TasteBuds

AI-powered dining recommendations shaped by your tastebuds and the people you trust.

Involvement

Product designer
UX researcher
UX/UI designer
Service designer
Interaction designer
UX writer

Disciplines

UX/UI design
Growth design
Branding
Product strategy
Monetization strategy

Timeline

Dec 2022 - June 2023

Tools

"Why is it so hard to find a place to eat?"

Diners trust friends over strangers online, but you can't wait for your group chat to respond every time you're hungry. You want somewhere you'll actually like, but sometimes you don't even know what you're craving. TasteBuds pairs AI recommendations based on preferences, dining choices, and input from friends to help diners decide faster. I designed the product end-to-end over 6 months: research, three connected user flows, a high-fidelity prototype, and a B2B monetization strategy.

So, how might we help diners make quick, confident decisions?

Dining recommendations shaped by your taste and the people you trust.

AI weighs each diner's preferences, dining history, and cravings against private reviews from friends, delivering a confident recommendation before the group chat responds.

A B2B revenue model restaurants opt into, not advertise on.

Restaurants subscribe for exclusive in-app features: promotional offers, performance analytics, and diner insights. TasteBuds runs no ads, so no restaurant buys its way into a diner's feed. Recommendations stay driven by preferences and friends.

Before conducting primary research, I reviewed secondary sources on consumer dining and choice paralysis, including industry reports, peer-reviewed studies, and Google Keywords.

Here are a few key things I learned about emerging and evolving diner habits and preferences in 2022:

73% of diners use reviews, yet only 24% are influenced by them.

Word-of-mouth is more trusted, but far less accessible.

Secondary Research Findings

Most diners visit a restaurant's website and online menu first.
Good food outweighs good service every time.
Promotions play a significant role in influencing decisions.
Friends' recommendations consistently outperform anonymous reviews.

These are the top 3 decision factors:

By the end of my secondary research, I began wondering: Why don't diners trust online reviews? Why do diners still resort to them?

Drawing from my secondary research, I formed a few assumptions to validate in user research.

Diners are most frustrated by limited discovery and slow decision-making.

↗ Refined

Overcommercialization and fake reviews create decision paralysis.

✓ Validated

Diners want a trustworthy, personalized, food-focused alternative.

✓ Validated

These helped me craft a screener survey that recruited diners for interviews and collected qualitative and quantitative data on 4 different aspects of dining out. I kept these assumptions at the back of my mind as I navigated the rest of user research, as I didn't want them to create bias.

I surveyed 20 participants and conducted 7 semi-structured interviews with the most dissatisfied diners.

At a high-level, my survey found that the largest painpoint for diners is making decisions. My findings really helped me structure my user interviews to delve deeper into diner experiences.

Diners' satisfaction levels: (1) Methods used for dining decisions, (2) time spent making decisions, (3) new dishes tried past 4 weeks, and (4) frequency of trying new restaurants.

Completely satisfied

Mostly satisfied

Somewhat satisfied

Neither

Somewhat dissatisfied

Mostly dissatisfied

Completely dissatisfied

The interviews were very insightful! Using FigJam, I synthesized the interview data with an affinity map.

Below is a summary of what I unconvered on diners' delights and painpoints:

There's a certain emotional trust about talking to someone about it...

If I had a magic wand, I’d want to receive suggested items that would suit my tastes.
If I had a magic wand, I’d want to receive
suggested items that would suit my tastes.

Conclusion

My research validated all of my initial assumptions except for one. Diners are more focused on their struggle to confidently making decisions than discovering new dining options. As I expected, discovery is still a significant pain point, but decision-making takes precedence because it is the most critical roadblock in their journey.

To understand how current tools shape the dining out experiences, I analyzed Yelp and Google Maps.

I assessed the features, user flows, content strategy, and information architecture of their search and review flows.

Competitor Analysis Findings

Yelp: Cluttered UI, excessive sponsored content, and complex navigation.
Google Maps: Intuitive interface but lacks filtering and sorting options.
Yelp & Google Maps: Becoming more system-centric (overcommercialization), not tailored specifically to diners.
Word-of-mouth recommendations: Most reliable but most inaccessible.

To better guide the Define stage, I created a mind map to connect my research findings and focus my insights.

In sum, diners lack confidence due to unreliable content and inaccessible word-of-mouth recommendations.

Given the diversity of diners' needs and behaviors, I created 3 user personas to better understand my target audience and guide my ideation phase: Alex, Brandon, and Cleo. The mind map I created earlier helped me clarify the distinct pain points and delights of each persona!

