From Visibility to Trust: The Impact of Agricultural Product Packaging Images in Livestreaming on Consumer Perception and Repurchase Intention

Abstract

This study investigates how the packaging image of agricultural products in livestreaming commerce influences consumers’ repurchase intentions and, on this basis, formulates effective packaging improvement strategies to enhance repurchase intention. Focusing on agricultural product packaging, we conducted an online survey of 392 consumers with livestream shopping experience and employed a combined approach of structural equation modeling (SEM) and artificial neural networks (ANNs) to analyze the effects of packaging image on repurchase intention. Starting from elements of packaging design—such as external appearance, cultural cues, imagery, and materials—we examined color schemes, typography, and illustration styles and further explored how these factors shape repurchase intention by elevating consumers’ perceived value. The findings indicate that the information-conveying functions of packaging—namely the communication of product culture, quality, and distinctive attributes—have a significant impact on repurchase intention. SEM results reveal that perceived value plays a pivotal mediating role between packaging image and repurchase intention. Complementarily, ANN analysis identifies visual appearance as the strongest predictor of repurchase intention among all packaging elements. Building on these insights, the study proposes concrete packaging design strategies, including optimizing visual appearance to highlight brand distinctiveness; distilling design symbols to narrate brand stories; and mining cultural connotations to strengthen cultural expression. This research not only verifies the importance of packaging design in boosting consumers’ repurchase intentions but also offers a practical strategic framework that can be extended to packaging design practices for other agricultural products. It provides new theoretical support and practical guidance for the field, offering a valuable reference for future research and practice.

1. Introduction

With the rapid development of internet technology and the widespread use of mobile smart devices, livestreaming has emerged as a new marketing approach that is increasingly favored by both consumers and businesses [1]. It not only offers a novel shopping experience but also opens up new marketing channels for enterprises. During the online sales process on livestreaming platforms, consumers can access information related to products or services, which leads to corresponding value judgments. As retail e-commerce develops rapidly, research on consumer responses has also attracted academic attention. In 2016, China’s online retail sales reached 5.16 trillion yuan, increasing by 32.2% to 7.18 trillion yuan in 2017, and further growing by 23.9% to 9.01 trillion yuan in 2018 [2]. Compared to traditional offline transactions, e-commerce platforms can bring profits and convenience to enterprises, merchants, and consumers alike [3]. However, as an emerging operational model, e-commerce platforms also face many business challenges, one of which is how to promote consumer purchasing behavior.

In the tide of globalization and digital networking, the agricultural products market is undergoing unprecedented transformation [4]. The explosive growth of livestreaming e-commerce—especially in China—has opened vast online sales channels for agricultural products while also bringing unique challenges. In online settings, consumers cannot assess product quality through traditional means such as touch or smell, making trust-building and value communication particularly difficult. Against this backdrop, packaging transcends its physical protective function to become a crucial visual proxy and carrier of information [5]. The packaging image of agricultural products—the overall visual impression formed by elements such as design, materials, color, and pattern [6]—has become the core basis on which consumers, amid information asymmetry, judge quality, build brand cognition, and ultimately make purchase decisions [7].

The livestreaming context further magnifies the importance of packaging image. In fast-paced, information-dense livestreaming sessions, packaging is key to capturing consumers’ “first-glance” attention [8]. An appealing package not only stands out amid a profusion of similar products but also swiftly conveys product distinctiveness and value propositions through visual language, thereby triggering immediate purchase intentions [9]. However, for agricultural e-commerce, true success does not stem from one-off impulse purchases; it is built on long-term trust and repeat purchases. Therefore, an excellent packaging image should not only facilitate the initial transaction but also leave a deep imprint in consumers’ minds, becoming a critical driver of repurchase intention [10].

Despite the growing scholarly interest in livestream marketing, the existing literature exhibits notable gaps. First, much research focuses on how the livestreaming environment affects consumers’ initial purchase intentions or impulsive buying (e.g., the roles of influencer traits, interactivity, or promotions), while largely neglecting “repurchase behavior,” which is essential for consumer loyalty and sustainable business models. Studies dedicated specifically to repurchase intention are already limited; systematic inquiries that situate it at the intersection of livestreaming e-commerce and agricultural packaging are exceedingly rare, constituting a theoretical gap in urgent need of attention. Second, methodologically, prior marketing studies have predominantly relied on traditional statistical approaches such as structural equation modeling (SEM) to validate pre-specified linear relationships. Yet consumers’ decision processes are often complex and nonlinear, subtly shaped by multiple psychological factors. As a powerful machine-learning tool, artificial neural networks (ANNs) excel at identifying and modeling complex nonlinear relationships and have shown substantial potential in market research domains such as consumer behavior prediction; however, their application to the study of agricultural product packaging and repurchase intention remains underexplored.

In response, this paper seeks to fill these gaps by focusing on the specific context of China’s livestreaming e-commerce and systematically examining how the packaging image of agricultural products influences consumers’ ultimate repurchase intentions through perceived value. The study’s innovations are twofold: first, it shifts the focus from the commonly examined “purchase intention” to the longer-term, more consequential “repurchase intention,” directly addressing a gap in the extant literature; second, it innovatively introduces an ANN framework to more precisely capture the complex, nonlinear relationships between packaging elements and consumers’ psychology and behavior. Specifically, the study explores the following core questions:

(1)

In livestreaming scenarios, how does the packaging image of agricultural products affect consumers’ repurchase intentions?

(2)

What mediating roles does the consumers’ perceived functional and emotional value play between packaging image and repurchase intention?

(3)

Compared with traditional models, can an ANN model more effectively uncover the complex influence patterns through which packaging design elements (e.g., color, material, pattern, function) shape consumers’ repurchase intentions?

By delving into these questions, this paper aims to provide a scientific basis for agricultural enterprises to develop more effective livestreaming marketing and packaging design strategies, helping them not only secure initial purchases amid fierce market competition but also build enduring brand loyalty.

