A physically active lifestyle contributes to a healthy and long life [1-3]. Being physically active is partly a choice and partly a consequence of external circumstances such as area of residence [4]. For example, “girls, women, older adults, underprivileged groups, and people with disabilities and chronic diseases” are among the most sedentary owing to limited access to affordable and safe opportunities for physical activities [5]. Thus, to increase equal opportunity and motivation for physical activity, both individual and societal aspects need to be addressed [4-6]. However, this study is limited to the individual perspective and the acknowledgment that, regardless of their personal circumstances, many people want to become healthier.

Digital health innovation can reduce health inequities [7]. Mönninghoff et al [8] claimed that digital innovations can be particularly useful for persons at risk and populations that are sick and present an affordable alternative for larger populations, whereas Western et al [9] claimed that those with low socioeconomic status are still left behind. Regardless, in 2021, the World Health Organization (WHO) issued the “Global strategy on digital health 2020-2025” [10] partly on the assumption that “digital innovations [can] promote and support people of all ages to be more active” [5]. Undoubtedly, finding effective and intersectional digital health innovations to support healthy behavior changes is of great importance.

Theories on learning and behavior change acknowledge the importance of both internal and external factors for behaviors. The classical social cognitive theory by Bandura [11] makes 3 basic postulations: first, people can learn through observation (ie, imitation); second, learning is influenced by mental states (eg, self-efficacy); and third, learning does not suffice for behavior change. Whether the observed behavior is adopted or not will depend on the observed reactions from the surroundings and feedback from imitating the behavior [11,12]; thus, the social cognitive theory shows the constant interaction between the person and the environment and the several critical points for disruption from observation to imitation.

Similarly, several theories that aim to describe the process of behavior change have acknowledged the interaction between the person and the environment [4,6,13,14]. One model that has combined several theories and techniques of behavior change is the Capability, Opportunity, Motivation–Behavior (COM-B) model [15,16]. The name of the model is an acronym for the 3 components deemed necessary for behavior change to occur, that is, capability, opportunity, and motivation [15]. Capability is the individual’s psychological and physical capacity to perform the desired behavior (eg, how to eat healthy or being aware that physical exercise is good for health); opportunity refers to the external conditions, such as living in a safe environment to exercise in or being able to afford healthy food, that promote desired behavior; and motivation includes all processes that direct behavior (eg, feelings and attitudes toward the behavior or the self). All 3 must occur simultaneously for the desired behavior to occur [15]. In other words, even if a person wants to change (ie, motivation) and knows what to do (ie, capability), their living situation can upturn the execution of the new behavior (ie, opportunity). In addition, behaviors are never fully extinguished, that is, the old behavior is only dormant and needs to be suppressed across a variety of situations, whereas the new behavior needs to be activated [13]. Thus, behavior change is unstable and can easily be disturbed [13]. In sum, changing a behavior is a challenge that goes beyond a willingness to change; opportunities for change and strategies for sustaining new, healthy behaviors should be determined on a daily basis [17,18].

There are numerous ways to support healthy behavior changes. For example, a systematic review and meta-analysis by Howlett et al [19] showed that behavior change techniques (BCTs) such as giving feedback about the behavior or strengthening capability and motivation were successful in helping adults who are inactive increase their level of physical activity. Samdal et al [20] obtained a similar result in a review and meta-analysis of physical activity and healthy eating among overweight or obese people. They pointed out that BCTs that help the person’s capabilities in combination with a person-centered method that enhances motivation are the driving principle for behavior changes in the reviewed studies. Their categorization of BCTs in the review was based on a taxonomy by Michie et al [16]. Bringing together the accumulated evidence and knowledge from years of research on BCTs in the context of fast-paced digitalization is a promising path forward [21].

Numerous digital health interventions (DHIs) have been designed to support behavioral changes [22]. DHI are health interventions delivered via digital devices or technologies, for example, smartphones, digital platforms, or SMS [23]. Most DHIs are designed to facilitate behavior changes using existing BCTs [24,25], where different types of feedback are the most commonly used strategy [21,26,27]. However, research on the effectiveness of DHIs and their effect on behavior changes is still unclear [21,28].

So far, the more personalized the DHI, the more effective it is for behavior change. Personalized DHIs—that is, interventions adapted to individual needs—“improve prevention, self-management and access to healthcare” [29]. For example, Wannheden et al [30] explored how self-monitoring and communication with health care via digital technology satisfy or frustrate basic psychological needs and found that individual preferences vary and that personalization of digital technological tools is essential. However, of the numerous digital health applications [31], few seem to offer personalized support.

