An Integrated KANO–AHP–DEMATEL–VIKOR Framework for Sustainable Design Decision Evaluation of Museum Cultural and Creative Products

Abstract

Museums widely regard cultural and creative products as both a major revenue stream and a means of revitalizing in-house cultural resources. However, traditional decision-making systems for Museum Cultural Creative Products (MCCP) design largely depend on subjective judgments, leading to inefficiency, resource waste, and weak market performance. To address these challenges and support sustainable design decision-making, this study proposes an integrated “KANO–AHP–DEMATEL–VIKOR” framework that combines qualitative and quantitative methods. First, consumer requirements are elicited through questionnaire-based interviews and literature review; the KANO model identifies key user needs, AHP determines their relative weights, and DEMATEL analyzes causal relationships among criteria. By integrating these results, the VIKOR method evaluates and ranks alternative designs, forming a comprehensive multi-criteria optimization process. To validate the framework, an empirical case of the Palace Museum’s refrigerator magnets is conducted, comparing computed rankings with actual sales data to verify predictive validity. The small discrepancy between the two indicates that the model effectively forecasts consumer acceptance across design alternatives. The major innovation of this research lies in its cross-method integration that bridges user perception analysis with quantitative sustainability evaluation, offering a replicable tool for early-stage decision-making of museum creative products. Accordingly, the framework enhances design efficiency, reduces evaluation subjectivity, and contributes to the economic and environmental sustainability of Museum Cultural Creative Products.

1. Introduction

Museum Cultural Creative Products (MCCP) generally refer to derivative goods developed by museums that combine creative design with cultural connotations. The term has only recently been formalized in Chinese and international scholarship; related expressions include “museum cultural and creative products,” “museum cultural products,” and “museum derivatives.” In this paper, we use the umbrella term Museum Cultural Creative Products (MCCP) [1]. As early as the mid-twentieth century, museum stores in the West had begun selling souvenirs, and in recent years the range of museum creative products has expanded markedly [2]. Unlike conventional tourist souvenirs, MCCP seek to embed museum collections and cultural elements into contemporary product design, thereby satisfying visitors’ commemorative and consumption needs while disseminating cultural meaning [3].

Prior research indicates that MCCP must strike a balance between cultural transmission and market responsiveness: on the one hand, they should embody the cultural depth and distinctive narratives of museum collections so that consumers perceive their cultural value; on the other hand, they must align with modern consumers’ aesthetic preferences and functional needs to enhance practicality and appeal [4]. For example, Chen proposes that the development of museum creative products should adhere to principles such as cultural relevance, aesthetic quality, practicality, playfulness, uniqueness, serialization, and diversity, to fully manifest their cultural and aesthetic value [5]. Research on the Palace Museum’s cultural and creative products similarly reports that consumers place particular emphasis on the cultural symbolism carried by the product and on novelty and distinctiveness in design [6]. Moreover, design aesthetics significantly influence purchase intention—an empirical study focusing on Palace Museum products finds that a high level of aesthetic quality meaningfully increases consumers’ propensity to buy [7]. In addition, the experiential value delivered by products has been receiving growing attention. Kotler and colleagues further argue that MCCP ought to offer social, recreational, and participatory experiences [8]. Accordingly, consumers also value the experiential benefits of creative products (e.g., emotional resonance and identity signaling). With the growth of the global creative industries, museums have increasingly leveraged product innovation to expand cultural influence and generate revenue. Integrating museums with the cultural and creative industries has been found to stimulate innovation and diversify income streams, thereby advancing the sustainable development of cultural heritage [9]. However, MCCPs still suffer from insufficient user appeal and a disconnect between cultural elements and contemporary needs, making it difficult to sustain consumers’ purchase intention and cultural resonance and resulting in direct economic and cultural losses for museums.

Against this backdrop, an increasing body of research has introduced consumer-behavior theories into MCCP development to enhance the specificity of product design and user satisfaction in product design. For instance, Chiou and Wang, using a genetic-algorithm model, classify museum consumers’ purchase motives into three “genes” (rational, emotional, etc.) and find that consumers prioritize cultural characteristics and connotations, followed by practicality/fun and reasonable pricing [10]. Ye and Huang likewise emphasize the importance of addressing consumers’ emotional appeals; because preferences vary across segments, design should be tailored to specific target groups [11]. User needs and satisfaction have thus become key guides for optimizing MCCP design. Satisfaction not only shapes purchase and word-of-mouth behaviors but also underpins the sustainable growth of museum creative businesses [12]. Recent studies in the MCCP field corroborate this point: Li et al. show empirically that consumer satisfaction is a critical driver of MCCP sustainability and can significantly affect the long-term success of AI-generated creative designs. In tourism management, Belenioti et al. similarly report that enhancing perceived value and experiential fulfillment in souvenir consumption strengthens purchase intention and loyalty [13].

However, systematic frameworks for the evaluation and optimization of MCCP design remain scarce. Most studies either emphasize qualitative analyses of consumer preferences or apply a single method to assess design alternatives, making it difficult to integrate user satisfaction with multidimensional indicators of sustainable design. Although scholars in manufacturing [14] and services [15] have explored combining customer-need analysis with multi-criteria decision-making (MCDM)—for example, Singh and Sarkar integrate fuzzy Delphi with DEMATEL for sustainable automotive product development [16], demonstrating the feasibility of such integrated approaches—related work in the MCCP domain is still limited. Existing studies often focus on isolated stages (e.g., need identification or weight/ranking alone) and lack an end-to-end methodology that connects need attributes, weight determination, relational structure among criteria, and alternative selection [17]. In addition, recent research in large group decision-making (LGDM) and cloud model-based decision systems has made notable progress in improving the handling of linguistic uncertainty and heterogeneous evaluation information. Jiang et al. proposed a rough integrated asymmetric cloud model under a multi-granularity linguistic environment to better capture decision-makers’ varying trust levels and preferences within large-scale groups, enhancing robustness and interpretability in uncertain contexts [18]. Similarly, Zhu et al. developed a cloud model-based multi-stage multi-attribute decision-making method under probabilistic interval-valued hesitant fuzzy environments, integrating stage and attribute weight optimization to strengthen decision stability and accuracy [19]. These approaches provide methodological insights that inform the present study’s attempt to construct a more comprehensive, integrated decision framework for MCCP evaluation. As Li et al.’s review notes, while research on perceived value and satisfaction spans multiple industries—hospitality [20], luxury [21], tourism products [22], and wetland parks [23]. However, rigorous, in-depth empirical work specific to MCCP is still insufficient. It is therefore necessary to fill this research gap.

To bridge this gap, it is essential to develop a methodological framework that can comprehensively capture both the qualitative and quantitative dimensions of user perception. Existing evaluation studies on MCCP often rely on a single decision-making technique, such as AHP, TOPSIS, or DEMATEL, which tends to explain only partial aspects of design performance or user satisfaction. However, decision-making for MCCP involves complex interactions between functional, emotional, and cultural attributes. Therefore, this research integrates four complementary methods—KANO, AHP, DEMATEL, and VIKOR—into a unified framework. Specifically, KANO identifies and categorizes user requirements; AHP quantifies their relative importance; DEMATEL uncovers causal interrelations among factors; and VIKOR synthesizes these outcomes to rank design alternatives through a compromise solution. This integration provides a holistic, systematic, and transparent decision process that bridges user satisfaction analysis with sustainable product optimization, addressing both theoretical and practical limitations in existing MCCP research.

Building upon this methodological foundation, the present study proposes a KANO–AHP–DEMATEL–VIKOR–based decision-making framework for MCCP design, focusing on the following research questions:

RQ1: How can user requirements for the cultural creative products of a specific museum be identified, and how do these requirements differentially affect user satisfaction?

RQ2: How can the relative influence of different user requirements on MCCP user satisfaction be determined, and what interrelationships exist among these requirements?

RQ3: How can multiple user requirements be jointly balanced to prioritize MCCP design alternatives and identify the option most suitable for production and development?