2. Theoretical Background and Research Hypotheses

2.1. Agricultural Product Packaging

In the field of agricultural products, due to the high degree of product homogeneity, differentiated packaging becomes particularly crucial. The packaging design of agricultural products plays a significant role in shaping consumers’ first impressions and purchase intentions. According to consumer psychology theory, attractive packaging can stimulate consumers’ desire to purchase [11], while high-quality and aesthetically pleasing packaging can enhance product credibility, thereby promoting consumer purchasing behavior [12]. The packaging design of agricultural products is not only an external display of the product’s image but also a comprehensive reflection of regional culture, rural characteristics, and the achievements of modern agricultural development. From a marketing perspective, appealing packaging can significantly enhance a product’s market competitiveness [13]. Silayoi’s research found that visual elements such as packaging color, images, and fonts directly influence consumers’ perceptions of product quality and brand value [14]. Especially in the highly competitive retail environment for agricultural products, packaging with regional cultural characteristics or emotional design is more likely to trigger consumers’ purchase desires. Ampuero also pointed out that packaging not only conveys product information but also serves as a tool for brand communication and market positioning, which is particularly important in the branding process of agricultural products [15].

From the perspective of economic development, innovative packaging design can bring more economic opportunities to rural areas [16]. A successful packaging design can boost the sales of agricultural products in an entire region and drive local economic development [8]. Through distinctive packaging design, agricultural products can better attract consumers’ attention and increase their purchase intentions. Therefore, further enhancing the innovativeness and practicality of agricultural product packaging design is of great significance for promoting rural economic development and cultural heritage.

2.2. Repurchase Intention

Jones posits that repurchase intention reflects a consumer’s tendency toward continued purchasing behavior following satisfaction and perceived value [17]. The formation of repurchase intention is not determined by a single factor but rather is the result of the combined effect of multiple factors [18]. For example, product quality, price, and cost-effectiveness serve as the foundation, but additional values perceived by consumers, such as brand image and service experience, also play a crucial role in the decision-making process [19]. These perceptions stem not only from the physical characteristics and functions of the product itself but also from the interactive experiences between consumers and the brand during the purchasing and usage processes, such as advertising, packaging design, sales service, and customer support [20]. Furthermore, the importance of consumer perception lies in its ability to significantly enhance consumers’ emotional connection and loyalty to the product. As demonstrated by Bruhn et al. [21], when the value perceived by consumers during online shopping exceeds their expectations, their satisfaction increases, thereby strengthening their repurchase intention, indicating that enhancing perceived value can be an effective strategy to boost repurchase intention. However, existing research has mainly focused on the impact on purchase intention, without further analyzing the influence on “repurchase” or “repeat purchase” behaviors; even studies that specifically examine repurchase behavior are rare. Therefore, this paper proposes a theoretical hypothesis, constructing a model of agricultural product packaging design, packaging function, and consumer repurchase intention, and incorporates perceived value as a mediating variable. By investigating the relationships between various dimensions of packaging design, packaging function, and consumer repurchase intention, this study aims to reveal the impact of perceived value on consumer repurchase intention.

2.3. Applications of Artificial Neural Networks in Consumer Behavior Research

Traditional consumer behavior research has predominantly employed linear structural equation modeling (SEM) to test causal relationships among variables. Although SEM excels at evaluating hypothesized theoretical frameworks, its inherent linearity constrains its ability to explain complex, nonlinear decision processes [22]. To overcome this limitation, artificial neural networks (ANNs), as a powerful machine-learning technique, have been increasingly adopted in marketing research [23]. ANNs emulate the connectivity of biological neurons and learn complex patterns from data, enabling effective modeling and prediction of nonlinear relationships without imposing a priori functional forms among variables. In market research, ANNs have been successfully applied to customer-churn prediction [24], market segmentation [25], advertising effectiveness assessment [26], and the prediction of purchasing behavior [27], among other tasks.

Despite substantial literature on the drivers of repurchase intention—and a growing number of studies employing advanced methods such as ANNs—important gaps remain. In the specific context of livestreaming e-commerce for agricultural products, consumers cannot physically inspect items; consequently, the visual presentation of packaging becomes a key proxy for building trust and perceived value. However, existing research provides limited evidence on the precise pathways through which different packaging dimensions (e.g., visual appearance, functional information) influence repurchase intention. Moreover, prior work has largely focused on establishing whether effects exist rather than uncovering the relative importance of competing drivers. When confronted with packaging information, is visual impact more decisive than functional cues? Are these effects linear or nonlinear? To address these gaps, we develop a theoretical model that integrates packaging design, packaging functionality, perceived value, and repurchase intention. Methodologically, our contribution lies in combining SEM with ANNs: we first use SEM to test the causal paths posited by the theoretical framework and then introduce an ANN model to identify and quantify—through a nonlinear lens—the key drivers of repurchase intention for agricultural products and their relative importance. This integrated approach provides deeper and more precise insights into consumer decision-making in livestreaming e-commerce contexts.

2.4. Stimulus–Organism-Response (SOR) Theory

The Stimulus–Organism-Response (SOR) theory is a widely applied theoretical framework in psychology and behavioral sciences, originally proposed by Mehrabian and Russell [28], used to explain how individuals experience psychological state changes under specific environmental stimuli, ultimately leading to behavioral responses. Its core assumption is that external Stimuli (S) influence internal psychological states (O), which in turn trigger behavioral Responses (R). In recent years, this theory has been increasingly applied in various research fields, such as consumer behavior, environmental psychology, information systems, and tourism management. Eroglu et al. explored how the interface design of online stores (Stimulus) affects consumers’ emotional states (Organism), thereby influencing their purchase intentions (Response) [29]. Sultan et al. examined how marketing communication channels for organic food and perceived value of organic food stimulate behavioral intentions of organic food consumers within the SOR framework [30]. Pereira et al. investigated impulse buying as a consumer behavioral outcome in omnichannel retailing using the SOR theory [31]. Therefore, based on the SOR theory, this paper identifies packaging as an external Stimulus (e.g., visual design, functionality), which must be evaluated through consumers’ internal perceptions (Organism) to trigger repurchase behavior (Response). Structural equation modeling and ANN models are employed to explore the interactions among these variables and their specific impacts on consumer repurchase intention.

2.5. Research Hypotheses

This study refines the packaging design factors influencing online consumers’ repurchase intentions into two variables: packaging design and packaging function. At the same time, perceived value is introduced as a mediating variable to construct a theoretical model of the factors influencing consumers’ repurchase intentions, as shown in Figure 1.

