Weathering the digital shift: how interaction spaces and weather conditions shape mobile app engagement Open Access

In today's technology-mediated consumption landscape, mobile apps have become essential to everyday consumer behavior, facilitating various activities from shopping and social interaction to health and lifestyle management. As mobile services become more deeply embedded in daily routines, understanding what drives consumers to engage with these apps becomes increasingly important. While existing research has extensively explored user-centric factors such as motivations and perceived value (Cui et al., 2024; Jiang et al., 2020), less is known about how broader contextual factors might shape mobile engagement behaviors. Recognizing the role of such external influences opens new avenues for understanding consumer–technology interactions in more ecologically valid settings.

Although mobile engagement research has traditionally prioritized internal drivers, less attention has been paid to external environmental contexts. Some studies have examined environmental factors such as weather effects on mobile promotions (Kyung et al., 2024; Li et al., 2017; Steinker et al., 2017), and offline retail presence on digital sales (Kim et al., 2024). However, the joint examination of how interaction spaces – physical environments that enable or constrain engagement – and weather conditions shape mobile app engagement remains limited. Addressing this gap can deepen our understanding of how everyday surroundings shape consumer–technology engagement beyond personal motivations or technological affordances.

Building on this gap, this study investigates how environmental and situational factors shape mobile app engagement within the pet care context. Although the pet care industry has expanded rapidly, prior research has largely focused on purchasing behavior (Di Cioccio et al., 2024; Jeong et al., 2024), overlooking how technology supports daily routines and health management beyond the point of sale. By focusing on real app usage behaviors rather than purchase outcomes, this study extends the understanding of consumer–technology–environment interactions into ongoing activity engagement. Specifically, we address the following research questions:

1. How do interaction spaces influence mobile app engagement?

2. How do weather conditions influence mobile app engagement?

Drawing on real location-based app usage from a pet physical activity-tracking application, we offer insights into how surrounding environments shape engagement behaviors. This study makes three key contributions. First, by positioning pet activity as an empirical setting, this study extends prior shopping-centered studies of mobile engagement to routine, health- and activity-oriented behaviors, offering a more holistic understanding of mobile service use. Second, we propose a typology of contextual influences – recreational, functional, and community spaces together with weather constraints – that map onto hedonic, utilitarian, social, and dynamic dimensions of engagement, thereby providing a transferable theoretical lens across app categories. Third, building on existing research that explores the impact of weather on consumer behavior, this study integrates weather dynamics into the domain of physical activity-tracking apps. By demonstrating how different weather conditions – heatwaves, heavy rain, and strong winds – shape engagement metrics, it uncovers the asymmetric and conditional nature of environmental effects on mobile service use, thereby enriching existing theories of mobile marketing and health behavior.

With the growth of digital technologies, researchers are paying more attention to the factors that shape mobile service use. Table 1 summarizes prior studies that examine how external contexts, such as interaction spaces and weather dynamics, affect mobile services. Prior studies have largely explored how environmental factors affect mobile shopping behavior, highlighting the role of offline store presence in driving digital sales (Cui et al., 2024; Kim et al., 2024; Jiang et al., 2020). Meanwhile, weather conditions have emerged as another prominent contextual factor, with research demonstrating how weather information can enhance the effectiveness of healthcare apps (Kyung et al., 2024), and influence consumer purchasing decisions across digital platforms (Li et al., 2017; Steinker et al., 2017).

Despite these advances, important gaps remain. Prior studies have largely examined environmental influences in commercial or transactional settings, leaving a limited understanding of how offline environmental factors, specifically interaction spaces and weather conditions, affect engagement with location-based mobile services in everyday contexts. This gap is particularly salient for physical activity–tracking apps, whose usage is inherently tied to users' surrounding environments. Understanding how such contextual factors shape engagement is therefore critical for explaining real-world patterns of mobile app use beyond shopping or information search. More broadly, environmental context in mobile and location-based services is multifaceted, encompassing not only natural conditions such as weather, but also built environments and social surroundings such as population density and crowding, all of which can shape engagement patterns in mobile settings (e.g. Pham et al., 2025).