In line with the above, the research objectives are as follows:

(1)

Using the Palace Museum as a case, to construct a user-requirement–oriented decision-making framework for creative products that spans functional, aesthetic, cultural, and related dimensions;

(2)

To propose an integrated KANO–AHP–DEMATEL–VIKOR evaluation method that quantitatively analyzes the importance and interdependence of user requirements and identifies their weights;

(3)

To validate the effectiveness of the method with museum creative products currently on the market, thereby deriving design-optimization recommendations for MCCP.

2. Methods

2.1. Theoretical Model

2.1.1. KANO Model

Proposed by Professor Noriaki Kanou of the Tokyo Institute of Technology in 1984, the KANO model is a qualitative method for effectively classifying functions or services [24]. It groups customer requirements into five “quality attributes” according to the nonlinear effects of product attributes on user satisfaction: must-be quality, one-dimensional (performance) quality, attractive quality, indifferent quality, and reverse quality [25]. Through these five attributes, the KANO model can effectively distinguish categories of product requirements. Internationally, it has been widely applied in product planning and service design to improve customer satisfaction [26]. In the cultural and creative domain, the KANO model likewise proves useful: Wang et al. used it to analyze visitor preferences for the Fang Zhimin Memorial Hall’s creative products, identifying perceptual needs that should be prioritized in the design of “red culture” products [27]. Xu and Song applied the KANO model to classify consumer requirements for sustainable fashion apparel and found that attributes such as wearing comfort, eco-friendly dyeing, and safety fall under high importance “one-dimensional” quality and should be prioritized [28]. These findings suggest that the KANO model can comprehensively reveal categories of user requirements and provide customer-centric directions for product improvement.

Applying the KANO model requires distinguishing consumer-requirement attributes into six categories: Attractive (A), One-dimensional (O), Must-be (M), Indifferent (I), Reverse (R), and Questionable (Q) [29]. This classification framework offers a systematic understanding of how different dimensions of consumer needs affect user satisfaction, enabling designers to differentiate which requirements should guide distinct design emphases.

In this study, the KANO model is employed to classify consumer requirements for MCCP. The meanings of each category are summarized in

Table 1. KANO attribute classification.

Attribute	Abbr.	Definition
Attractive Quality	A	Refers to factors exceeding user expectations. If unmet, they do not affect satisfaction; if well-implemented, they significantly enhance satisfaction.
One-dimensional Quality	O	Refers to expected factors. If unmet, satisfaction decreases; if fulfilled, satisfaction increases proportionally.
Must-be Quality	M	Basic requirements users take for granted. If unmet, they cause severe dissatisfaction; if fulfilled, they do not increase satisfaction.
Indifferent Quality	I	Factors with neutral impact on satisfaction regardless of implementation.
Reverse Quality	R	Factors that generate negative perceptions. Their implementation reduces satisfaction.
Questionable Result	Q	Indicates misinterpreted questions or contradictory responses from participants.

. The calculation formulas for each category are shown in

Table 2. Calculation method of KANO attribute.

Positive Question	Negative Question
	Must-Be	One-Dimensional	Neutral	Live with	Unacceptable
Must-be	Q	A	A	A	O
One-dimensional	R	I	I	I	M
Neutral	R	I	I	I	M
Live with	R	I	I	I	M
Unacceptable	R	R	R	R	Q

and Figure 1.

2.1.2. Analytic Hierarchy Process (AHP)

The Analytic Hierarchy Process (AHP), proposed by Saaty, is a multi-criteria decision-making method used to determine the relative weights of evaluation criteria via pairwise comparisons [30]. AHP decomposes a decision problem into a hierarchical structure comprising the goal, criteria, sub-criteria, and so on. Decision-makers assign pairwise comparison judgments to elements at the same level, construct judgment matrices, and compute eigenvectors to obtain weights. AHP assumes that decisions involve multiple criteria of differing importance and employs a consistency test to ensure the reliability of judgments. Owing to its clear logic and operational simplicity, AHP has been widely applied in design, engineering, and environmental management [31,32,33], and is regarded as a reliable tool for complex decision tasks. In the cultural and creative product domain, AHP helps assess the importance of different design factors. For example, Zhan et al. integrated KANO and AHP to improve home-renovation design decisions, using AHP-based weighting to identify the optimal design alternative [34]. Overall, AHP translates subjective judgments in the evaluation process into quantitative weights, laying the groundwork for comprehensive multi-criteria assessment.

The application of AHP proceeds as follows:

Step 1: Build the AHP hierarchy. Based on the characteristics and categories of MCCP consumer requirements obtained from the KANO method, define the Goal layer (G), the Criteria layer (C), and the Alternatives layer(A).

Step 2: Separately construct the judgment matrices for the Goal layer (G) and the Criteria layer (C). Using Saaty’s 1–9 scale, perform pairwise comparisons to form the matrices; the procedure is given in Equation (1), and the nine-point scale is summarized in

Table 3. Scale table of nine-level judgment matrix.

Value	Definition
1	Equal importance: Two factors contribute equally to the objective
3	Moderate importance: Slightly favors one factor over another
5	Strong importance: Strongly favors one factor over another
7	Very strong importance: Dominantly favors one factor over another
9	Extreme importance: The highest possible affirmation of favoring one factor
2, 4, 6, 8	Intermediate judgments between adjacent scales
Countdown	Reciprocals (1/n) represent inverse relationships.

.

$$K=\left(a_{ij}\right)n\times n=\left[\begin{array}{cccc}{a}_{11}& a_{12}& \dots & a_{1n}\\ a_{21}& a_{22}& \dots & a_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ a_{n1}& a_{n2}& \dots & a_{nn}\end{array}\right]$$

(1)

In Equation (1), $a_{ij}$ denotes the result of the importance comparison between consumer requirements $i$ and $j$ within the same group, with $a_{ij}>0$. Here, $n$ is the number of subsets (i.e., items being compared) included in the judgment matrix $K$.

Step 3: Compute hierarchical weights. Calculate the weights and rank the requirements by importance. For the MCCP consumer-requirement evaluation hierarchy, first conduct pairwise comparisons among all requirement elements and judge their relative importance to obtain the multiplicative product $M_i$; the computation is given in Equation (1).

$$M_i={\displaystyle \prod }_{j=1}^{\mathrm{n}}{a}_{ij} (i,j=1,2,\cdots n)$$

(2)

Based on ${\mathrm{M}}_{\mathrm{i}}$, the weights ($w_i$) for Alternatives layer (A) (C) and the Alternatives layer (A) are derived as follows:

$$W_i=\sqrt[n]{M_i} (i=1,2,\cdots n)$$

(3)

Normalize the results $W_i$ obtained in the previous step to obtain the weights for the Criteria layer (C) and each Alternatives layer (A)$W$:

$$W={\displaystyle \frac{W_i}{{{\displaystyle \sum }}_{i=1}^n{W}_i}} (i,j=1,2,\cdots n)$$

(4)

Step 4: During the pairwise comparison of importance, to ensure consistency among evaluators and to determine whether the constructed judgment matrix contains logical errors, a consistency test is required [35]. First, obtain the matrix’s maximum eigenvalue, ${\lambda }_{\mathrm{m}\mathrm{a}\mathrm{x}}$:

$${\lambda }_{\max }={\displaystyle \sum }_{i=1}^n{\displaystyle \frac{{{\displaystyle \sum }}_{i=1}^n{a}_{ij}{W}_i}{n{W}_i}} (i,j=1,2,\cdots n)$$

(5)

where $w_i$ is the weight of the $i$-th element and $n$ is the order of the matrix.

Compute the consistency index ($C_I$) of each judgment matrix:

$$C_{\mathrm{I}}={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$$

(6)

Finally, obtain the consistency ratio ($C_R$), as in Equation (7):

$$C_R={\displaystyle \frac{C_I}{R_I}}$$

(7)

Here, $R_I$ is the random consistency index (see

Table 4. Random Consistency Index ($R_I$) values.

n	3	4	5	6	7	8	9	10
RI Value	0.52	0.89	1.12	1.26	1.36	1.41	1.46	1.49

). If $C_R<0.10$, the consistency test is passed.