Willcox et al [32] argued for the use of SMS for personalization; it is a widely adopted and inexpensive mobile function. The effectiveness of different types of SMS support has been studied quite extensively in different settings—for example, smoking cessation, medication adherence, and self-management of long-term conditions and health, including diabetes and weight loss [32,33]—and with varying degrees of personalization from simple to more tailored SMS prompts [34-37]. In the meta-analysis with a randomized control trial by Head et al [38], the SMS was automated and tailored to the participant’s characteristics (eg, age, health status, or motivation) or targeted the intended behavior (eg, smoking cessation). Most studies combined automated SMS with at least 1 more intervention (websites, print materials, or human counselors). Only half of the included studies were theory-based interventions, and personalization across studies varied considerably. “Personalization” was defined based on whether the SMS included the participant’s name or not. The result showed that studies that tailored the SMS according to demographics and psychosocial factors (ie, motivation or self-efficacy) and requested frequency were more effective than studies using only targeted SMS and a set SMS frequency [38]. Head et al [38] speculated whether the tailored SMS indicates a “social presence” that augments the intervention [39].

The importance of the relationship between the therapist and the client is well known in psychology [40]. A systematic review conducted by Chatterjee et al [21] revealed that digital interventions held promise for effectiveness, with a particular emphasis on the significance of personalization and tailoring for success. Among these interventions, video-based approaches were found to be the most well received. Thus, the user perspective needs to be considered, and the authors pushed for “in-depth qualitative research to gradually improve interventions to satisfy users” [21]. They also stressed the importance of knowing the patient “to ensure both relationship communication and goal-oriented coaching” [21]. Taken together, there is substantial evidence supporting the effectiveness of using personalized SMS as a DHI to facilitate behavior change. However, additional research is required to delve into the individual and social dimensions of DHIs.

The aim of this study was 2-fold: to test the effectiveness of theory-driven personalized SMS coaching in a naturalistic setting in Sweden using the COM-B model [15] and to explore user experiences of personalized SMS coaching at a Swedish nonprofit fitness organization; here, the users are both fitness consultants and clients. This mixed methods study used 3 types of data: quantitative self-reported data from the clients, pseudonymized SMS conversations between the fitness consultants and clients from the intervention, and qualitative interviews with the fitness consultants. The purpose of the mixed methods design is to (1) investigate the effect of personalized SMS coaching and its underlying process and (2) to further explain and validate these findings with follow-up interviews.

Personalized SMS is defined as content and frequency that are highly tailored to a client’s specific needs and wants.

Fitness in Värmland (FiV; Friskvården i Värmland in Swedish) is a nonprofit organization in a rural region in middle western Sweden with a longstanding mission to support the citizens in the region in making changes regarding unhealthy dietary habits and sedentary lifestyles. Approximately 20 fitness consultants use individual health coaching, group training, or diet courses to support their clients’ lifestyle changes. The main focus is to help clients find and sustain motivation to become healthier; for this, motivational interviewing (MI) and SMART (specific, measurable, attainable, realistic, timely) goal setting are used. MI is a person-centered conversation methodology that aims to enhance individuals’ decisions to initiate behavior changes [41]. According to MI, ambivalence is a necessary drive for change and starts when a person starts to question their current way of living, and MI aims to help the client formulate their own drive for change [41]. Therefore, the fitness consultants address both the aspects of capability and motivation.

The vast majority of FiV clients (75%) are prescribed to increase their physical activity by licensed medical staff (eg, physicians, nurses, and physiotherapists) to prevent or reduce symptoms of a clinical diagnosis such as diabetes, heart and vascular diseases, and anxiety. The remaining 25% voluntarily seek help to achieve healthier lifestyle choices. Most clients are female (65%), aged >40 years (77%), and many persons have comorbidities (the percentages are from 2019 and were provided by FiV).

When the clients first come to 1 of the 16 FiV centers in the region, they are tested for basic health indicators (eg, weight, aerobic fitness, and physical and dietary habits) [42]. Some of these tests were included in this study and are described in more detail in the Materials section. The fitness consultants also start an MI-based dialogue (health talks) with the clients about lifestyle choices and possible changes. The health talks are then offered on a regular basis, often every fourth to sixth week.

