As nutrition apps increasingly shape young consumers’ daily health management, sustaining customer satisfaction remains a challenge. This study aims to examine how perceived competence predominance relative to warmth influences satisfaction with nutrition apps and whether this effect is strengthened by self-focus framing.
Three popular nutrition apps, MyFitnessPal, Lose It! and Fitbit, were identified through a preliminary survey of 305 young consumers aged 18–26. A total of 146,661 user reviews of these apps were collected over a 30-month period. Using a data-mining approach, a balanced subsample of 18,000 reviews was analysed. Linguistic Inquiry and Word Count software was used to extract language-based constructs, followed by regression analyses to test the hypotheses.
Customer satisfaction is higher when nutrition apps are perceived as competence- predominant rather than warmth-predominant. This positive relationship is stronger among consumers who adopt a self-focus framing of their app experiences.
This research contributes to the literature on the stereotype content model and customer satisfaction in the context of consumer–technology-enabled health service interactions.
The findings provide practical guidance for designing and marketing nutrition apps targeting young consumers, indicating that emphasising competence-related cues and self-focused framing in apps can enhance customer satisfaction.
By integrating the stereotype content model with construal level theory, this research clarifies the role of social cognition and psychological framing in shaping satisfaction with nutrition apps.
1. Introduction
Mobile health applications (apps) have transformed how individuals manage health by providing accessible, data-driven tools to monitor diet, exercise, mental health and other health behaviours. Industry report notes that millennials dominate the market for mobile health apps, underscoring young consumers’ leading role in app adoption and engagement (Business of Apps, 2024). Among these technologies, nutrition apps warrant distinct scholarly attention because they target every day, habitual eating behaviours that require sustained self-regulation while offering limited immediate gratification (Charry et al., 2024). For young consumers aged 18–26, who are typically far from chronic disease onset, healthy eating often lacks perceived urgency, leading to declining motivation despite early adoption of nutrition apps (Huang and Lee, 2023; Nour et al., 2018). Consequently, although nutrition apps are designed to facilitate healthier eating behaviours, customer satisfaction frequently declines over time (Charry et al., 2024; Yin et al., 2022).
Research identifies satisfaction as a key challenge for sustaining engagement, highlighting the need to understand the factors that foster long-term satisfaction and strengthen young consumers’ relationships with nutrition apps (Charry et al., 2024; Kahriz et al., 2023). Understanding customer satisfaction with nutrition apps at this life stage is therefore particularly significant, as satisfaction underpins continued use and sustained engagement (Charry et al., 2024; Yin et al., 2022). Such insights can inform how these tools better support healthy eating among young consumers and guide app designers, educational institutions, families and public health stakeholders seeking scalable solutions aligned with World Health Organisation targets to reduce all forms of malnutrition by 2030 (World Health Organization, 2026).
To explore the factors influencing young consumers’ satisfaction with nutrition apps, previous studies have mainly focused on functional design features such as self-tracking and gamification, often overlooking the social cognition of these apps. To address this gap, the stereotype content model (SCM; Fiske et al., 2002) describes a key form of social cognition that explains how people evaluate both human and non-human entities along two universal dimensions: warmth and competence. Warmth reflects perceived intentions, including qualities such as friendliness, sincerity and empathy, whereas competence captures perceived capability and effectiveness, encompassing attributes such as intelligence, skill and reliability (Abele and Bruckmüller, 2013; Fiske et al., 2007). In health service contexts, this model helps explain how consumers perceive health service providers as competent (“gets it”) in demonstrating efficiency, knowledge and skill, and warm (“gets me”) in showing personal engagement, connection and care towards the consumer (Howe et al., 2019). As digital natives, young consumers perceive nutrition apps not merely as functional tools but as everyday health service providers that facilitate health-related decision-making (Marconi et al., 2024), which invites broader social-cognitive evaluations rather than only functional assessments. Importantly, when forming overall evaluations of nutrition apps, consumers may not weigh warmth and competence equally, indicating that the relative emphasis placed on these dimensions may shape satisfaction (McKee et al., 2024; Zheng et al., 2022). Accordingly, this study moves beyond identifying whether warmth or competence is present to examine how their relative salience influences young consumers’ satisfaction with nutrition apps.
