With the rapid increase in the use of mobile technologies and smartphones for health information [1,2], mobile apps present one possible solution for communicating preventive health information to the public [3-5]. Over the past decade, hundreds of health mobile apps have been produced—many designed by public health interventionists and researchers for cancer and other chronic disease prevention by encouraging healthy eating and physical activity [6-8]. While it remains unclear how successful these apps have been in reducing the incidence of cancer or improving health outcomes for other chronic diseases, there is a call for an increase in the accountability, reliability, and standardizations of evidence-based health apps developed by the research community [8-10].

Despite the potential of mobile health (mHealth) apps for communicating up-to-date, evidence-based prevention information and helping users maintain or implement healthy habits, there is very little guidance on how these intervention apps should be designed to ensure adoption [11]. Designing apps so they are appealing and used is a critical first step for apps to have an impact [12]. Visual and interactive design influences initial user evaluations, which are made within milliseconds, and serve as gateways for subsequent user engagement (eg, use) of apps as mHealth interventions [13-15]. Ignoring design can detrimentally impact the communication of evidence-based science to health consumers and undercut the effectiveness of mHealth interventions; yet, few mHealth interventions mirror the look and function of popular, industry-developed apps. Thus, our study objective was to explore app features expectations and examine how meeting expectations with high- (vs low-) prototypicality apps may influence predictors of app adoption.

How apps are designed (visual display) and the features they include (interactivity) can influence users’ experience of and willingness to engage with apps. Individuals use salient cues that match their expectations, or mental models, to evaluate web-based information [16,17]. These expectations are met (or not) by the level of prototypicality or the degree to which an app resembles others in its comparative group [17,18]. Based on included design cues, in the form of interactive features, apps can range from having high prototypically (looks like others and meets expectations well) to low prototypicality (does not resemble others nor meet expectations) [19]. Users are often quicker and more willing to attend to apps that have high prototypicality—when designs align with one’s mental models for how an app should look and function [19-21]. Indeed, users look for and pay attention to expected, salient features as guides to orient themselves to novel apps and platforms [21]. When these expected features are present, they increase familiarity and potential use of the app [19-21]; however, little is known on how attention for specific features translates into individual feature use versus overall app use.

The perceived affordances, or perceived action possibilities (eg, learn health tips), that users sense from app features also directly impact a user’s experience and likelihood to engage with a design [22-24]. Specifically for mediated communication, including apps, design communicates what the viewer can do or gain from the use of an app, through interface symbols. Thus, not only must mHealth interventions have evidence-based content to drive use, but also apps must incorporate an evidence-based design to appeal to and engage audiences.

Design features influence the appeal or perceived aesthetics of the app and the likelihood for use [25,26]. To be effective, health apps must surely be used. It is necessary to understand how objective design features (the visible objects or designs in an app) influence subjective evaluations for initial appeal on the basis of theories of aesthetics [27-29] and antecedents for technology adoption in the Technology Acceptance Model (TAM); that is, perceived ease of use, perceived usefulness, and intentions to use [30,31]. Aesthetics, including facets for how information is organized and displayed, function as a precursor to perceptions for technology acceptance [28,31]. Accounting for users’ expectations of features and placements within apps will shed light on how prototypicality impacts evaluations critical for future adoption.

Utility also drives evaluation of an app’s usefulness and potential adoption, according to Nielsen et al’s [32] well-established usability study. Utility refers to the inclusion of necessary features—whether an app provides the elements an individual needs or wants. When utility is paired with usability—when features are perceived as easy (perceived ease of use) and pleasant (aesthetics) to use—individuals are encouraged to engage or interact. In other words, interactivity is dependent on a user’s willingness to engage with specific design features, if present (utility) and function properly (usability). In our work, we focus on the former—how app features that are needed (utility) or expected (prototypical) are the gateway to potential adoption.

In sum, engagement with and use of an app is driven by initial impressions and perceptions of what the app can do for the user. Top-rated industry-developed apps often incorporate a user-focused sleekness and are feature loaded; in comparison, pared-down mHealth interventions—despite the inclusion of theory-based content—may not appeal to audiences who need them [33]. When resources are not abundant, health researchers and interventionists need evidence-based guidance for design investments. Thus, we explored app expectations for the presence and placement of potential features, how these features garner attention and predict use, and how high-prototypicality apps (vs low-prototypicality apps) may influence app adoption through app choice and predictors of use. We asked the following research questions: What features do people expect and where do they expect these features to be placed (RQ1)? What specific features are associated with attention and predicted use of the app features (RQ2)? Last, we also examined whether high prototypicality, resembling that of top downloaded apps (vs low-prototypicality apps, resembling research intervention apps) would increase app choice (H1), aesthetics (H2), perceived ease of use (H3), perceived usefulness (H4), intentions to use the app (H5), and perceived affordances or action possibilities with the app (H6).