2.1.3. The DEMATEL Method

The Decision-Making Trial and Evaluation Laboratory (DEMATEL) method was developed by the Battelle Geneva Research Center in 1972 to analyze causal influence relationships among factors in complex systems [36]. DEMATEL constructs an influence matrix via expert scoring and, using graph theory and matrix operations, quantifies the intensity of interactions among factors to derive indices such as cause degree, effect degree, and centrality, thereby generating a relational diagram that visualizes the causal network [37]. Because product-design evaluation involves interactive effects among multiple criteria, introducing DEMATEL helps uncover the latent connections among different user requirements or design elements [38]. Prior studies have begun to apply DEMATEL to design decisions: for example, Lin et al. used DEMATEL to identify factors shaping museum consumers’ purchase intention, finding that “culture” and “experience” are the two core criteria influencing consumer decisions. Karasan et al. integrated DEMATEL with QFD and AHP to analyze links between customer requirements and technical characteristics in automotive seat design, developing a design-optimization approach that accounts for factor interactions [39]. These studies indicate that DEMATEL effectively reveals the causal mechanisms among requirement indicators, compensating for the limitations of traditional independent-weight analyses. For enhancing the systematicity and specificity of museum creative-product design, DEMATEL offers an intuitive way to determine which user requirements drive others, thereby indicating the primary directions for design optimization. In this study, the DEMATEL process was implemented as follows: (1) experts were asked to rate the direct influence between each pair of user-requirement factors on a 0–4 scale, forming the initial direct-influence matrix; (2) the matrix was normalized to derive the total influence matrix; (3) the cause (D−R), effect (R−D), and prominence (D+R) values were calculated to determine each factor’s role and intensity within the causal network; and (4) a causal diagram was plotted to visualize the directional relationships and identify the dominant “driving” and “driven” attributes. This systematic computation enabled the identification of key user requirements that exert the greatest influence on others, providing a quantitative foundation for subsequent design prioritization.

The core of DEMATEL is to compare design elements against one another through expert ratings to obtain the direct-relation matrix and to analyze interrelationships among indicators. The specific computational steps are as follows:

(1)

First, construct the normalized direct-relation matrix $X$, where $b_{ij}$ denotes the degree of influence of requirement $i$ on requirement $j$.

$$S=\max {\displaystyle \sum }_{j=1}^n{b}_{ij} (1\le i\le n)$$

(8)

In the equation, $B_{ij}$ denotes the direct influence strength of factor $i$ on factor $j$; $n$ is the total number of factors; and $S$ is the maximum of all row sums $\sum b_{ij}$, used as the normalization baseline.

$$X={\displaystyle \frac{B}{S}}$$

(9)

In the equation, $B={\left(b_{ij}\right)}_{n\times n}$ is the original direct-relation matrix.

(2)

Construct the total-relation matrix ($T$), where ($I$) denotes the identity matrix:

$$T=\mathrm{X}{\left(I-X\right)}^{-1}$$

(10)

(3)

Obtain the influence degree, the influenced degree, the centrality (prominence), and the cause degree among the design factors; then, in accordance with the interpretation of centrality, normalize it to derive the weight of each element.

$$R={\displaystyle \sum }_{i=1}^n{t}_{ij},i,j=1,2,3,\dots ,n$$

(11)

$$C={\displaystyle \sum }_{j=1}^n{t}_{ij},i,j=1,2,3,\dots ,n$$

(12)

$$D_i=C_i\times R_i$$

(13)

$$W_2={\displaystyle \frac{D_i}{{{\displaystyle \sum }}_{i=1}^n{D}_i}}$$

(14)

In the equation, $W_2$ denotes the normalized weight vector of the factors computed via DEMATEL. Let $D_i$ be the influence-degree vector, where $R$ is the total influence of element $i$ on all other elements, and let $C$ be the influenced-degree vector, where $C$ is the total influence received by element $i$ from all others. Centrality is defined as the sum of the influence and influenced degrees, $R+C$; causality is defined as their difference, $R-C$. Finally, the Influence Relation Map (IRM) can be created by plotting the dataset in the $(R+C, R-C)$ plane.

2.1.4. AHP–DEMATEL Integrated Weight Calculation

AHP determines the initial weights of the criteria by constructing a hierarchical structure, whereas DEMATEL analyzes interrelationships among the criteria to derive relational weights. Combining the two enables a comprehensive consideration of the importance of different consumer-requirement indicators, yielding a more objective weight allocation and mitigating the pronounced subjectivity of AHP [40]. This integrated approach helps objectively assess the relative importance of each indicator in shaping consumers’ purchase intentions. In the AHP–DEMATEL integration, the initial weights obtained from the two methods are first multiplied and then normalized.

$${W^0}_{c,i}=W_{1,i}\times W_{2,i}$$

(15)

$$W_{c,i}={\displaystyle \frac{{W^0}_{c,i}}{{{\displaystyle \sum }}_{j=1}^n{W^0}_{c,j}}}$$

(16)

where ${W^0}_{c,i}$ is the initial combined weight, $W_{c,i}$ is the final comprehensive weight after normalization; in the formula, i represents each consumer demand element obtained based on the KANO method, and j represents other consumer demand elements except i.

This integrated mechanism also provides scalability for larger datasets or group-based evaluations. When the number of criteria or participants increases, hierarchical decomposition and group aggregation techniques can be applied to maintain computational efficiency. Additionally, the AHP–DEMATEL structure can be expanded with fuzzy or cloud-model extensions to accommodate uncertain or multi-granularity linguistic data, thereby enhancing its applicability to large-scale decision-making scenarios.

2.1.5. The VIKOR Method

The VIKOR method, proposed by Opricovic and colleagues, is a multi-attribute optimization and compromise-ranking approach. It simultaneously pursues the maximization of group utility and the minimization of individual regret, thereby offering notable advantages in ranking stability, objectivity, and credibility [41]. In the context of cultural and creative product design, VIKOR can be introduced to conduct a comprehensive evaluation of alternative designs or improvement strategies. By considering multiple criteria—such as the fulfillment of user requirements, design cost, and environmental impact—VIKOR helps identify the option that is comparatively well-balanced across dimensions and best meets overall requirements. For example, Yazo-Cabuya et al. demonstrate that VIKOR is effective for prioritizing organizational sustainability risks, yielding rankings comparable to those of AHP and ANP, with sensitivity analyses confirming result robustness [42]. Thus, VIKOR has practical value for complex decision optimization and provides a scientific basis for selecting MCCP design alternatives. The computation process involves four key steps: (1) constructing the decision matrix of alternatives and criteria based on expert or user evaluations; (2) determining the best (f) and worst (f–) values for each criterion; (3) calculating the group utility (S) and individual regret (R) measures, followed by the compromise index (Q) that integrates S and R through the parameter v; and (4) ranking all alternatives according to their Q values, where a smaller Q indicates a more preferable solution. In this study, the parameter v was set to 0.5 to balance the decision-makers’ preference between collective satisfaction and individual regret, while sensitivity analysis was later conducted to examine the robustness of the results under different v settings. The main steps are as follows:

In the evaluation system, suppose there are $\mathrm{m}$ design alternatives and $\mathrm{n}$ evaluation criteria; the $n$ criteria of the $m$-th alternative form a set. Accordingly, the original evaluation matrix ${\mathit{X}}^{\ast }$ is constructed:

$${\mathit{X}}^{\ast }=\left[\begin{array}{cccc}{x}_{11}^{\ast }& x_{12}^{\ast }& \dots & x_{1n}^{\ast }\\ x_{21}^{\ast }& x_{22}^{\ast }& \dots & x_{2n}^{\ast }\\ ⋮& ⋮& ⋮& ⋮\\ x_{m1}^{\ast }& x_{m2}^{\ast }& \dots & x_{mn}^{\ast }\end{array}\right]$$

(17)

(1)

Normalize the original matrix using the sum-of-squares method to obtain the normalized decision matrix X:

$$X=\left[\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1n}\\ x_{21}& x_{22}& \dots & x_{2n}\\ ⋮& ⋮& ⋮& ⋮\\ x_{m1}& x_{m2}& \dots & x_{mn}\end{array}\right]$$

(18)

(2)

Compute the positive and negative ideal solutions, ($f_n^{+}$) and ($f_n^{-}$):

$$\begin{array}{l}{f}_n^{+}=\max \left\{f_{nm}\right\}\hfill \\ f_n^{-}=\min \left\{f_{nm}\right\}\hfill \end{array}$$

(19)

In the formula, $f_{nm}$ denotes the group decision value of all respondents for the (m)-th alternative on the (n)-th evaluation criterion.