2.5.1. Packaging Design and Repurchase Intention

In the contemporary market environment, the packaging design of agricultural products is not only a fundamental element of product display but also a key factor in shaping product image, enhancing brand value, and ultimately influencing consumers’ repurchase intentions. Excellent packaging can add vitality to products and highlight their uniqueness and advantages, thereby attracting consumers’ attention and expanding the market [32]. Research shows that innovation and uniqueness in packaging design are important drivers for consumers’ purchasing decisions [11]. Packaging design serves as a bridge connecting products with consumers’ psychology, not only increasing the market competitiveness of products but also promoting repurchase by influencing consumers’ psychology and behavior, thus providing momentum for brand building and market expansion [33]. Packaging design is the overall aesthetic perception that individuals form of a package’s visual elements (e.g., color, shape, typeface, and imagery) [18]. In the information-overloaded livestreaming context, superior visual design is the primary means of capturing consumer attention [34]. Repurchase intention refers to the likelihood that, following a purchase experience, consumers choose the same brand or product again; it is a key indicator of customer loyalty and a firm’s long-term profitability [6]. In packaging design, cultural connotations, innovativeness, and practicality should be fully considered to meet consumers’ multilayered needs. This paper divides packaging design into two dimensions: visual appearance and visual elements. Based on the foregoing, we propose the main hypothesis H1 and, given the multidimensionality of packaging design, further advance sub-hypotheses H1a and H1b for specific testing:

H1. 

The packaging design of agricultural products has a positive impact on online consumers’ repurchase intentions.

H1a. 

The visual appearance of agricultural product packaging has a positive impact on online consumers’ repurchase intentions.

H1b. 

The design elements of agricultural product packaging have a positive impact on online consumers’ repurchase intentions.

2.5.2. Packaging Function and Repurchase Intention

The functionality of product packaging can directly affect consumers’ perception of product performance. If the packaging can effectively protect the product from damage, extend its shelf life, or facilitate storage and transportation, it can enhance consumers’ recognition of the product’s functional value [35], thereby increasing the likelihood of repurchase. The functionality of packaging not only improves the actual use value of the product but also indirectly influences consumers’ repurchase decisions by meeting their expectations for convenience and quality assurance. These aspects should be considered in packaging design to promote consumers’ repurchase intentions and enhance brand loyalty. Therefore, this paper divides packaging function into two dimensions; this study advances the main hypothesis H2 and, considering the multidimensional nature of packaging design, further proposes sub-hypotheses H2a and H2b for specific empirical validation:

H2. 

The packaging function of agricultural products has a positive impact on online consumers’ repurchase intentions.

H2a. 

The portability and protection function of agricultural product packaging has a positive impact on online consumers’ repurchase intentions.

H2b. 

The information communication function of agricultural product packaging has a positive impact on online consumers’ repurchase intentions.

2.5.3. Consumer Perceived Value and Repurchase Intention

In the current consumer market, consumer perceived value (CPV) is an important factor influencing purchasing decisions and repurchase intentions. Scholars have conducted extensive research on this topic, proposing perceived value theories from different dimensions, which provide a multi-perspective understanding of consumers’ complex purchasing behaviors. Drawing on existing research, this paper adopts the framework of the scholar Kato [36] and divides consumer perceived value into two dimensions: perceived functional value and perceived emotional value.

Perceived functional value mainly focuses on the practicality and performance of the product, such as quality, reliability, and cost-effectiveness, which are fundamental considerations for consumers when contemplating repurchase. Functional value is a key criterion for consumers to judge whether a product is “worth the price,” directly affecting their satisfaction and loyalty. Perceived emotional value, on the other hand, encompasses the emotional experiences brought about during the purchase and use of the product, such as pleasure, self-fulfillment, or a sense of belonging. The enhancement of this value often stems from brand image, product design, or alignment with consumers’ values. Emotional value is particularly important in today’s consumer culture, as it is closely related to the expression of personal identity and lifestyle. In actual purchasing processes, if a product can simultaneously meet consumers’ functional and emotional needs, its perceived brand value will be significantly enhanced. Consumers’ overall satisfaction and trust in the brand will increase, which not only directly raises the initial purchase rate but also significantly boosts repurchase intentions.

H3. 

Consumers’ perceived value mediates the relationship between the dimensions of packaging design and online consumers’ repurchase intention.

H3a. 

Consumers’ perceived functional value mediates the relationship between appearance and consumers’ repurchase intention.

H3b. 

Consumers’ perceived functional value mediates the relationship between visual elements and consumers’ repurchase intention.

H3c. 

Consumers’ perceived emotional value mediates the relationship between appearance and consumers’ repurchase intention.

H3d. 

Consumers’ perceived emotional value mediates the relationship between visual elements and consumers’ repurchase intention.

H4. 

Consumers’ perceived value mediates the relationship between the dimensions of packaging functionality and online consumers’ repurchase intention.

H4a. 

Consumers’ perceived functional value mediates the relationship between portability–protection and consumers’ repurchase intention.

H4b. 

Consumers’ perceived functional value mediates the relationship between information communication and consumers’ repurchase intention.

H4c. 

Consumers’ perceived emotional value mediates the relationship between portability–protection and consumers’ repurchase intention.

H4d. 

Consumers’ perceived emotional value mediates the relationship between information communication and consumers’ repurchase intention.

3. Research Design

3.1. Construction of the Research Model

Based on the Stimulus–Organism-Response (SOR) theoretical framework and incorporating the multidimensional attributes of agricultural product packaging, this study constructs the theoretical model shown in Figure 1. The packaging image of agricultural products is systematically deconstructed into two dimensions: the visual layer (aesthetic appeal and recognizability of packaging design) and the functional layer (fresh-keeping performance and convenience). These dimensions serve as external stimuli (Stimulus), which influence consumers’ perceived value (Organism) and ultimately affect their repurchase intention (Response).

Regarding the operational definition of variables (see

Table 1. Operational definitions and measurement basis of core variables.

Variable Category	Core Variable	Measurement Dimension	Theoretical Basis
Stimulus (S)	Packaging Visual Layer	Design Aesthetics, Recognizability	Multidimensional Brand Image Model
	Packaging Functional Layer	Preservation Performance, Convenience of Use	Technology Acceptance Model (TAM)
Organism (O)	Perceived Functional Value	Confidence in Quality and Safety, Information Credibility	Perceived Risk Theory
	Perceived Emotional Value	Sense of Pleasure, Identification, Belonging	Emotional Value in SOR Framework
Response (R)	Repurchase Intention	Repeat Purchase Tendency in Livestreaming, Willingness to Recommend	Repurchase Intention Scale

), the measurement scales are adapted from established scales and revised to fit the livestreaming context. The visual layer of packaging draws on the five-dimensional measurement system for agricultural product brand image (function, value, emotion, culture, and visual dimension), covering indicators such as color matching, graphic and text layout, and material texture. The functional layer incorporates core constructs from the Technology Acceptance Model (TAM), introducing measurement items such as fresh-keeping effectiveness, ease of opening, and information readability. The mediating variable, “consumer perceived value,” primarily reflects consumers’ confidence in product quality and safety, drawing on perceived value theory in livestreaming shopping.

