Driving Sustainable Consumption and Word of Mouth Through AI Fitness Apps: The Role of Customer Engagement

Abstract

This study investigates the factors influencing customer engagement in AI-powered fitness applications and the subsequent impact on behavioral outcomes such as word of mouth and sustainable consumption. A quantitative research design was employed, with data collected through a structured survey from users of AI fitness applications, and analyzed using Structural Equation Modeling (SEM). Drawing on the Information Systems Success Model and engagement theory, the research examines the roles of service quality, system quality, information quality, health consciousness, anthropomorphism, and personal innovativeness. Findings reveal that higher perceived quality across service, system, and information dimensions, coupled with health consciousness and human-like features, significantly enhances user engagement. Engagement, in turn, drives both advocacy and sustainable behaviors, while personal innovativeness selectively amplifies the effect of system quality. The study advances theoretical understanding by adapting and extending established models to the context of AI-driven health technologies, while also providing practical insights for the development of intelligent, user-centric fitness applications that promote sustained engagement and responsible use.

1. Introduction

Artificial intelligence (AI) is rapidly redefining digital service environments, particularly in the health and fitness sector [1,2]. Recent advances in digital health technologies have significantly transformed how individuals engage with health and fitness services [3,4,5]. Earlier generations of digital health systems primarily functioned as passive tools, offering static information, basic tracking features, and limited personalization. In contrast, AI-powered fitness applications represent a new generation of intelligent systems that continuously learn from user data, adapt recommendations in real time, and provide personalized feedback [3]. This shift marks a transition from simple information delivery platforms to adaptive systems that actively shape user behavior and decision-making processes. This transformation has shifted digital fitness platforms from passive tools to active, data-driven agents capable of influencing behavioral decisions and lifestyle patterns [6]. A key distinction of AI fitness applications lies in their ability to simulate human-like interaction through features such as conversational interfaces, virtual coaching, and emotionally responsive feedback [7].

As a result, users do not merely interact with these applications as tools but increasingly perceive them as interactive partners that guide, motivate, and respond to their needs [8]. This fundamentally changes the nature of user engagement, moving it from occasional or task-based usage toward a continuous, relationship-like interaction. Such sustained engagement is particularly important in promoting long-term behavioral outcomes, including healthier and more sustainable consumption patterns.

The relevance of AI fitness applications extends beyond personal health management. Digital fitness solutions potentially reduce reliance on physical infrastructure, transportation, and printed materials, thereby contributing to more sustainable consumption practices [9,10]. AI-enabled fitness applications contribute to sustainable consumption behavior in ways that extend beyond the conventional notion of digital substitution [7,11,12]. While earlier digital health solutions primarily reduced the need for physical resources such as gym infrastructure or travel, AI-driven platforms actively shape user behavior through continuous interaction and adaptive feedback. By leveraging real-time data, these applications provide personalized recommendations, predictive insights, and context-aware nudges that encourage consistent engagement in health-related activities. This ongoing interaction supports the formation of long-term habits rather than one-time behavioral changes, thereby fostering more sustainable lifestyle patterns. Additionally, engaged users of AI fitness apps may act as advocates, spreading positive word of mouth within their social networks [11]. Despite the growing adoption of AI-enabled fitness technologies, existing research has not fully captured the unique mechanisms through which these systems influence user behavior. Studies grounded in traditional digital health or information systems frameworks often assume static system characteristics and focus primarily on usage or satisfaction outcomes [13,14]. Similarly, engagement research tends to emphasize user interaction without adequately considering how intelligent system features drive such engagement [11].

To examine these relationships, this study builds upon the Information Systems Success (ISS) model, one of the most widely recognized frameworks for evaluating system effectiveness. The ISS model traditionally proposes that system quality, information quality, and service quality determine system use and user satisfaction, which subsequently influence net benefits [15]. However, the emergence of AI-enabled systems necessitates theoretical refinement. AI systems are adaptive, interactive, and capable of simulating intelligent responses, thereby altering user perceptions and interactions.

Transforming the ISS model into an AI Information Systems Success framework enables a more context-sensitive examination of AI-specific attributes and their impact on user engagement within fitness applications. While the ISS model provides a foundational structure, it does not fully capture the behavioral and psychological dimensions that are central to interactions with AI systems. Therefore, insights from complementary theoretical perspectives are also relevant. For instance, the Technology Acceptance Model (TAM) highlights perceived usefulness and perceived ease of use as key determinants of technology adoption. In AI fitness contexts, system and information quality may indirectly shape perceived usefulness, while system usability reflects perceived ease of use [16,17]. Similarly, the Unified Theory of Acceptance and Use of Technology (UTAUT) highlights performance expectancy and effort expectancy as drivers of behavioral intention, reinforcing the importance of technical and informational attributes in AI-based systems [18]. The ISS model is more suitable for this study because it focuses specifically on evaluating system performance through core dimensions such as system quality, information quality, and service quality, which directly align with AI fitness application characteristics. Unlike TAM, and UTAUT, which primarily explain technology adoption intentions, ISS provides a more comprehensive framework for assessing post-adoption outcomes such as user engagement and behavioral impacts. Therefore, it offers a stronger foundation for examining how AI system attributes translate into engagement and sustainable consumption behavior in this context.

However, adoption-focused models such as TAM and UTAUT primarily explain initial acceptance rather than sustained engagement [19]. In contrast, customer engagement theory provides a relational perspective, emphasizing cognitive, emotional, and behavioral investment in ongoing interactions with a platform [20]. Within AI fitness applications, engagement can be reflected in regular app usage, emotional attachment to AI-driven coaching, active involvement in challenges, and the incorporation of recommendations into everyday routines. Engagement is particularly significant because it serves as a mechanism through which system qualities translate into behavioral outcomes, including word of mouth and sustainable consumption behavior [21,22].