From a theoretical standpoint, the influence of environmental contexts on mobile engagement can be understood through a stimulus-organism-response (S-O-R) mechanism. In this framework, environmental cues such as physical space and weather conditions serve as stimuli that shape users' affective and cognitive states, which in turn drive behavioral responses, including engagement with mobile services (Mehrabian and Russell, 1974). This perspective aligns closely with the atmospherics literature in marketing, which emphasizes that physical environments systematically influence consumer emotions and actions (Kotler, 1973).

At the same time, prior research cautions that mobile engagement is not uniformly positive or stable. Quantifying everyday activities can undermine intrinsic enjoyment and long-term persistence (Etkin, 2016), and design strategies that encourage early adoption may backfire over time, as evidenced in studies of anthropomorphized trackers (Fronczek et al., 2023). Weather effects are also context-dependent and asymmetric, exerting differential influences on mobile responses depending on conditions and user circumstances (Li et al., 2017; Schlager et al., 2020). These findings point to important boundary conditions that limit the generalizability of prior results. Building on this literature, the present study examines how interaction spaces and weather conditions jointly shape engagement with location-based physical activity–tracking apps.

In marketing research on location-based apps, prior studies have primarily focused on commercial services such as geo-targeted promotions near stores and restaurants (Bernritter et al., 2021). Beyond mobile commerce, however, numerous apps leverage location-based features for healthcare purposes, particularly physical activity tracking (Kim et al., 2025; Kyung et al., 2024). To comprehensively understand location-based apps that track physical activity, it is essential to consider the surrounding environment. Human spatial cognition, the internal representation of environmental information often called a “cognitive map,” influences how individuals perceive and interact with their surroundings (Golledge, 1999), which can shape their engagement with location-based applications.

Building on this perspective, prior research shows that these spaces influence user motivation in different ways (Cui et al., 2024; Mirzaei et al., 2018). We classify them into three types: recreational spaces, functional spaces, and community spaces. This framework builds on Bonaiuto et al.’s (2003) model, which highlights how different qualities of local environments influence behavior. Specifically, recreational spaces include architectural-planning features, the organization and accessibility of spaces, and the presence of green areas; functional spaces encompass welfare, commercial, and transportation services; and community spaces cover social interactions and community relations (Bonaiuto et al., 2003). This perspective also parallels the servicescape literature in marketing, which emphasizes that physical surroundings systematically shape user experiences and engagement with services (Bitner, 1992). From this view, interaction spaces can be understood as service environments that cue effective and behavioral responses during mobile app usage, reinforcing the long-standing insight that physical contexts influence service engagement.

Daily walking activities represent a fundamental form of physical activity in everyday life (Ham and Epping, 2006). These habitual behaviors become routine through repetition and are triggered by contextual cues rather than environmental enjoyment (Wood and Neal, 2007). From a habit-formation perspective, repeated engagement with activity-tracking apps in stable environmental contexts can reinforce automatic usage patterns (Labrecque et al., 2017). Walking trails exemplify recreational interaction spaces that support daily activity patterns. Well-designed trails offer designated paths, safe areas for movement, and spatially pleasing surroundings, enriching the walking experience (Frank et al., 2005; Handy et al., 2005; Saelens et al., 2003). Such spatially attractive environments encourage users to spend more time engaging with activity-tracking apps. However, while nearby recreational spaces may enhance motivation for hedonic activities, they may not provide sufficient incentive to alter daily routines or increase frequency. Thus, we posit the hypothesis:

Another characteristic of the local environment is its association with functional services. For users engaged in location-based physical activities, services such as veterinary clinics, grooming centers, and pet supply stores are crucial for maintaining regular activity routines and fulfilling practical needs. The convenience of having these functional spaces nearby serves as a strong motivator for users to engage in outdoor activities, as they regularly access these facilities for check-ups, grooming appointments, or to purchase supplies like food and essentials (Frank et al., 2005). The need to regularly access these services ensures that users increase the frequency of their physical activity, resulting in more frequent, but purpose-driven, movements. Given that these services are often in urban areas, visits are typically focused on completing specific tasks efficiently, reflecting the utilitarian nature of activity (Ewing and Cervero, 2010). This fosters more frequent activities driven by practical requirements yet is less likely to impact session duration (Cao et al., 2006). Therefore, we propose the hypothesis:

Community spaces such as pet-friendly restaurants and cafés serve as social and leisure spaces where users can relax and enjoy time with their pets. These spots create environments where users stay longer and engage in social interactions, contributing to longer outings (Christiansen et al., 2016; Sugiyama et al., 2014). However, visits to these places are often leisure-oriented rather than part of a regular routine. As a result, usage frequency may not be significantly affected. Hence, we hypothesize:

Research on weather and consumer behavior has a long history in marketing and management (Hu et al., 2024; Steele, 1951). Weather is especially influential compared to other contextual factors, as elements like sunshine, temperature, and rain strongly affect people's mood (Howarth and Hoffman, 1984). These weather conditions influence psychological states, which can affect motivation for outdoor activities, and also impact consumer behavior by altering travel costs, shaping decision-making processes, and even driving changes in retail shopping patterns (Kyung et al., 2024).

Moderately high temperatures and increased sunlight exposure can have a positive impact on mood (Howarth and Hoffman, 1984; Murray et al., 2020), motivating users to maintain regular outdoor activity patterns despite the heat. However, the physical discomfort of extreme heat, including overheating and dehydration, can make prolonged outdoor activities unappealing and even risky (Sherwood and Huber, 2010; Zivin and Neidell, 2014). Therefore, while users may still engage in outdoor activities to maintain routine, session duration is expected to shorten for safety and comfort. As a result, we propose the hypothesis:

Heavy rain conditions have a significant psychological impact on outdoor activities. Rainy weather can lead to negative moods (Denissen et al., 2008) and reduced motivation to go outside (Keller et al., 2005; Spinney and Millward, 2011), thereby decreasing the frequency of app engagement. Physically, heavy rain increases difficulty and inconvenience through reduced visibility, slippery surfaces, and potential hazards (Spinney and Millward, 2011). These factors discourage both the initiation and prolongation of outdoor activities. Accordingly, we put forward the hypothesis:

Strong winds are less likely to have significant psychological or physical impacts on outdoor activity engagement. Research indicates that wind velocity does not generate significant negative mood effects (Howarth and Hoffman, 1984). Unlike rain or extreme temperatures, wind is generally perceived as less intrusive. Physically, while strong winds can cause minor discomfort, people can adapt by choosing sheltered routes or using protective clothing (Sherwood and Huber, 2010). Therefore, strong wind is unlikely to deter regular engagement patterns. Accordingly, we assert the hypothesis:

Figure 1 presents the conceptual framework integrating interaction spaces and weather conditions with mobile app engagement.

The evolution of human-pet relationships has significantly reshaped consumer behavior, fueling the rapid growth of the global pet care industry. Valued at USD 304.4 billion in 2023, the market is projected to grow at a Compound Annual Growth Rate of 6.8% through 2032, driven by the trend of pet humanization and the demand for personalized experiences (Kim et al., 2025; Global Market Insight, 2024). Alongside this expansion, technological innovations such as mobile apps and smart devices have increasingly supported pet health management and daily routines (Kim et al., 2025). Despite this technological integration, prior research has primarily focused on purchasing behaviors, overlooking how technology facilitates ongoing engagement. The pet industry, where daily care and emotional bonding are essential, presents a valuable context for investigating engagement behaviors beyond transactions (Li et al., 2026). Building on this setting, this study examines pet healthcare apps, particularly activity-tracking services, focusing on heavy users to better understand how technology shapes everyday consumer-pet interactions.