(3)

Calculate the group-utility value (S) and the individual-regret value (R), which reflect the overall performance of each alternative relative to the others [43].

$$S_m={\displaystyle \sum }_{n=1}^n{w}_n{\displaystyle \frac{f_n^{+}-f_{nm}}{f_n^{+}-f_n^{-}}} R_m=\underset{1\le n\le J}{\max }{w}_n{\displaystyle \frac{f_n^{\ast }-x_{nm}}{f_n^{\ast }-f_n^{-}}}.$$

(20)

Here, ($S_m$) denotes the group-utility value of the (m)-th alternative. ($R_m$), termed “individual regret,” reflects—for alternative (m)—the maximum distance from the ideal solution across all criteria (i.e., its relatively worst-performing criterion).

(4)

By jointly considering the group-utility value (S) and the individual-regret value (R), compute the final compromise solution index (Q). A smaller (Q) indicates a shorter distance to the optimal solution and thus a better alternative:

$$Q_m=\lambda {\displaystyle \frac{S_m-S^{+}}{S^{-}-S^{+}}}+(1-\lambda ){\displaystyle \frac{R_m-R^{+}}{R^{-}-R^{+}}}$$

(21)

In the formula, $Q_m$ denotes the compromise solution for design alternative $m$. The parameter $\lambda$ represents risk preference; its default value is 0.5. If $\lambda >0.5$, it indicates risk seeking, whereas $\lambda <0.5$ indicates risk aversion [44].

$$S^{+}=\max \left(S_i\right);S^{-}=\min \left(S_i\right);R^{+}=\max \left(R_i\right);R^{-}=\min \left(R_i\right)$$

2.2. Research Framework

This study aims to construct a consumer-requirement-based integrated evaluation model to support sustainability-oriented design decision-making for museum cultural and creative products (MCCP). To this end, we integrate multiple qualitative and quantitative decision techniques to systematically and rigorously identify consumer needs and to establish a precise weighting system for those needs, thereby enabling sustainability evaluation of product design alternatives.

Refrigerator magnets constitute an important subcategory of MCCP [45]. These small magnetic ornaments adhere to refrigerators or other metal surfaces, combining practical utility with cultural communication. They are typically made of metal, resin, or enamel with an embedded magnet. Owing to their strong communicative power, low purchase threshold, and high emotional attachment, they have become a widely popular vehicle for cultural dissemination; virtually all museums in China that operate MCCP shops design and sell refrigerator magnets [46]. At the same time, because such products are numerous and updated frequently, issues related to resource consumption, market acceptance, and long-term environmental impact are especially salient, creating an urgent need for a precise, scientific evaluation method to guide sustainable optimization of the design process.

Leveraging nearly six centuries of imperial heritage and a collection exceeding 1.86 million objects, the Palace Museum has created China’s largest cultural-creative IP. Nevertheless, its product development process also exhibits problems such as resource waste, product homogenization, and unstable market performance [47]. Accordingly, this study selects the Palace Museum’s creative refrigerator magnets as a case for empirical analysis, a choice with strong theoretical and practical representativeness.

This study first identifies fourteen user-requirement elements for the Palace Museum’s refrigerator magnets through a literature review, analysis of online consumer reviews, and field interviews; based on the KANO model, we then screen ten key factors that affect consumer satisfaction and group them into three categories—social, emotional, and functional. Next, we use AHP and DEMATEL to calculate, respectively, the relative importance of requirements and their interrelationships, yielding integrated weights for each requirement indicator. Finally, we apply VIKOR to rank ten refrigerator-magnet designs currently sold in the Palace Museum’s Taobao store and compare the ranking with actual sales data to validate the effectiveness of the evaluation model. The overall research framework is shown in Figure 2.

The integration of KANO, AHP, DEMATEL, and VIKOR in this study is grounded in the complementary nature of their theoretical foundations. KANO theory, derived from motivation-hygiene psychology, distinguishes between different types of user satisfaction drivers, providing the qualitative basis for requirement classification. AHP, based on hierarchical systems theory and pairwise comparison logic, quantifies the relative importance of these requirements. DEMATEL, rooted in graph theory and causal inference, reveals interdependencies and influence pathways among criteria, ensuring that structural relationships are incorporated rather than assumed independent. Finally, VIKOR, developed from multi-criteria compromise theory, synthesizes collective satisfaction and individual regret to determine balanced solutions. Together, these methods form a coherent theoretical chain that moves from qualitative perception identification to quantitative weighting, relational modeling, and compromise-based optimization, reflecting the full cognitive process of user-centered decision-making.

Compared with traditional decision-making approaches such as standalone AHP, TOPSIS, or fuzzy comprehensive evaluation, the proposed KANO–AHP–DEMATEL–VIKOR framework demonstrates notable methodological advantages. Conventional single-method approaches often focus on either the weighting of criteria or ranking of alternatives, lacking the ability to capture interrelationships among user requirements or reconcile trade-offs between collective satisfaction and individual regret. By contrast, the integrated framework developed in this study combines KANO’s capability to classify user needs, AHP’s hierarchical weighting mechanism, DEMATEL’s causal analysis of factor interdependence, and VIKOR’s compromise-based optimization. This holistic structure not only enhances evaluation accuracy and interpretability but also improves robustness by linking qualitative user insights with quantitative optimization. Consequently, the framework provides a more systematic and transferable model for sustainability-oriented design decision-making in museum cultural-creative product development.

3. Results

3.1. Establishing KANO-Based Design Evaluation Indicators

Between 20 and 26 December 2024, the research team used multiple channels—literature review, analysis of online user comments, in-person interviews, and on-site observation—to synthesize fourteen consumer requirements applicable to the Palace Museum’s creative refrigerator magnets. Based on their characteristics, these requirements were grouped into three dimensions: social, emotional, and functional. The social dimension comprises gift significance (X1), commemorative significance (X2), and social identity (X3); the emotional dimension includes pattern/design recognizability (X4), visual aesthetic comfort (X5), accuracy of appearance/form (X6), brand recognition (X7), conveyance of cultural content (X8), material and craftsmanship quality (X9), and conveyance of auspicious symbolism (X10); the functional dimension covers durability in use (X11), tactile comfort (X12), size compatibility (X13), and interactive functionality (X14). Explanations of each requirement, developed from the fieldwork, are provided in

Table 5. Consumer demand factors for cultural and creative refrigerator magnets from the Palace Museum.

No.	Category	Factor	Description
X1	Social	Gift-giving suitability	The refrigerator magnet is designed to clearly reflect its suitability as a gift, effectively expressing the giver’s sentiments and enhancing emotional exchanges among people.
X2		Memorability	The refrigerator magnet embodies memories of visits to the Palace Museum, integrating unique cultural elements of the Palace Museum into its design, allowing tourists to relive pleasant travel experiences.
X3		Social recognition	The design of the refrigerator magnet resonates with others, encouraging social interactions and providing the holder with a sense of recognition and belonging.
X4	Emotional	Distinctive pattern design	The pattern design of the refrigerator magnet demonstrates distinctive features and innovation rooted in Palace Museum culture, enabling strong visual identification and attracting consumer attention.
X5		Visual aesthetic harmony	The refrigerator magnet’s color matching is harmonious and visually pleasing, aligning with mainstream aesthetic preferences, thereby enhancing its visual attractiveness.
X6		Accuracy of appearance modeling	The refrigerator magnet accurately replicates elements or symbols of the Palace Museum, displaying detailed craftsmanship and reflecting high-quality standards.
X7		Palace Museum brand identifiability	The refrigerator magnet clearly displays recognizable symbols such as “Palace Museum” or “GUGONG,” helping consumers easily identify the product’s cultural origin and brand value.
X8		Cultural content communication	The refrigerator magnet effectively conveys cultural meanings and historical context of the Palace Museum, enriching consumer cultural experiences during use.
X9		Material quality and craftsmanship	The refrigerator magnet exhibits high-quality materials and sophisticated craftsmanship, elevating the overall sense of quality and artistic value.
X10		Expression of auspicious symbolism	The refrigerator magnet incorporates auspicious symbols and elements of blessing, offering consumers psychological satisfaction and conveying good wishes.
X11	Functional	Product durability	The refrigerator magnet is manufactured from durable materials that ensure long-lasting use, effectively meeting consumers’ needs for product durability.
X12		Tactile comfort	The refrigerator magnet provides a pleasant tactile experience, enhancing consumer willingness to frequently interact with and use the product.
X13		Size suitability	The refrigerator magnet is suitably sized for household appliances such as refrigerators, ensuring visibility without occupying excessive space.
X14		Interactive functionality	The refrigerator magnet features interactive or movable components, increasing consumer engagement by enhancing the product’s enjoyment and interactivity.