Moreover, AI systems incorporate human-like interaction features that go beyond traditional system evaluation frameworks. In this context, perceived anthropomorphism—the extent to which users attribute human-like qualities to AI systems—emerges as a key determinant [23,24]. When users perceive AI fitness applications as conversational, empathetic, or coach-like, they may develop stronger emotional connections, enhancing engagement levels [25,26]. This relational dynamic is not explicitly addressed in the traditional ISS model but becomes increasingly relevant in AI-driven environments. Health consciousness, derived from health behavior theories, reflects the degree to which individuals are motivated to maintain and improve their well-being [27]. Health-conscious users may be more inclined to actively engage with AI fitness platforms that provide structured guidance and continuous monitoring [28]. Additionally, personal innovativeness, rooted in innovation diffusion theory, explains why some individuals are more willing to experiment with new technologies [29,30]. Highly innovative users may respond more positively to advanced AI features, strengthening the impact of perceived system, service, and information quality on engagement.

Although prior research has separately explored digital health applications, AI adoption, and customer engagement, important gaps remain in the literature. First, there is a limited integration of the Information Systems Success (ISS) model with engagement theory within the specific context of AI fitness applications. Second, sustainable consumption behavior as an outcome of AI-enabled service usage has not been sufficiently examined. Third, constructs such as anthropomorphism and health consciousness remain underexplored within AI system success frameworks, resulting in an incomplete understanding of the psychological drivers of engagement. To address these gaps, this study develops an AI Information Systems Success framework that integrates system-related quality dimensions with key psychological and relational constructs. Specifically, it investigates how perceived service quality, system quality, and information quality of AI fitness applications influence customer engagement, alongside the roles of health consciousness and perceived anthropomorphism.

By doing so, it extends existing perspectives by positioning AI fitness applications not only as technological tools but as intelligent, interactive systems capable of influencing behavioral outcomes. This approach offers a more comprehensive understanding of how digital health innovations can contribute to both user engagement and sustainability. It further investigates how customer engagement drives word of mouth and sustainable consumption behavior. Additionally, the moderating role of personal innovativeness is assessed to understand how individual differences shape engagement formation. Accordingly, this study seeks to answer the following research questions: RQ1: How do AI fitness app quality dimensions, health consciousness, and perceived anthropomorphism influence customer engagement? RQ2: How does customer engagement with AI fitness applications affect word of mouth and sustainable consumption behavior, and how is this relationship moderated by personal innovativeness?

By integrating the ISS model with engagement theory, technology acceptance perspectives, anthropomorphism insights, and innovation diffusion concepts, this study offers a comprehensive framework for understanding AI fitness applications. The findings are expected to contribute theoretically by extending system success models into AI-driven environments and practically by guiding developers and marketers in designing AI fitness platforms that foster engagement, advocacy, and sustainable behavioral outcomes.

2. Literature Review

2.1. AI and Information Systems Success (ISS) Model

Artificial intelligence (AI) has redefined modern information systems by introducing learning capabilities, predictive analytics, and adaptive decision-making [31]. Unlike traditional systems that primarily process and store information, AI-enabled systems dynamically personalize content and improve performance based on user interactions [32]. These intelligent features require a more nuanced evaluation of system effectiveness. The Information Systems Success (ISS) model provides a strong theoretical foundation for assessing system success through three key dimensions: system quality, information quality, and service quality [33]. In AI contexts, system quality reflects technical reliability and algorithmic efficiency; information quality captures accuracy, relevance, and personalization of AI-generated insights; and service quality represents responsive, adaptive support embedded within the system [34]. The Information Systems Success (ISS) model has been widely used to evaluate the effectiveness of digital systems through key dimensions such as system quality, information quality, and service quality [35,36]. Traditionally, this model has been applied in contexts where systems are relatively stable, predictable, and primarily designed to deliver information or services consistently [37,38]. Within such settings, system success is typically assessed through user satisfaction and usage behavior, assuming that users interact with systems in a largely functional and task-oriented way.

However, the emergence of AI-powered applications challenges several of these underlying assumptions. Unlike traditional systems, AI fitness applications are dynamic and adaptive, continuously learning from user interactions and modifying their outputs accordingly [7,9]. This means that system quality is no longer limited to technical performance or ease of use, but also includes the system’s ability to provide intelligent, personalized, and context-aware responses. Similarly, information quality extends beyond accuracy and relevance to encompass the degree of personalization and usefulness of insights generated through data-driven algorithms. Service quality, in turn, evolves from basic support functions to include interactive, real-time, and often human-like assistance delivered through AI-enabled interfaces. Given these developments, there is a need to reinterpret the ISS model within the context of intelligent systems. In AI environments, system quality dimensions play a more active role in shaping user experiences, not only by facilitating use but also by influencing ongoing engagement. Rather than viewing system success as a function of satisfaction and usage alone, it becomes important to consider how these systems foster deeper, continuous interaction with users. This is particularly relevant in AI fitness applications, where sustained engagement is critical for achieving long-term behavioral outcomes.