Beyond the nutrition app stereotypes, individual differences in self-focused framing could also shape satisfaction with these apps. Drawing on construal level theory (Trope and Liberman, 2010), self-focused framing refers to the extent to which individuals process experiences through a self-referential lens. Consumers high in self-focus framing are more likely to interpret their interactions with nutrition apps from a personal and goal-oriented perspective, whereas those low in self-focus framing may perceive their app use as driven by external guidance or social influence that leads to beneficial outcomes (Wang and Yim, 2025). These different ways of interpreting app experiences may influence how strongly competence-warmth stereotypes shape satisfaction. However, existing research has rarely examined how individual-level framing tendencies condition the effects of stereotype-based app evaluations on satisfaction, particularly in digital health contexts involving young consumers. As a result, it remains unclear when relative competence- versus warmth-based app evaluations translate into higher or lower satisfaction. Thus, understanding how such framing tendencies interact with nutrition app stereotypes provides insight into the psychological processes underlying satisfaction with nutrition apps.
Against this backdrop, the present study investigates how young consumers evaluate nutrition apps based on the relative salience of competence and warmth, and how such evaluations influence customer satisfaction. Specifically, the study conceptualises this relative evaluation as perceived competence predominance in nutrition app stereotypes and examines how its effect on customer satisfaction is moderated by self-focused framing. To address these questions, the study collected 146,661 user reviews from the three most popular nutrition apps among young consumers, namely MyFitnessPal, Lose It! and Fitbit, spanning a 30-month period (March 2023 to August 2025). A data-mining approach was then applied to analyse a balanced subsample of 18,000 reviews, enabling systematic examination of the relationships among key constructs. This large-scale field data provides ecologically valid insights into young consumers’ genuine app experiences.
By integrating the SCM with construal level theory, this research aims to shed light on the underlying processes shaping young consumers’ satisfaction with nutrition apps. It contributes to the literature on the SCM and customer satisfaction within the context of consumer–technology-enabled health service interactions. The findings also offer managerial implications for the effective design and marketing of nutrition apps targeting young consumers.
2. Literature review
2.1 Stereotype content model in the health context
A stereotype refers to a simplified and generalised judgement that carries evaluative meaning (Aaker et al., 2010). Within the SCM, such judgments are understood through two principal dimensions: warmth and competence (Fiske et al., 2002). The warmth dimension reflects perceived intentions, encompassing traits such as friendliness, sincerity, and kindness. In contrast, competence pertains to perceived capability and effectiveness, characterised by attributes such as intelligence, skilfulness and efficiency (Fiske et al., 2007; Liao et al., 2023). Warmth and competence are central traits governing social judgments of individuals and groups, thereby influencing attitudes and behaviours.
For health consumers, the SCM has been widely applied to explain how individuals perceive health service providers through the dimensions of competence and warmth. For example, Howe et al. (2019) demonstrate that patients perceive a provider’s competence by judging whether the doctor “gets it”, reflecting expertise, accuracy and professional capability and warmth by judging whether the doctor “gets me,” reflecting empathy, care and interpersonal understanding. The model’s applicability extends beyond interpersonal settings, as many contemporary health services now operate as technology-enabled health service providers, including digital health-related brands (Kim and Magnini, 2020), technology-mediated healthcare services (Wang et al., 2022a), and AI as a source of health information (Wang et al., 2025). Despite this growing body of research, limited attention has been given to whether these stereotypes extend to mobile health apps, particularly nutrition apps, which have increasingly functioned as frontline health service providers for young consumers.
Importantly, the stereotype content model also allows for comparative judgments, whereby individuals evaluate whether competence or warmth is more salient in shaping overall impressions, rather than assessing each dimension in isolation. This comparative perspective is theoretically grounded in the warmth–competence trade-off, which suggests that emphasising one dimension often reduces the relative salience of the other (McKee et al., 2024; Zheng et al., 2022). Such relative evaluations are especially likely for technology-enabled health service providers, where interactions are mediated through functional features and interface cues rather than rich interpersonal signals. Consequently, mobile health apps, including nutrition apps, invite comparative judgments about whether competence or warmth is more salient in users’ evaluations, providing a theoretical basis for examining how such relative perceptions shape consumer responses in this context.
2.2 Nutrition app stereotypes and customer satisfaction
As a specific type of technology-enabled health service provider, nutrition apps differ from interpersonal health services in that users primarily infer service qualities from system design, functional outputs and interface cues rather than direct human interaction. Nutrition apps can be regarded as health service providers that aim to convey both competence and warmth. Consumers can perceive warmth through social and emotional design features. Social design features such as in-app communities, teamwork challenges and leaderboards help create a sense of belonging and shared purpose (Chung et al., 2017; Slazus et al., 2022). Similarly, encouragement messages or personalised recommendations can make consumers perceive the app’s friendly intent, feel understood and foster the impression that the app “cares” about their progress (Ahn and Park, 2023; Zhai et al., 2022). Literature suggests that perceived warmth can increase consumers’ positive attitudes towards the app, such as customer satisfaction.