.

To effectively screen the core requirements that affect consumer satisfaction, the study applies the KANO model to the fourteen items above. A KANO questionnaire employing a five-point Likert scale was designed with paired questions from both positive and negative perspectives. The questionnaire was distributed through both online and offline channels: 92 responses were collected via the “Wenjuanxing” (Questionnaire Star) platform, and an additional 18 were obtained through on-site recruitment at Beijing Institute of Technology, where participants were invited to complete the survey under researcher supervision. The survey was randomly distributed to 110 consumers with prior purchasing experience of Palace Museum creative products, yielding 103 valid responses. All respondents confirmed that they had previously purchased at least one museum cultural and creative product (MCCP), ensuring data relevance. For each consumer-requirement factor, the KANO attribute was classified as Attractive (A), One-dimensional (O), Must-be (M), Indifferent (I), Reverse (R), or Questionable (Q), and the attribute with the highest frequency was adopted as that factor’s KANO category. The classification of the refrigerator-magnet requirements derived from the KANO responses is summarized in

Table 6. KANO attribute classification of each factor.

No.	Factor	A%	O%	M%	I%	R%	Q%	Result
X1	Gift-giving suitability	44.66%	0.97%	0.97%	25.24%	9.71%	18.45%	A
X2	Memorability	13.59%	3.88%	38.83%	32.04%	9.71%	1.94%	M
X3	Social recognition	46.60%	1.94%	0%	36.89%	5.83%	8.74%	A
X4	Distinctive pattern design	41.75%	1.94%	0%	35.92%	3.88%	16.5%	A
X5	Visual aesthetic harmony	9.71%	4.85%	33.01%	36.89%	7.77%	7.77%	I
X6	Accuracy of appearance modeling	16.50%	1.94%	28.16%	34.95%	9.71%	8.74%	I
X7	Palace Museum brand identifiability	44.66%	0.97%	0.97%	27.18%	14.56%	11.65%	A
X8	Cultural content communication	39.81%	0.97%	1.94%	27.18%	14.56%	15.53%	A
X9	Material quality and craftsmanship	19.42%	2.91%	29.13%	33.98%	6.8%	7.77%	I
X10	Expression of auspicious symbolism	42.72%	0.97%	0%	28.16%	13.59%	14.56%	A
X11	Product durability	20.39%	1.94%	26.21%	30.1%	7.77%	13.59%	I
X12	Tactile comfort	24.27%	1.94%	33.98%	27.18%	2.91%	9.71%	M
X13	Size suitability	21.36%	4.85%	33.01%	32.04%	4.85%	3.88%	M
X14	Interactive functionality	14.56%	1.94%	35.92%	23.3%	9.71%	14.56%	M

.

The results in Table 6 show that within the social dimension, “commemorative significance” ranks highest, ahead of “gift significance” and “social identity,” indicating that consumers primarily regard magnets as travel mementos rather than generic gifts or social symbols—a preference that aligns closely with the context of offline tourist purchases. In the emotional dimension, “pattern/design recognizability,” “Palace Museum brand recognition,” “conveyance of cultural content,” and “conveyance of auspicious symbolism” are clearly prioritized over “visual aesthetic comfort,” “accuracy of appearance/form,” and “material and craftsmanship quality.” This suggests that consumers value the cultural information and innovative symbolism embodied in the product more than form and workmanship per se; at the same time, emphasizing brand elements can further strengthen consumers’ emotional attachment to the Palace Museum IP. In the functional dimension, “tactile comfort,” “size compatibility,” and “interactive functionality” are generally seen as basic requirements, whereas “product durability” is relatively less salient, indicating that users are more concerned with immediate use-experience and playfulness.

In sum, the KANO analysis reveals the consumption motives and core pain points of Palace Museum refrigerator magnets as cultural souvenirs. Accordingly, the study retains ten key requirement indicators belonging to the Attractive (A), One-dimensional (O), and Must-be (M) categories, and excludes Indifferent (I), Reverse (R), and Questionable (Q) items, as shown in

Table 7. Results of factor screening based on KANO method.

Screening Result	KANO Attribute	Factors Retained or Eliminated
Retained	A	X1 Gift-giving suitability	X3 Social recognition	X4 Distinctive pattern design	X7 Palace Museum brand identifiability	X8 Cultural content communication	X10 Expression of auspicious symbolism
	M	X2 Memorability	X12 Tactile comfort	X13 Size suitability
Eliminated	I	X5 Visual aesthetic harmony	X6 Accuracy of appearance modeling	X9 Material quality and craftsmanship	X11 Product durability

.

3.2. AHP-Based Evaluation of the Relative Importance of Requirement Indicators

After identifying and screening the key consumer requirements, we further clarified the relative importance of each requirement within the overall design evaluation system. Building on the preceding KANO results, we constructed a three-tier hierarchical model comprising the Goal layer (G), the Criteria layer (C), and the Alternatives layer (A). As shown in Figure 3, this structure reflects the specific manifestation of consumers’ focus on cultural and creative refrigerator magnets in the three dimensions of society, emotion and function. We then designed an AHP questionnaire to quantify users’ subjective emphasis on each key requirement.

In total, 65 users with actual MCCP purchasing experience completed the survey, providing pairwise-comparison ratings for the ten screened key requirements. Participants were recruited through the same hybrid approach as in the KANO stage, combining the online “Wenjuanxing” platform with limited on-site distribution at Beijing Institute of Technology. Before data collection, respondents were screened by asking whether they had purchased or used Palace Museum creative products to ensure the validity of the AHP judgments. Using Saaty’s standard 1–9 scale, and following the AHP computation procedure detailed in Section 2.1.2, we constructed judgment matrices for the Criteria layer (C1–C3) and the Alternatives layer (A) (A1–A10), and subsequently computed the eigenvectors and weights for each level. The detailed matrices and results are reported in

Table 8. Goal layer P judgment matrix and weights.

Goal (G)	C1	C2	C3	Eigenvector Component	Priority Weight
C1	1	1/4	3	0.909	0.22554
C2	4	1	5	2.714	0.67383
C3	1/3	1/5	1	0.405	0.10062

,

Table 9. Criteria layer B1 judgment matrix and weights.

Criteria (C1)	A1	A2	A3	Eigenvector Component	Priority Weight
A1	1	1/7	3	0.754	0.14886
A2	7	1	9	3.979	0.78539
A3	1/3	1/9	1	0.333	0.06575

,

Table 10. Criteria layer B2 judgment matrix and weights.

Criteria (C2)	A4	A5	A6	A7	Eigenvector Component	Priority Weight
A4	1	5	1/3	4	1.606	0.27395
A5	1/5	1	1/7	1/3	0.312	0.05327
A6	3	7	1	6	3.350	0.57133
A7	1/4	3	1/6	1	0.595	0.10145

and

Table 11. Criteria layer B3 judgment matrix and weights.

Criteria (C3)	A8	A9	A10	Eigenvector Component	Priority Weight
A8	1	4	1/3	1.100	0.27962
A9	1/4	1	1/5	0.368	0.09362
A10	3	5	1	2.446	0.62675

. All valid responses were checked for consistency, and duplicate IP entries were excluded to ensure reliability.