This study used a within-subject design with an ABBA design, set in a naturalistic setting using completion data only, that is, the study only included participants who followed through the intervention, and no imputations were made for missing data. Using an ABBA design offers several advantages. It is straightforward to put into practice, ensures robust experimental control, counters order effects, and acknowledges that certain behaviors such as acquiring new skills may not be reversible [43]. The recruited clients were randomized to start with either the 8-week SMS intervention (condition A) or placed on the waiting list (condition B). Table 1 presents the overview and timeline. The clients who started with the SMS intervention (ie, AB group) were placed on the waitlist after the 8-week intervention, and the clients who started on the waitlist (ie, BA group) were given the intervention after the 8-week waitlist. Thus, the clients acted as their own controls (Table 1). As Table 1 shows, the quantitative self-reported data from the clients were collected on 3 different occasions, time 1 (T1) to time 3 (T3), where the order of the interventions varied.

This study follows the ethical principles for research involving human subjects as stated by the World Medical Association, and all participants gave their informed consent to participation. The study was approved by an ethical advisor at Karlstad University (HS 2020/102). No clinical trial registration has been made. All participants provided their informed consent and were advised that all participation was voluntary. If they chose not to participate or later wanted to opt out, they were informed that this would not affect their regular support from the fitness consultants. All participation data from the clients was anonymous to the researchers. The researchers only had contact with the fitness consultants and all their data has been pseudonymized (deidentified). No compensation was offered to the participants, clients, or fitness consultants.

The mixed method used three types of data: (1) quantitative self-reported data, (2) SMS conversations between the fitness consultants and clients during the intervention, and (3) qualitative interviews with the fitness consultants after the intervention. Using different data, we accumulated different viewpoints on the effects of personalized SMS on behavior changes and the participants’ experiences from SMS coaching. The integration of quantitative self-report data and qualitative SMS conversations is conducted in a parallel mixed design fashion [44]. To be more specific, this approach can be considered an embedded design [45]. In this context, the qualitative data (SMS conversations) serves to elucidate the underlying processes that contribute to the effects observed in the quantitative data (the impact of the intervention). The outcomes of the intervention are subsequently combined with the qualitative data obtained from interviews with the fitness consultants, following a sequential mixed design approach [44]. This is executed as an explanatory design, where the qualitative data are used to provide explanations and additional insights into the results obtained from the intervention [45]. The mixed methods design also improves the construct validity of the study, that is, how well the findings from the analysis process reflect reality [46].

Owing to ethical considerations, the participating clients were kept anonymous to the researchers, and all client data were pseudonymized by the fitness consultants before they were handed over to the researchers. All participation was voluntary, and informed consent was obtained from all participants. SMS coaching was chosen because SMS is a widely adopted mobile function and an inexpensive mobile phone feature and has been proven to be effective in supporting behavior changes [32,38]. In addition, according to the Swedish Internet Foundation, Sweden is highly digitalized (92% own a smartphone [47], and 48% of the population uses a digital health service or mobile app on a weekly basis [48]). Thus, SMS is an inclusive method regardless of the type of mobile phone the participant owns, and the focus is not on the technical device but on the intervention itself [5].

The study was cocreated with fitness consultants after a pilot study in 2019 that aimed to explore the usefulness of SMS coaching for FiV and its clients. The results of the pilot study were evaluated in a workshop where the fitness consultants provided their input based on their experiences. The pilot study resulted in an SMS guide with guidelines for SMS coaching and explicit SMS examples such as encouragements, reinforcements, and challenges, with MI and the COM-B model being the points of reference. This SMS guide was a starting point for the SMS intervention and was used to create a highly personalized coaching protocol. The SMS guide included personalization regarding frequency (eg, daily, every other day, or weekly), when to send the SMS (eg, the days or specific time points), and whether and when the clients were expected to respond. The SMS guide also regarded the target behavior (eg, “eat less sweets,” “exercise three days a week,” and “take a walk weekly”), an action plan for how to achieve the goal, and if the SMS could include pictures or links to web pages.

All fitness consultants were invited to participate in the study and were presented with a folder and a web page providing information about the study, informed consent contracts for clients, the SMS guide, and questionnaires for all assessment points. The information clearly stated that participation for all parties, both clients and fitness consultants, was voluntary; that they could discontinue their participation at any time; and that nonparticipation or discontinuation would not impact their continued contact or employment with FiV. No specific inclusion criteria for participation were stated by the researchers, and the fitness consultants were free to choose participants they deemed fit for SMS coaching. The web page also provided a randomization list where fitness consultants could randomize the recruited clients to start with either the SMS intervention or the waitlist (ie, the ABBA design).