2.1.1. AI Fitness App Service Quality

Service quality, within the Information Systems Success framework, reflects the extent to which a system provides responsive, reliable, and user-oriented support [39,40]. In the context of AI fitness applications, service quality may extend beyond traditional customer support to include intelligent assistance, adaptive coaching responsiveness, real-time feedback, and personalized interaction. Service quality in AI-driven fitness applications may reflect the extent to which users receive timely, interactive, and supportive assistance through features such as virtual coaching and responsive feedback. This ongoing support enhances user motivation and accountability, thereby reinforcing long-term engagement and contributing to sustained behavioral change aligned with sustainability goals. Unlike static digital platforms, AI fitness apps simulate continuous service encounters through automated guidance, reminders, and performance adjustments [7]. When users perceive these AI-enabled services as supportive, dependable, and empathetic, they are more likely to develop a positive relational orientation toward the application [40]. Perceived service quality plays a critical role in shaping users’ willingness to interact consistently with the system [41]. High-quality AI service experiences can reduce uncertainty, enhance perceived value, and foster trust in algorithmic recommendations [42]. As a result, users are likely to devote greater cognitive and emotional resources to the platform. This heightened involvement enhances customer engagement, which is reflected in continuous interaction, active participation, and emotional attachment to the AI fitness application. Accordingly, higher perceived service quality of AI fitness apps is expected to lead to a significant increase in customer engagement. We propose:

H1a: 

Perceived AI fitness App service quality positively impacts customer engagement.

2.1.2. AI Fitness App System Quality

System quality refers to the technical performance and functional characteristics of an information system, including usability, reliability, efficiency, and accessibility [43]. In AI fitness applications, system quality may encompass the app’s interface design, ease of navigation, loading speed, data synchronization accuracy, and overall technical stability. System quality in AI-enabled fitness applications may extend beyond technical performance to include adaptability, responsiveness, and intelligent functionality. High system quality enables seamless interaction and real-time personalization, which supports continuous user engagement and facilitates the development of consistent, sustainability-oriented habits over time [44]. Because AI fitness apps are frequently used in dynamic and time-sensitive environments—such as during workouts—technical consistency becomes particularly important [45]. A system that functions smoothly and intuitively minimizes user frustration and enhances perceived control [46]. When users experience seamless integration between wearable devices, performance tracking tools, and AI-generated insights, they are more likely to perceive the application as dependable and professionally designed [47]. High system quality reduces cognitive effort and allows users to focus on their fitness goals rather than technical barriers [48]. This positive functional experience encourages repeated use and deeper interaction with app features. Within the ISS framework, superior system quality enhances user involvement with the platform. In AI fitness contexts, this involvement translates into greater behavioral participation, emotional connection, and sustained usage—key dimensions of customer engagement. Therefore, higher perceived AI fitness app system quality is expected to positively influence customer engagement. We propose:

H1b: 

Perceived AI fitness App system quality positively impacts customer engagement.

2.1.3. AI Fitness App Information Quality

Information quality represents the relevance, accuracy, timeliness, personalization, and clarity of content provided by an information system [15]. In AI fitness applications, information quality is particularly critical because users rely on AI-generated insights to guide physical activity, health decisions, and performance improvements. Information quality in AI fitness platforms maybe characterized by the accuracy, relevance, and personalization of insights generated from user data. By delivering tailored recommendations and actionable feedback, high-quality information encourages informed decision-making and promotes more mindful and sustainable health-related behaviors. Unlike generic fitness advice, AI-powered apps provide data-driven recommendations tailored to individual health metrics, preferences, and progress patterns [49]. When users perceive the information generated by the AI system as accurate, customized, and actionable, they are more likely to trust and value the application. High-quality information reduces ambiguity in fitness planning and strengthens users’ confidence in the app’s guidance [50]. Moreover, personalized insights can create a sense of being understood by the system, fostering a stronger psychological connection. Within the ISS model, information quality is a central determinant of system success because it shapes user perceptions of usefulness and credibility. In AI fitness contexts, superior information quality may encourage users to actively explore recommendations, track progress, and implement guidance in daily routines. This ongoing interaction enhances cognitive involvement and behavioral participation, thereby increasing customer engagement. We propose:

H1c: 

Perceived AI fitness App information quality positively impacts customer engagement.

2.2. Health Consciousness

Health consciousness reflects the degree to which individuals are aware of, concerned about, and actively involved in maintaining their health and well-being [11,51]. It represents a motivational orientation that shapes how individuals seek, interpret, and respond to health-related information [27,52]. In the context of AI fitness applications, health consciousness becomes particularly relevant because these platforms are designed to support preventive health behaviors, structured exercise routines, and data-driven lifestyle improvements. Individuals with high health consciousness tend to be proactive in monitoring their physical condition, tracking progress, and adopting technologies that enhance their well-being [27,53]. Unlike users who engage with fitness apps sporadically, health-conscious individuals are more likely to explore app features thoroughly, pay attention to AI-generated insights, and integrate recommendations into daily routines [8]. Their intrinsic motivation to maintain or improve health strengthens their willingness to interact consistently with the platform. AI fitness applications provide performance analytics, personalized guidance, and feedback loops that align closely with the needs of health-conscious users [7]. When users perceive the application as a valuable tool for achieving their health goals, they are more likely to invest cognitive attention, emotional energy, and behavioral effort, which are key dimensions of customer engagement. Accordingly, individuals with higher levels of health consciousness are expected to exhibit stronger engagement with AI fitness applications. We propose:

H2: 

Health consciousness positively impacts customer engagement.

2.3. Perceived Anthropomorphism

Perceived anthropomorphism refers to the degree to which users attribute human-like characteristics, intentions, or emotions to non-human entities, such as AI systems [54]. In the context of AI fitness applications, this perception is often shaped by design features such as conversational interfaces, human-like voices, virtual coaches, and personalized, emotionally responsive feedback [55]. These elements make the system appear more relatable and socially interactive, allowing users to engage with it as if it were a human partner rather than a purely technical tool. This perception plays an important role in shaping user experience and behavior. When users perceive an AI system as more human-like, they are more likely to develop trust, emotional connection, and a sense of companionship with the application [7]. This, in turn, enhances their willingness to interact with the system more frequently and to follow its recommendations more closely. As a result, perceived anthropomorphism can strengthen user engagement and increase the effectiveness of AI-driven guidance.