To ensure the consistency and logical validity of the ratings, we applied Saaty’s consistency test, substituting the corresponding random consistency index (RI) values to calculate the consistency ratio (CR) for each level. As shown in

Table 12. Consistency test results.

Level	Maximum Eigenvalue	Consistency Index	Random Index	Consistency Ratio	Result
G	3.086	0.043	0.520	0.082	Pass
C1	3.080	0.040	0.520	0.077	Pass
C2	4.173	0.058	0.890	0.065	Pass
C3	3.086	0.043	0.520	0.082	Pass

, all judgment matrices yield (CR < 0.1), indicating good agreement among participants and high reliability of the questionnaire data.

After obtaining the relative weights at the Alternatives layer (A), we aggregated them with the upper-level Benchmark weights and normalized the results to derive the initial AHP weight vector (W₁) for subsequent integrated-weight analysis. The relevant results are presented in

Table 13. AHP initial weights and ranking.

Primary Criterion	Weight	Secondary Criterion	Weight	AHP Weight	Rank
C1	0.22554	A1	0.14886	0.03357	7
		A2	0.78539	0.17714	3
		A3	0.06575	0.01483	9
C2	0.67383	A4	0.27395	0.18460	2
		A5	0.05327	0.03589	6
		A6	0.57133	0.38498	1
		A7	0.10145	0.06836	4
C3	0.10062	A8	0.27962	0.02814	8
		A9	0.09362	0.00942	10
		A10	0.62675	0.06306	5

.

3.3. DEMATEL-Based Analysis of Interrelated Influences Among Requirement Indicators

Building on the identified key user requirements, we further investigate their interrelationships and causal roles within the system by employing DEMATEL (Decision-Making Trial and Evaluation Laboratory) to model structural linkages among requirement elements and adjust weights accordingly. This approach identifies each element’s pathways and influence weights within the overall network, enabling a more comprehensive understanding of how consumer preferences are formed and improving the accuracy and explanatory power of the evaluation method.

To construct the DEMATEL influence matrix, the research team designed a comparative questionnaire covering ten requirement factors (A1–A10) and invited ten experts and user representatives with in-depth knowledge of museum-culture–based creative design to provide ratings. Experts were selected through purposive sampling to ensure professional diversity and representativeness. Data were collected between 5 and 10 January 2025, via email and in-person sessions held at the School of Design and Arts, Beijing Institute of Technology. The sample included master’s and doctoral students in design as well as university professors, ensuring a balance between professional expertise and user perspectives(

Table 14. Demographic of the Expert Panel of the DEMATEL method.

Expert ID	Gender	Age	Identity	Professional Background
E1	Female	48	Professor	Industrial Design
E2	Female	39	Associate Professor	Industrial Design
E3	Male	38	Associate Professor	Cultural Heritage
E4	Female	35	Associate Professor	Environmental Design
E5	Male	28	PhD Student	Industrial Design
E6	Female	30	PhD Student	Industrial Design
E7	Female	26	PhD Student	Industrial Design
E8	Male	25	Master’s Student	Industrial Design
E9	Male	23	Master’s Student	Environmental Design
E10	Female	23	Master’s Student	Environmental Design

). Each expert provided independent ratings to avoid mutual influence, and all responses were aggregated using arithmetic means to form the influence matrix.

A 0–4 Likert-type scale was adopted, corresponding to five levels— “no influence,” “slight influence,” “influence,” “strong influence,” and “very strong influence.” Respondents judged the direct influence intensity between any two requirements. The resulting (10 × 10) raw rating matrix was formed from the pairwise arithmetic means, reflecting the overall pattern of cognition.

In this study, the arithmetic mean of the ten respondents’ scores for each pairwise relation was taken as the value of that relation. Following the computational procedure in Section 2.1.3, we derived the Normalization direct impact matrix (X) (

Table 15. Normalization direct impact matrix X.

Item	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10
A1	0.000	0.241	0.172	0.034	0.155	0.000	0.276	0.034	0.069	0.017
A2	0.069	0.000	0.103	0.034	0.190	0.414	0.000	0.138	0.034	0.017
A3	0.000	0.000	0.000	0.121	0.241	0.103	0.086	0.017	0.000	0.000
A4	0.000	0.069	0.000	0.000	0.345	0.310	0.052	0.034	0.069	0.069
A5	0.000	0.052	0.121	0.034	0.000	0.276	0.052	0.052	0.052	0.086
A6	0.000	0.000	0.069	0.155	0.017	0.000	0.259	0.069	0.017	0.052
A7	0.103	0.000	0.086	0.017	0.086	0.069	0.000	0.000	0.017	0.000
A8	0.034	0.086	0.086	0.000	0.052	0.103	0.052	0.000	0.345	0.241
A9	0.000	0.052	0.000	0.052	0.017	0.052	0.017	0.207	0.000	0.121
A10	0.034	0.069	0.017	0.172	0.103	0.000	0.052	0.034	0.172	0.000

) and the Integrated Impact matrix (T) (

Table 16. Integrated Impact Matrix T.

Item	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10
A1	0.092	0.352	0.385	0.242	0.501	0.470	0.520	0.219	0.256	0.193
A2	0.133	0.135	0.317	0.283	0.499	0.798	0.344	0.320	0.264	0.235
A3	0.039	0.072	0.124	0.242	0.422	0.373	0.252	0.114	0.114	0.116
A4	0.064	0.182	0.203	0.227	0.601	0.702	0.341	0.216	0.264	0.261
A5	0.051	0.133	0.252	0.216	0.237	0.538	0.275	0.182	0.202	0.221
A6	0.059	0.079	0.187	0.274	0.248	0.258	0.404	0.162	0.152	0.165
A7	0.127	0.065	0.177	0.108	0.226	0.231	0.136	0.069	0.089	0.070
A8	0.104	0.226	0.256	0.238	0.356	0.469	0.299	0.230	0.563	0.442
A9	0.048	0.143	0.115	0.185	0.211	0.293	0.170	0.318	0.196	0.270
A10	0.082	0.172	0.154	0.312	0.351	0.333	0.232	0.188	0.326	0.158

), and then obtained each element’s influence degree, influenced degree, centrality, and the initial DEMATEL weights (

Table 17. Influence, Influenced, Centrality and Causality of Factors.

Item	Influence Degree	Influenced Degree	Centrality	Causality	Dematel Weight
A1	3.231	0.799	4.030	2.431	0.082
A2	3.328	1.558	4.886	1.770	0.100
A3	1.868	2.170	4.038	−0.302	0.082
A4	3.061	2.327	5.388	0.734	0.110
A5	2.307	3.652	5.960	−1.345	0.122
A6	1.989	4.466	6.455	−2.476	0.132
A7	1.297	2.973	4.270	−1.676	0.087
A8	3.181	2.018	5.199	1.162	0.106
A9	1.949	2.426	4.375	−0.476	0.089
A10	2.308	2.130	4.438	0.177	0.091

).

Through this analysis, we not only reveal the interaction mechanisms among consumer requirements for the Palace Museum’s refrigerator magnets, but also provide a basis for dynamically adjusting the integrated weights, thereby enhancing the model’s ability to capture the complexity of real-world design decisions.

3.4. Computation and Determination of AHP–DEMATEL Integrated Weights

After establishing user-requirement weight systems separately with AHP and DEMATEL, we further integrate the two sets of results to obtain more interpretable composite weights. Single method approaches to multi-criteria preferences can be one-sided: AHP emphasizes hierarchical structuring and expert preference modeling, which may overlook interactions among indicators; DEMATEL highlights inter-factor influence relationships but is relatively less precise in capturing global hierarchical weights. Accordingly, we integrate the two to balance subjective judgment with system-level causality, thereby improving the scientific rigor and robustness of weight allocation.

Specifically, following the formulas presented in Section 2.1.4, we fuse the initial AHP weights with the structural influence weights derived from DEMATEL and then normalize the results, thereby determining the FAHP–DEMATEL integrated weights for the eleven design elements, as shown in

Table 18. AHP-DEMATEL combined weights and ranking.