In studying app utilization, focusing on heavy users is insightful for several reasons (Twedt, 1964). Heavy users engage extensively with app features and exhibit consistent behavioral patterns, providing rich data for analysis. They also serve as lead users who demonstrate stronger opinion leadership and influence the adoption decisions of other consumers (Schreier et al., 2007). Therefore, understanding heavy users' behaviors can provide valuable insights for enhancing user experiences and marketing strategies.

We compile data from multiple sources: the activity app, the Ministry of the Interior and Safety's public data portal, the Korea Meteorological Administration (KMA), and the Korean Statistical Information Service (KOSIS). The raw dataset from the activity app includes walking records from the top 200 users, with a total of 48,261 observations collected between July and October 2023. To ensure accuracy, the data undergoes preprocessing: outliers are removed based on a z-score threshold, records with abnormal app termination are excluded, only single-pet users are retained, and only observations within the same administrative region (Sigungu ^[1]) are included. We aggregate the data on a daily basis, summarizing walking sessions into daily metrics: frequency and duration. The final dataset comprises 19,143 observations from 175 users. This daily aggregation includes only days of walking activity, resulting in an unbalanced panel dataset.

Mobile app engagement is measured by monitoring physical activities, as the app is primarily designed for this purpose. The dataset provides detailed records for each pet walk session, including start and end times, enabling precise calculation of walking frequency and duration. Usage Frequency is measured by the number of times a pet is taken for a walk by user i in Sigungu s within Sido d on day t, and Session Duration is measured by the length of time, in minutes, spent on each walk session by user i in Sigungu s within Sido d on day t.

Interaction Spaces. Data from the Public Data Portal of the Ministry of the Interior and Safety ^[2] was used to analyze the Sigungu-level regional service environment related to pet care. This dataset includes the availability of natural walking spots (parks and trails), pet-care services (veterinary clinics, pharmacies, shops, grooming salons), and pet-friendly amenities (restaurants and cafes). Number of Pet-Walking Trails, Number of Pet-Care Services, and Number of Pet-Friendly Places are measured by the count of these facilities available in Sigungu s within Sido d where user i resides.

Weather conditions. Daily temperature, precipitation and wind velocity data at the Sido-level are collected from the KMA ^[3]. Binary variables were created based on weather warning criteria: A Severe Weather Warning for Heatwave is defined when the highest perceived temperature is expected to be 33 °C or higher for two consecutive days, indicating a heatwave in region s where user i resides on day t. Similarly, Severe Weather Warning for Heavy Rain and Severe Weather Warning for Strong Wind are triggered by accumulated rainfall of 110 mm or more in 12 h, and wind speeds of 50.4 km/h or gusts of 72.0 km/h or more, respectively, and are indicated as binary variables in Sido d where user i resides on day t.

User and pet information controls. The app provides demographic information for users and basic characteristics for pets. User-level variables include age, gender, login method (i.e. Apple ID, email ID, or messenger ID), and subscription status for marketing, event and reward notifications. Pet-level variables include age, sex, and weight. Incorporating these user- and pet-specific controls allows the analysis to account for individual-level variations influencing walking behaviors.

Local environment controls. We collected additional regional indices for control variables from the KOSIS [⁴]. These control variables include Sigungu-level data on district areas, capturing geographical scope for outdoor activities; population data, reflecting population density; and housing types, indicating the proportion of single-family versus multi-family residences. Incorporating these regional indices into our analysis allows for a more nuanced understanding of environmental factors impacting activity app usage. Summary statistics for these variables are presented in Table 2.