In the context of AI fitness applications, anthropomorphism manifests through interactive features like conversational interfaces, personalized coaching messages, empathetic feedback, or adaptive responses that simulate human behavior [55,56]. By rendering AI systems more relatable and socially responsive, anthropomorphism can significantly influence users’ emotional and cognitive evaluations, thereby shaping their overall engagement with the platform. When an AI fitness application exhibits behaviors such as understanding, encouragement, or personalized attention, users are more likely to perceive it as a social actor rather than a mere technological tool [7]. This perception fosters relational closeness, trust, and emotional attachment, all of which are critical components of customer engagement.

Anthropomorphism also complements health consciousness in shaping engagement. Health-conscious users may respond more positively to AI applications that provide personalized guidance with a “human touch,” as these features reinforce motivation and accountability [57]. Similarly, users are more likely to explore features, follow recommendations, and integrate AI guidance into their routines when they feel a psychological connection with the system [58]. In line with engagement theory, perceived anthropomorphism stimulates cognitive, emotional, and behavioral investment [59]. Users who consider AI systems as relatable and socially intelligent are more likely to devote attention, interact consistently, and participate actively within the platform [60]. We propose:

H3: 

Perceived anthropomorphism positively impacts customer engagement.

2.4. Customer Engagement

Customer engagement represents a multidimensional construct that captures the cognitive, emotional, and behavioral investment of users in their interactions with a platform [61]. In the context of AI fitness applications, customer engagement may emerge as a central mechanism linking perceived system characteristics—such as service quality, system quality, information quality, health consciousness, and anthropomorphism—to meaningful outcomes, including word of mouth and sustainable consumption behavior. Cognitively, engagement involves the attention and mental effort users allocate to interacting with AI fitness apps [62]. Emotionally, it encompasses feelings of attachment, enjoyment, and satisfaction derived from the use of the platform [20]. Behaviorally, engagement is reflected in repeated interaction, exploration of advanced features, adherence to AI-generated guidance, and participation in challenges or social functionalities within the application [63]. Engaged users are more likely to experience the benefits of personalized guidance, adaptive support, and human-like interaction, reinforcing their continued investment in the platform [64]. Through personalized feedback, adaptive goal-setting, and real-time progress tracking, AI fitness apps encourage users to remain consistently involved in their health routines. This sustained engagement strengthens users’ commitment and reinforces positive behavioral patterns over time.

The outcomes of customer engagement extend beyond personal health and app usage. One key behavioral outcome is word of mouth (WOM), which represents the voluntary dissemination of positive information about the application to others. Engaged users who perceive the AI fitness app as valuable, reliable, and supportive are more likely to share their experiences within social networks, contributing to organic promotion, brand advocacy, and community building [11]. Another important outcome is sustainable consumption behavior. Engaged users often internalize the guidance provided by AI fitness apps, leading to long-term health-conscious practices and lifestyle choices that reduce dependence on resource-intensive alternatives, such as gym visits, printed fitness materials, or excessive equipment use. By integrating AI-recommended routines and behaviors into daily life, engaged users not only achieve personal well-being but also contribute to environmentally and socially sustainable patterns of consumption [65].

In sum, customer engagement acts as a pivotal construct in AI fitness ecosystems. It transforms the technological and psychological qualities of the application into tangible outcomes, facilitating both social advocacy through word of mouth and responsible behavioral change through sustainable consumption. The depth of engagement reflects the strength of the user–system relationship and indicates the extent to which AI fitness platforms create value for both individual users and wider societal outcomes. Accordingly, the following hypotheses are proposed:

H4a: 

Customer engagement positively impacts word of mouth.

H4b: 

Customer engagement positively impacts sustainable consumption behavior.

2.5. Moderating Role of Personal Innovativeness

Personal innovativeness reflects an individual’s willingness to adopt, explore, and experiment with new technologies [30]. In the context of AI fitness applications, personal innovativeness plays a crucial role in determining how perceived system features, service quality, and information outputs translate into customer engagement. Highly innovative individuals are more likely to explore advanced app features, trust AI-generated guidance, and integrate recommendations into daily routines [66]. They perceive technological complexity as an opportunity rather than a barrier, which amplifies their engagement with AI fitness applications. Conversely, users with lower innovativeness may require simpler or more familiar interactions to engage effectively, even when app quality is high [67]. Users with high personal innovativeness are more likely to actively explore advanced AI features such as intelligent recommendations, adaptive coaching, and real-time feedback systems [29]. As a result, they tend to respond more strongly to improvements in system capabilities, particularly in terms of system quality, as they perceive such innovations as valuable and engaging rather than complex or risky. Figure 1 represents the conceptual framework of the study. It highlights independent and dependent variables, as well as the flow of moderation.

Accordingly, personal innovativeness strengthens the relationship between AI system features and user engagement by enhancing users’ openness to technological novelty and increasing their willingness to interact with the platform [29]. This moderating effect suggests that the impact of AI fitness application quality on engagement is not uniform across all users but is amplified among those who are more technologically innovative and receptive to change.

From a theoretical standpoint, personal innovativeness acts as a moderating factor within the Information Systems Success (ISS) framework. Although perceived service quality, system quality, and information quality provide the foundational drivers of engagement, the degree to which users actively engage with the platform differs based on their level of innovativeness. For example, users with higher innovativeness are likely to exhibit stronger engagement responses to high-quality services and personalized insights compared to less innovative users, thereby positioning personal innovativeness as a key boundary condition in the model [68]. Moreover, personal innovativeness can enhance the effects of other psychological factors, such as health consciousness and perceived anthropomorphism [69]. Users with a high tendency to experiment are more receptive to AI’s human-like cues and personalized health recommendations, reinforcing the impact of system qualities on engagement.