3.2. Research Methods and Procedures

This study employs a hybrid analytical approach that integrates structural equation modeling (SEM) with artificial neural networks (ANNs). This design leverages complementary strengths: SEM is first used to test theory-driven causal paths and mediation effects among variables and to validate the model’s theoretical constructs; subsequently, the ANN is employed to capture potential complex nonlinear relationships among variables and to predict consumers’ repurchase intention with higher precision. The overall procedure of this study is depicted in Figure 2. We first establish the theoretical foundation through a literature review, from which we develop the conceptual model and formulate the research hypotheses. In the first stage, structural equation modeling (SEM) is employed to test the theoretically derived causal relationships. This includes selecting key variables, designing the questionnaire, conducting rigorous data preprocessing, and subsequently performing model estimation, evaluation, and interpretation of path relationships. The second stage applies an artificial neural network (ANN) model to uncover complex nonlinear patterns in the data and to identify the relative importance of predictors without imposing linearity assumptions. Finally, the results from the SEM and ANN are integrated. This dual-track approach not only validates the theoretical model but also, in a data-driven manner, deepens our understanding of the key factors shaping consumers’ repurchase intention, thereby yielding more nuanced conclusions.

Structural equation modeling (SEM), a mature multivariate statistical technique, can simultaneously handle observed and latent variables and effectively test direct, indirect, and total effects among variables [37]. Given that the influence of agricultural product packaging design on consumer decision-making constitutes a complex system involving multiple factors, such as cognition and emotion, SEM provides a robust analytical tool for validating the theoretical model proposed in this study. In this study, the SEM analyses will be conducted using IBM SPSS AMOS 26.0, following a two-stage approach: first, confirmatory factor analysis (CFA) will be used to assess the reliability and validity of the measurement model; then, path analysis and mediation testing will be performed for the structural model.

An artificial neural network (ANN) is a nonlinear modeling technique that simulates the operation of biological neural systems and is noted for its strong self-learning and adaptive capabilities [38]. Unlike traditional statistical models that rely on strict linear assumptions, the ANN does not require pre-specifying functional relationships among variables; it can automatically learn and approximate highly complex nonlinear patterns from data and typically achieves higher accuracy in prediction tasks. In this study, the ANN will be used to complement and deepen the SEM results, aiming to more precisely quantify the relative importance of antecedent variables for repurchase intention and to build a high-accuracy predictive model. To ensure transparency and reproducibility, we will explicitly specify the ANN model configuration. The Neural Networks module in IBM SPSS Statistics 26.0 will be used to build and train the model. The specific hyperparameter settings are shown in Figure 3.

3.3. Questionnaire Design and Variable Measurement

This study collected data using a structured questionnaire, which consisted of two parts: (1) respondents’ demographic characteristics (such as gender, age, income, and frequency of livestream shopping); (2) measurement items for core variables, all of which were assessed using a five-point Likert scale (1 = “strongly disagree,” 5 = “strongly agree”), as shown in

Table 2. Scale design.

First-Level Dimension	Second-Level Dimension	Questionnaire Items
Packaging design	Appearance (AP)	Livestreaming presentation effect
		Livestreaming unboxing experience
		Material texture expressiveness
	Visual elements (VE)	Color coordination
		Reasonable text and image layout
		Illustration style innovation
Packaging function	Portability and protection (PP)	Portability
		Protection
		Freshness preservation
	Conveying information (CI)	Conveying product culture
		Conveying product quality
		Conveying product features
Consumer perceived value	Perceived functional value (PFV)	Trustworthiness of preservation performance
		Reliability of traceability information
		Consistency between live broadcast and actual product
	Perceived emotional value (PEV)	Aesthetic pleasure
		Environmental awareness
		Sense of community belonging
Repurchase intention (RI)		Livestreaming repurchase intention
		Livestreaming recommendation intention

. All measurement items used in this study are detailed in Supplementary Materials. To ensure content validity, the research team conducted two rounds of pre-testing: the first round involved in-depth interviews with 12 livestream shopping consumers to optimize the wording of the items; the second round distributed 450 questionnaires online, and the reliability and validity of the scale were tested using Cronbach’s α coefficient and exploratory factor analysis.

3.4. Sample Selection and Data Collection

This study adopted a stratified quota sampling strategy, covering agricultural product livestreaming consumers across China’s seven major geographic regions (such as North China, East China, South China, etc.). Field surveys were conducted through a professional research platform, with three screening criteria: (1) having purchased agricultural products via livestreaming in the past three months; (2) a single purchase amount of at least 50 yuan; (3) age between 18 and 55 years old. The quotas were set according to the ratio of Generation Z (18–28 years old) to middle-aged and young adults (29–55 years old) internet users published by the National Bureau of Statistics, at 1:1.2, while also balancing gender distribution. A total of 450 questionnaires were collected. After excluding responses with patterned answers and those failing attention check questions (e.g., “Please select ‘strongly agree’”), 392 valid questionnaires were retained, resulting in an effective rate of 87.1%. The sample structure showed that Generation Z consumers accounted for 46.5%, females for 57.3%, households with a monthly income of 8000–15,000 yuan for 62.7%, and high-frequency livestream shoppers (≥2 times per week) for 41.2%. These sample characteristics align with the core consumer profile of current agricultural product livestreaming.

4. Analysis of Design Influencing Factors Based on SEM Model

4.1. Reliability and Validity Testing

Table 3. Data reliability analysis table.

