Interaction Design
User Research
Usability Testing
Team of 4
Role
Yet the burden goes deeper than logistics. Restaurant discovery is fundamentally social, shaped by group outings and peer recommendations, and being the person with restrictions means contending with real anxiety and quiet exclusion before the meal even begins.
People with cultural, religious, or medical dietary restrictions still want to enjoy eating out, but finding the right place often means doing extra work with incomplete, inconvenient, or unreliable information.
People with dietary restrictions love eating out too — the tools just haven't caught up.
What Beli Hasn't Solved Yet
Extending
Beli
Restaurant discovery, for everyone at the table
Because your diet is not an edge case
What prompted us to reimagine Beli?
Eating out is one of the ways people build relationships, celebrate, work, date, and spend time with friends, but for people with dietary restrictions, that same social ritual can become stressful and isolating.
Journal of Personality and Social Psychology
Food restrictions are not only medical; they can also be cultural, religious, ethical, and deeply personal, making this more than a menu-filtering problem. It is a social inclusion problem. Beli is a growing social food app built around the idea that the best restaurant recommendations come from people you trust. Yet despite that promise, it leaves users with dietary needs to manually scan menus rather than benefit from those same community-driven recommendations. The people who need trusted recommendations the most are the ones least served by them.
Our team believes that people with dietary restrictions shouldn't feel like exceptions or need a separate app from the one their friends are using. The goal is to extend Beli so that dietary needs are centered in the experience, not treated as edge cases. This problem space is also personal to us: Three out of four team members are vegetarian due to cultural backgrounds, and all have close relationships with people managing allergies, intolerances, or chronic dietary needs
increase in reported loneliness
19%
Food restrictions limit the ability to bond with others through shared meals, making eating out a source of isolation, not connection.
52% of U.S. diners with exclusion diets struggle to find compliant meals when eating out, and half say their eating habits negatively affect their social life
(Restaurant Dive)
Food allergies alone affect an estimated 22 million people in the U.S. (AAFA), and when layered with vegetarian, vegan, gluten-free, religious, and chronic-health diets, the audience becomes far larger than a traditional "allergy filter" use case
The U.S. restaurant industry is projected to reach $1.55 trillion in sales in 2026, and discovery is becoming increasingly digital and social — 74% of diners use social media to decide where to eat
When someone cannot confidently tell whether a restaurant fits their needs, they may skip the restaurant, avoid the group plan, or disengage from the platform entirely. That creates lost trust for users, lost engagement for discovery platforms, and lost revenue opportunities for restaurants.
Why this matters
Business Insights
User Research
Key insights that shaped our design
Our Participants
To better understand this problem space, we conducted 3 semi-structured interviews (30–45 min each) with people managing ongoing dietary restrictions, spanning food allergies, vegetarian and vegan diets, halal and kosher requirements, and gluten intolerance. We chose semi-structured interviews to give participants space to share personal, nuanced experiences while allowing us to probe deeper into moments of uncertainty or stress. Sessions were held in familiar environments to encourage openness around what is often an anxiety-laden topic.
We wanted to understand three things: how people currently find restaurants, how they feel throughout that process, and what builds enough trust for them to actually commit to a choice. That meant looking at which tools and people they rely on, what workarounds they've developed, and where the experience, social and practical, breaks down. While three participants is a small sample, the consistency of patterns across very different restriction types gave us confidence in the themes that emerged.
Work-arounds: Cross-references Google, TikTok, Instagram, Reddit, and Yelp before committing to a single restaurant. "It's a whole chore."
Medical Restriction: Gluten intolerant, dairy intolerant, walnut allergy
1. Not all recommendations carry the same weight Word of mouth is already the most trusted form of restaurant discovery but for people with dietary restrictions, that trust becomes even more specific. A recommendation only feels reliable when it comes from someone who understands the same stakes. "If someone else is scared for their well-being the same way you are — you're going to trust them." This directed us toward restriction-aware social filtering within Beli's existing recommendation layer, so users can surface reviews from people who share their exact needs.
2. Critical safety information can't be found on a menu Knowing a dish is gluten-free means nothing if the kitchen doesn't handle cross-contamination seriously. Users with severe allergies depend heavily on staff knowledge and restaurant practices, information that no menu, website, or general review reliably surfaces. This directed us toward a way for users to signal and share restriction-specific experiences beyond just a star rating, so that safety-critical context travels with the recommendation.
3. Finding a safe restaurant is a fragmented multi-platform ordeal Users were cross-referencing Google, TikTok, Reddit, Yelp, and restaurant websites before committing to one meal. "I wish I didn't have to go to a separate app to verify if it was Halal or not." The opportunity for Beli is consolidating that trust into one place.