H5a: 

Personal Innovativeness moderates the relationship between perceived AI fitness app service quality and customer engagement.

H5b: 

Personal Innovativeness moderates the relationship between perceived AI fitness app system quality and customer engagement.

H5c: 

Personal Innovativeness moderates the relationship between perceived AI fitness app information quality and customer engagement.

3. Methodology

3.1. Research Design

This study adopted a quantitative research design to examine the effects of AI fitness application characteristics on customer engagement, word of mouth, and sustainable consumption behavior. A survey-based approach was employed, as it is suitable for collecting standardized data from a relatively large and diverse group of respondents and enables statistical analysis using structural equation modelling techniques. This design allows for testing the hypothesized relationships among constructs systematically and empirically. Reliability and validity were assessed using Cronbach’s alpha, composite reliability, factor loadings, and average variance extracted (AVE). Discriminant validity was checked with the Fornell–Larcker criterion. Hypotheses were tested using Partial Least Squares Structural Equation Modelling (PLS-SEM), suitable for predictive models with reflective constructs. Bootstrapping with 5000 resamples evaluated path coefficients and moderation effects [70]. This methodology ensures a rigorous and contextually relevant examination of how AI fitness app qualities, individual characteristics, and psychological factors influence customer engagement, word of mouth, and sustainable consumption behavior in Saudi Arabia. This analytical technique allows for the assessment of both direct and indirect effects among constructs, providing a comprehensive understanding of how AI fitness application characteristics influence customer engagement, word of mouth, and sustainable consumption behavior.

Table 1. Demographic Profile of Respondents.

Variable	Category	Frequency (n)	Percentage (%)
Gender	Male	270	59.5
	Female	184	40.5
Age	18–25	120	26.4
	26–35	200	44.1
	36–45	90	19.8
	46+	44	9.7
Education	High school or below	80	17.6
	Bachelor’s degree	290	63.9
	Postgraduate degree	84	18.5
Occupation	Student	90	19.8
	Employed	280	61.7
	Self-employed/Other	84	18.5
Income (SAR)	Less than 5000	120	26.4
	5000–10,000	200	44.1
	10,001–15,000	90	19.8
	Above 15,000	44	9.7

represents the demographic profiling of the respondents.

3.2. Study Context

The study was conducted in the Kingdom of Saudi Arabia, which provides a highly relevant and contemporary context for examining AI-driven fitness applications. Over the past decade, the country has undergone substantial digital transformation, supported by widespread internet access, high smartphone penetration, and strong investment in smart technologies [71,72]. These developments have created a favorable environment for the adoption of mobile health and fitness applications, particularly those enhanced with artificial intelligence features such as personalized training plans, real-time feedback, and behavioral tracking. In addition, Saudi Arabia has placed significant emphasis on health and wellness through national-level strategies and public initiatives aimed at improving physical activity levels and reducing lifestyle-related health risks [73,74]. Campaigns promoting active living, alongside large-scale programs encouraging fitness participation across different age groups, have contributed to growing public awareness of the importance of maintaining healthy lifestyles [73]. These contextual factors make the Kingdom an especially suitable setting for investigating user interaction with AI-enabled fitness platforms. As a result, the Saudi context offers valuable insights into emerging digital health trends and their behavioral implications in a rapidly evolving market environment.

3.3. Population and Sampling Procedure

The target population for this study consisted of adults residing in Saudi Arabia who have access to smartphones and represent potential users of AI fitness applications. A non-probability purposive sampling approach was adopted to ensure that participants had relevant exposure or interest in digital fitness technologies. This approach is appropriate for studies focusing on technology usage behavior, where identifying relevant users is more important than random selection. To enhance sample diversity, participants were approached through multiple online channels, including social media platforms, health and fitness forums, and e-mail distribution lists. This multi-channel approach helped ensure inclusion of respondents from different demographic backgrounds such as age, gender, occupation, and income levels.

3.4. Survey Instrument and Language

The questionnaire was designed based on validated measurement scales adapted from prior studies in the fields of information systems, digital health, and consumer behavior. All constructs were measured using a structured Likert scale to ensure consistency and comparability across responses. To ensure clarity and accuracy, the survey was initially developed in English and then translated into Arabic using a back-translation procedure. This process helped ensure that the meaning of the items remained consistent across both language versions. The instrument was further reviewed to ensure cultural appropriateness and ease of understanding for respondents in the Saudi context. A structured questionnaire was used to measure all constructs. Items were adapted from validated scales in prior studies, with modifications for AI fitness app contexts. All items were measured using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Perceived AI fitness app service, system, and information quality were adapted from [75]. The health consciousness variable was adapted from [76]. Perceived anthropomorphism was adapted from [25]. The customer engagement variable was adapted from [61,77]. The perceived innovativeness was adapted from [78]. The sustainable consumption behavior was adapted from [79], and word of mouth was adapted from [80]. The questionnaire items are available in the Appendix A.

3.5. Data Collection Process

Data were collected using an online self-administered questionnaire distributed electronically to potential respondents. A total of 650 questionnaires were disseminated to individuals who were considered relevant to the study context. Out of these, 479 responses were received, indicating a strong overall response rate and suggesting good participant engagement with the survey. Following data collection, the responses were carefully screened to ensure data quality and reliability. This process involved checking for missing values, identifying incomplete submissions, and removing responses that exhibited inconsistent or patterned answering behavior that could compromise the validity of the results. After this rigorous screening process, 454 valid responses were retained and used for the final statistical analysis. The data collection procedure was designed with ethical considerations in mind. Participation was entirely voluntary, and respondents were informed about the purpose of the study prior to completing the questionnaire. Measures were also taken to ensure anonymity and confidentiality, which helped to reduce social desirability and response bias. This ethical approach encouraged participants to provide honest and accurate responses, thereby enhancing the overall credibility and robustness of the dataset.

