For this study, we turned to Buddy Project, an online peer-to-peer support system that aims at fostering relationships that provide support to those struggling with mental illness. Buddy Project was founded in 2015 by Gabby Frost, a young person who wanted to prevent suicide and self-harm while advocating for mental health. Buddy Project’s mission to prevent suicide is now displayed on the website’s home page and in its Twitter and Instagram biographies (Figure 1). So far, the nonprofit movement has paired more than 236,000 adolescents aged between 12 and 25 years. Participants are matched based on age difference and shared interests and connected using their social media accounts (Twitter or Instagram). The sign-up process is shown in Figure 2, where users choose the social media account they want to use. Peers are then redirected to a Google form. On the first page, they provide their email address, first name, and Twitter or Instagram username, depending on the sign-up process they selected. On the second page, peers rank their first to fifth interests. A complete list of the interest options is included in Multimedia Appendix 1. We developed higher-level themes for these interests to gain a better sense of what they include. Interests categories include arts and entertainment (eg, musicians, television shows, books), identity (eg, religion, political ideologies, gender, and sexuality), and time zones. The third page prompts peers to provide their age. The last page informs peers what to expect once they submit the Google form. It states that they will receive an email once paired with their buddy, that their selection is not automatic, and that peers are welcome to sign up for multiple buddies, along with the links to Buddy Project’s website, Instagram, Twitter, and Facebook accounts.

Buddy Project’s founder manually pairs all of the buddies using a Microsoft Excel spreadsheet. When pairing buddies, they take into consideration the interests they selected when signing up and their age. If someone wants to refuse a buddy, they can simply do so by not connecting with them once paired, and they can sign up to be paired with someone else. There is currently no formal training offered to buddies because buddies sign up to make friends who understand them (and thus can provide them with peer support), not provide or receive any type of professional help or crisis counseling.

We conducted in-depth semistructured interviews with 13 Buddy Project users.

The recruitment process began with a screening survey that the Buddy Project organization shared on their Twitter and Instagram accounts. To participate in the survey, participants needed to (1) be a social media user, (2) live in the United States, (3) be at least 18 years old, and (4) be a current or previous user of Buddy Project. Being a social media user was important for us to situate Buddy Project within other online support systems that participants may have used. If they met these qualifications, participants were asked to complete the second portion of the survey that asked questions about demographics, social media use, length of longest relationship with buddies, number of buddies they had matched with, month and year when they first used Buddy Project, and overall experience with their buddies. We did not screen participants on the basis of mental health status or diagnosis, as we were interested in learning experiences with Buddy Project, and Buddy Project does not match buddies based on the diagnosis. The survey was open from June 12 to 25, 2019, and received 123 responses. From the responses, 63 participants met the initial criteria, 38 were invited to participate in the study, and 13 were interviewed. We sent an interview invitation to all qualified participants who had consistent survey responses (eg, the time of their longest relationship with a buddy did not extend past the date of when they started using Buddy Project) and had used Buddy Project within the last 2 years. We also purposefully recruited participants from diverse demographics to the extent possible. The invitation included study details and an online consent form. A total of 13 individuals completed these forms and participated in the interview. Table 1 includes the details of participants. We continued recruiting participants through the interview process and stopped when no new themes emerged. We offered a US $15 gift card to interview participants. This study was approved by our institutional review board.

Apart from assisting with recruitment, Buddy Project had no other role in any part of this study. We plan to share this paper with the founder after peer review.

Participants included 10 women, 1 man, 1 nonbinary, and 1 gender-fluid person. The average age of participants was 20.3 (SD 2.25) years (range 18-25 years). In total, 6 participants lived in rural areas, and 7 lived in urban areas; 9 participants were White, 1 was Black, 1 was Latina, 1 was Asian, and 1 was South Asian. A total of 7 participants had previously used Buddy Project but were no longer active users and 6 were current users. The number of buddies ranged from 1 to 10, and the length of contact with a buddy ranged from 1 month to 4 years. A total of 11 participants described their experience with Buddy Project as being positive overall, whereas 2 described it as both positive and negative.

We conducted in-depth semistructured interviews that allowed participants to provide details about their experiences with their buddies while still covering topics regarding their experience with Buddy Project. The interviews were conducted via participants’ preferred method of voice or video call, and the audio was transcribed for analysis. The average length of the interviews was 47 min (range 28-56 min).