Spoiler alert: In hindsight, this was the beginning of my scope creep. Eventually, I narrowed my focus to Alex and Brandon to ensure a more targeted and effective product experience.

Alex

I love dining out. I get to spend quality time with friends and family while enjoying and discovering delicious food . Sure, I use Yelp and Google Maps, but I trust word-of-mouth recommendations over online reviews to know what's actually good . Authenticity and quality are key!

Brandon

I prioritize convenience over discovery due to my hectic work schedule. I dine out or order takeout for quick, good-value options , especially when working late or when I don’t have time or energy to cook.

Cleo

I love dining out to discover new, trendy places and spend quality time with friends and family. Convenience doesn't matter —I’m willing to go the extra mile if the experience is worth it! So I don’t dine out just for the food: As a remote worker, I look for aesthetic coffee shops with WiFi and outlets, so accurate online details and high-quality photos essential.

Despite their differences, Alex, Brandon, and Cleo have something in common: They feel overwhelmed by endless search results that are irrelevant to their needs and desire more personalization in their search methods.

How might I empower diners to confidently discover dining options and make decisions?

To narrow the scope, I created user stories to help me empathize with the needs of Alex, Brandon, and Cleo and start envisioning how my solution could empower them. Here are the highlights:

I want to easily get recommendations from friends, so I can confidently try new places.

I want personalized recommendations, so I can discover and explore options that suit me.

I want to search for options with more customization like filters, so I can avoid decision paralysis.

I want to write reviews that are visible to my friends only, so I can share my opinions comfortably.

Since diners typically explore options with Yelp and Google, I decided to prototype a mobile-app for my solution. Using the new diner stories, I brainstormed features and conceptualized 3 user flows to visualize how diners like Alex, Brandon, and Cleo would interact with the product to enhance their dining experience.

Diner signs up and sets up their account

Create account via email or social
Input food preferences
Add friends

After a standard sign-up process, diners are guided to input their food preferences and add their friends by linking social media accounts and contacts. This facilitates personalized dining recommendations and access to recommendations from friends.

Diner explores options

Saved search presets
Personalized recommendations
Compatibility score (% Match)
Limited refreshable results

To explore dining options, diners navigate a search flow facilitated by a new or saved search with food-focused and user-based filters. After inputting preferences, diners will find a finite list of results sorted by a compatibility score.

Diner reviews their dining experience

Rate restaurant and dishes
Strengths and weaknesses
Recommendation feedback
Written review (optional)

Diners follow a step-by-step review flow focused on providing feedback on the restaurant and dishes, featuring a unique 0-5 star rating system in 0.5 increments. Since reviews contribute to AI recommendations, diners can make any review private, allowing them to improve suggestions without sharing select opinions with friends.

Using the user flows, I began developing my app's information architecture. I planned on implementing a bottom navigation bar, so I first brainstormed the app's top-level destinations with a few secondary-level content:

Home

Cuisine quick links
AI recommendations
Offers & promotions

Social feed

Friends' reviews
Inbox
Write a review

New search
Saved search
Search results

saved

Saved restaurants
Create collections
Saved reviews

profile

User's reviews
Food preferences
Account settings

View full site map.

While sketching how my ideas would come together for diners, the challenge was ensuring the user flows would compose an efficient and resilient ecosystem for the solution. Essentially, all of my user flows are connected: Each user flow supply and utilize user-generated data from each other. For example, I spent a lot of time figuring out the user data needed to tailor dining recommendations and exploring how the user flows would individually and collectively support these data sets.

Guiding questions

What critical user data is needed to tailor dining recommendations?
How can the search feature optimize and empower dining decisions?
How can I minimize cognitive overload while maximizing engagement for the structured review flow?

How can the search feature optimize and empower dining decisions?

Four sketches of a mobile app interface for finding Taiwanese restaurants.

Before moving onto low fidelity explorations, I quickly prototyped my sketches and conducted 5 guerilla usability tests for the search and review flows.

Successes

All users praised the personalized recommendations
All users mentioned and affirmed the restaurant "deal" feature
All users found the search and review flows intuitive
All users wished this tool existed

Pain points

3 users didn't notice the map behind the list of results
4 users struggled with the slider design to rate stars

Notes

1 user mentioned they avoid ingredients they dislike
1 user said they use the map for location-based searches

View complete test report ->

Using my testing insights, I iterated my sketches into low-fidelity wireframes and designed new screens.

I was worried about drop-off rates during onboarding given the multistep process, so I designed a new linear process supported by a progress bar and a modal checkpoint . To further support AI recommendations, I added a step for users to input ingredients they simply dislike but are not allergic to.