Work-arounds: Eats before group dinners rather than risk navigating menus under social pressure. "I don't say anything because I'm one person in the friend group and I don't want to change everyone's plans."
Medical Restriction: Severe nut allergy
Sahana
Cultural Restriction: Vegetarian
Work-arounds: Defaults to eating vegetarian at most restaurants rather than attempting to verify Halal status. "I just eat vegetarian, but I don't bend it."
Cultural Restriction: Halal
Aya
Maya
Preview
Eaters Like You
Eaters Like You opens Beli's social layer to a broader community of diners who share the same restrictions.
Friend groups rarely include someone with the exact same dietary need, so this feature lets users feed off recommendations from a wider, more relevant community.
Sahana — "If someone else is scared for their well-being the same way you are. You're going to trust them."
A better Onboarding
Users with dietary restrictions arrive already cautious. They need to feel heard by the app before they can trust it.
User Profiles
Knowing what someone ate isn't enough. You need to know who they are to know how much weight to give their review. We updated profiles to make dietary restrictions clearly visible, so users can instantly gauge how relevant someone's recommendation is to their own situation.
Perspective 1 — Passive, social discovery designed for the browsing mindset when you're not looking for somewhere to eat right now, but building trust in people whose recommendations you can rely on later. Users follow others with the same restrictions and organically discover places through their saves and ratings.
Perspective 2 - Active, goal-directed search designed for intent when you need a restaurant tonight and can't afford to get it wrong. The challenge here isn't just finding somewhere safe; it's finding somewhere safe that your friends will also want to go. Our users don't want to be the reason the group settles. So rather than separating dietary filtering from the core discovery experience, this perspective brings restriction-matched reviews directly into the search flow — so users can evaluate a restaurant for themselves and for the group in the same moment, without the extra research.
Eaters Like You Evolution
Sahana: "It's like a fun activity with friends because now, for example, I said another friend of mine has said she wants to go here. I can literally be like, oh, this is on your Beli. I literally eat out with her once a week. So now we can just find places we have in common and go there."
Maya: "If my friends want to go somewhere, I have to go in and check the menu myself to eat there. And if it's not something, if it's not a place that I can find food, then I just feel bad."
Our user tests revealed that participants preferred perspective 2. We learnt that our users rely on the following information when making a decision:
What dietary restrictions does the reviewer have?
Does the review provide information that helps determine whether the restaurant handles cross-contamination safely?
The difference between our before and after designs reflects our effort to make these two pieces of information more visible and accessible.
After
Made the dietary labels more noticeable
Bolded the keywords of reviews so that users grasp understand the take aways faster
Before
The questions that shaped our thinking
Before exploring directions, we identified three core tensions we needed to resolve:
Does this app even understand me?
How do I know it has understood me?
How much reassurance does someone who is ultra-careful about their diet actually need — and at what points in the flow do they need it most?
We also had to resolve a key interaction logic question: if a user selects a lifestyle, should its ingredients be selected by default? We landed on yes — selecting a lifestyle pre-selects its ingredients, and users deselect what doesn't apply. But adding a new ingredient outside that lifestyle requires an active selection. This distinction matters because it respects the difference between refining a known diet and flagging a specific allergy.
Version 1 — Checkbox list with real-time summary A more transparent, text-forward approach. Users move through lifestyle first, then ingredients, with a live summary updating as they select — giving constant visual confirmation that the app is registering their needs. This directly addressed our core design question: "How do I know the app understood me?" For users like Sahana, who described feeling anxious about repeating herself to waiters and being "so perceived" at the dinner table, seeing their restrictions reflected back in real time offered the reassurance that existing apps never provided.
Version 2 — Visual lifestyle cards with end summary A more guided, visually driven flow. Users select a lifestyle card and then refine from within it — deselecting ingredients that don't apply to them rather than building from scratch. The summary appears only at the end as a confirmation moment. This approach leaned into the nuance our research surfaced — that restrictions rarely fit neatly into one label. As Maya noted, she filters for gluten first because "it does more of the heavy work" before thinking about dairy separately. Version 2 mirrors that layered thinking, letting users start broad and get specific rather than confronting every option at once.
Onboarding Evolution
Since our tests revealed that users preferred a simple checklist onboarding process, we updated our flow to better fit this process- rather than using simply a toggle to input both lifestyle and ingredient restrictions, we adapted it into a process that takes users' through both, ensuring that they do not accidentally miss inputting important information. Based on feedback, we also added a confirmation screen that displays selected restrictions using both icons and written labels, making profiles easier to review and reducing ambiguity.
After
Before
A special thanks to Professor Philip Guo for crafting COGS 127 Data-Driven UX/Product Design. This course has helped me overcome a long-standing anxiety of creating a case study and has helped me develop my skills in presenting my work :)