Customer satisfaction refers to the consumer’s emotional reaction to the perceived difference between app performance and expectations (Hennig-Thurau et al., 2002) and is often used to evaluate the impact of app design (Yin et al., 2022). However, empirical findings on the influences of perceived warmth on satisfaction are mixed. While some studies suggest that perceived warmth is a key factor for satisfaction (Ahn and Park, 2023), other evidence shows that perceived warmth can have a negative effect on satisfaction, as it may divert consumers from their original purpose of using nutrition apps to achieve health goals (Wang et al., 2022b) or even cause frustration (Sun et al., 2025).
In contrast, perceived competence reflects consumers’ assessment of the app’s capability to help them achieve their health goals effectively. Nutrition apps convey perceived competence through functional and informational design features such as goal tracking, practical health knowledge and structured guidance (Lu et al., 2018; Samoggia and Riedel, 2020). Prior studies have shown that perceived competence can enhance customer satisfaction, as it reinforces consumers’ belief that the app facilitates goal attainment and strengthens their self-efficacy in managing health goals (Charry et al., 2024; Sun et al., 2025). The impact of perceived competence on satisfaction could be particularly important among young consumers, who often experience difficulties in health goal management due to limited healthy eating capability (Nour et al., 2018; Tsai et al., 2022).
Consistent with the warmth–competence trade-off established in the broader health literature, both competence and warmth remain relevant attributes in health apps. In nutrition app contexts, however, this trade-off becomes practically consequential, as emphasising one dimension can reduce the salience of the other in users’ evaluations, shaping whether the app is perceived as more effective or more socially engaging in supporting health objectives. Building on this perspective, research on technology-enabled health service providers suggests that satisfaction judgments may be asymmetric rather than balanced (McKee et al., 2024; Zheng et al., 2022). When consumers engage with nutrition apps primarily to manage their health, competence-related cues that signal effectiveness, accuracy and capability are likely to carry greater evaluative weight than warmth-related cues. This relative emphasis provides a theoretical basis for expecting that satisfaction is shaped not by competence or warmth in isolation, but by which dimension predominates in users’ perceptions of the app.
In line with this reasoning, research on technology-enabled health service providers suggests that satisfaction tends to depend more on perceived competence than on perceived warmth (Deng and Yan, 2025; Samoggia and Riedel, 2020). This tendency, referred to as perceived competence predominance, indicates that young consumers are more likely to experience greater satisfaction when nutrition apps demonstrate higher perceived competence than warmth. Accordingly, the following hypothesis is proposed:
Perceived competence predominance of nutrition apps is positively associated with customer satisfaction.
2.3 Self-focus framing as a moderator
Although perceived competence predominance is expected to positively associate with customer satisfaction, this effect may depend on individual differences in how consumers frame their app experiences, particularly through self-focused framing. Drawing on construal level theory (Trope and Liberman, 2010), individuals with high self-focused framing process information at a lower construal level, emphasising concrete, proximal and personally relevant details. In contrast, those low in self-focus adopt a higher construal level, focusing on abstract, collective or socially oriented meanings (Giles and Ogay, 2007). Prior research in health consumption contexts demonstrates that such framing differences systematically shape how individuals evaluate health-related services and value (Ervas, 2025; Gelbrich et al., 2021).
In the context of nutrition apps, young consumers with high self-focused framing tend to evaluate their experiences through a self-referential lens, focusing on personal growth, capability development and progress towards health goals (Wang and Yim, 2025). They are more likely to highlight individual benefits such as enhanced skills, improved appearance or a sense of personal improvement. Extant research suggests that when consumption experiences are framed around self-focus, competence-related cues become especially diagnostic for satisfaction judgments (Ervas, 2025; Shin et al., 2020). When these consumers perceive nutrition apps as high in perceived competence predominance, they interpret their experiences as directly advancing their health goal management and self-regulation, thereby experiencing greater satisfaction with the apps.
Conversely, young consumers with low self-focused framing view their app experiences from a more collective or other-focused perspective, prioritising social connection and shared outcomes (Giles and Ogay, 2007). For these consumers, perceived competence predominance may be less salient, as it does not align with their focus on social belonging or mutual encouragement. Research indicates that consumers with a lower self-focus place greater weight on interpersonal warmth-related cues when forming evaluations, rather than competence alone (Gelbrich et al., 2021; Smorti et al., 2022). Consequently, low self-focused framing is less likely to strengthen the positive relationship between perceived competence predominance of nutrition apps and customer satisfaction among this group.