We build models to investigate the effects of interaction spaces and weather conditions on mobile app engagement, focusing on the two key outcomes: usage frequency and session duration. Before estimating the models, we assess the correlation between these outcomes and find it to be −0.17, indicating a low association. Given this weak correlation, we choose to model walking frequency and duration separately, allowing us to better capture the distinct influences on each aspect of pet-activity app usage. The usage frequency, Frequency_is(d)t for user i in Sigungu s within Sido d on day t, follows a Poisson distribution with parameter λ_is(d)t. The model is specified as follows:

In this model, Recreation_1,is(d), Function_1,is(d) and Community_1,is(d) represent the availability of pet-walking trails, pet-care services and pet-friendly cafes and restaurants, respectively, in Sigungu s within Sido d where user i lives. The variables WarningHeat_1,idt, WarningRain_1,idt and WarningWind_1,idt represent the severe weather warnings of heatwave, heavy rain, and damaging wind, respectively, in Sido d where i lives on day t.

We control for user-specific, pet-specific, and regional characteristics as previously defined, including demographic factors, pet attributes, and local environmental indices. To address user-time-specific unobservables, we include the previous week's walking duration as a control variable. X_1,is(d)t represents the vector of all control variables. A random intercept is incorporated to capture unobserved heterogeneity at the user level, accounting for individual differences that may influence outdoor behavior beyond observed variables. Additionally, Sido-fixed effects (δ_1,d) and time-fixed effects (τ_1,t) capture regional and temporal shocks, respectively.

The natural logarithm of the walking duration, log(Duration_is(d)t) for user i in region s and day t, follows a normal distribution. The model is specified as follows:

This section presents the results of the model estimation, with Table 3 providing a detailed summary. The models offer insights into how various factors influence app usage frequency and session duration.

The results confirm H1, showing that Recreational Spaces do not significantly affect Usage Frequency (β_1,1 = −0.008, p > 0.05), indicating no meaningful change in how often users walk their pets. By contrast, Recreational Spaces significantly increase Session Duration (β_2,1 = 0.064, p < 0.01), which corresponds to an increase of approximately 6.6% in walking time. This indicates that for dedicated users who already walk frequently due to intrinsic motivation, trails do not change the frequency. However, aesthetically pleasing and well-maintained walking-trails enhance the enjoyment of walking, leading to longer walking durations (Handy et al., 2005; Saelens et al., 2003).

The findings support H2, demonstrating a significant positive effect of Functional Spaces on Usage Frequency (β_1,2 = 0.245, p < 0.01). This corresponds to an approximate 27.8% increase in frequency, suggesting that access to pet-care services, such as veterinary clinics and grooming salons, prompts more frequent outdoor activity as users incorporate these errands into their routines (Cao et al., 2006; Frank et al., 2005). However, no significant effect is observed on Session Duration (β_2,2 = −0.097, p > 0.05), indicating that the time spent does not increase because the focus remains on completing tasks efficiently rather than extending leisure (Ewing and Cervero, 2010).

The analysis confirms H3, indicating that Community Spaces do not significantly affect Usage Frequency (β_1,3 = −0.017, p > 0.05). However, they have a significant positive effect on Session Duration (β_2,3 = 0.042, p < 0.05), translating into an approximate 4.3% increase in walking time. This suggests that visits to pet-friendly locations are typically leisure-oriented rather than routine, which has little impact on frequency. However, these service environments enhance the walking experience by offering social and leisure opportunities, thereby encouraging users to extend their outings, driven by the hedonic value of such activities.

The results support H4, indicating that severe Heatwave warnings do not significantly affect Usage Frequency (β_1,4 = −0.005, p > 0.05). This suggests that users continue their walking routines despite the heat, possibly due to the positive mood effects of moderate warmth and sunlight (Howarth and Hoffman, 1984). However, Session Duration is significantly reduced during heatwaves (β_2,4 = −0.053, p < 0.01), corresponding to an approximate 5.2% decrease in walking time. The discomfort and health risks associated with extreme heat prompt users to shorten walks, consistent with prior research on heat-related reductions in outdoor activities (Zivin and Neidell, 2014). This asymmetry suggests that pet owners maintain essential routines for their animals but adjust behavior by shortening walking duration to manage physical risk, highlighting the trade-off between habit persistence and environmental stress.

