College campuses are filled with ambitious, intellectually curious students, yet meaningful peer-to-peer mentorship rarely happens organically. Even when students want to connect, reaching out feels socially risky. Cold messaging someone you might see in class or on campus the next day carries a heightened fear of rejection or awkwardness.Existing platforms don’t solve this tension. LinkedIn feels overly formal and performance-driven, while social platforms like Instagram lack academic or professional context. There is no shared space that normalizes peer-driven intellectual outreach within the same campus community.
Our user ressearch revealed that even when friction is lowered through thoughtful design, engagement does not automatically follow. Sparse profiles, hesitation to reach out to strangers, and limited reciprocity from highly sought-after users reinforced emerging norms of minimal participation. The challenge was not an aesthetic UI or technical feasibility — it was social confidence, incentives, and the dynamics of a two-sided marketplace.
Project Description
A campus-based networking platform that connects students through curated coffee chats, matching them based on shared interests, goals, and academic pursuits.
Team
Katherine Sullivan
Nicole Esibov
Merve Ondogan
Darynne Lee
Nicole Esibov
Merve Ondogan
Darynne Lee
Role
Product Designer
User Researcher
User Researcher
Duration
March - June 2025
About
Problem
Opportunity
The opportunity was to create a campus-specific platform where reaching out to peers is encouraged rather than intimidating. Cardinal Connect was designed to transform “cold messaging” into mutual intellectual curiosity by establishing a shared space where outreach is normalized and socially safe. When both sides signal openness to conversation, the psychological barrier of initiating contact begins to shrink.
By surfacing academic interests, hosted chat counts, and availability, the platform aimed to make credibility and receptivity visible. This transparency reframed coffee chats from awkward requests into purposeful exchanges between peers with shared passions.
Beyond scheduling conversations, the broader opportunity was cultural. A thoughtfully designed system could increase communication across class years, reduce inequities in access to informal mentorship, and make intellectual exploration a visible part of campus life.
Our testing also revealed that sociotechnical systems must be cultivated, not simply launched. Early participation patterns shape norms quickly, and without incentives for highly sought-after users ("the hard side"), engagement can stall. We sought to maintain our platform's success by deliberately engineering reciprocity, social proof, and sustained participation.
By surfacing academic interests, hosted chat counts, and availability, the platform aimed to make credibility and receptivity visible. This transparency reframed coffee chats from awkward requests into purposeful exchanges between peers with shared passions.
Beyond scheduling conversations, the broader opportunity was cultural. A thoughtfully designed system could increase communication across class years, reduce inequities in access to informal mentorship, and make intellectual exploration a visible part of campus life.
Our testing also revealed that sociotechnical systems must be cultivated, not simply launched. Early participation patterns shape norms quickly, and without incentives for highly sought-after users ("the hard side"), engagement can stall. We sought to maintain our platform's success by deliberately engineering reciprocity, social proof, and sustained participation.
Challenge
.png)
Interviews
Research Methodology
We used a mix of quantitative and qualitative research methods to understand how students engage with peer-to-peer coffee chat systems.
First, we launched Cardinal Connect within a small atomic network of 31 students on campus. This allowed us to observe real usage patterns in a semi-controlled environment while avoiding the “ghost town” effect of a fully public launch. Through Supabase backend analytics, we tracked quantitative signals such as profile completion rates, availability updates, coffee chat requests, and request acceptance patterns. This enabled us to study actual behavior rather than relying solely on stated intent.
Second, we gathered qualitative feedback through structured surveys and informal follow-up conversations. We asked users about their emotional experience requesting and accepting chats, what influenced their decision-making, what stood out in the interface, and what friction points they encountered. These responses helped us identify underlying psychological barriers such as fear of rejection, preference for familiar ties, and hesitation around incomplete profiles.
Finally, we analyzed the system through the lens of sociotechnical theory from CS278, interpreting patterns using concepts such as descriptive norms, social proof, signaling theory, and two-sided marketplace dynamics. This theoretical framing helped us understand why lowering UI friction alone did not guarantee engagement and how early participation behaviors shaped emerging norms on the platforrm.
Together, these methods allowed us to evaluate not only whether the platform functioned technically, but whether it successfully shifted social behavior.
First, we launched Cardinal Connect within a small atomic network of 31 students on campus. This allowed us to observe real usage patterns in a semi-controlled environment while avoiding the “ghost town” effect of a fully public launch. Through Supabase backend analytics, we tracked quantitative signals such as profile completion rates, availability updates, coffee chat requests, and request acceptance patterns. This enabled us to study actual behavior rather than relying solely on stated intent.
Second, we gathered qualitative feedback through structured surveys and informal follow-up conversations. We asked users about their emotional experience requesting and accepting chats, what influenced their decision-making, what stood out in the interface, and what friction points they encountered. These responses helped us identify underlying psychological barriers such as fear of rejection, preference for familiar ties, and hesitation around incomplete profiles.