Construct	Item	Cronbach’s α
AP	Livestreaming presentation effect	0.896
	Livestreaming unboxing experience
	Material texture expressiveness
VE	Color coordination	0.866
	Reasonable text and image layout
	Illustration style innovation
PP	Portability	0.882
	Protection
	Freshness preservation
CI	Conveying product culture	0.924
	Conveying product quality
	Conveying product features
PFV	Trustworthiness of preservation performance	0.856
	Reliability of traceability information
	Consistency between live broadcast and actual product
PEV	Aesthetic pleasure	0.907
	Environmental awareness
	Sense of community belonging
RI	Livestreaming repurchase intention	0.939
	Livestreaming recommendation intention

Note. AP: Appearance; VE: Visual elements; PP: Portability and protection; CI: Conveying information; PFV: Perceived functional value; PEV: Perceived emotional value; RI: Repurchase intention. Cronbach’s α: A high value (>0.7) indicates good internal consistency of the scale, while a low value (<0.6) indicates poor consistency.

presents the measurement dimensions, specific items, and internal consistency reliability (Cronbach’s α) coefficients for each research variable. Overall, the Cronbach’s α values for all dimensions are above 0.85, indicating good internal consistency among the scale items and high reliability of the measurement tools, thus providing a solid data foundation for subsequent empirical analysis.

Table 4. KMO and Bartlett’s test.

KMO		0.849
Bartlett’s sphericity	spherical test	5670.722
	df-value	190
	p-value	0.000

Note. KMO: A high value (>0.8) indicates that the sample is suitable for factor analysis, while a low value (<0.6) indicates that it is not suitable.

reports the results of the KMO measure of sampling adequacy and Bartlett’s test of sphericity. The KMO value is 0.849, well above 0.8, indicating that the sample data are suitable for factor analysis and possess good structural validity. The approximate chi-square value for Bartlett’s test of sphericity is 5670.722, with 190 degrees of freedom and a significance level of 0.000 (p < 0.001), suggesting that the correlation matrix is not an identity matrix and that there are strong correlations among the variables. This further supports the feasibility of factor analysis. In summary, the data meet the prerequisites for factor analysis, providing a solid foundation for subsequent structural modeling and variable extraction.

4.2. Confirmatory Factor Analysis

Table 5. Validated factor analysis model fit.

Fitting Index	Acceptable Range	Measured Value
CMIN		358.901
DF		150
CMIN/DF	1–3	2.393
GFI	≥0.8	0.916
AGFI	≥0.8	0.883
RMSEA	<0.08	0.060
IFI	≥0.9	0.963
NFI	≥0.8	0.938
TLI (NNFI)	≥0.9	0.953
CFI	≥0.9	0.963

presents the main fit indices and their values for the structural equation model. The results show that the CMIN/DF is 2.393, which falls within the acceptable range of 1 to 3, indicating a good model fit. The GFI and AGFI are 0.916 and 0.883, respectively, both exceeding 0.8, suggesting a high overall model fit. The RMSEA is 0.060, below the threshold of 0.08, further demonstrating excellent model fit. The values for IFI, NFI, TLI (NNFI), and CFI are 0.963, 0.938, 0.953, and 0.963, respectively, all above the standard of 0.9 or 0.8, indicating strong explanatory power and adaptability of the model. In summary, all fit indices meet or exceed acceptable standards, indicating that the structural equation model constructed in this study has a good fit and can effectively reflect the data structure and theoretical assumptions.

4.3. Hypothesis Testing

Table 6. Model measurements.

Hypothesis				Estimate	S.E.	C.R.	p	Testing the Hypothesis
H1a	RI	<---	AP	0.210	0.055	3.792	***	Established
H1b	RI	<---	VE	0.159	0.058	2.732	0.006	Established
H2a	RI	<---	PP	0.150	0.057	2.623	0.009	Established
H2b	RI	<---	CI	0.169	0.048	3.523	***	Established
H3	RI	<---	PFV	0.197	0.074	2.646	0.008	Established
H4	RI	<---	PEV	0.171	0.052	3.316	***	Established
H3a	PFV	<---	AP	0.239	0.046	5.250	***	Established
H3b	PFV	<---	VE	0.141	0.049	2.868	0.004	Established
H3c	PEV	<---	PP	0.171	0.064	2.670	0.008	Established
H3d	PEV	<---	CI	0.318	0.052	6.165	***	Established
H4a	PEV	<---	AP	0.166	0.059	2.815	0.005	Established
H4b	PEV	<---	VE	0.193	0.065	2.964	0.003	Established
H4c	PFV	<---	PP	0.136	0.048	2.824	0.005	Established
H4d	PFV	<---	CI	0.111	0.039	2.886	0.004	Established

Note. *** indicates p < 0.001; AP: Appearance; VE: Visual elements; PP: Portability and protection; CI: Conveying information; PFV: Perceived functional value; PEV: Perceived emotional value; RI: Repurchase intention.

and Figure 4 present the test results for each path hypothesis in the structural equation model. All estimated coefficients are positive, with relatively small standard errors (S.E.), critical ratios (C.R.) all greater than 2, and p-values all less than 0.01 (many of which are highly significant, p < 0.001), indicating that all path relationships are statistically significant and all hypotheses are supported. Specifically, variables such as the visual appearance (AP) and visual elements (VE) of packaging design, the portability and protection (PP) and information communication (CI) of packaging function, as well as consumer perceived functional value (PFV) and perceived emotional value (PEV), all have significant positive effects on repurchase intention (RI). At the same time, each dimension of packaging design and packaging function also has a significant positive effect on perceived functional value (PFV) and perceived emotional value (PEV). Overall, the results indicate that packaging design and function significantly promote repurchase intention by enhancing consumer perceived value, thus validating the rationality of the theoretical model and the effectiveness of the path mechanisms.

4.4. Mediation Analysis

Table 7. Analysis of intermediation effects.

Mediation Path	Effect	Estimate	Lower	Upper	p
AP—PFV—RI	Indirect Effect	0.047	0.004	0.103	0.026
	Direct Effect	0.210	0.082	0.329	0.001
	Total Effect	0.257	0.144	0.372	0.001
VE—PFV—RI	Indirect Effect	0.028	0.002	0.068	0.028
	Direct Effect	0.210	0.082	0.329	0.001
	Total Effect	0.238	0.112	0.361	0.001
PP—PFV—RI	Indirect Effect	0.028	0.000	0.066	0.047
	Direct Effect	0.159	0.008	0.306	0.037
	Total Effect	0.186	0.035	0.334	0.017
CI—PFV—RI	Indirect Effect	0.033	0.003	0.077	0.022
	Direct Effect	0.159	0.008	0.306	0.037
	Total Effect	0.192	0.044	0.347	0.010
AP—PEV—RI	Indirect Effect	0.027	0.001	0.065	0.045
	Direct Effect	0.150	0.006	0.302	0.041
	Total Effect	0.176	0.034	0.324	0.014
VE—PEV—RI	Indirect Effect	0.029	0.001	0.066	0.040
	Direct Effect	0.150	0.006	0.302	0.041
	Total Effect	0.179	0.036	0.333	0.019
PP—PEV—RI	Indirect Effect	0.022	0.001	0.054	0.034
	Direct Effect	0.169	0.056	0.283	0.006
	Total Effect	0.171	0.027	0.323	0.019
CI—PEV—RI	Indirect Effect	0.054	0.013	0.110	0.009
	Direct Effect	0.150	0.006	0.302	0.041
	Total Effect	0.224	0.128	0.326	0.001