The interviews began with the interviewer sharing the goals of the study and asking for permission to record the conversation. The participants were then asked about internet and social media use before diving into their stories and experiences with Buddy Project. We asked to hear the story of how the participant found Buddy Project, why they started participating, why they continued or stopped connecting with their buddies, what the relationship with their buddy was like, and what topics they discussed. We asked questions about perceptions of the shared interest feature and their best and worst buddy experiences. We continued with questions regarding how their buddies compared with other friends, online and offline, and when relevant, they were probed about how they used Buddy Project to cope with mental illness or distress and how this compared with their other coping mechanisms (if any). It was important to us that participants only discussed experiences they felt absolutely comfortable with; therefore, we only probed on specific mental health experiences only if they mentioned the topic organically. We paid particular attention to a unique feature of Buddy Project, that is, the way buddies are matched based on the interests they select when they sign up. The interview protocol is available in Multimedia Appendix 2.

We started coding the data with an open coding procedure using Dedoose—a software for analyzing qualitative data. We followed the constant comparative approach [54], where we looked for patterns, consistencies, and differences in the data in an iterative manner. Specifically, one author coded one interview. Then, the 2 authors met with each other to discuss and refine the codes. The same author then coded 4 more interviews, for a total of 5, and the authors met to sort codes into themes and to further discuss and refine each code and theme. The same author then coded the next 8 interviews with these codes in mind, and the 2 authors met frequently to continue refining the themes. No further codes or themes emerged in the latter process.

Mental illness can prevent people from feeling included and connected with peers, often categorizing those diagnosed as minorities, creating even more disconnection from one’s broader social environment. Overcoming stigma is often the first step to seeking and receiving support for mental illness [58]. Creating a safe space for discussing mental illness often requires a commitment to destigmatization, and Buddy Project is doing exactly that with their explicit mission to prevent suicide and commitment to raise awareness for mental health via their social media outlets. When asked what she thinks about the positive messages on Buddy Project’s Twitter and Instagram, P10 responded as follows:

Relatedly, P3 described Wallpaper Wednesday as “weekly wallpapers for your phone and your mobile devices.” He continued as follows:

He added the wallpapers remind him that his feelings were valid:

Buddy Project’s mission and how it was exemplified in the organization’s social media accounts in practice had immediate importance to many participants.

The mission statement is the first piece of information that one sees upon entering the website, and the statement is embedded in both their Twitter and Instagram bios. P5 testified to the need for a safe conversation space when she said:

Echoing a similar point, P4 described how having a mission related to suicide prevention, or mental health, allows discussion of it to become less of a taboo topic with her buddies compared with other people she may meet online.

She said the following:

The explicit framing of Buddy Project allowed less perceived stigma between buddies to discuss mental health, resulting in more discussions of mental health compared with participants’ other social settings online and offline. We see how having a mission to raise mental health awareness and demonstrating that in action (eg, through providing resources or social media posts) is an important characteristic for technology-mediated peer-to-peer support systems that want to encourage participants to discuss mental health and exchange support.

We found that another feature that participants deemed helpful with Buddy Project was how peers are matched on the basis of shared interests and identities (eg, music, television shows, identity). Our analysis suggests that shared interests act to promote conversation between new buddies who are otherwise technically 2 strangers at the time of being paired. In addition, salient shared identities help with ensuring that one will be understood when discussing mental health and intersections of their identities with mental health.

Participants’ accounts suggest that shared mental health diagnosis may not always be the best way to match peers who need social support to cope with mental illness. We found that buddies typically base their first conversations off their stated shared interests. As previously described, buddies’ experiences have been positive with the shared interest feature, whereas the mental health orientation of Buddy Project creates a shared understanding that discussing mental health with buddies is not taboo. Participants noted that when the shared diagnosis replaces the shared interest, the dynamic of the relationship changes. The friendship aspect is lost, as discussing mental health diagnosis is not how people begin to form friendships with one another; of course, once a relationship exists, discussing mental health is something participants feel more comfortable with. In this section, we describe how participants reflected on the perceived benefits and drawbacks of being matched with someone that shares their same mental illness diagnosis. These include concerns about differences in the severity of buddies’ experiences, engaging in self-comparison, and sharing unhealthy coping mechanisms.

