Food Blend in Beli: a friend’s profile with Create a Blend, swiping to build a blend profile, the shared blend list and the blend saved to the profile

FOOD BLEND

Role
Product Designer
Timeline
Feb - May 2026 (4 months)
Tools
Figma
FigJam
FigmaMake
Notion
Slack
TurboScribe
Team
Just me!

Overview

Beli is an app that allows users to share food recommendations with friends, personalizing the food experience and building a community among foodies. As someone who loves venturing new foods, I wanted to reduce decision fatigue when deciding a place to eat. This idea was a part of an independent case study in my Digital Product Design class at Cornell.

Problem & research

HOW DOES BELI HELP PEOPLE DECIDE WHERE TO EAT?

After interviewing 4 individuals, I got a better idea on how Beli is used and how food discovery currently looks like.

INSIGHTS FROM INTERVIEW:

REVISITING THE PEOPLE PROBLEM

When I am deciding on a place to eat, I want to choose a restaurant based on personalized recommendations, so I can confidently choose a place that I will enjoy. But this is hard because:

  1. Hard to connect personalized recommendation system with friend’s activity to make a decision.

  2. Existing solutions has a place for multiple user input without a specific direction in where to take them

Market research

HOW DO EXISTING PRODUCTS PROVIDE RECOMMENDATIONS?

Yelp provides anonymous reviews with details for ratings and specific notes from random users.

Google Maps provides recommendations based on location, having the users to access reviews through location search.

Spotify Blend provide recommendations through themed and shared playlists with friends. Many of the recommended songs are based on the user’s current listening experience.

Brainstorming

From the research, I centered my solutions around answering a “How Might We” question. I identified three opportunity areas to brainstorm more specific solutions, coming up with two specific solutions in each of the spaces - six solutions total.

How might we maintain and prioritize trust while streamlining the process of transforming personalized recommendations into meaningful decisions?

After completing SWOT and feasibility & impact analysis of each solution, I decided to focus on the Food Blend and Plate Off feature where users can compare recommendations to come to a decision. Food Blend invites the user’s friends to generate a list and Plate Off allows users to compare each of the restaurants by swiping to generate a ranked list.

VISUALIZING THE SOLUTION

Hand-drawn sketches of the blend creation flow
Low fidelity sketches
Information architecture diagram for the blend feature
Information architecture
Greyscale mid-fidelity screens of the entry point, swiping and results
Medium fidelity iterations

Iterating

EXPLORATION AND DESIGN DECISIONS

1. Entry Point

Entry point option: a blend banner at the top of the feed
  • Visible and clear blurb of feature
  • Disrupts the feed
  • Placement is unrelated to feature’s function
Entry point option: a create-blend tile inside Guides
  • Clear indication of function
  • Similar to repeated components
  • Redundant of the “+” box next to it
  • Nested/hidden within “Guides”
Chosen entry point: a Create a Blend card on a friend's profile
  • Chooses the friend to create a blend with
  • Easy to revisit to see shared places
  • Visible and clear blurb of feature
  • Might be similar to “Places you both want to try”

2. Card Swiping

Swiping option: several stacked restaurant photos
  • Shows multiple images at once
  • Limited information shown at a time
  • Confusing layout because cards are split
Swiping option: one card with tags and top dishes
  • Variety of information show
  • Access to more information
  • Cluttered and too much information at once
Chosen swiping card: one large photo with the restaurant name and tags
  • Images condensed
  • Distinguished card to visually represent swiping
  • Simple and less cluttered
  • Multiple interactions: tapping and swiping

3. Friend Selection

Friend selection option: pick several friends, then share out of the app
  • Select multiple users and “create a party”
  • Leaves the app
  • Too many clicks to choose more than one person
Friend selection option: a grid of friends to tap inside the app
  • User stays within the app
  • Automatically allows multiple people selection
  • New users are hard to invite

FEASIBILITY REASSESSMENT

Since there is no messaging system to share with multiple people on Beli, I decided that friend selection would be made through each friend’s profile. Although this limits to creating the blend with one friend, it reduces the number of steps it took to create a blend.

4. Output Playlist

Output option: a blend completed preview with the top three places
  • Preview to make quick decision
  • Frictionless discovery
  • Extra step before seeing the whole blend
Output option: the full blend list with match percentages
  • Variety of choices
  • Changes between the list and map view
  • Percentage match to align with user’s taste
  • Extra details about each place

User testing

FLOW FOR TESTING

USER FEEDBACK

Based on the feedback from user testing, I adjusted and revised my solution to ensure that my design choices were intentional in solving the original problem.

Revised swiping card for Gangnam Station New restaurant details screen with scores and popular dishes
  • Multiple stacked pictures replaced with separate screen to show more details
  • Distinguished actions for more information and inputting preferences
Friend profile with the blend card Guides tab with a blend section
  • Blend available through friend’s profile and separate section in guides
  • Feature is accessible in social aspect and through multiple entry points
Revised blend list with friend avatars instead of match percentages
  • Percent number over condense opinions / could be ambiguous
  • User profiles to represent preferences

FINAL PRODUCT

Reflection

IF I HAD MORE TIME...

WHAT I LEARNED

User Empathy

It was important for me to recognize that the problem I identified early on did not encompass the full user experience. By revisiting my people problem, I am able to brainstorm solutions that center the user.

Continue Iterating!

There were moments where I was stuck when designing, but creating more iterations and talking to my friends gave me ideas that helped me brainstorm new directions of the project.

Adapt to Changes

When iterating and designing my solution, Beli developed new features that pushed me to reassess my people problem. Although the design process can sometimes be unpredictable, it has pushed me to be more creative.