Building on the above reasoning, self-focused framing offers a meaningful lens for understanding when competence predominance in nutrition apps becomes especially influential for young consumers’ satisfaction, as consumption in such contexts is strongly oriented towards personal capability and self-improvement (Nour et al., 2018; Tsai et al., 2022). Consistent with this logic, prior research shows that self-focused framing enhances the positive association between perceived competence predominance of nutrition apps and consumers’ emotional responses towards such apps (Berger and Jung, 2024; Charry et al., 2024). Accordingly, the following hypothesis is proposed:
Self-focus framing strengthens the positive effect of perceived competence predominance of nutrition apps on customer satisfaction.
3. Methods
This study investigates how users’ perceptions of competence versus warmth stereotypes in nutrition apps and their self-focus affect customer satisfaction. To identify the nutrition apps most popular among young consumers, we collected 305 valid survey responses via Prolific (50.5% female; Mage= 22.757, SD = 2.261) from US participants aged 18–26 who had used a nutrition app at least once in the past month on Prolific. Data was collected in December 2023. Respondents indicated the primary nutrition app they had used in the past month and then completed demographic questions. Ethics approval was obtained in September 2022.
The three most frequently reported apps were MyFitnessPal (33.8%), Lose It! (20.3%), and Fitbit (14.4%), which together accounted for 68.3% of usage. Based on these results, we retrieved user reviews of these apps from the Apple App Store and Google Play over a 30-month period (March 2023–August 2025), yielding a data set of 146,661 reviews. Online reviews were used to capture young consumers’ perceptions and behaviours, as this form of field data reflects actual rating behaviour that indicates satisfaction rather than self-reported attitudes. Moreover, young consumers constitute the most active segment in online reviewing across age groups (D'Souza, 2024). Therefore, this field data provides ecologically valid insights into young consumers’ genuine app experiences. Ethics approval for using field data was obtained in May 2025.
To avoid bias towards specific applications and ensure balanced representation across app types, we adopted the sampling strategy of Yi et al. (2025), randomly selecting 100 reviews per app per month from each platform, Google and Apple. This balanced sampling strategy was specifically adopted to minimise the influence of app- and platform-level effects that could otherwise introduce unexplained variability into the analysis. By drawing an equal number of reviews across apps, platforms and time periods, this approach reduces the likelihood that the observed results are disproportionately driven by any single app or platform, while remaining aligned with the review-level unit of analysis. This approach produced a balanced data set of 18,000 reviews (3,000 per app from each platform across 30 months) for analysis.
3.1 Construct measures
The language-based constructs were derived using the Linguistic Inquiry and Word Count (LIWC) software (LIWC, 2022; Pennebaker et al., 2022). LIWC analyses text by matching words against predefined dictionaries and calculating the proportion of words that fall within specified categories, producing indices that range from 0 to 100 (Boyd et al., 2022). This tool has been widely applied in text analysis, including studies examining perceived competence predominance of app (e.g. Hoang et al., 2023) and self-focus framing of app experiences (e.g. Mulcahy et al., 2024), and provides validated measures suitable for the constructs examined in this study.
3.1.1 Customer satisfaction.
Customer satisfaction was operationalised using review ratings, measured as the user-generated star ratings in the Apple App Store and Google Play. Ratings range from 1 (lowest) to 5 (highest), reflecting users’ overall satisfaction with the app experience. This approach is consistent with prior research using review data, where star ratings have been commonly used to capture customer satisfaction (Xu, 2020; Yousaf and Kim, 2023).
3.1.2 Perceived competence predominance of app.
Following prior research (Hoang et al., 2023), this study assessed perceived competence predominance by integrating linguistic indicators of competence and warmth with the valence of consumer online reviews. Competence and warmth were measured using the agency and communion dictionaries from LIWC (Pietraszkiewicz et al., 2019), which capture two fundamental dimensions of social cognition that shape how individuals evaluate both people and products (Abele and Bruckmüller, 2013; Louvet et al., 2019). The relative emphasis on perceived competence predominance was calculated by subtracting communion scores from agency scores, producing a competence–warmth differential. Higher levels of agency-related language relative to communion-related language suggest higher perceived competence predominance, consistent with the warmth–competence trade-off identified in social cognition literature (McKee et al., 2024; Zheng et al., 2022).
However, this differential alone does not capture the valence, which refers to whether the expressed evaluation is positive or negative. For example, the phrases “a highly competent app” and “an awfully incompetent app” contain similar competence-related terms but convey opposite valence. In naturalistic review data, agency-related language often appears in evaluative or complaint-oriented contexts, such that competence–warmth differentials may reflect critical appraisal rather than affirmative competence endorsement (Hoang et al., 2023). To account for this, the positive tone and negative tone variables from LIWC were used to measure the overall valence of each review (Boyd et al., 2022). A valence score was computed by subtracting negative tone from positive tone and the two components were multiplied to yield an integrated measure of perceived competence: (Agency − Communion) × (Positive tone − Negative tone).