To explore app features expectations and examine how meeting expectations with high-prototypicality apps (vs low-prototypicality apps) may influence predictors of app adoption, we conducted a web-based survey with an embedded within-subjects experiment. Participants first responded to survey items about expectations for specific app features to answer RQ1-2 and an app choice (preview of apps with high vs low prototypicality) to address H1. Participants were then asked to rate their perceptions of the app overall, with the exposure order of condition (high vs low) randomized, to address H2-6.

Using G*Power, our a priori power analysis indicated a required sample of at least 450 participants to detect a small-to-medium effect (Cohen f=0.14) for within-subjects comparison of the high and low prototypicality apps. Participants (n=462) were recruited from Amazon’s Mechanical Turk (MTurk), a web-based crowdsourcing platform often used for social science research [34-36], through a link open to individuals over the age of 18 years. Participants were eligible if they were aged 18 years or older, resided in the United States, and had a task approval rate of 85% or higher on the MTurk platform, which indicates valid participation or completion of previous tasks. Participants received US $3 as compensation for their time (approximately 15 minutes). The institutional review board of University of North Carolina approved this study.

Following consent, participants selected features (from a list) they would expect to find in a health app. For all expected features, participants were shown an outline of a smartphone and asked where that feature would be located in a typical health app. Participants were then randomly assigned to 1 of the 2 app types for the remainder of the study: fitness apps or nutrition apps. Participants selected the app they would most like to download from 2 previews (prototypical: high vs low). On subsequent pages, participants indicated what features grabbed their attention and what features they predicted they would use (predicted use) on their preferred app. Participants were shown the app previews again (one at a time, in a random order) and asked closed-ended items for perceived aesthetics, ease of use, usefulness, intentions to use the app in the future, and perceived affordances. Lastly, demographic, health, and health app information were collected from all participants. Closed-ended items and response options are described below (see Measures) and provided in Multimedia Appendix 1.

To assess the impact of prototypicality on app perceptions, app previews were created for four fictitious brands: 2 fitness and 2 nutrition health apps (Figure 1). We designed previews for each app as they would appear if searched for in a mobile app store, including the app icon, brand name, and 2 preview screens of the app. High-prototypicality apps were developed on the basis of structure and content from top rated apps (Aaptiv, Lifesum) in the Health & Fitness section of the App Store. Low-prototypicality apps were designed to mirror the mobile interface of an interactive intervention (Carolina Health Assessment and Research Tool) for data collection and tailored feedback for preventive health behaviors [37].

We used n (%) values to describe app feature expectations, placement, app choices, attention, predicted use, and perceived affordances. Frequencies for attention vs predicted use and for perceived affordances of the high- vs low-prototypicality apps were compared with McNemar chi-square tests. Prior to this analysis for direct effects of prototypicality, a multivariate analysis of variance (MANOVA) was used to determine if there are any significant differences in perceptions among the app types (fitness and nutrition) across aesthetics and TAM outcomes. No differences were observed for high prototypicality (aesthetics outcomes: Wilks λ=0.98; F_4,454=1.08; P=.10; TAM outcomes: Wilks λ=0.99; F_3,454=1.08; P=.36) or low prototypicality (aesthetics outcomes: Wilks λ=0.99; F_3,452=0.68; P=.61; TAM outcomes: Wilks λ=1.00; F_3,454=0.10; P=.96), so data within conditions (high vs low prototypicality) were combined for analyses. Two repeated measure (RM) MANOVAs and analyses of variance (ANOVAs) were then conducted with high vs low prototypicality as the predictor; 1 for aesthetic outcomes (simplicity, diversity, colorfulness, and craftsmanship) and 1 for technology acceptance outcomes (perceived ease of use, usefulness, and intentions to use).