4. Data Analysis

4.1. Results and Analysis

In the Measurement Model,

Table 2. Factor Loadings and Variance Inflation Factor.

Items	Factor Loadings	VIF
CENG1	0.879	2.465
CENG2	0.877	2.461
CENG3	0.741	1.262
HCON1	0.828	2.466
HCON2	0.841	3.054
HCON3	0.806	2.336
HCON4	0.694	1.142
PAIINQ1	0.924	3.678
PAIINQ2	0.914	3.057
PAIINQ3	0.908	3.353
PAIINQ4	0.615	1.362
PAISQ1	0.785	1.210
PAISQ2	0.782	1.781
PAISQ3	0.831	1.887
PAISYQ1	0.930	3.848
PAISYQ2	0.926	3.471
PAISYQ3	0.918	3.328
PAISYQ4	0.771	1.986
PANTHM1	0.705	1.515
PANTHM2	0.824	1.765
PANTHM3	0.832	2.140
PANTHM4	0.825	2.105
PINNO1	0.826	2.276
PINNO2	0.818	2.007
PINNO3	0.687	3.066
PINNO4	0.748	3.422
SUSCB1	0.793	1.597
SUSCB2	0.875	1.912
SUSCB3	0.890	2.220
WOFM1	0.916	2.282
WOFM2	0.924	2.531
WOFM3	0.699	1.494

shows the factor loadings and the values of the variance inflation factor (VIF) for all reflective constructs in the model. These results provide a good indication of the reliability of the indicators and demonstrate that multicollinearity is not a problem. With the exception of a few items, all measurement items show loadings that greatly exceed 0.70, providing further indication that the indicators capture a large amount of variance in the latent constructs they measure [81]. These loadings are close to 0.70 for items such as HCON4, PAIINQ4, PINNO3, and WOFM3 but are above the 0.60 cutoff, and therefore justify retention of these items due to the contribution they make to the measure, as well as to the construct validity of the measure [70,82]. Variance Inflation Factor (VIF) is used to assess the multicollinearity between constructs. Values shown in Table 2 indicate that multicollinearity is not a problem, as all the VIF values are below the threshold of 5.

The results for all latent constructs’ internal consistency, reliability, convergent validity, and discriminant validity are presented in

Table 3. Reliability and HTMT Ratio.

Constructs	α	CR	AVE	1	2	3	4	5	6	7	8	9
Customer Engagement	0.778	0.873	0.697
Health Consciousness	0.809	0.872	0.631	0.708
Perceived AI Fitness App Information Quality	0.867	0.911	0.723	0.718	0.515
Perceived AI Fitness App System Quality	0.913	0.937	0.790	0.532	0.310	0.347
Perceived Ai fitness App Service Quality	0.725	0.842	0.640	0.633	0.502	0.369	0.353
Perceived Anthropomorphism	0.809	0.875	0.637	0.725	0.570	0.518	0.399	0.587
Personal Innovativeness	0.782	0.854	0.596	0.612	0.613	0.407	0.413	0.403	0.633
Sustainable Consumption Behavior	0.814	0.889	0.729	0.723	0.582	0.586	0.398	0.516	0.611	0.573
Word of Mouth	0.819	0.887	0.727	0.487	0.753	0.413	0.293	0.372	0.494	0.345	0.487

α = Cronbach’s alpha; CR = Composite reliability; AVE = Average variance extracted.

. Internal consistency was measured by Cronbach’s alpha (α) and composite reliability (CR). All constructs, based on the results, surpass the necessary 0.70 for both α and CR, thus confirming the internal consistency and construct reliability [83]. CR’s ranged from 0.842 to 0.937, which confirms the reliability and stability of the measurement scales even more. Convergent validity is assessed by the average variance extracted (AVE). All constructs are within the accepted range, which indicates that the constructs can explain more than half of the variance of their indicators [84] and thus confirms convergent validity for all constructs of the study, such as AI fitness app quality dimensions, customer engagement, health consciousness, anthropomorphism, personal innovativeness, and behavioral outcomes. The HTMT ratio of correlations was used to measure discriminant validity. All HTMT values are below potential thresholds of 0.85 and 0.9, which shows that the constructs are empirically distinct from one another [85]. This indicates that though customer engagement, anthropomorphism, and AI fitness app quality dimensions are closely related in theory, they are distinct from one another. The results indicate evidence of internal consistency reliability, convergent validity, and discriminant validity, which confirms that the measurement model is solid enough to proceed to structural model estimation and hypothesis testing.

4.2. Structural Model Results and Hypotheses Testing

For each proposed relationship, significance and effect strength were assessed using bootstrapping. Path coefficients, t-values, and significance levels are reported in

Table 4. Hypothesis Testing Bootstrapping @5000 subsamples.