Note. AP: Appearance; VE: Visual elements; PP: Portability and protection; CI: Conveying information; PFV: Perceived functional value; PEV: Perceived emotional value; RI: Repurchase intention.

presents the results of the mediation effect analysis, specifically examining the indirect effects of each dimension of packaging design and function (AP, VE, PP, CI) on repurchase intention (RI) through perceived functional value (PFV) and perceived emotional value (PEV). The results show that all indirect effects of the mediation paths are positive, and the confidence intervals (Lower–Upper) do not include zero, with p-values all less than 0.05, indicating that the mediation effects are significant. At the same time, both the direct effects and total effects are also statistically significant, suggesting that packaging design and function not only directly influence repurchase intention but also indirectly promote repurchase intention by enhancing consumers’ perceived value. Overall, perceived functional value and perceived emotional value play a partial mediating role in the process by which packaging design and function affect repurchase intention, further validating the rationality of the theoretical model and the diversity of the path mechanisms.

5. Construction of the ANN Model

To further optimize the model and enhance its predictive accuracy, this section integrates the results of structural equation modeling (SEM) analysis with artificial neural network (ANN) methods to construct an SEM–ANN-based model for agricultural product repurchase intention. Referring to the research of Liébana et al. [39], and based on the results of SEM testing and the principles of ANNs, three artificial neural network models—A, B, and C—were constructed, as shown in Figure 5. The input layer of Model A includes appearance (AP), visual elements (VE), portability and protection (PP), and information communication (CI), with the output being perceived functional value (PFV). The input layer of Model B consists of appearance (AP), visual elements (VE), portability and protection (PP), and information communication (CI), with the output layer being perceived emotional value (PEV). The input layer of Model C integrates appearance (AP), visual elements (VE), portability and protection (PP), information communication (CI), perceived functional value (PFV), and perceived emotional value (PEV), with the output layer being repurchase intention (RI).

5.1. Root Mean Square Error Test

Table 8. Root mean square error test for artificial neural network models.

	Model A		Model B		Model C
	Input: AP, VE, PP, CI		Input: AP, VE, PP, CI		Input: AP, VE, PP, CI, PFV, PEV
	Output: PFV		Output: PEV		Output: RI
Neural network	Training	Testing	Training	Testing	Training	Testing
ANN1	0.3910	0.3809	0.3518	0.3620	0.2658	0.2380
ANN2	0.3871	0.4363	0.3150	0.2826	0.2594	0.2655
ANN3	0.3505	0.4551	0.3237	0.2682	0.2965	0.2572
ANN4	0.3561	0.2819	0.3494	0.3182	0.3043	0.2005
ANN5	0.3544	0.3374	0.2976	0.1915	0.2891	0.3150
ANN6	0.4708	0.2019	0.3225	0.2877	0.3622	0.2888
ANN7	0.3457	0.3140	0.3366	0.2953	0.2586	0.3285
ANN8	0.3969	0.3954	0.3244	0.3753	0.2878	0.2029
ANN9	0.3539	0.3065	0.3581	0.2321	0.3001	0.1484
ANN10	0.3721	0.3289	0.2738	0.2521	0.3331	0.3731
Mean	0.3779	0.3438	0.3253	0.2865	0.2957	0.2618
SD	0.1887	0.2679	0.1567	0.2304	0.1761	0.2532

Note. AP: Appearance; VE: Visual elements; PP: Portability and protection; CI: Conveying information; PFV: Perceived functional value; PEV: Perceived emotional value; RI: Repurchase intention.

presents the training and testing mean square error (MSE) performance of the three neural network models (Model A, B, and C) under different input and output settings. Models A and B use packaging design and functional dimensions (AP, VE, PP, CI) as inputs to predict perceived functional value (PFV) and perceived emotional value (PEV), respectively; Model C incorporates PFV and PEV as inputs to predict repurchase intention (RI). In terms of mean values, the training and testing errors of all three models are relatively low, and the training errors are close to the testing errors, indicating good generalization ability. Among them, the testing mean square error of Model C (0.2618) is slightly higher than that of Model B (0.2865), but the overall performance is stable, suggesting that the model’s predictive ability for repurchase intention is improved after introducing mediating variables. The standard deviation (SD) values are moderate, reflecting small fluctuations in the results of each experiment. Overall, the neural network models can effectively capture the complex nonlinear relationships among variables, providing strong empirical support for the theoretical model’s multiple paths and configuration analysis.

5.2. Sensitivity Analysis

Table 9. Analysis of the importance of normalization in artificial neural network models.

Neural Network		ANN1	ANN2	ANN3	ANN4	ANN5	ANN6	ANN7	ANN8	ANN9	ANN10	Average Relative Importance	Normalized Relative Importance (%)
Model A (Output: PFV)	AP	0.477	0.270	0.354	0.354	0.329	0.313	0.370	0.451	0.350	0.379	0.365	100.000
	VE	0.262	0.231	0.220	0.206	0.221	0.241	0.252	0.137	0.275	0.208	0.225	61.644
	PP	0.101	0.195	0.225	0.271	0.290	0.223	0.218	0.054	0.209	0.215	0.200	54.800
	CI	0.160	0.305	0.201	0.169	0.160	0.222	0.159	0.358	0.166	0.198	0.210	57.530
Model B (Output: PEV)	AP	0.225	0.193	0.156	0.015	0.201	0.203	0.258	0.227	0.058	0.187	0.172	55.128
	VE	0.416	0.205	0.275	0.235	0.184	0.306	0.389	0.295	0.335	0.217	0.286	91.667
	PP	0.303	0.304	0.313	0.466	0.381	0.225	0.270	0.242	0.255	0.357	0.312	100.000
	CI	0.057	0.298	0.256	0.284	0.233	0.266	0.083	0.235	0.352	0.239	0.230	73.718
Model C (Output: RI)	AP	0.200	0.183	0.190	0.207	0.148	0.171	0.137	0.193	0.197	0.094	0.172	85.149
	VE	0.147	0.180	0.217	0.173	0.231	0.027	0.212	0.202	0.194	0.268	0.185	91.584
	PP	0.108	0.183	0.208	0.150	0.170	0.402	0.135	0.051	0.099	0.232	0.174	86.139
	CI	0.183	0.143	0.045	0.175	0.146	0.245	0.195	0.189	0.219	0.075	0.162	80.198
	PFV	0.112	0.050	0.143	0.081	0.130	0.102	0.110	0.145	0.089	0.096	0.106	52.475
	PEV	0.250	0.260	0.197	0.215	0.175	0.053	0.212	0.219	0.202	0.235	0.202	100.000