Participants (n=462) were aged 18 to 70 years (mean age 35.03 years, SD 10.02 years) and half of them were female (50%, 232/462). Participants identified as White (78%, 358/462), African American (13%, 58/462), Asian (8%, 35/462), or multiracial/other; additionally, 48 of 462 participants (10%) reported their ethnicity as Hispanic. Education levels included high school to some college (33%, 153/462), associate degree (13%, 60/462), bachelor’s degree (43%, 197/462), master’s degree (10%, 45/462), and doctoral or professional degree (2%, 7/462). Most participants reported their health as good (48%, 220/462) or very good (17%, 78/462), although some did report that their health was fair (30%, 138/462), poor (4%, 20/462), or very poor (1%, 4/462). Over half of the participants (53%, 248/462) reported currently using health apps.

Each of the 12 features was selected by at least 44% (203/462) of participants (RQ1). The majority of participants (92%, 425/462) selected a menu, settings options, and user profile; notably, these features (ie, menu, settings option, and user profile) were selected an equal number of times but not by the same respondents. Additional features were expected, including the following: login (88%, 406/462), summary graph/chart (86%, 396/462), summary statistics (86%, 395/462), input data feature (80%, 368/462), calendar (77%, 354/462), logo, (77%, 357/462), search (69%, 321/462), page title (62%, 286/462), and an option to “share with friends” (44%, 203/462).

Most features were expected in similar locations (Figure 2) among participants who had expected features (n=425). Menus were consistently expected to be in the top-left, while search and login options are placed in the top-right corner. Other features—title, logo, profile, and settings—were expected along the top, in the center, or either side. Sharing capability was expected to appear in the bottom-right of the app, although expectations of where to log input data were more diffuse. Users expect summary statistics, graphs, and calendars to be shown across the center of the app.

Respondents selected features of their preferred app, which caught their attention and they would use (Table 1 and Multimedia Appendix 1). Attention and predicted use patterns of the high-prototypicality apps indicate 4 distinct categories of mHealth app features. Mundane features are those that have similar low attention and predicted use values. In the fitness app, the footer menu options “Discover” and “Saved” represent mundane features. Functional features have higher predicted use than attention, but predicted use remains low (<50%, <231/462) among participants, such as the settings icon in both apps. Flashy features are elements identified as attention-capturing by most participants (>50%, >231/462), and attention is significantly higher than the predicted use. In the nutrition app, large photo-based links for the “Ketogenic Easy” and “Ketogenic Medium” diets represent flashy features. Essential features are elements that most participants (>50%, >231/462) thought they would use, and where predicted use is higher than or similar to attention, as with the “Calorie Tracker” in the nutrition app. Not included in these 4 categories are elements that have higher attention than predicted use, but the attention remains low (<50%, <231/462); the only features with these characteristics were logos and app titles, as well as 2 features partially obscured in the design.

When asked to choose between the high-prototypicality app and one designed to look more like a typical health intervention (low prototypicality), 347 of 462 (75%) participants indicated they would download the high-prototypicality app (H1).

Prototypicality had a significant main effect on all facets of aesthetics and technology acceptance outcomes (Table 2). High-prototypicality apps (vs low-prototypicality apps) had significantly higher ratings of aesthetics for simplicity (F_1,455=291; P<.001), diversity (F_1,455=578; P<.001), colorfulness (F_1,455=295; P<.001), and craftsmanship (F_1,455=462; P<.001). Similarly, the high-prototypicality app was rated higher than the low prototypicality app for perceived ease of use (F_1,455=84; P<.001), usefulness (F_1,455=116, P<.001), and intentions to use the app (F_1,455=170; P<.001). H2-5 were supported.

Participants reported that the app would allow them to carry out various actions in both the high- and low-prototypicality design (Table 3). Almost all perceived affordances had significantly higher endorsement for the high-prototypicality (vs low-prototypicality) apps (P<.01), partially supporting H6; to “learn health tips” was the only affordance endorsed similarly in both conditions. The most highly endorsed affordances (>60% across conditions or >277/462) were the following: “track my progress” (high: 93%, 430/462; low: 70%, 325/462), “set health goals” (high: 88%, 405/462; low: 73%, 339/462), “improve my health” (high: 74%, 342/462; low: 63%, 293/462), “learn health tips” (high: 73%, 336/462; low: 76%, 353/462), and “give me more information about my health” (high: 70%, 325/462; low: 63%, 292/462).