		β	SD	t	p Values	Decision
H1a:	Perceived AI Fitness App Service Quality → Customer Engagement	0.136	0.045	2.999	0.003	Supported
H1b:	Perceived AI Fitness App System Quality → Customer Engagement	0.221	0.056	3.976	0.000	Supported
H1c:	Perceived AI Fitness App Information Quality → Customer Engagement	0.279	0.056	4.973	0.000	Supported
H2:	Health Consciousness → Customer Engagement	0.212	0.052	4.064	0.000	Supported
H3:	Perceived Anthropomorphism → Customer Engagement	0.149	0.055	2.698	0.007	Supported
H4a:	Customer Engagement → Word of Mouth	0.419	0.061	6.814	0.000	Supported
H4b:	Customer Engagement → Sustainable Consumption Behavior	0.581	0.063	9.279	0.000	Supported
H5a:	Personal Innovativeness x Perceived AI fitness App Service Quality → Customer Engagement	−0.005	0.047	0.111	0.911	Not Supported
H5b:	Personal Innovativeness x Perceived AI Fitness App System Quality → Customer Engagement	0.083	0.038	2.197	0.028	Supported
H5c:	Personal Innovativeness x Perceived AI Fitness App Information Quality → Customer Engagement	−0.069	0.057	1.206	0.228	Not Supported

. All three dimensions of quality showed significant positive effects on customer engagement, supporting H1a, H1b, and H1c. Among them, information quality had the strongest impact (β = 0.279, p < 0.001), followed by system quality (β = 0.221, p < 0.001) and service quality (β = 0.136, p < 0.01). Health consciousness also had a significant positive effect on customer engagement (β = 0.212, p < 0.001), supporting H2, indicating that individuals with higher health awareness are more likely to engage with AI-based fitness applications. Similarly, perceived anthropomorphism positively influenced customer engagement (β = 0.149, p < 0.01), supporting H3, suggesting that human-like features enhance emotional connection and sustained engagement.

Customer engagement, in turn, significantly predicted both word of mouth (β = 0.419, p < 0.001) and sustainable consumption behavior (β = 0.581, p < 0.001), supporting H4a and H4b. The stronger effect on sustainable consumption highlights engagement as a key psychological mechanism through which AI fitness apps encourage long-term, responsible health-related behavior beyond immediate usage. Regarding moderation effects, personal innovativeness significantly strengthened the relationship between system quality and engagement (β = 0.083, p < 0.05), supporting H5b. However, its moderating effects on service quality (H5a) and information quality (H5c) were not significant. This suggests that while innovative users respond more strongly to advanced system features, perceptions of service and information quality remain consistently important across user groups. Finally,

Table 5. Coefficients of Determination (R2).

Constructs	R-Square	R-Square Adjusted
Customer Engagement	0.615	0.607
Sustainable Consumption Behavior	0.337	0.335
Word of Mouth	0.174	0.172

presents the R² values, providing further insight into the explanatory power of the proposed model. Figure 2 represents measurement model pathways.

5. Discussion

The findings provide strong support for most of the proposed hypotheses and are broadly consistent with prior research in information systems, digital service quality, and technology-enabled engagement. First, the positive effects of perceived service quality, system quality, and information quality on customer engagement align with earlier studies based on the Information Systems Success Model [50,86]. Previous research has repeatedly shown that when users perceive a digital platform as reliable, responsive, and capable of delivering accurate and useful information, they are more likely to interact actively and develop stronger psychological involvement [87]. In the context of AI fitness applications, the results reinforce the argument that quality dimensions remain central drivers of user engagement even in advanced, intelligent systems [88]. The positive relationship between health consciousness and customer engagement also corresponds to the existing health behavior literature [89]. Earlier studies suggest that individuals who are more aware of and concerned about their well-being tend to adopt and consistently use health-related technologies [90]. The present findings reinforce this perspective by showing that health motivation significantly enhances users’ cognitive, emotional, and behavioral involvement with AI-based fitness applications.

Similarly, the significant influence of perceived anthropomorphism on engagement aligns with human–computer interaction research [91,92]. Prior studies have demonstrated that when digital systems display human-like characteristics, users tend to perceive them as more relatable and trustworthy, which enhances interaction and emotional connection [93]. The findings confirm that anthropomorphic cues in AI fitness apps can foster stronger engagement. Furthermore, the strong relationship between customer engagement and both word of mouth and sustainable consumption behavior supports engagement theory. Earlier research has established that engaged customers are more likely to recommend services and demonstrate loyal, responsible behaviors [94,95]. The results reaffirm engagement as a key mechanism translating perceptions into behavioral outcomes. Regarding moderation, the findings show that personal innovativeness strengthens the effect of system quality but does not significantly influence the relationships involving service and information quality. This partially aligns with innovation diffusion research, which suggests that innovative individuals are particularly responsive to advanced technical features, though not necessarily to all quality dimensions equally [96].

The findings of this study demonstrate that AI fitness applications play a meaningful role in promoting sustainable consumption behavior by encouraging users to adopt consistent and resource-efficient health practices. Through continuous interaction, personalized feedback, and adaptive guidance, these platforms support the development of long-term behavioral patterns that align with sustainability objectives. This study contributes to the literature by showing that sustainable consumption is not merely an indirect outcome of digital substitution, but is actively shaped through user engagement with AI-enabled features. In this regard, AI fitness applications function as behavioral enablers that guide users toward more mindful, disciplined, and efficient lifestyle choices. Furthermore, the results extend existing research by highlighting that sustainability in digital health contexts can be achieved through sustained engagement and intelligent system design. This positions AI-driven fitness platforms as important tools for fostering not only individual well-being but also broader sustainability outcomes.

5.1. Theoretical Implications

The study reinforces the relevance of the Information Systems Success Model in AI-powered fitness applications. Traditional applications of the model focused on general IT systems or e-commerce platforms, but this research demonstrates its applicability in intelligent, AI-driven environments where dynamic content, adaptive algorithms, and personalized user interactions are critical [7,97,98]. This highlights that quality dimensions remain central to user satisfaction and engagement even as technology evolves, supporting the theoretical proposition that perceived system effectiveness is a key determinant of behavioral outcomes. By showing that technological perceptions directly influence behavioral outcomes such as word of mouth and sustainable consumption through customer engagement, the study reinforces engagement theory in the context of digital health technologies. It highlights that engagement is an active process, where users’ interaction with AI fitness applications transforms their perceptions of technology quality into meaningful actions [21,99]. This supports the view that engagement involves both cognitive and emotional involvement, effectively linking user experiences with social and behavioral outcomes.