Consistent with Hoang et al. (2023), this multiplicative formulation treats emotional tone as an implicit conditioning factor that situates competence-related language within its evaluative context, rather than as a standalone construct. By conditioning competence–warmth differentials on evaluative orientation, this approach improves interpretive precision by distinguishing endorsed competence from competence referenced in evaluative critique, which is common in unsolicited online reviews and cannot be reliably separated using category frequencies alone (Boyd et al., 2022). This composite score captures both the extent to which competence outweighs warmth and the valence through which it is expressed, providing a more accurate representation of how users linguistically convey their perceptions of app competence.
3.1.3 Self-focus framing of app experiences.
Self-focus framing was measured using the first-person singular pronoun category (“I” code) from the LIWC software (Boyd et al., 2022). This category captures the frequency of words such as “I” and “me”, indicating the extent to which users describe their app experiences from a self-referential perspective. The frequent use of first-person pronouns is recognised as a linguistic marker of self-identity and self-expression (Pennebaker et al., 2003). Conversely, lower use of such pronouns reflects a greater sense of shared identity, community and commonality (Giles and Ogay, 2007).
3.1.4 Control variables.
Consistent with prior research using online review data, two control variables, namely review length and positive tone, were included to minimise potential confounding effects. Review length was measured as the total word count using LIWC, representing the amount of information contained in each review. Longer reviews often include more detailed reasoning and elaborate descriptions, which may influence app evaluations (Mulcahy et al., 2024). Controlling for word count ensures that the observed relationships are not driven by differences in the level of detail across reviews (Pham et al., 2023).
The positive tone of each review was measured from LIWC (Boyd et al., 2022), which captures the proportion of words expressing positive affect, such as “happy”, “love”, or “good”. This variable reflects the overall positivity embedded in the reviewer’s language. Controlling for positive tone accounts for individual differences in emotional expressiveness, ensuring that the hypothesis tests are not confounded by general positivity in linguistic style (Borghi and Ratcharak, 2025).
3.2 Empirical models
To test the proposed hypotheses, the following linear regression model was estimated:
where denotes the error term, capturing unobserved factors that could influence customer satisfaction. The coefficients () indicate the expected change in customer satisfaction corresponding to a one-unit increase in each independent variable, with other variables held constant. The model examines the effect of perceived competence predominance on customer satisfaction and tests whether self-focus framing of app experiences moderates this relationship.
Table 1 presents the descriptive statistics. The mean value of perceived competence predominance (−8.83) is negative, indicating that nutrition apps tend to be described with more warmth-related language than competence-related language. The average customer satisfaction is 3.00 out of 5, suggesting that reviews are generally positive. The mean of positive tone (8.26) also reflects an overall upbeat emotional style across reviews. Table 1 further reports the correlation matrix, showing that none of the correlation coefficients exceed the 0.70 threshold (Ratner, 2009). The variance inflation factor (VIF) values, ranging from 1.03–1.42, remain well below the accepted threshold of 10 (Hair et al., 2018), suggesting the absence of multicollinearity concerns within this data set.
Descriptive statistics and Pearson correlation coefficient of key variables
| Variable | M | SD | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|---|
| 1. Customer satisfaction | 3.00 | 1.71 | ||||
| 2. Perceived competence predominance | –8.83 | 341.27 | –0.03*** | |||
| 3. Self-focus framing | 5.79 | 6.16 | 0.17*** | –0.02* | ||
| 4. Word count | 38.12 | 38.27 | –0.16*** | 0.04*** | 0.08*** | |
| 5. Positive tone | 8.26 | 13.85 | 0.38*** | –0.16*** | –0.11*** | –0.27*** |
| Variable | M | 1 | 2 | 3 | 4 | |
|---|---|---|---|---|---|---|
| 1. Customer satisfaction | 3.00 | 1.71 | ||||
| 2. Perceived competence predominance | –8.83 | 341.27 | –0.03 | |||
| 3. Self-focus framing | 5.79 | 6.16 | 0.17 | –0.02 | ||
| 4. Word count | 38.12 | 38.27 | –0.16 | 0.04 | 0.08 | |
| 5. Positive tone | 8.26 | 13.85 | 0.38 | –0.16 | –0.11 | –0.27 |
*p < 0.05; **p < 0.01; ***p < 0.001
3.3 Results
Table 2 presents the empirical results of the regression analysis. Model 1 serves as the baseline specification, including only the control variables log(word count) and positive tone. The results show that log(word count) is negatively and significantly associated with customer satisfaction (α = −0.09, SE = 0.01, p < 0.001), indicating that longer reviews tend to be associated with lower customer satisfaction. In contrast, positive tone is positively and significantly associated with customer satisfaction (α = 0.04, SE = 0.001, p < 0.001), suggesting that reviews written in a more positive emotional style are likely to yield higher customer satisfaction.