My iterations for the search flow focused on facilitating intuitive explorations of dining options, prioritizing scanability and above-the-fold design. I designed the home page (for you), a new split-view for results, a new restaurant preview carousel, and wireframes for each restaurant subpage. The most challenging screen to design was the reviews page due to the amount of content each review needs to display.

My iterations for the review flow aim to make the process more engaging and streamlined. I reduced the flow from 7 screens to 4 screens. I enhanced and added a few features to support the process, such as a progress bar, option to save and finish later, and natural language copy. I also redesigned the star rating interaction and standardized the rating system with descriptive labels for each level.

Iteration

In branding my solution, I wanted to foster warmth, friendliness, trust, excitement, and a sense of community with my color palette.

Initially, I considered red, which is used by McDonald's and Yelp to stimulate appetite and spending. However, since my app focuses on building trust through personal recommendations, I chose a sunflower yellow to encourage social interaction and exploration, paired with sky blue for familiarity in social media.

Spoiler Alert: I later realized my typography was too small. It was a pain to readjust my wireframes since I didn't know to use Figma's public library tool. This was a hard lesson to learn!

Color Palette

Typography

Buttons

Search bar

Badges

8-point-grid

With my new style guide, I designed high-fidelity wireframes and prototyped them for my next round of usability tests. A key objective in my high-fidelity explorations was enhancing accessibility, so I reviewed Apple's Human Interface Guidelines for Accessibility and Web Content Accessibility Guidelines (WCAG) . My learnings guided me to significantly increase spacing and physical interactions when possible , supporting both users with limited mobility and power users.

I implemented more conversational UI and visual aids and further streamlined the user flow.

Introduction screens

Food preferences

I enhanced my business goals by conceptualizing restaurant partnerships and their role in the diner journey! I enhanced social proof elements, added more engagement features, and developed the 'Redeem offer' screen.

Home page

Search process

Search results

Restaurant listing

Reviewing menu items

Diners can add media and tag them with menu items and/or friends so users can quickly look up dishes.

Tag media with dishes and people

To test my high-fidelity prototype and gather feedback for iterations, I conducted 5 remote moderated usability tests. The participants were diners between the ages 21 and 30 who dined out at least once a week and used at least one online method to find dining options.

Tasks

Onboarding : Create an account using your Facebook login.
Search : Find a restaurant to dine at within 3 miles of your current location.
Review : Write a 4.5-star review with a photo for Olives Branch Bistro, where you had a Mixed Kebab Plate.

Successes

All users praised the visual and interaction design.
All users prefer TasteBuds over tools like Yelp.
All users loved the personalization and customization features
All users found TasteBuds informative, diverse, and inclusive.
All users reported the visual aids enhanced scanability and encouraged them to complete the flows

Pain points

All diners misinterpreted the app premise by the end of onboarding.
4 diners overlooked and/or were confused by the onboarding copy.
3 diners struggled to rate stars.
3 diners struggled to find the review page and found the flow lengthy.
3 diners missed the change between ingredient preference steps
2 diners assumed preferences limit discovery.

Notes

2 users explored the home page to search for dining options
1 user searched for dining via cuisine filter buttons on the home page
Most users expressed they prefer detailed reviews
Findings remain consistent with primary and search research insights

With new insights, I enhanced my prototype with iterations to resolve the user pain points and confusion, while expanding my business partnership concept and its integration into the user flows.

Methodology

For my final round of testing, I conducted 5 remote moderated tests with my iterated prototype. The participant profile and assigned tasks was the same as the previous test. I was very interested in seeing if there are any unintended use cases.

Affinity Mapping

For data synthesis, I wanted to challenge myself and generate some quantitative data. While 5 tests don't yield substantial data, I wanted to experiment with Notion to identify patterns and themes via keywords.

Using Notion's variety of ways to view data, I generated affinity maps for each flow:

This turned out to be quite productive because Notion allowed me to filter and arrange data to quickly identify patterns. For example, using the count property, I could tally participants and sort data to visualize the most recurrent issues. The tags helped me pinpoint themes and locate data quickly, making the process much easier.

Feel free to take a look at my database here !