Note. AP: Appearance; VE: Visual elements; PP: Portability and protection; CI: Conveying information; PFV: Perceived functional value; PEV: Perceived emotional value; RI: Repurchase intention.

shows the relative importance and normalized percentages of each input variable in the neural network models under different output variables. The results indicate that in Model A, with perceived functional value (PFV) as the output, appearance (AP) has the most significant impact on PFV (normalized importance 100%), followed by visual elements (VE), information communication (CI), and portability and protection (PP), suggesting that the appearance and design of packaging are the core factors influencing functional value perception. In Model B, with perceived emotional value (PEV) as the output, portability and protection (PP) and visual elements (VE) are the most important, indicating that the practicality and innovative design of packaging play a prominent role in enhancing emotional value. For Model C, with repurchase intention (RI) as the output, perceived emotional value (PEV) and visual elements (VE) are the most critical, while appearance (AP), portability and protection (PP), and information communication (CI) also have considerable influence, and perceived functional value (PFV) is relatively less important. Overall, each dimension of packaging design and function plays an important role in consumer perceived value and repurchase intention, with emotional value being particularly prominent in driving repurchase intention, further enriching the theoretical explanation of how packaging influences consumer behavior.

6. Discussion and Conclusions

6.1. Discussion of Results

With the rapid growth of livestreaming e-commerce, the sales channels and consumption scenarios for agricultural products have undergone profound changes. When purchasing agricultural products on livestreaming platforms, consumers no longer engage with physical goods offline; instead, they rely on product information and visual representations conveyed through the screen. In this emerging consumption environment, the packaging image of agricultural products—serving as a crucial medium between products and consumers—has an increasingly salient impact on repurchase intention. Accordingly, this paper systematically examines how packaging visual elements enhance consumers’ perceived value and thereby influence repurchase behavior, and, by combining structural equation modeling (SEM) with artificial neural networks (ANNs), provides an in-depth analysis of the mechanisms and relative influence weights of individual packaging elements. The specific answers to the research questions are as follows:

(1)

How does packaging design affect repurchase intention? SEM results indicate that visual appearance, cultural expression, and material selection all exert significant positive effects on repurchase intention. In particular, the effective communication via packaging of product culture, quality, and distinctive attributes represents a key pathway influencing consumers’ repurchase intention (both H1 and H2b are supported).

(2)

What are the mechanisms of information transmission and perceived value? The results show that packaging indirectly strengthens repurchase intention by elevating consumers’ perceived value of the product, thereby confirming the mediating effects (H3/H4).

(3)

How do the influence weights of different packaging elements differ? ANN analysis reveals that visual appearance (normalized importance = 85.149%, Model C) has the most pronounced impact on repurchase intention, far exceeding that of other variables such as emotional value or portability.

6.2. Research Contributions

This study finds that, within livestreaming contexts, the visual appearance and cultural expression of packaging play a dominant role in shaping consumers’ repurchase intentions. Compared with the “experience–trust–repurchase” pathway typical of offline retail, livestreaming exhibits a more pronounced “visual–trust–repurchase” pattern. Because livestream shopping lacks physical contact, consumers’ reliance on packaging increases markedly, making visual elements the primary gateway to trust formation and decision-making. This result aligns with prior research suggesting that “visual impact enhances online consumption trust,” while our study further quantifies the relative importance of specific elements via an ANN model, thereby enriching the theoretical linkage between packaging design and consumer behavior.

Practically, firms should emphasize visual innovation and cultural expression in packaging, strengthening the brand-identification functions of color, illustration, and material. Packaging design should also prioritize clear information transmission, enabling consumers to grasp core product value quickly and thereby boosting purchase confidence. For brand building, packaging is not merely the product’s exterior but a carrier of brand culture and product value. Rigorous, research-informed packaging design helps elevate perceived value and enhance brand loyalty. Notably, the effect of visual appearance on repurchase intention far exceeds that of emotional value or portability. A likely reason is that, in the livestreaming environment, consumers’ initial perceptions are driven primarily by visual impact, whereas emotional and functional values are established gradually through subsequent experience. In addition, cultural symbols and brand storytelling can rapidly evoke emotional resonance and strengthen brand memory, which further explains the pronounced influence of visual elements.

In sum, this study provides new evidence that, in the emerging commercial ecosystem of livestreaming, the packaging image of agricultural products is a key strategic determinant of consumers’ repurchase intentions. The findings indicate that packaging not only directly affects repurchase through its information-conveying function and overall design aesthetics; more importantly, it operates via consumers’ perceived value, which mediates this relationship. Especially noteworthy is that the media characteristics of livestreaming render visual appearance the single most influential factor. These insights offer empirical support for targeted livestreaming marketing and packaging strategies and contribute preliminary extensions to brand management theory in the digital era. We hope this study provides useful guidance for industry practice and inspires subsequent academic inquiry.

7. Recommendations and Countermeasures

This paper systematically investigates how the packaging image of agricultural products elevates consumers’ perceived value and thereby shapes their repurchase behavior. The results show that elements such as visual appearance, information communication, and cultural connotations exert direct positive effects on consumers’ repurchase intentions. Packaging profoundly influences repurchase intention via the pathway “visual appeal → information transmission → cultural identification.” Beyond its theoretical significance, the study offers clear and actionable strategic guidance for agricultural producers, brand marketers, and packaging designers competing in the highly contested livestreaming market. As illustrated in Figure 6, the paper advances strategic countermeasures and recommendations from three perspectives.