Linear regression results
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variables | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. |
| Constant | 2.91*** | (0.05) | 2.90*** | (0.05) | 2.70*** | (0.05) |
| Controls | ||||||
| Log (wordcount) | –0.09*** | (0.01) | –0.09*** | (0.01) | –0.14*** | (0.01) |
| Positive tone | 0.04*** | (0.001) | 0.04*** | (0.001) | 0.05*** | (0.001) |
| Main effect | ||||||
| Competence | 0.0002*** | (0.00003) | 0.0002*** | (0.00003) | ||
| Interaction | ||||||
| Competence × self-focus | 0.00002** | (0.000007) | ||||
| R2 | 0.15 | 0.15 | 0.20 | |||
| Adjusted R2 | 0.15 | 0.15 | 0.20 | |||
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variables | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. |
| Constant | 2.91 | (0.05) | 2.90 | (0.05) | 2.70 | (0.05) |
| Controls | ||||||
| Log (wordcount) | –0.09 | (0.01) | –0.09 | (0.01) | –0.14 | (0.01) |
| Positive tone | 0.04 | (0.001) | 0.04 | (0.001) | 0.05 | (0.001) |
| Main effect | ||||||
| Competence | 0.0002 | (0.00003) | 0.0002 | (0.00003) | ||
| Interaction | ||||||
| Competence × self-focus | 0.00002 | (0.000007) | ||||
| R2 | 0.15 | 0.15 | 0.20 | |||
| Adjusted R2 | 0.15 | 0.15 | 0.20 | |||
*p < 0.05; **p < 0.01; ***p < 0.001
Model 2 extends the baseline by adding perceived competence predominance as a predictor. The results indicate that perceived competence predominance is positively and significantly associated with customer satisfaction (α = 0.0002, SE = 0.00003, p < 0.001). Thus, H1 is supported.
Model 3 introduces the interaction term between perceived competence predominance and self-focus framing. The results show that perceived competence predominance remains positively and significantly associated with customer satisfaction (α = 0.0002, SE = 0.00003, p < 0.001). Moreover, the interaction between perceived competence predominance and self-focus framing is positive and significant (α = 0.00002, SE = 0.000007, p < 0.001). Thus, H2 is supported. Overall, Model 3 explains 20% of the variance in customer satisfaction (R2 = 0.20, Adjusted R2 = 0.20), representing an improvement over Models 1 and 2. The results of robustness checks (Appendix Table A1) confirm that coefficient patterns remain stable and consistent with the main results, supporting the reliability of the findings.
4. General discussion
Grounded in the SCM and construal level theory, this research examined how perceived competence–warmth stereotypes of nutrition apps and consumers’ self-focus framing of app experiences shape customer satisfaction. Using large-scale review data from young consumers’ top three nutrition apps, namely MyFitnessPal, Lose It! and Fitbit, we found that customer satisfaction increased when the apps were described with higher perceived competence predominance, that is, when competence-related language outweighed warmth-related language in online reviews. Moreover, this positive effect was amplified among reviews showing stronger self-focused framing, indicating that consumers who interpreted their app experiences in a self-referential manner derived greater satisfaction from nutrition apps perceived as more competence-predominant. These findings were robust across multiple estimation techniques, confirming the stability and reliability of the observed effects.
4.1 Theoretical implications
This research makes several theoretical contributions. Firstly, this study extends the SCM (Fiske et al., 2002) to the technology-enabled health service context by demonstrating that consumers perceive both competence and warmth in mobile health apps such as nutrition apps. While prior research has predominantly examined these stereotypes in interpersonal health service providers, such as doctors and other medical professionals (e.g. Howe et al., 2019; Zhang and Zhang, 2021), the present findings reveal that consumers also ascribe these social-cognitive dimensions of competence and warmth to non-human, technology-enabled service providers. This suggests that consumers evaluate technology-enabled health services using the same social cognition dimensions that typically guide judgments of human agents. This study advances understanding by demonstrating that the SCM can be meaningfully applied to human–technology interactions in the mobile health context.