Item	AHP Weight	DEMATEL Weight	Integrated Weight (AHP–DEMATEL)	Rank
A1	0.034	0.082	0.025	8
A2	0.177	0.100	0.157	3
A3	0.015	0.082	0.011	9
A4	0.185	0.110	0.180	2
A5	0.036	0.122	0.039	6
A6	0.385	0.132	0.451	1
A7	0.068	0.087	0.053	4
A8	0.028	0.106	0.027	7
A9	0.009	0.089	0.007	10
A10	0.063	0.090	0.051	5

.

3.5. Comprehensive Evaluation of Design Alternatives and Empirical Validation Based on the VIKOR Method

To further verify the accuracy and practicality of the KANO–AHP–DEMATEL approach in assigning weights to consumer requirements, this study introduces the VIKOR method to conduct a multi-attribute comprehensive ranking of actual design alternatives and compares the results with product sales performance, thereby evaluating the model’s predictive validity.

After excluding newly launched items and products without sales records, we selected ten refrigerator magnets with complete sales data from the Palace Museum’s online store (https:/gugong1925.jiyoujia.com (accessed on 20 May 2025)) as sample design alternatives to ensure representativeness and data comparability. Sample images are shown in Figure 4. Based on the integrated weighting system established above and the key consumer-requirement indicators A1–A10, we then carried out systematic scoring and ranking of these ten alternatives.

Ten expert reviewers were invited, including two scholars specializing in cultural heritage, six master’s/doctoral students in design, and two consumer representatives with purchasing experience of Palace Museum refrigerator magnets. The recruitment followed the same procedure as the DEMATEL stage, with invitations issued via institutional email and follow-up validation of participant expertise. Experts were briefed on the evaluation framework and scoring criteria prior to participation to ensure consistency of understanding. The demographic and professional information of the experts is shown in

Table 19. Demographic of the Expert Panel of the VIKOR method.

Expert ID	Gender	Age	Identity	Professional Background	Affiliation
E1	Male	52	Professor	Visual Communication	University
E2	Female	36	Associate Researcher	Cultural Heritage	University
E3	Female	38	Consumer Representative	Cultural Product Buyer	Design Company
E4	Female	35	Consumer Representative	Cultural Product Buyer	Tourism Company
E5	Male	28	PhD Student	Visual Communication	University
E6	Female	30	PhD Student	Industrial Design	University
E7	Male	24	Master’s Student	Industrial Design	University
E8	Female	25	Master’s Student	Industrial Design	University
E9	Male	23	Master’s Student	Environmental Design	University
E10	Female	22	Master’s Student	Environmental Design	University

. Using a unified 10-point scale, they rated each alternative’s performance on every criterion, with higher scores indicating closer alignment with user expectations. The resulting data were used to construct the original evaluation matrix (

Table 20. Evaluation matrix of the VIKOR method.

Sample	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10
Sample 1	8.6	8.4	7.8	8.6	8.5	8.6	8.3	6.9	8	5.5
Sample 2	7.9	8.3	6.8	7.8	7.6	7.9	6.6	6	7.4	5.1
Sample 3	7.1	7.4	5.1	7.5	7.1	7.6	7.5	6	6.5	4.6
Sample 4	7.9	8	6.4	8.4	7.8	7.4	7.8	6.7	7.7	7.6
Sample 5	6.7	7.1	6.2	7.2	6.5	7.5	7	6.7	6.4	7.4
Sample 6	6.5	6.4	4.4	6.6	5.8	6.8	6.3	5.4	6.4	3.7
Sample 7	5.6	5.9	4.6	6.3	5.7	5.8	5.1	5	5.8	3.8
Sample 8	5	4.7	5.1	5.1	5	4.3	4.7	4.9	5.3	3.4
Sample 9	4.5	5.2	4.8	5.3	5.6	5	4.7	4.7	5.4	3.2
Sample 10	4	4	4.7	5.7	4.1	4.4	3.5	4.2	5	3.4

). The evaluation sessions were conducted between 10 and 15 May 2025, both online (through shared spreadsheets) and offline at Beijing Institute of Technology, and all ratings were anonymized prior to aggregation. By integrating the positive and negative ideal-solution data in

Table 21. Positive and negative ideal solutions for each consumer demand.

Item	Positive Ideal Solution	Negative Ideal Solution
A1	0.415	0.193
A2	0.396	0.189
A3	0.433	0.244
A4	0.391	0.232
A5	0.413	0.199
A6	0.406	0.203
A7	0.415	0.175
A8	0.381	0.232
A9	0.391	0.245
A10	0.479	0.202

with the relative weights of the performance indicators and following the computational steps in Section 2.1.5, we computed the (S) values, (R) values, and the final (Q) values for the ten samples.

To enhance the credibility of the analysis, the VIKOR ranking was further compared with the corresponding sales rankings of the samples in the Palace Museum’s Taobao store (

Table 22. VIKOR method ranking and actual sales ranking of samples.

Sample	S	R	Q	VIKOR Rank	Sales Rank
Sample 1	0.0243	0.0243	0.0000	1	1
Sample 2	0.1911	0.0733	0.1471	2	2
Sample 3	0.2823	0.1048	0.2329	4	3
Sample 4	0.1730	0.1257	0.1988	3	4
Sample 5	0.2891	0.1153	0.2489	5	5
Sample 6	0.4949	0.1886	0.4454	6	6
Sample 7	0.6555	0.2934	0.6546	7	7
Sample 8	0.9361	0.4505	0.9897	10	8
Sample 9	0.8372	0.3772	0.8505	8	9
Sample 10	0.9554	0.4401	0.9877	9	10

). The results show a high degree of concordance between the two sequences; in particular, the top five and the bottom-ranked samples exhibit similar positions in both rankings. This indicates that the consumer-requirement evaluation model constructed via KANO–AHP–DEMATEL has good fit and explanatory power in predicting market performance. Furthermore, to verify the robustness of the VIKOR results, a sensitivity analysis of the parameter v was performed by varying its value from 0.25 to 0.75. The ranking outcomes under different v settings showed only marginal changes, with the top five alternatives remaining identical across all scenarios. This finding confirms that the proposed model is stable and not significantly influenced by the risk preference of decision-makers. The empirical validation not only demonstrates the model’s stability and predictive capacity in practical design evaluation, but also offers operational guidance for subsequent development of cultural and creative products: prior to market launch, designers can use requirement-oriented weighted rankings to identify alternatives with high potential acceptance, effectively avoiding resource waste and design-failure risk, thereby improving overall decision efficiency and sustainable value.

4. Discussion

This study proposes a user-centered sustainable design decision framework that integrates the KANO, AHP, DEMATEL, and VIKOR methods to optimize the front-end design evaluation process for MCCP. The framework quantifies the expected market performance of design alternatives at the early stage of product development, thereby improving design sustainability, minimizing material and financial waste, and alleviating the production and research burdens on museum institutions.

Beyond methodological integration, the framework also reflects underlying psychological mechanisms that drive consumer preferences for MCCPs. Specifically, users’ differentiated responses captured by the Kano model correspond to distinct motivational structures—basic functional requirements that prevent dissatisfaction, and higher-level emotional or symbolic needs that generate delight. These patterns align with the dual-process perspective of consumer behavior, in which utilitarian and hedonic motivations jointly shape perceived value and decision satisfaction. In the context of museum cultural and creative products, consumers’ preferences are further influenced by psychological factors such as cultural identity, nostalgic attachment, and perceived authenticity, which amplify emotional resonance and purchase intention. Recognizing these psychological drivers provides a deeper understanding of the empirical results and underscores the importance of integrating behavioral insights into design evaluation frameworks.

Previous studies have widely applied the combination of KANO and AHP methods in design evaluation, as these approaches effectively transform user perceptions into weighted priorities of requirements. However, most of these studies focus solely on the relative importance of requirements and neglect the interdependencies among them. Such simplification can lead to incomplete understanding of user needs, since consumer preferences are often influenced by complex causal relationships between functional, emotional, and cultural dimensions. To overcome these limitations, the present study integrates DEMATEL and VIKOR into the KANO–AHP system. DEMATEL identifies the causal interactions among requirement attributes, while VIKOR transforms multidimensional evaluation results into a unified ranking of alternatives. By combining weight hierarchies and relational effects, the framework generates composite weights that more accurately reflect consumer perceptions, thereby reducing subjectivity and enhancing model robustness.