The analysis supports H5, showing that Heavy Rain warnings significantly reduce both Usage Frequency (β_1,5 = −0.064 (p < 0.01) and Session Duration (β_2,5 = −0.137, p < 0.01). The effect size indicates that heavy rain decreases walking frequency by about 6.2% and shortens walking duration by approximately 12.7%. This result suggests that heavy rain discourages users from walking their pets, consistent with the hypothesis that rainy weather diminishes motivation for outdoor activities (Denissen et al., 2008; Keller et al., 2005). Moreover, the substantial reduction in session duration reflects the dual impact of negative mood and physical discomfort, as rainy conditions create unsafe and unpleasant outdoor activity environments (Spinney and Millward, 2011). Unlike heatwaves, which only shorten activity length, rain imposes both psychological barriers and immediate physical constraints, explaining why its effects suppress both frequency and duration.

The analysis validates H6, showing that severe weather warnings for Strong Winds do not significantly affect Usage Frequency (β_1,6 = −0.003, p > 0.05) or Session Duration (β_2,6 = −0.005, p > 0.05). Both coefficients are close to zero and statistically insignificant, indicating that the effect sizes are negligible. This suggests that wind conditions do not deter users from walking their pets or influence the duration of these walks. As Howarth and Hoffman (1984) noted, wind velocity has minimal impact on mood, and unlike heavy rain or extreme heat, it does not create substantial physical discomfort or barriers. This pattern illustrates that not all weather-stressors carry the same behavioral weight – while heat and rain trigger clear behavioral adjustments, wind is perceived as a manageable inconvenience with limited influence on engagement.

The supplementary analysis, reported in Table 4, examines how interaction spaces and weather conditions influence walking behaviors across different age groups by including interaction terms between environmental features and user demographics. For interaction spaces, the main effects remain consistent with the primary analysis. The age interaction with functional spaces reduces walk frequency (β = −0.006, p < 0.05), indicating that younger users respond more strongly to these facilities. Conversely, the age interaction with community spaces positively influences duration (β_2,6 = 0.005, p < 0.01), suggesting that older users extend walking time more in socially oriented spaces. No significant age interactions were found for recreational spaces. For weather conditions, the effects of heatwave, heavy rain, and strong wind warnings remain consistent across age groups, with no significant age interactions observed. This indicates that weather-related behavioral adjustments are uniform regardless of user age.

To validate the findings, we conducted robustness checks (Table 5). First, we re-estimated the models using user fixed effects instead of random effects; results remained consistent. Second, we repeated analyses after excluding the top and bottom 1% of users based on walking duration; findings were largely unchanged. Third, we included user tenure (i.e. the length of time each user had been active on the platform) as an additional control variable; this did not materially alter the results. Across all checks, the relationships remained reliable and robust to changes in model specification and sample composition.

The findings reveal the varying impacts of interaction spaces and weather conditions on mobile app engagement. Firstly, recreational spaces like outdoor tracks and community spaces such as accessible places do not significantly influence app usage frequency but do extend the session duration. These recreational spaces cater to the hedonic needs of users by making walks more enjoyable. Secondly, functional spaces, such as convenience services, increase app usage frequency, reflecting their role in meeting the utilitarian needs of users. Regular visits to pet-care services for grooming or health check-ups lead to more frequent outings, though these are viewed as practical tasks. Thirdly, heatwave warnings shorten session durations due to health concerns, whereas heavy rain significantly reduces both usage frequency and session duration. Conversely, strong wind warnings have a minimal impact on mobile app engagement.

This study enriches the theoretical understanding of how environmental characteristics, specifically, interaction spaces and weather conditions, influence mobile user behavior in activity-tracking apps and pet-care industry. While previous research has primarily examined mobile apps in shopping and retail contexts (Cui et al., 2024; Jiang et al., 2020), this study extends this perspective into the mobile app engagement behavior, moving beyond transactional outcomes toward everyday activity patterns. By examining the distinct effects of recreational spaces, functional spaces, and community spaces on both usage frequency and session duration, this research contributes to a more nuanced understanding of how environmental motivations shape mobile engagement behavior (Bonaiuto et al., 2003; Mirzaei et al., 2018). This differentiation clarifies the complex interplay between external factors and mobile app usage, underscoring the importance of aligning app features with diverse user needs.