As presented in Table 1, the intervention started (T1) with the SMS guide, self-rated questionnaires, and the basic health indicators (weight, waist circumference, and aerobic fitness). A similar protocol was followed at both time 2 (T2) and T3, where possible revisions could be made to the SMS guide. At T3, the fitness consultants also held a concluding talk with the clients about their experiences from the SMS coaching to allow the clients to give feedback and express their thoughts about their experiences. These concluding talks were not recorded or included in this study. The SMS conversations were with the consent from the clients pseudonymized and forwarded to the researchers.

After T3, the researchers interviewed the fitness consultants. The aim of these semistructured interviews was to learn about the fitness consultants’ experiences from the SMS coaching. The interviews were conducted in person or digitally via Zoom (Zoom Video Communications) or Facetime (Apple Inc) owing to the COVID-19 pandemic restrictions. The audio was digitally recorded with the consent of the interviewees and transcribed verbatim by professional transcribers.

The materials included descriptions for the SMS guide and the quantitative data collected at baseline and after the SMS intervention and waitlist. Mean values were computed for all Likert scales.

Taken together, the SMS intervention showed positive effects on clients’ behavior changes, both with regard to the quantitative and qualitative data. The clients reported positive experiences from the SMS intervention and would have liked to continue with the SMS coaching (mean 5.86, SD 1.28; range 2.50-7.00; 25/26, 96%). The fitness consultants also wanted to continue the SMS coaching and considered it complementary to the ordinary work routine. The SMS conversations verified the results from both the questionnaires and interviews, showing how personalized SMS supported clients’ behavioral changes.

The paired sample t test showed that the participants’ waist circumference was reduced after the SMS intervention (t₂₀=3.23, 95% CI 0.82-3.80; P=.004) and after the waitlist (t₂₂=2.22, 95% CI 0.13-3.87; P=.04), in comparison with the baseline. Table 2 presents all descriptive data for the significant differences in the health data at baseline, SMS intervention, and waitlist.

The ratings for the lifestyle question, level of overall health, was improved after the SMS intervention in comparison with baseline (t₂₄=−2.09, 95% CI −1.31 to −0.01; P=.047). An improvement was also observed in the clients’ level of dietary habits for the waitlist in comparison with baseline (t₂₃=−2.66, 95% CI −1.96 to −0.24; P=.01). All differences had acceptable effect sizes (Cohen d), ranging from 0.42 to 0.70 (Table 2). No other differences were observed.

The paired sample t test of the FiV questionnaire showed that both capability (t₂₇=−2.64, 95% CI −0.70 to −0.09; P=.01) and opportunity (t₂₇=−2.29, 95% CI −1.61 to −0.03; P=.03) increased after the SMS intervention in comparison with the baseline. Ratings for the relationship with FiV were higher in the SMS intervention than in the waitlist (t₂₅=2.59, 95% CI 0.06-0.52; P=.02). All differences had acceptable effect sizes (Cohen d), ranging from 0.43 to 0.51 (Table 3). No other significant differences were observed. Table 3 shows the descriptive data for the significant differences for capability, opportunity, and relationship with FiV.

The analysis of interviews with the fitness consultants revealed 3 major themes: supplement, lessons learned, and benefits.

The deductive analysis of the fitness consultants’ SMSs showed that all their SMSs can be categorized according to the COM-B model. Some SMSs are overlapping, that is, they can be categorized into 2 or even 3 of the COM-B concepts. Table 4 presents the main patterns.

The results from the inductive analysis of the clients’ SMSs revealed interesting patterns regarding their behavioral change processes. One pattern relates to failures and negative circumstances in the clients’ everyday lives, which were readily addressed when they emerged owing to the continuity of SMS coaching and seems to have contributed positively to the clients’ motivation and willingness to stick to the plan and not disrupt the behavioral change process. One of the clients stated the following:

Another finding was that the SMS coaching seems to have contributed positively to the participants’ motivation and mindset regarding healthy lifestyles, as illustrated in the following SMS examples:

The intervention also allowed for a more informal relationship between the fitness consultants and their clients and for the clients to get to know themselves better. Overall, the SMS conversations took on a friendly or even familiar tone, with some conversations ending with “hugs,” “good luck,” “wish you a happy birthday,” and so on. The SMS conversations also disclosed how the clients became aware of personal strengths and weaknesses, what worked well and what did not, and the type of support they needed:

In addition, the SMS conversations indicate that after some weeks, clients started to take responsibility for their own change work by taking initiatives and setting up goals on their own, as illustrated in the following quote:

This also demonstrates the client’s belief in their own ability to change their lifestyle and achieve the desired behavior. Overall, the SMS conversations show an improvement over time in the clients’ capabilities and opportunities (ie, what to do and when) and their motivation (ie, how they feel and think about their behaviors).