The empirical validation using the Palace Museum’s refrigerator magnets demonstrates the practical utility of the framework. In the KANO analysis, “commemorative value” and “basic comfort of use” were identified as must-be requirements that ensure functional satisfaction, whereas “innovation” and “cultural symbolism” were found to be attractive requirements that significantly enhance emotional and aesthetic satisfaction. This indicates that MCCP design should not only meet utilitarian and commemorative demands but also emphasize emotional expression and symbolic resonance to elicit deeper user engagement and cultural identification.

Further, the AHP results revealed that the emotional dimension carried greater weight than the functional and social dimensions, underscoring the pivotal role of affective design in shaping consumer satisfaction and purchase intention. The DEMATEL analysis visualized the intricate network of inter-factor influences, showing that emotionally driven factors—such as cultural symbolism and innovation—serve as causal drivers that stimulate other dimensions of user perception. When these relational weights were integrated with the AHP-derived priorities, a comprehensive weighting system was obtained, providing a more objective reflection of real consumer cognition.

Finally, the VIKOR-based evaluation ranked multiple design alternatives of the Palace Museum’s refrigerator magnets. The ranking results closely matched the actual market sales performance, exhibiting only marginal deviations. This high consistency confirms that the integrated evaluation model possesses strong predictive validity and can accurately capture the key determinants of consumer behavior in MCCP markets. The single composite score (Q value) provided by VIKOR also offers a transparent and comparable index for product evaluation, enabling design teams to determine market readiness and prioritize high-potential design concepts. Beyond empirical validation, the methodological robustness of the framework was further clarified through comparison with TOPSIS. While both VIKOR and TOPSIS are compromise-based methods, TOPSIS evaluates alternatives only by their relative distance from the ideal and negative-ideal solutions, which may overlook decision-makers’ risk attitudes and the balance between group utility and individual regret. In contrast, VIKOR introduces the parameter (v) to reflect risk preference and control ranking sensitivity, providing higher adaptability and robustness in complex decision contexts. This distinction highlights why the proposed VIKOR-based integration is more suitable for sustainability-oriented and perception-driven product evaluations.

By applying this integrated framework at the early stage of MCCP design, museums can avoid resource-intensive development of low-performing products and focus on optimizing alternatives with stronger market potential. This systematic decision process aligns product development with both user demand and sustainability goals, ensuring the efficient allocation of resources and the minimization of production waste. Consequently, the proposed KANO–AHP–DEMATEL–VIKOR model not only advances theoretical understanding of user-driven sustainable design decision-making but also provides an operational tool for museums to enhance design quality, economic efficiency, and environmental responsibility in cultural and creative product development. Moreover, given the model’s structured and transferable nature, its applicability can extend beyond museum contexts to various domains such as green consumer electronics, sustainable packaging, and fashion design. These sectors similarly face the challenge of integrating customer preferences with sustainable innovation goals, making the framework a valuable decision-support tool across a broad range of product design environments.

The proposed KANO–AHP–DEMATEL–VIKOR pipeline offers actionable guidance for museum product management. First, at the concept-screening stage, managers can use the requirement-driven weights and the Q-based ranking to prioritize high-potential alternatives and avoid resource-intensive development of low-performing designs. Second, the decomposed social–emotional–functional dimensions and their causal relations provide a common language for design, marketing, and operations to align narratives, packaging, and production plans. Third, the sensitivity analysis of the VIKOR parameter v supports risk control by revealing the stability range of rankings under different risk preferences, enabling robust “go/hold” decisions. Fourth, the weights and Q thresholds can be translated into supplier briefs and stage-gate criteria (e.g., minimum acceptable Q), improving transparency and accountability in procurement and prototyping. Finally, by benchmarking model rankings against sales and iteratively updating the requirement set with new user feedback, museums can build a closed-loop decision system that raises forecast accuracy, reduces waste, and enhances sustainable value creation.

5. Conclusions

This study, grounded in empirical research, carries significant theoretical and practical implications. This paper provides an interpretable, reproducible, and implementable evaluation approach for front-end decision-making in MCCP. Compared with traditional ex post reviews and experience-based judgments, this pipeline shifts decision-making upstream to the concept-formation stage, markedly reducing futile iterations, prototyping, and material consumption; meanwhile, the composite weights and the (Q) threshold furnish a common language and a traceable basis for cross-functional design–marketing–operations teams, improving coordination efficiency and transparency.

Specifically, this project’s research will deliver the following contributions:

(1)

Domain application: This study is the first to apply an integrated evaluation approach to the assessment and prioritization of museum cultural and creative products (MCCP), filling the gap of a systematic methodological framework in this field. It will help enhance user satisfaction and market competitiveness, thereby fostering a virtuous cycle of cultural dissemination and museum operations.

(2)

Methodological innovation: This study proposes an integrated evaluation method of KANO + AHP × DEMATEL + VIKOR, providing a computable, auditable, and communicable methodological basis for “requirement identification—weight determination—ranking-based evaluation,” reducing arbitrariness in design decisions and improving the objectivity and consistency of conclusions.

The current framework still relies on expert and user ratings to derive weights and influence degrees; the conclusions may be affected by sample composition and parameter settings (e.g., the (v) parameter in VIKOR). Moreover, each technique within the integrated model (KANO, AHP, DEMATEL, VIKOR) has distinct assumptions and limitations that may lead to discrepancies in model validation outcomes. For instance, AHP is sensitive to inconsistencies in pairwise comparisons, while DEMATEL’s reliance on expert judgment may introduce subjectivity. To minimize potential expert bias, the evaluation process adopted an anonymous scoring mechanism and iterative cross-validation among experts to ensure independence and reduce mutual influence. Furthermore, mean normalization was applied to aggregate expert ratings, thereby mitigating individual deviation and enhancing objectivity. Although the current study ensured inter-rater reliability through repeated validation and independence of expert scoring, the relatively small sample size and disciplinary homogeneity of the expert panel may still constrain representativeness. To improve robustness, we plan to enlarge the sample, optimize expert panel composition, and conduct consistency and parameter-sensitivity analyses. In future work, the framework could also incorporate statistical tests such as Cronbach’s α and consistency ratio assessments to quantify internal reliability, thereby strengthening the empirical validity of expert evaluations. In addition, future research will further extend the proposed framework by integrating alternative quantitative techniques such as the Best-Worst Method (BWM), fuzzy and probabilistic MCDM approaches, and emerging models including rough cloud models and multi-granularity linguistic representations. These enhancements will enable the framework to manage uncertainty, heterogeneity, and large-scale participant data more effectively, thereby improving the robustness, adaptability, and generalizability of decision-making across different museums and product categories. Moreover, as the current empirical validation focuses on a single case—the Palace Museum’s refrigerator magnets—future studies will extend the framework to multiple museums and varied cultural and creative product types to further examine its adaptability and cross-context generalizability. Future directions also include (i) developing a decision-support dashboard that operationalizes the pipeline and visualizes weights/Q thresholds for cross-functional teams; (ii) incorporating large-scale text-mining of online user reviews to dynamically update requirement sets; (iii) conducting multi-site, longitudinal studies to track pre-launch rankings versus post-launch performance; and (iv) embedding cost–benefit and environmental impact metrics alongside user-requirement scores to better support sustainable portfolio decisions.

Overall, the integrated KANO–AHP–DEMATEL–VIKOR framework unifies methodological strengths with empirical validation across the full chain of “requirement identification—weight determination—ranking-based evaluation.” It provides reliable quantitative tools and a clear course of action for pre-launch option screening, resource allocation, and sustainable operation of museum cultural and creative products. This approach not only jointly considers the objective performance of product design and consumers’ subjective perceptions, but also computes the relative importance of evaluation criteria across decision layers and reveals inter-factor linkages, thereby enabling designers to make more scientific and well-reasoned decisions in complex cultural-creative design contexts.