The study provides empirical support for the role of anthropomorphism in AI interactions, showing that human-like cues in AI fitness apps enhance user engagement. This aligns with theories of social presence and media equation, which posit that users treat computers or AI agents as social actors when they display human-like attributes [25,40]. The findings extend these theories by demonstrating their applicability in health-focused AI applications, highlighting how design features that mimic empathy, personalization, or conversational abilities can foster emotional connection and sustained engagement [100,101]. The partial moderating effect of personal innovativeness provides nuanced insights for innovation adoption literature. While previous studies often suggest that innovative individuals consistently respond more positively to new technologies, this research shows that innovativeness enhances engagement selectively—specifically with system quality but not with service or information quality [102,103]. This finding suggests that user traits interact with specific technological dimensions differently, refining theoretical models of adoption by emphasizing the conditional nature of personal innovativeness on technology acceptance and engagement.

5.2. Practical Implications

For AI fitness app developers, ensuring smooth functionality, fast response times, and reliable system performance is important for supporting consistent user interaction. From a theoretical perspective, this relates to the system quality dimension of the Information Systems Success Model, suggesting that technical reliability provides a foundation for user engagement. When system performance is weak, it may negatively affect user experience and reduce continued usage, whereas stable and efficient performance is more likely to support repeated interaction and sustained use. Providing accurate, relevant, and easy-to-understand information further contributes to user engagement by addressing users’ informational needs and supporting health-related decision-making. This aligns with the information quality dimension of the IS Success framework, indicating that well-structured and useful content can influence users’ engagement and encourage positive behavioral outcomes such as continued usage and word-of-mouth sharing.

In addition, incorporating anthropomorphic design elements such as conversational interfaces, empathetic feedback, and personalized encouragement may enhance users perceived social connection with the application. This is consistent with social presence and media equation perspectives, suggesting that human-like features can positively shape user experience and support ongoing interaction in AI-driven environments. From a managerial perspective, insights related to personal innovativeness suggest that users may respond differently to AI fitness applications depending on their openness to new technologies. Users with higher innovativeness may be more receptive to advanced system features, while others may rely more on perceived usefulness and clarity of information. This aligns with innovation adoption theory and highlights the importance of considering user heterogeneity when designing communication and engagement strategies.

6. Conclusions

This study provides robust evidence that the success of AI fitness applications is jointly determined by technological quality dimensions and user-specific characteristics. In particular, perceived service quality, system quality, and information quality, alongside individual factors such as health consciousness and perceived anthropomorphism, play a pivotal role in fostering customer engagement. The findings demonstrate that when users perceive AI fitness applications as reliable, informative, and human-like in their interactions, their level of psychological and behavioral engagement significantly increases.

Moreover, customer engagement emerges as a critical mechanism through which these antecedents translate into favorable behavioral outcomes. Specifically, higher levels of engagement lead to stronger word of mouth intentions and more sustainable consumption behaviors, highlighting the broader societal and environmental implications of AI-driven health technologies. The moderating analysis further reveals that personal innovativeness selectively strengthens the relationship between system quality and engagement, underscoring the nuanced influence of individual differences in shaping users’ responses to advanced digital systems.

Theoretically, this study advances and integrates key perspectives from Information Systems Success, customer engagement theory, human–computer interaction, and innovation adoption literature within the emerging domain of AI-enabled health applications. By doing so, it extends existing knowledge on how technological attributes and human factors interact to drive meaningful digital experiences and outcomes. From a practical standpoint, the findings emphasize the need for developers and practitioners to prioritize system reliability, seamless functionality, and the delivery of accurate, personalized, and up-to-date information. Additionally, incorporating anthropomorphic features that simulate human-like interaction can further enhance user connection and engagement. Importantly, the results highlight the value of segmenting users based on their level of innovativeness and tailoring strategies accordingly to maximize effectiveness. Overall, the study underscores that the alignment of technological excellence with user motivations and psychological drivers is essential for cultivating deep engagement, encouraging positive behavioral diffusion through word of mouth, and promoting sustained, responsible usage of AI fitness applications over time.

7. Limitations and Future Directions

This study has several limitations that should be acknowledged. First, the research is based on a cross-sectional design, which captures user perceptions at a single point in time. As a result, it does not fully capture changes in user engagement or sustainable consumption behavior over time. Future studies could adopt longitudinal designs to better examine how continued interaction with AI fitness applications influences behavioral outcomes in the long run. Second, the data used in this study is based on self-reported responses, which may introduce common method bias and social desirability bias. Although appropriate procedural and statistical measures were considered to minimize these issues, future research could incorporate objective usage data from AI fitness platforms or behavioral tracking metrics to improve accuracy and robustness. Third, the study focuses on a single-country context, which may limit the generalizability of the findings to other cultural or geographical settings. Since perceptions of technology and sustainability behavior may vary across regions, future research could replicate this model in different countries to enhance external validity and enable cross-cultural comparisons. Fourth, while this study focuses on key constructs such as system quality, information quality, service quality, engagement, and sustainable consumption behavior, other potentially relevant factors were not included. Future research could extend the model by incorporating additional variables such as trust in AI, perceived privacy risk, habit formation, or social influence to provide a more comprehensive understanding of user behavior in AI-driven environments.