Finally, we analyzed the system through the lens of sociotechnical theory from CS278, interpreting patterns using concepts such as descriptive norms, social proof, signaling theory, and two-sided marketplace dynamics. This theoretical framing helped us understand why lowering UI friction alone did not guarantee engagement and how early participation behaviors shaped emerging norms on the platforrm.
Together, these methods allowed us to evaluate not only whether the platform functioned technically, but whether it successfully shifted social behavior.
Goal
The goal of our study was not simply to test whether students liked the UI. It was to evaluate whether a campus-specific sociotechnical system could meaningfully reduce the social risk of peer-to-peer outreach. Specifically, we aimed to:
- Test whether visible signals (bios, interests, hosted chat counts, availability) would increase confidence in initiating coffee chats
- Observe whether students would reach out to weak ties when mutual intent was made explicit
- Analyze engagement patterns between high-intent requesters and highly sought-after upperclassmen
- Examine how early participation behaviors shape emerging norms within a closed campus network
- Identify whether design alone could shift social behavior, or whether incentives were required
Personas
- The Curious Explorer
- The Experience Upperclassman ("Hard Side")
Biography
Annika is a freshman exploring a potential shift from Computer Science to Economics. She is intellectually curious and wants insight from upperclassmen who have gone through recruiting and coursework before her. She is comfortable browsing profiles but feels hesitant initiating direct outreach, especially when she might see the person around campus afterward.
Goals
- Learn from upperclassmen about academic and career paths
- Explore potential major changes with low social risk
- Connect with peers who share overlapping interests
- Feel confident that outreach is welcomed
Needs
- Clear signals that others are open to coffee chats
- Complete profiles with bios and shared interests
- Visible availability to reduce scheduling friction
- A structured environment that normalizes outreach
Pain Points
- Fear of rejection or awkwardness on campus
- Sparse or incomplete profiles that reduce trust
- Uncertainty about whether someone will respond
- Hesitation to reach out to total strangers
Biography
Joey is a junior who has completed multiple recruiting cycles and has strong academic and professional insight. He is open to helping underclassmen but is time-constrained. He is more likely to receive chat requests than initiate them and evaluates requests based on effort and alignment.
Goals
- Share insight with genuinely curious students
- Protect his time and avoid low-effort interactions
- Maintain credibility and professionalism
Needs
- Clear context behind incoming chat requests
- Signals that the requester has read his profile
- Efficient scheduling and low coordination
- Incentives or recognition to validate his contribution
Pain Points
- Receiving vague or low-effort requests
- Lack of meaningful reciprocity
- Overwhelming number of requests without filtering
Social Computing Theory & Design
Design Process
Our design process began with a central tension: students are surrounded by potential mentors and collaborators, yet peer outreach feels socially risky. We approached Cardinal Connect as a sociotechnical intervention rather than just a scheduling tool. Instead of focusing purely on usability, we asked how interface decisions could shape norms, signal intent, and reduce perceived vulnerability.
We rapidly prototyped and deployed the platform using Lovable, allowing us to “vibe-code” the front end and iterate quickly, while Supabase powered authentication, database management, and row-level security on the backend. This combination enabled us to move from idea to live deployment within weeks and observe real behavioral data from a 31-person atomic network.
Throughout development, we continuously tested whether visible signals, structured profiles, and campus-specific aesthetics could shift behavior — not just interface interaction. Our goal was to design for trust, reciprocity, and norm formation from the outset.
We rapidly prototyped and deployed the platform using Lovable, allowing us to “vibe-code” the front end and iterate quickly, while Supabase powered authentication, database management, and row-level security on the backend. This combination enabled us to move from idea to live deployment within weeks and observe real behavioral data from a 31-person atomic network.
Throughout development, we continuously tested whether visible signals, structured profiles, and campus-specific aesthetics could shift behavior — not just interface interaction. Our goal was to design for trust, reciprocity, and norm formation from the outset.
Norms
Social norms played a central role in our system design. Drawing from descriptive norms and social proof theory, we aimed to create a space where outreach felt expected rather than awkward. By requiring Stanford email sign-in and mirroring Stanford’s visual identity, we intentionally signaled shared community membership. Profile bios, photos, and hosted chat counts were designed as credibility signals to normalize participation.
Affordances
We designed Cardinal Connect with intentional affordances to reduce friction and clarify mutual intent. Filters by major, interests, and class year made discovery purposeful. Recurring availability signaled openness. Hosted chat counts and reviews acted as trust signals. The “Request Coffee Chat” flow structured outreach in a way that reduced ambiguity and social risk.