Although Buddy Project is not the most technically sophisticated peer support system, we found that its design principles are effective in facilitating relationships that help young adults cope with mental illness by engaging in peer support. By taking Buddy Project as a case study that uses technology to pair up young people to cope with mental illness, we provide insights into the qualities possessed by effective technology-mediated peer support systems. We found that designing a technology-mediated peer-to-peer support system with a commitment to mental health awareness is an effective way to create a coping tool for young adults facing mental health challenges. This clear commitment lowers the stigma attached to mental health discussions between buddies and provides a shared understanding that sharing mental health–related experiences is a legitimate need that can be met within this peer support system.

Future digital peer support tools for mental health can use a combination of the features that Buddy Project users found helpful: (1) semiguided chat spaces made possible through matching based on interests and identities that peers self-identify with and having an explicit mental health–related mission coupled with social media and other online presence that makes it clear that discussing mental health is safe within the peer support ecosystem; (2) not matching based on broad mental health diagnosis; however, if diagnosis is important to peers to incorporate such matching with educating peers on how to provide helpful support and how to avoid unhelpful coping mechanisms and accounting for symptom severity in matching; (3) allowing for some degree of anonymity and control over how peers present themselves to each other; (4) allowing potential peers to assess the fit and whether they would like to embark on a relationship with their peers based on browsing public social media profiles of one another or innovating privacy-preserving means of assessing initial fit and trust. We describe these in detail in the remainder of this section.

The participants in this study expressed concern about the potential costs of being matched based on shared diagnosis—comparing one’s self and severity of illness and sharing unhealthy coping mechanisms. In fact, individuals with mental illnesses do not always prefer for their peers to have the same diagnosis as theirs [15]. This finding resonates with previous studies, suggesting that sharing unhealthy coping mechanisms is one of the most likely disadvantages of online support groups and peer support [38,55,60,66]. This can include suicidal ideations being shared among peers, and even the possibility of a suicidal pact, and an overall diminished self-esteem and well-being [55]. In the case of depression, which affects an estimated 25% of young adults [67], hearing a peer’s depressive thoughts can cause a downward spiral [56]. Being matched based on the same diagnosis can also lead to lower self-esteem and hope when people engage in self-comparison and feel as though others are doing better [16], which counteracts the peer support that one should be receiving [17].

However, with the right system and the right peer match (eg, facilitated through shared interests and identities), sharing healthy coping mechanisms could be a positive experience for participants as it acts as an outlet to seek and provide helpful mental health resources and provide a sense of solidarity and connection to peers [22]. For example, if a participant has been seeking help from a therapist, they could suggest their peer to do the same and support their claim with personal experience, making it more likely that their peer seeks additional mental health support. For this to be possible, we argue that individuals would have to be educated on what healthy and unhealthy coping mechanisms are. In addition, when matched based on diagnosis, peers can provide support to each other around common stressors, symptoms, stigma, and other challenges, resulting in them feeling less alone [58].

A useful approach that addresses the concerns and allows users to receive the benefits of shared diagnosis pairing could be offering educational material to peers that explains to them the potential dangers of these unhealthy actions and instead educating them on healthy coping mechanisms. There is further evidence for an unmet demand for such educational programs, as many individuals express interest in training to become a peer counselor [68]. For instance, the 7 Cups of Tea website educates users on active listening and support provision; similarly, Crisis Hotlines trains their volunteers.

Technological spaces such as Buddy Project or other services dedicated to mental health advocacy and awareness can use their online presence (eg, social media posts) to provide such educational material. They can also provide a set of resources to buddies once they are matched, educating them on healthy coping mechanisms and how to be a supportive, compassionate buddy. The effectiveness of any such approach should be investigated in future research. Our study provides initial insights that there is room for improvement when it comes to engaging in healthy behavior in technology-mediated peer support systems for mental health when a shared diagnosis is a factor in matching peers.