This study found that personalized SMS coaching is effective in supporting behavior changes grounded in the COM-B model and that both fitness consultants and clients had positive experiences and wanted to continue using SMS coaching. The SMS coaching resulted in reduced waistlines and improved overall health. The results also show that personalized SMS coaching enhances personal relationships (human connection) and strengthens the behavior change process (added value).

The results from all 3 data sets (ie, self-reported questionnaires, interviews, and SMS conversations) and Figure 1 show how personalized SMS contributes to the clients’ behavior changes. The connections depicted in Figure 1 pertain to the findings of this study, but they do not entirely align with the interactions and directions outlined in the COM-B model by Michie et al [15].

Personalized SMS coaching seemed especially effective in empowering the clients, as shown in the quantitative analysis where the clients’ capabilities and opportunities increased during the intervention. Thus, the intervention using the COM-B model as a point of reference [15] helped the clients feel more capable of performing the behavior and finding opportunities to do so. Fitness consultants noticed that the clients became more accountable for their behavior change process (eg, taking more of their own initiative and changes in attitudes toward behavior changes). The analysis of the SMS conversations validated this result, with several examples of how the clients displayed more initiative and willingness even under suboptimal circumstances.

Furthermore, the findings show that personalized SMS coaching complements the existing work methods at FiV and enriches the relationship between fitness consultants and their clients. The intervention established continuity and flexibility, facilitating a closer rapport between the health talks. As Figure 1 illustrates, the value of personalized SMS coaching is referred to as added value and includes aspects such as the complementary method as well as continuity, flexibility, and the turnaround situation created by the SMS coaching.

An obvious strength of this study is the use of a naturalistic sample, which facilitates the exploration of user experience and environment [21,30]. Another strength is the use of 3 types of data (longitudinal self-reported quantitative data from questionnaires and health assessments, qualitative interviews with fitness consultants after the interventions, and pseudonymized qualitative SMS conversations), which enabled us to understand the effects of the SMS coaching as well as follow the process and explore the experiences of the participants. This triangulation allowed for indirect validation of the results and a deeper understanding of the process.

A possible limitation is the use of an unvalidated COM-B instrument. As Table 3 shows, the internal consistency, as assessed by Cronbach α, was low at baseline for capability, opportunity, and motivation but increased over time. However, internal consistency can be affected by small sample size or seen as an indication of the clients’ understanding of the items, or in this case, their resources. Hence, the initial low and subsequently increasing levels of internal consistency might reflect the learning curve of the participants rather than a psychometric issue. The data triangulation supports this explanation, as both the interviews and SMS conversations point to the development of clients and their behavior changes. Therefore, the internal consistency of items (Cronbach α) is not entirely an accurate measure in this context. Low internal consistency can be a reliability problem that needs to be addressed in future studies.

The results of this study are in line with previous studies that show how strengthening a person’s capability is an important driver for behavior change [19,20,29,36]. This is also in line with several studies that have testified that personalized SMS in terms of demographics, motivation, and requested SMS frequency support behavior changes [29,32,38].

An aspect of the added value that emerged during the interviews was the importance of learning about the clients’ everyday life and the turnaround created by the SMS coaching. The clients no longer came to a meeting with the fitness consultants, but the fitness consultants became a “virtual visitor” of the clients, which created a different dynamic in their relationship. A turnaround that mainly seems to be related to the continuity and closeness of the SMS intervention created. This turnaround situation aligns with the findings of Bäccman et al [51], who explored the user experience of an automatic shower, where the users, both personnel and clients, experienced a shift in power and control when using the automatic shower. The clients’ perception of the whole shower situation changed even if most aspects apart from the actual shower remained the same (eg, getting to the shower, dressing and undressing, and getting help to dry). This small change in perception was essential for the clients’ sense of control and perceived independence. These results demonstrate the importance of user involvement in the context of use where unexpected aspects of using digital technology can be discovered [52].