Secondly, the findings advance understanding of the competence–warmth trade-off in the health service context. Specifically, this study moves beyond treating competence and warmth as parallel attributes by demonstrating that their relative predominance constitutes a meaningful evaluative mechanism shaping customer satisfaction. Inconsistent with some studies that emphasise providing both competent and warm services to enhance customer satisfaction (Howe et al., 2019; Zhang and Zhang, 2021), this study suggests that prioritising competence predominance is more effective in increasing satisfaction. This emphasis aligns with prior research indicating that health consumers often face imperfect information and information asymmetry between service providers and consumers, as they typically lack the specialised knowledge required to make informed health decisions (Arrow, 2012). Consequently, highlighting competence predominance becomes essential for strengthening consumers’ confidence in service quality. The finding that competence predominance drives satisfaction is also consistent with research on mobile health apps for young consumers in mental health, fitness and health management (Beltzer et al., 2023; Grech et al., 2024; Sun et al., 2025). For young consumers, who are in early adulthood and beginning to make independent health decisions, competence predominance is particularly valued because it enhances confidence and reduces uncertainty (Nour et al., 2018). Overall, the results suggest that young consumers primarily use nutrition apps for effective health goal management, where perceiving the app as capable and reliable contributes more to satisfaction than perceiving it as friendly or empathetic.
Thirdly, this study contributes to the consumer satisfaction literature by integrating the SCM (Fiske et al., 2002) and construal-level theory (Trope and Liberman, 2010) to explain satisfaction with mobile health services. By introducing self-focus framing as a moderator, the study reveals that consumers’ self-referential framing determines whether competence predominance is interpreted as emotionally positive (i.e. customer satisfaction) towards mobile health apps. Consistent with prior research grounded in self-determination theory and goal orientation theory, consumers who adopt a more self-focused perspective, emphasising personal growth, goal progress, capability development and self-improvement, are more likely to experience satisfaction with health service providers (Beltzer et al., 2023; Grech et al., 2024; Rockmann and Gewald, 2019). This finding complements prior studies on satisfaction with mobile health services, which have primarily examined moderators related to app functions, such as goal customisation (Fronczek et al., 2023) and gamification (Six et al., 2021), but have seldom considered individual differences such as self-focus framing. Thus, this study advances understanding of how individual differences in self-focus framing enhance the positive relationship between competence predominance and consumer satisfaction in mobile health contexts.
4.2 Marketing implications
The findings of this study provide actionable insights for those developing and marketing nutrition apps targeting young consumers. For developers, nutrition apps designed to promote healthy eating, balanced diets or weight management should prioritise competence predominance in functionality. Features such as structured guidance, goal-tracking dashboards, progress visualisations and data accuracy summaries can enhance young consumers’ perceptions of capability and control over their health goals (Grech et al., 2024; Sun et al., 2025). Integrating self-focused communication can further strengthen this effect. Personalised dashboards showing progress (e.g. “Your weekly results show steady improvement”) or achievement messages (e.g. “You reached your target intake today”) can encourage consumers to view their experiences from a self-referential perspective, leading to greater satisfaction with the nutrition app.
For marketing and platform managers, the findings suggest strategies to increase the visibility and appeal of nutrition apps among young consumers. Marketing communications introducing new apps or features should highlight competence-enhancing elements (e.g. adaptive challenges, structured guidance) and self-focused framing (e.g. customised services, individual recommendations). Campaigns can use competence-oriented and self-focused language such as “Achieve your own goals faster”, “Track your progress and see results” or “Take control of your health journey”. Platform managers, such as those overseeing app store promotions, can enhance visibility by prioritising nutrition apps that demonstrate competence-supportive features or include consumer testimonials reflecting individual experiences. Framing app value around competence and self-focus can therefore cultivate stronger satisfaction among young consumers.
For young consumers, the findings provide guidance for making informed app choices and maximising usage benefits. When selecting nutrition apps from app platforms or ranking systems, they should look for user reviews emphasising competence-related and self-focused language. Examples include comments describing personal experiences of capability building, such as “I became more confident in managing my diet” or “This app helped me build better eating habits”. Apps with such language are more likely to deliver satisfying experiences. During usage, young consumers can prioritise competence-enhancing functions such as goal tracking and adaptive challenges (Charry et al., 2024). Regular engagement and provision of personal data can also help the app learn from individual eating behaviours, generating more personalised and self-focused recommendations (Cheng and Ebrahimi, 2023).