Findings

Overall, diners were impressed by TasteBuds! They praised the personalization features and how TasteBuds leverages word-of-mouth recommendations. My iterations resolved all previous errors. The only issue was that some users found nonessential parts of the review UI confusing. One diner said:

I usually use Google Maps. . . I don't think there’s really any filtering. Sometimes you can be overloaded with all the results. No other apps have filters as extensive… If I had this app, I'd totally use it all the time. I really like the different filters.
Participant 02

I like how there's a lot of options in general. It’s very detailed. The whole point of the review is to provide details to people. The way it’s set up can help people understand someone’s experience at the place. It was very comprehensive, but also gave you the optionals. You can decide how much of a review you wanted to write.
I like how there's a lot of options in general. It’s very detailed. The whole point of the review is to provide details to people. The way it’s set up can help people understand someone’s experience at the place. It was very comprehensive, but also gave you the optionals. You can decide how much of a review you wanted to write.
Participant 01

Successes

All pain points and confusions from the previous test were resolved!
All diners praised the personalization features and inclusive/diverse content.
All diners found the new visual aids helpful and engaging to help content comprehension and task completion.
3 diners intuitively explored dining options outside of the primary flow.
All diners prefer TasteBuds over their existing tools!

Pain points

3 users found nonessential parts of the review UI confusing

Discover dining experiences with personalized recommendations.

The final solution empowers diners to confidently discover dining options with people they trust. Leveraging word-of-mouth recommendations from friends, Tastebuds provides personalized recommendations based on diners' preferences and dining history.

Explore individual dining experiences from friends.

Diners can explore individual recommendations from friends on their following page and visit other diner profiles for

B2B business model that provides restaurant subscribers access to user data and feedback

To generate a revenue stream without overcommercialization and improve the overall diner journey, TasteBuds invites restaurants to join as business partners to offer exclusive in-app promotions and stay connected with diners.

Play with my prototype! Select any user flow from the left hand panel.

If I had more time, I would collaborate with engineers and data scientists to assess technical feasibility, iterate on the prototype, and test it in real-world contexts—such as how diners review their experiences with TasteBuds. My goal would be to evolve the prototype into a true MVP, refine the UX ecosystem, and balance innovation with established mental models. I'm also interested in exploring features like a thumbs up/down rating system and designing for edge cases.

If I had the opportunity to ship this product with a team, I’d prioritize refining the business strategy and developing a prototype for restaurant partners. This would integrate them more meaningfully into the user journey and optimize partner touchpoints based on additional research. Through interviews and surveys with restaurant owners, we could tailor features to support seamless, high-value interactions on both sides of the platform.

This usability test taught me a lot about my design solutions and design thinking. I uncovered some of my most significant mistakes and weaknesses as a designer and as an individual that contributed to the issues:

Rely on testing and avoid excessive prototyping

This project highlighted the power of usability testing in optimizing the design process. Initially, I made the mistake of spending long hours prototyping alternative user flows instead of focusing on testing. Effective and strategic testing can significantly save valuable resources!

Mental models can change

I learn that we don't need to follow mental models so closely as they can cost innovation. I was pleased to learn that people are willing to challenge their mental models for compelling alternatives, especially for painkiller products. After all, industry standards are always evolving. As a designer who values innovation, this was a critical and motivating lesson.

New love for UX ecosystems

I developed a deep appreciation for UX ecosystems and symbiotic user flows. Although navigating them for the first time was challenging, it revealed my natural ability to consider technological constraints in design—something I’m eager to apply in a cross-functional team.

Iterate user flows

Reviewing other UX case studies made me realize a lot of designers tend to ideate in a linear process, but I found that it's valuable to reiterate techniques like user flow maps and user personas through testing to maintain scope. This also helps track iterations, which would save so much time on case studies.

TasteBuds

The Challenge

Dining recommendations shaped by your taste and the people you trust.

A B2B revenue model restaurants opt into, not advertise on.

secondary and market research
Exploring consumer dining and choice paralysis

primary research
Talking to dissatisfied diners

Conclusion

Competitor Analysis
Learning from my competitors

Defining the problem

synthesis
Connecting the dots with a mind map

user personas
Meet the diners:
Alex, Brandon, and Cleo

USER STORIES
What functionalities do Alex, Brandon, and Cleo desire?

user flow
How do I make diner stories a reality?

Optimizing navigation for the diner journey

Exploring and refining through sketching

How can the search feature optimize and empower dining decisions?

test
Rapid prototyping and testing new sketches

low-fidelity
Iterating sketches through low-fidelity wireframes

prototype
Branding and prototyping high-fidelity wireframes

branding
Style guide

High-fidelity prototype

test
Testing and iterating

Final round of testing

Discover dining experiences with personalized recommendations.

Explore individual dining experiences from friends.

B2B business model that provides restaurant subscribers access to user data and feedback

What's next?

Learnings