For mHealth to have an impact on reducing risk for chronic disease, intervention apps must be designed to effectively reach wide audiences to promote preventive health behaviors. Identifying the impact of prototypicality—the extent to which apps meet expectations—on app reception and adoption is a critical step in mHealth intervention research. Designs that match users’ perceptions of organization and content evoke prototypicality and can influence intentions to use web-based tools, including health resources [21,31,43]. Our study on prototypicality serves as an antecedent to positive app reception and technology acceptance in preventive health apps. We also found designs that contradict what users typically expect from apps (eg, low prototypicality), leading to a suboptimal first impression and diminishing users’ expectations [19].

It is likely that the actual use of multiple apps influences preventive behavior [44]; thus, identifying key features, or classes of features, to increase orientation and facilitate ease of use and usefulness are needed to guide intervention development. Our findings for user attention and predicted use of features point to 4 distinct types of mHealth features that should be considered when developing mHealth. Of these, 3 categories serve as useful features of mHealth: driving attention, perceived use, or both.

Functional features have higher predicted use than attention, and a majority “would expect to find” these sorts of features in a health app. To meet expectations, salient functional features such as search options, settings, and menus should be included, in their expected corner placement. Even if these features do not draw attention as much as others, users still expect to see them in mobile apps, and meeting baseline expectations can reduce time and cognitive demand for initial orientation and web-based information processing [21]. Arguably, these functional features constitute a sort of prototypical milieu or background environment for mHealth apps to likely help users orient themselves within new and unfamiliar apps.

Flashy features garner significantly more attention from users; these attention-capturing features may be most influential for positive initial impressions. Flashy features often incorporated photographs or novel design elements, which have been shown to increase attention and appeal [43,45]. Beyond meeting expectations, flashy features represent the unique category that should be treated differently in designs: using visuals to highlight salient benefits and perceived affordances.

Essential features—including those selected by most users as features that they predict to use and garner their attention—are also important components of mHealth designs. It is important to note, however, that the essential features seen in this study are all familiar: calendar, calorie counter, and performance tracker. Even though some designers may assume that features as basic as a calendar are not worth the time and effort to include, respondents strongly indicated that these features remain important components of mHealth apps.

Our findings also highlight a distinct category that can be skipped or given little attention in development: mundane features. Mundane features, such as app title and tabs for discovering or saving, elicited little attention and predicted use and are a good indication not to waste precious resources on these elements.

Potential mHealth users had consistent expectations for some features by region (eg, middle or top corner), but not necessarily a specific location. Essential features, such as a calendar, were expected to be shown across the center of the app. Other features, such as function features including search and settings, had more narrow placement expectations. Understanding these location expectations is critical to ensure that feature placement matches individual models [21].

Higher prototypicality led to higher ratings for aesthetics, perceived ease of use, usefulness, and intentions to use apps. Individuals also expect greater function, possibilities, and valuable outcomes from apps with higher prototypicality. Low prototypicality led to lower rankings for aesthetics, perceived ease of use, and perceived usefulness. Additionally, low prototypicality runs the risk of users initially dismissing the app. Negative product evaluations—where expectations are not met—can also lower satisfaction with product interaction [46].

This study is limited to the specific health apps manipulated herein; these apps do not represent all available mHealth strategies. Although we evaluated placement, attention, and predicted use, we could have reviewed more features within apps. Our findings are also limited to a convenience sample of participants of a web-based panel. It is possible that our participants have more digital literacy or skills than the general population or diverse subgroups.

Future studies should consider assessing actual use after download, instead of solely predicted use. Replication with more diverse audiences, varied app designs, and expanded methodological approaches are needed to generalize our findings. Notably, future research should account for additional personal characteristics, such as health literacy or the ability to obtain, process, and understand health information [47], to examine how these skills affect both first impressions for app adoption and actual use to determine the effectiveness of health apps.

Mobile apps can communicate critical health information for preventive health behaviors through readily available and consumer-friendly tools. Apps that are thoughtfully designed to match potential users’ expectations, with increased prototypicality, will support app use. Conversely, designs that do not include a threshold of expected features will be dismissed, thus undermining the potential of app-based interventions. Designing mHealth apps to account for user expectations will increase the likelihood of adoption and impact from actual use. Prototypicality is positively related to favorable reception and expectations for future use of health apps. These findings provide guidance for user expectations of feature presence and location.