5. Limitations and future research
This study has several limitations that open avenues for future research. Firstly, the online review data for these apps may also include inputs from other demographic segments. Although this diversity enhances the ecological validity of the findings by reflecting how young consumers interact with apps within a broader market environment, future research could complement this field-based study with survey or experimental methods specifically targeting young consumers to further validate these insights. Secondly, this study focuses solely on nutrition apps, as young consumers often display low motivation to maintain healthy eating behaviours (Nour et al., 2018). Similar motivational tendencies may also be observed in other health-related domains, such as physical activity, mental wellbeing and overall lifestyle management, partly because young consumers typically perceive themselves as physically resilient and less vulnerable to health risks (Bibbins-Domingo and Burroughs Peña, 2010). Therefore, future research could extend these findings to other health app categories to examine the generalisability of the results. Thirdly, while this study adopts a balanced sampling strategy to reduce the influence of app- and platform-level effects, future research could explicitly model such effects using multilevel designs or platform-specific analyses. For example, hierarchical modelling could be used to disentangle review-level linguistic patterns from app- or platform-level characteristics, allowing for a more fine-grained examination of cross-platform or cross-app variability (Ren et al., 2023). Fourth, although the model explains a modest proportion of variance (20%), this level of explanatory power is consistent with prior studies using naturalistic online review data (e.g. Brandes and Dover, 2022; Yi et al., 2025), reflecting the influence of unobserved factors. Future research could enhance explanatory power by incorporating additional variables, such as individual differences. Finally, as self-focused framing plays a central role in this study, future research could explore alternative framing strategies, such as gain or loss framing and examine how these interact with app stereotypes.
6. Conclusion
By integrating the Stereotype Content Model with construal level theory, this field-based study demonstrates that customer satisfaction with nutrition apps reflects stereotypical perceptions grounded in social cognition and consumers’ self-focused framing of their app experiences. Collectively, these findings advance theoretical understanding of the Stereotype Content Model and consumer satisfaction literature and provide practical guidance for marketers in designing and promoting nutrition apps that target young consumers.
References
Appendix
Table A1 presents the results of the robustness checks conducted to verify the stability of the estimated models. Four alternative estimation techniques were used: (1) heteroskedasticity-consistent standard errors, which correct for potential heteroskedasticity in the residuals; (2) quantile regression (median regression, τ = 0.5), which assesses whether the relationships hold across the conditional distribution of customer satisfaction; (3) bootstrap standard errors (1,000 replications), which provide bias-corrected estimates through repeated resampling; and (4) robust regression using the M-estimator, which down-weights the influence of outliers. Across all robustness checks, the direction, magnitude, and significance of the coefficients remain consistent with the main findings, confirming the reliability of the estimated effects.
Robustness check results
| Heteroskedasticity-consistent standard errors | Quantile regression (median regression, τ = 0.5) | Bootstrap Standard Errors (1,000 replications) | Robust regression (M-estimator, down-weights outliers) | |||||
|---|---|---|---|---|---|---|---|---|
| Variables | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. |
| Constant | 2.70*** | (0.05) | 2.24*** | (0.07) | 2.70*** | (0.05) | 2.27*** | (0.07) |
| Controls | ||||||||
| Log (wordcount) | –0.14*** | (0.01) | –0.15*** | (0.02) | –0.14*** | (0.01) | –0.13*** | (0.01) |
| Positive tone | 0.05*** | (0.001) | 0.07*** | (0.003) | 0.05*** | (0.001) | 0.07*** | (0.002) |
| Main effect | ||||||||
| Competence | 0.0002*** | (0.00004) | 0.0003*** | (0.00009) | 0.0002*** | (0.00004) | 0.0003*** | (0.00005) |
| Interaction | ||||||||
| Competence × self-focus | 0.00002*** | (0.000005) | 0.00006** | (0.00002) | 0.00002*** | (0.000005) | 0.00005*** | (0.000009) |
| R2 | 0.20 | – | – | – | ||||
| Adjusted R2 | 0.20 | – | – | – | ||||
| Heteroskedasticity-consistent standard errors | Quantile regression (median regression, τ = 0.5) | Bootstrap Standard Errors (1,000 replications) | Robust regression (M-estimator, down-weights outliers) | |||||
|---|---|---|---|---|---|---|---|---|
| Variables | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. | Estimate | S.E. |
| Constant | 2.70 | (0.05) | 2.24 | (0.07) | 2.70 | (0.05) | 2.27 | (0.07) |
| Controls | ||||||||
| Log (wordcount) | –0.14 | (0.01) | –0.15 | (0.02) | –0.14 | (0.01) | –0.13 | (0.01) |
| Positive tone | 0.05 | (0.001) | 0.07 | (0.003) | 0.05 | (0.001) | 0.07 | (0.002) |
| Main effect | ||||||||
| Competence | 0.0002 | (0.00004) | 0.0003 | (0.00009) | 0.0002 | (0.00004) | 0.0003 | (0.00005) |
| Interaction | ||||||||
| Competence × self-focus | 0.00002 | (0.000005) | 0.00006 | (0.00002) | 0.00002 | (0.000005) | 0.00005 | (0.000009) |
| R2 | 0.20 | – | – | – | ||||
| Adjusted R2 | 0.20 | – | – | – | ||||
*p < 0.05; **p < 0.01; ***p < 0.001

