveals that optimizing these indicators can directly improve the user experience of the charging station while also exerting a significant synergistic effect on other related indicators, thereby further enhancing the overall performance of the system.

According to comprehensive weighted value sorting, the indicators B _l1 (consistency), B ₁₂ (information hierarchy), B _l3 (product attribute), B ₂₂ (constraints), B ₂₃ (cognitive transfer), B ₃₂ (service facilities), B ₄₁ (segregated circulation), and B ₅₁ (notification scope) should be prioritized for implementation. This prioritization aims to enhance the overall performance of the system and improve user experience.

5 Discussion

This study demonstrates that users possess elevated expectations regarding the recognizability, ease of use, and anxiety relief associated with charging services. When users perceive that a charging service provides clear informational guidance, a straightforward operational process, and an effective mechanism for waiting and soothing, they are more inclined to cultivate a positive attitude toward the charging experience. Furthermore, when users hold favorable expectations concerning the convenience, reliability, and comfort of the charging process, they are likely to develop a more positive attitude toward its use, thereby increasing their willingness to continue utilizing the service (Zeng et al., 2024; Suraj et al., 2025).

Service-oriented auxiliary facilities and the segregation of pedestrians and vehicles have emerged as significant factors influencing the utilization of charging stations. Public charging stations must consider operational efficiency, user-friendliness, and the enhancement of the user experience. Given that electric vehicles typically require 30 min or longer to charge, drivers often experience anxiety during this waiting period (Ma et al., 2025). To address this concern, a vertical space separation strategy is implemented to effectively distinguish between pedestrian and vehicle areas, as illustrated in Figure 4. The non-charging functional zone is positioned above the charging area. This three-dimensional layout ensures a safe distance between users and charging equipment, thereby creating an environment conducive to both physical and mental rejuvenation.

Research indicates that minimizing users’ cognitive load and decision-making costs within charging service scenarios can effectively foster their understanding and trust in the system (He et al., 2022). The alignment between visual and operational logic aids users in developing a coherent cognitive model, thereby decreasing the costs associated with learning and recognition. The structured presentation of information and feedback effectively directs users’ attention and mitigates anxiety stemming from ambiguous charging statuses. Furthermore, the clear articulation of product attributes significantly influences users’ comprehension of the functions and the perceived credibility of the service. These findings suggest that systematically enhancing recognition performance in the design of charging services may align with user expectations and improve service quality.

Constraints guide and standardize user behavior through explicit physical characteristics, effectively reducing the risk of misoperation while enhancing the safety and reliability of the interaction process. Cognitive transfer highlights the alignment between product forms, interface elements, and users’ existing mental models; effective mapping can significantly lower learning costs and facilitate intuitive understanding of operating logic. This study reveals that both factors collaboratively simplify the interaction flow between users and charging facilities, which is a critical determinant of users’ perceived ease of use and operational confidence.

Research indicates that the scope and accuracy of information dissemination are directly linked to users’ efficiency and confidence during the initial planning phase of charging. An optimal information system should deliver highly relevant and meticulously curated content at appropriate times, thereby facilitating informed decision-making. By implementing an intelligent information recommendation mechanism and effectively filtering out irrelevant or redundant data, the cognitive load on users can be substantially diminished, enhancing their perceptual effectiveness and decision-making satisfaction, while simultaneously increasing the overall value of the service.

The research indicates that the scope and accuracy of information push are directly correlated with the efficiency and trustworthiness of users during the initial charging phase (Yang et al., 2018). An optimal information system should deliver highly relevant and meticulously curated content at appropriate moments, thereby facilitating informed decision-making. By implementing an intelligent information recommendation mechanism and effectively filtering out irrelevant or redundant data, the cognitive load on users can be substantially diminished, enhancing their perceptual effectiveness and decision-making satisfaction, while simultaneously increasing the overall value of the service.

6 Conclusion

6.1 Research findings and contributions

This paper examines the specific challenges present in the current charging processes for EVs. Utilizing the user journey mapping, we identify user demands. To establish a comprehensive weight ranking table of these demands, we use the AHP-DEMATEL hybrid decision-making method, which allows us to ascertain the importance and interrelationships among each demand, thereby achieving the intended research objectives.

This article enhances the existing body of knowledge in several significant ways. First, our research investigates the behaviors, psychology, and needs of EV users throughout the entire charging process. Second, by integrating the AHP and the DEMATEL methods, we address the limitations associated with relying on a single method, thereby rendering the calculation results more robust. This approach not only allows for the determination of the weights of each index but also facilitates the identification of indicators requiring immediate improvement. Given the extensive volume of research data and the emphasis on objectivity, the comprehensive application of AHP and DEMATEL for index weighting enhances the scientific rigor and objectivity of the weight distribution. Thirdly, this model incorporates five core dimensions—recognizability, usability, trustworthiness, reliability, perceptibility ofInformation—as the criterion layer, along with 18 indicators forming the sub-criterion layer (Figure 3). This framework yields a comprehensive weight reference for the design of charging stations, thereby offering a quantifiable basis for decision-making related to the layout planning and enhancement of service capabilities at these stations.

6.2 Research limitations and future directions

This study offers a preliminary exploration of user experience; however, the advancement of EVs remains limited by several factors, including charging costs, battery range, and charging speed, which were not addressed in this research. Furthermore, the findings require validation through the assessment of current or future charging behaviors to facilitate ongoing testing and refinement of the design scheme.

Climatic conditions, the density of charging infrastructure and usage habits in northern China may differ from those in other regions. Consequently, the weighting results of this study may reflect the specific needs of these areas and may not be applicable to southern regions, where the charging infrastructure is more developed. Future studies should expand the geographical scope and conduct comparative analyses across regions with differing levels of infrastructure to assess the stability of the methodological model. The conclusions of this study are directly relevant to the design of public charging services in northern China. Researchers in other regions can replicate the methodology of this study and recalibrate it using local user data. Furthermore, although the expert interviews and observed user behavior primarily originate from the northern regions, the initial online questionnaire also included participants from other areas, thus some core indicators demonstrate a degree of universality. Additionally, the design and planning of charging stations will incorporate the perspectives of operators to optimize economic benefits.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Rui Esteves Araújo, University of Porto, Portugal

Reviewed by: Francesca Bruno, University of Salerno, Italy

Muhammad Ishfaq, Lanzhou University of Technology, China

Pedro Pereira, New University of Lisboa, Portugal