As illustrated in Figure 1, human connection—and its importance for the participants in this study—can be observed in all 3 data sets. The questionnaires showed that the SMS intervention positively contributed to the clients’ relationship with the fitness consultants in comparison with the baseline and waitlist. This improved relationship was also articulated by the fitness consultants and emerged as a theme from the analysis of the interviews. In the SMS conversations, human connection was manifested as an increasingly familiar and easy tone between the fitness consultants and their clients. This result can be compared with the importance of the therapeutic alliance in psychotherapy, where the bond between the therapist and the client is a well-known predictor of the therapeutic outcome [40,53]. Thus, forming a bond with the client seems to be an important factor, regardless of the setting or type of professional relationship. However, it is still unclear whether the relationship is a mediating or moderating aspect of the SMS coaching [54].

The high level of personalization corresponds with Spark et al [37], who found that DHI should include “an element of human connection to foster ongoing participant satisfaction and accountability.” A similar result was reported by Godino et al [36], who found that SMS was most effective when combined with monthly phone calls. It is important to consider whether the human connection is sufficient to support behavior change on its own. For example, a future study can use the following two conditions: (1) personalized SMS with BCTs similar to this study and (2) only personalized coaching where the SMS is mere encouragement and reminders, thus not actually providing BCTs. If the effect of these 2 conditions is the same, personal contact can be concluded to be effective, regardless of the content or the competence of the coach. If so, personalized SMS could be used more freely and would not require personnel trained in either MI or BCT, making it easier to scale up.

In contrast to several other studies [19,20,29,36,38], we did not observe any differences in motivation between the conditions. However, as Figure 1 illustrates, the qualitative data, both the interviews and SMS conversations, imply that the SMS intervention increased motivation. An explanation for this discrepancy may be that client motivation was already high at baseline, making incremental changes difficult to detect. Thus, the difference observed is a reflection of what is known as “selection effects,” where the motivation level among the participants was higher than that of the general population, which can make it more challenging to detect changes. This is akin to the concept of “restriction of range” [55]. Another related explanation may be that the questionnaire is unable to detect changes, thus not being sensitive enough.

A third plausible explanation for the inconsistency between the quantitative and qualitative data may be that motivation was something that occurred in the interaction between the fitness consultant and the clients, making it more difficult to assess quantitatively at a specific time (ie, at T2 or T3). In other words, the noted motivation was a product of the intervention mode, that is, the fitness consultants’ experience with MI and personalized SMS [16,20,37]. Hence, it is essential to recognize that the emotional and cognitive facets of motivation are dynamic states shaped by the actions and conditions fostered by the design of a DHI, rather than being tied to a specific BCT. As noted, personalized SMS is “a mode of delivery rather than a BCT” [16]. Similarly, learning is influenced by the feedback process, where the behavior is reinforced by the observed reactions from the surrounding [11,12], again pointing toward the ever-changing possibility of motivation. However, this separation of the “the what” (eg, BCT) and “the how” (eg, personalized SMS or MI) is merely a theoretical separation, and the practical aspects are much more entwined [40].

This study supports the studies that claim that DHI needs to support social interactions [20,37]. Here, we used personalized SMS grounded in the COM-B model and personalized SMS to the client’s needs as well as used MI as a mode for delivery. To fully understand the effective aspects of SMS and any DHI, it is important to further investigate how human connections, personalization, and BCTs interact during a DHI to optimize its potential for behavior change. So far, it seems as if real or perceived human connection is an essential part of DHI.

User involvement is necessary to fully understand the effects of DHI. The results from the studies pressing for user involvement and the human connection of personalization [21,29,30] were obtained after the completion of our study, which validates the results of our study. Not involving end users in the process will likely result in more digital solutions that no one uses.

This study makes a significant contribution to the literature because it uses (1) personalized SMS, which is theoretically grounded (ie, COM-B) and tailored to the client’s specific needs; and (2) 3 types of data, allowing for indirect validation of the results and a deeper understanding of the process. We have not come across any other study with this design.

The main findings of this study show that personalized SMS coaching is effective in supporting behavior changes. This is promising because SMS is an affordable digital solution that many already have access to [5,32,33,38,47]. The fact that FiV has implemented personalized SMS coaching as an ordinary work method after this study bears witness to the scalability and functionality of personalized SMS. In other words, personalized SMS coaching is scalable and accessible to a large portion of the population, and intricate digital solutions to support behavior change are not always necessary.