Privacy concerns and the risk associated with sharing information are key challenges for internet users because they can make them more reluctant to sharing information, asking for support, and building connections with other users [69]. It can also be difficult to reach out to strangers with whom one has no pre-existing ties to find support. One of the reasons Buddy Project is a popular choice among young people seeking online peer support is the sense of anonymity and separation from participants’ existing and known networks (eg, family, classmates). They are not completely anonymous, as buddies have access to their peers’ Twitter or Instagram account and share varying levels of personal information with each other; however, there is a sense that they are unlikely to have a pre-existing tie with their buddy. Therefore, their buddy would also not know anyone in their existing social network. There are also no requirements for sharing physical world names or identities, as they are in platforms such as Facebook. As a result, buddies feel they can safely share their mental health struggles in confidence, without worrying about others’ undesired access to their personal information and struggles.

A sense of anonymity facilitates openness in self-expression, support seeking, and support provision around sensitive topics [70-72]. Our findings suggest that control over how to present oneself (eg, using a pseudonym or just the first name) coupled with separation from networks of known ties (ie, existing social connections such as friends and family) and some contextual information through access to the prospective buddy’s social media accounts, or shared interests, helped peers to take the first steps in a relationship with a buddy. Once there was more trust developed in the relationship, participants revealed more and deeper information about themselves, consistent with the social penetration theory [73]. This theory suggests that relationships develop and become more intimate over time as people share more intimate information with each other and move away from solely shallow information [73]. Here, we see how access to prospective buddies’ social media profiles before conversing with them for the first time in tandem with the mental health mission and shared interests matching provide a fruitful sociotechnical space for developing deep supportive relationships with buddies.

Giving individuals who sign up to be paired an option to add their social media information is how Buddy Project conducts its matching process; however, users do not need to have public profiles. We learned that being able to assess their proposed buddy before connecting with them through the availability of some signals on their social media was helpful to participants. However, this would be harder for those whose online profiles are private or do not include much information. We suggest that Buddy Project or similar systems provide ways for their users to assess their buddies before deciding to connect; this does not have to occur using social media platforms. For example, an alternative would be creating a profile within the system that has an about me page, how long the person has been a user, buddies paired with, and any other information that users choose to share about themselves.

Some drawbacks of connecting online to find peer-to-peer social support are privacy concerns and cyberbullying [31]. Our findings demonstrate privacy concerns but not much about cyberbullying concerns—an important area for future research. In summary, our findings suggest that peer support systems should consider allowing some degree of anonymity, separation from existing social networks, and control over presentation, along with providing tools to assess and maintain safety and initial fit with prospective buddies before connecting with them.

We contribute to an understanding of desirable features for digital mental health peer support systems: (1) matching peers on the basis of interests and identities they identify with; (2) having an explicit mental health mission coupled with other online presence to signal that discussing mental health is not a taboo within the digital peer support system; and (3) not matching peers based on broad mental health diagnosis; however, if diagnosis is crucial to account for, accounting for illness severity and educating peers on how to provide support while avoiding suggesting unhelpful coping mechanisms; (4) allowing for some anonymity and control over how peers present themselves to each other; and (5) providing relevant information and tools to potential peers to aid in their decision in connecting with the proposed peers before connecting with them.

As is common in interview studies, this study’s sample was not representative of Buddy Project users or their experiences with the service, even though we sought a diverse participant group in demographics and experiences. Future work is needed to evaluate our findings with a larger and representative population, possibly through other methods (eg, survey of Buddy Project users). Nonetheless, following best practices in interview research, our goal is not generalizability [74], rather generating conceptual insights. The challenges faced by individuals with multiple intersecting marginalized identities are worthy of future exploration. For example, although we note that connecting over identities is helpful, future research should explore what an ideal connection would look like when multiple identity facets are concerned.

Our study does not account for all kinds of mental illnesses, and further research is needed to examine the similarities and differences in designing for various mental health peer support systems. Nevertheless, we identify the factors that such designs should consider.

Within our interview participants, we did not encounter anyone who mentioned that they had encountered fake profiles or those with malicious intentions. However, this does not mean that such harmful interactions do not happen. It is possible that those willing to interview with us had more positive experiences. For example, Buddy Project shared our study link on their social media accounts. It is possible that those who follow Buddy Project on social media have had better experiences with the platform (although we attempted to recruit individuals with both positive and negative experiences). However, examining how to design technologies to facilitate trust and safety while reducing harmful behavior is an ongoing area of research that is beyond the scope of this study.