Cold Start
To address the cold-start problem, we intentionally launched within a small atomic network of 31 users. This approach aimed to seed high-quality engagement and prevent the “ghost town” effect common in early marketplaces. We hypothesized that visible early activity would generate social proof and encourage participation.
While this strategy helped create initial momentum, it also revealed a limitation: when early adopters joined out of personal obligation rather than intrinsic motivation, engagement data became skewed. Once participation extended beyond our immediate network, the platform’s true engagement dynamics emerged. This reinforced that cold start solutions must balance density with authentic incentive alignment.
While this strategy helped create initial momentum, it also revealed a limitation: when early adopters joined out of personal obligation rather than intrinsic motivation, engagement data became skewed. Once participation extended beyond our immediate network, the platform’s true engagement dynamics emerged. This reinforced that cold start solutions must balance density with authentic incentive alignment.
The Hard Side
Our system exhibited classic two-sided marketplace dynamics. The “easy side” — high-intent seekers — actively requested chats. The “hard side” — experienced upperclassmen whose profiles attracted interest — had less intrinsic incentive to engage. Without enough high-value participants actively responding, overall momentum slowed.
We attempted to account for this by including hosted chat counts, public reviews, and profile visibility to provide social validation and legitimacy. However, our rollout revealed that functional transparency alone does not guarantee participation. Conquering the hard side requires clear incentive structures, recognition mechanisms, and perceived return on time investment.
This insight became one of our strongest theoretical takeaways: sociotechnical systems are not simply built through features, but sustained through balanced incentives and cultivated norms.
We attempted to account for this by including hosted chat counts, public reviews, and profile visibility to provide social validation and legitimacy. However, our rollout revealed that functional transparency alone does not guarantee participation. Conquering the hard side requires clear incentive structures, recognition mechanisms, and perceived return on time investment.
This insight became one of our strongest theoretical takeaways: sociotechnical systems are not simply built through features, but sustained through balanced incentives and cultivated norms.
Demo
Reflection
What did I learn working on this project?
This project fundamentally changed how I think about social platforms. I learned that building a sociotechnical system requires more than clean UI and functional features — it requires intentional norm design and governance from day one.
Through deploying Cardinal Connect, I saw how quickly descriptive norms emerge and solidify. Sparse profiles became contagious. Minimal effort participation shaped expectations. Engagement asymmetry between seekers and the “hard side” slowed momentum. These behaviors were not bugs; they were predictable outcomes grounded in social computing theory.I
learned that moderation, norms, and incentives cannot be afterthoughts. Trust signals, reciprocity, and participation standards must be designed deliberately from Day 1.
Through deploying Cardinal Connect, I saw how quickly descriptive norms emerge and solidify. Sparse profiles became contagious. Minimal effort participation shaped expectations. Engagement asymmetry between seekers and the “hard side” slowed momentum. These behaviors were not bugs; they were predictable outcomes grounded in social computing theory.I
learned that moderation, norms, and incentives cannot be afterthoughts. Trust signals, reciprocity, and participation standards must be designed deliberately from Day 1.
What would I have done differently?
Looking back, timing was one of our biggest constraints. Launching near the end of the quarter meant users lacked bandwidth to fully engage. In a future iteration, I would align rollout with a moment of genuine demand, such as the beginning of a quarter when students are actively seeking connection.
I also would have enforced stronger profile completion standards during onboarding. Allowing incomplete bios and missing photos unintentionally normalized minimal participation and weakened trust signals. More intentional norm-setting early — such as seeded exemplar profiles or completion requirements — could have shaped behavior differently.
Finally, I would design stronger incentives for the “hard side.” Hosted chat counts and public reviews were not sufficient motivation. Clear value exchange and visible recognition mechanisms would be necessary to sustain high-value contributors..
I also would have enforced stronger profile completion standards during onboarding. Allowing incomplete bios and missing photos unintentionally normalized minimal participation and weakened trust signals. More intentional norm-setting early — such as seeded exemplar profiles or completion requirements — could have shaped behavior differently.
Finally, I would design stronger incentives for the “hard side.” Hosted chat counts and public reviews were not sufficient motivation. Clear value exchange and visible recognition mechanisms would be necessary to sustain high-value contributors..
How has this experience prepared me for my next project?
Bootstrapping a live social platform grounded explicitly in social computing theory was a uniquely formative experience. Rather than abstractly learning about norms, signaling, cold start dynamics, and two-sided marketplaces, I observed them unfold in real time.
This experience sharpened my ability to think beyond feature design and toward behavioral systems design. I now approach social products with a deeper understanding that early participation patterns shape long-term culture, that incentives must be balanced across sides of a network, and that governance structures determine sustainability.
As I move into future projects — especially those involving networks, multi-sided marketplaces, or community-driven platforms — I will carry these principles forward. Sociotechnical systems are not simply built through functionality; they are cultivated through intentional design of norms, incentives, and trust.