Beyond documenting these empirical relationships, the study strengthens its conceptual contribution by outlining a typology of contextual influences that can be generalized across mobile app categories. Specifically, recreational spaces reflect hedonic and experiential cues, functional spaces highlight utilitarian and task-oriented affordances, and community spaces represent socially oriented contexts. Weather conditions, in turn, introduce dynamic constraints that moderate the influence of these spatial cues. This typology provides a transferable framework for analyzing how environmental contexts shape digital engagement across domains, thereby extending theories of mobile marketing and health behavior.

Furthermore, building on existing research that explores the impact of weather on consumer behavior (Denissen et al., 2008; Howarth and Hoffman, 1984; Zivin and Neidell, 2014), this study integrates weather dynamics into the domain of physical activity-tracking apps. By demonstrating how different weather conditions – heatwaves, heavy rain, and strong winds – influence engagement metrics, it enriches understanding of how environmental fluctuations can shape health-related behaviors through mobile platforms. Nevertheless, our findings should be interpreted with caution. Prior work shows that quantification and design features can reduce motivation (Etkin, 2016; Fronczek et al., 2023), and that weather effects are context-dependent rather than uniform (Li et al., 2017; Schlager et al., 2020). This suggests that our effects are conditional rather than universally applicable.

The findings of this study offer valuable insights for marketers and developers of mobile healthcare and activity-tracking apps, particularly those operating in location-based service environments. Integrating features that leverage interaction spaces and weather conditions can significantly enhance user engagement by aligning app experiences with users' physical contexts. Effectively analyzing the characteristics of nearby interaction spaces, and matching them to users' behavioral patterns can substantially increase the perceived relevance of the app. For example, providing recommendations based on access to pet-friendly cafes or parks can make it easier for users to plan outings, enhancing both the functional and emotional value of the service. Such integration strengthens the connection between app features and users' real-world activities.

Meanwhile, offering real-time notifications about severe weather conditions, such as heatwave alerts or recommendations for shaded walking routes, can further enhance user satisfaction and retention by helping users sustain their routines during adverse conditions. Beyond these baseline features, app design can be advanced through adaptive user interfaces that dynamically respond to environmental cues and predictive engagement models that anticipate user needs based on local contexts. Thus, these strategies enable healthcare apps to deliver more adaptive and context-sensitive experiences, reinforcing trust in the platform and promoting healthier, more consistent activity patterns for both users and their pets.

Beyond practical implications, the findings hold broader societal and policy relevance. Apps that support activity maintenance under adverse conditions can contribute to public health campaigns promoting healthier lifestyles. The importance of recreational and community spaces highlights how urban design decisions, investments in green areas and walkable neighborhoods, can shape consumer well-being. From a policy perspective, mobile engagement data can serve as an evidence base for city planners helping them evaluate existing investments and identify underserved areas. Overall, the implications extend beyond marketers to inform urban planning and health policy.

Despite offering important insights, this study has several limitations that future research should address. First, regarding sample and scope, the focus on heavy users of a specific app may not represent the broader population, as their behaviors may differ systematically from light or occasional users. Additionally, the four-month study period within a single geographic context may restrict generalizability across seasons and regions. Future research could include more diverse user samples and extend the observation window across different temporal and regional contexts. Second, regarding methodology, the observational nature of the data limits strong causal inference despite extensive controls for potential self-selection concerns. Future research could adopt experimental methodologies to more directly examine how environments influence engagement. Furthermore, the spatial aggregation at the Sigungu and Sido levels may obscure micro-level variations; finer-grained spatial data could reveal whether access to recreational spaces buffers adverse weather effects on engagement.

The authors are grateful to Jung Eun Kim, the Chief Executive Officer of the activity-tracking mobile app, for making the data available for this study.