7.1. Optimize Visual Appearance: Creating Visual Identity for Livestreaming Scenarios

To effectively enhance brand recognition and appeal in the livestreaming context, agricultural product packaging should not be limited to the optimization of static aesthetics but should instead construct a “dynamic visual hammer” adapted to the characteristics of the livestreaming medium. The core of this strategy lies in viewing packaging design as a medium that “performs” in front of the camera. Specific countermeasures include the following: First, enhancing the “unboxing ceremony” design. The structure of the packaging should be meticulously designed so that the process of it being opened by the live-streamer is layered and full of surprise. This can be achieved by adopting pull-out or layered designs, or those with tear-off “Easter egg” tags, transforming the act of “opening the package” itself into captivating and easily shareable visual content, thereby implanting a deep and positive memory in the consumer’s mind. Second, applying “media-adaptive” colors and materials. Considering the lighting in livestreams and screen color discrepancies, a color system with high saturation and high contrast should be selected to ensure that the visual impact is not diminished under various lighting conditions. At the same time, the choice of material should consider both tactile feel and sound effects. Third, endowing the packaging with a “second life” that transcends its single-use mission. Designing it as a reusable storage box or decorative item and cleverly integrating brand logos not only extends the brand’s exposure in the consumer’s daily life but also provides the live-streamer with additional product value to narrate, effectively stimulating the consumer’s desire for ownership and expectation of repurchase. This design not only extends the packaging’s life cycle but also projects the brand image from the moment of consumption into the consumer’s daily life scenarios, becoming a continuous emotional asset that effectively stimulates the desire for possession and repurchase expectations.

7.2. Refine Symbols: Building a Brand IP Symbol System

Transcending simple information delivery, agricultural product packaging should serve as an encapsulated medium for the brand’s narrative, effectively transforming the packaging into a brand IP by constructing a comprehensive “symbolic narrative system.” This strategy aims to elevate the consumer’s purchasing behavior from “buying a product” to “buying a story and a sense of identity.” Specific strategies include the following: First, unearth and interpret the core narrative motif. Brands must delve deep into their unique story core—such as the founder’s craftsmanship, the legend of a specific region of origin, or a unique farming philosophy—and distill it into a set of highly recognizable core visual symbols (e.g., abstracted mountain shapes, water patterns, or agricultural tool totems). Second, integrate this set of symbols throughout the packaging design of the product series, forming a unified yet dynamic visual language. In a livestream, the host no longer simply introduces “this apple is very sweet” but can “decode” the symbols on the packaging: “This pattern represents the unique sun-facing slope of our orchard, which is the secret to its sweetness.” This interactive interpretation endows the product with depth and soul. Third, instill value through “serialization and collectability.” Design the packaging for different batches or seasonal products as a continuous, collectible series, where each package tells a new chapter of the story or showcases a unique symbol variant. This “narrative hook” employs the psychological mechanism of “Collect-to-Complete” incentives, transforming a single purchase into a quest for the complete story experience. It encourages consumers to move from “trying something new” to “collecting them all,” thereby significantly enhancing user stickiness and the drive for repeat purchases. Furthermore, marketers can plan to design packaging for different product series as a collectible set, leveraging the consumer’s “stamp-collecting” mentality to convert a single purchase into a sustained pursuit of the complete brand narrative. Finally, by embedding QR codes or NFC tags on the packaging that link to immersive content such as panoramic VR tours of the place of origin, the packaging is transformed from a static surface into a “digital portal” connecting online and offline experiences. This effectively converts the traffic from initial purchasers into highly engaged brand fans.

7.3. Explore Culture: Integrating Regional Culture with Modern Aesthetics

To establish high-dimensional brand barriers amidst homogenized competition, agricultural product packaging must deeply explore and vitalize the “cultural terroir” of its region, crafting the packaging into a “cultural business card” that carries regional culture and collective memory. The innovativeness of this strategy lies in its shift from the superficial appropriation of cultural elements to a deep resonance with cultural values. Specific strategic recommendations include the following: First, conduct in-depth ethnographic research on the “local culture.” The research team should move beyond the shallow replication of traditional patterns to deeply investigate the intangible cultural heritage, festive customs, folklore, and even unique expressions in local dialects of the product’s place of origin, seeking cultural DNA that spiritually aligns with the product’s attributes. Second, promote the “co-creation and regeneration of cultural IP.” Collaborate with local cultural inheritors, young artists, or design schools to “redesign” traditional cultural elements—such as local opera masks, traditional weaving textures, or ancient agricultural murals—in a modern and artistic manner. This ensures they retain the recognizability of their cultural roots while appealing to the aesthetic tastes of contemporary consumers. Third, undertake the “aesthetic and cultural interpretation of transparent traceability.” This involves transforming the functional element of product traceability information into a cultural and aesthetic expression through the language of design. For instance, temperature and humidity data from key growth stages can be presented as artistic infographics, or the authentic signatures and handprints of farmers can be incorporated as design elements, merging the traditional cultural value of “integrity” with the modern consumer’s pursuit of “authenticity.” This “ethical aesthetics” approach to design visualizes trust, elevating a functional requirement into an opportunity for cultural identification. Fourth, during livestreams, the host can narrate the cultural stories and value propositions behind the packaging. This endows the act of consumption with a social significance that transcends material satisfaction, thereby building the most steadfast brand loyalty.

7.4. Limitations and Prospects

Although this study systematically explores the impact of agricultural product packaging image on consumers’ repurchase intentions in the context of livestreaming and proposes practical packaging design strategies, there are still some notable limitations. First, the research sample mainly comes from specific livestreaming platforms and a limited consumer group, with relatively concentrated regional and demographic distribution, which may affect the generalizability of the findings. Consumers from different regions and age groups, or with different consumption habits, may have different perceptions and preferences regarding packaging image. Future research can further expand the sample scope to include more diverse consumer groups and different types of agricultural products, thereby improving the external validity of the results. Second, this study collected data through questionnaires. Although efforts were made to ensure the scientific rigor and validity of the questionnaire design and implementation, the subjectivity of self-reported data may lead to some discrepancies between consumers’ actual purchasing behavior and their responses. In addition, consumer decisions in livestreaming environments are often influenced by multiple factors, such as the host, atmosphere, and promotional activities, while this study mainly focuses on the packaging image itself and does not incorporate these external variables into the analytical framework. Future research could combine experimental methods and behavioral tracking to further enhance the objectivity and depth of the study.