Streamlining U.S. Real Estate Discovery
Over six months, I led the design and product development of Cozying.ai from the ground up, transforming 800+ MLS data points into an intuitive real estate platform with a 61.58% engagement rate. I built the UX/UI and design system, integrated and customized Google Maps API for our platform, and shaped the product strategy. By designing an interactive CTA system optimized for U.S. and Korean markets, we achieved a 22.9% return visit rate in early 2025.
This case study highlights key milestones and strategic solutions that drove these outcomes.
Categories
Business Development
Data Visualization
Product Design
Interactive Design
UI/UX
Tools
Figma
HTML/CSS
Google Maps API
Spreadsheet
Company / Position
Habit Factory USA
Product Lead (Intern)
3 Designer / 1 Developer
Period
6 months,
Aug. 2024 - Feb. 2025
Table of Contents
Table of Contents
Table of Contents
Designing Cards: The Essence of Real Estate Design
Designing Cards:
The Essence of Real Estate Design
From 800 Columns to One Clear View
From 800 Columns to One Clear View
4 min read
100% contribution
4 min read
100% contribution
Intro
Intro
The first step to owning your dream home is finding the perfect one among countless listings.
Each property comes with unique attributes to compare—but how many attributes exactly?
The first step to owning your dream home is finding the perfect one among countless listings.
Each property comes with unique attributes to compare—but how many attributes exactly?
You can find original data examples like this on Google.
You can find original data examples like this on Google.
Property data is stored in Multiple Listing Service (MLS) tables by providers.
While companies use similar formats, manual data entry across over 800 scattered data columns often causes inconsistencies, leading to fragmented information.
Real estate platforms must convert this scattered data into consistent, meaningful sets.
Property data is stored in Multiple Listing Service (MLS) tables by providers.
While companies use similar formats, manual data entry across over 800 scattered data columns often causes inconsistencies, leading to fragmented information.
Real estate platforms must convert this scattered data into consistent, meaningful sets.
User Interview & Data Analysis
User Interview & Data Analysis
What aspects of real estate listings do people consider most important?
What aspects of real estate listings do people consider most important?
To extract design insights from this question,
I conducted interviews with 6 industry professionals and 10 potential buyers.
To extract design insights from this question,
I conducted interviews with 6 industry professionals and 10 potential buyers.
Based on the research,
the following factors were ranked in order of importance when selecting a property:
Based on the research,
the following factors were ranked in order of importance when selecting a property:
Key Home Buying Factors
Hover to focus on each value
Location
School District
Size & Layout
Price
Parking
Building Condition
Exterior & Lot
HOA
Total Interviewees
16
Industry Experts
6
Potential Buyers
10
Key Home Buying Factors
Hover to focus on each value
Location
School District
Size & Layout
Price
Parking
Building Condition
Exterior & Lot
HOA
Total Interviewees
16
Industry Experts
6
Potential Buyers
10
Key Home Buying Factors
Hover to focus on each value
Location
School District
Size & Layout
Price
Parking
Building Condition
Exterior & Lot
HOA
Total Interviewees
16
Industry Experts
6
Potential Buyers
10
Location
Price
School District
Size & Layout
Parking
Building Condition
Exterior & Lot
HOA (Homeowners Association)
Location
Price
School District
Size & Layout
Parking
Building Condition
Exterior & Lot
HOA (Homeowners Association)
To assess how each attribute influences purchasing decisions, participants were shown randomized listings and asked to rate each attribute while making their final decision.
To assess how each attribute influences purchasing decisions, participants were shown randomized listings and asked to rate each attribute while making their final decision.
Property Feature Impact Matrix
With a purchase threshold set at 100, the impact of each attribute on the final decision was analyzed.
Property Feature Impact Matrix
With a purchase threshold set at 100, the impact of each attribute on the final decision was analyzed.
Attribute
Attribute
Location
Location
Price
Price
School
District
School
District
Size &
Layout
Size &
Layout
Parking
Parking
Building
Type
Building
Type
Built Year &
Condition
Built Year &
Condition
View &
Exterior
View &
Exterior
HOA
HOA
Best
Best
Good
Good
Bad
Bad
Worst
Worst
+35
+35
+25
+25
-20
-20
-25
-25
+30
+30
+15
+15
-10
-10
-30
-30
+20
+20
+10
+10
-12
-12
-15
-15
+15
+15
+5
+5
-10
-10
-25
-25
+8
+8
+5
+5
-8
-8
-15
-15
+12
+12
+3
+3
-8
-8
-12
-12
+7
+7
+5
+5
-5
-5
-20
-20
+13
+13
+10
+10
-3
-3
-20
-20
+10
+10
+7
+7
0
0
-10
-10
*The "worst-case" scenario assumed a minimum legally accessible environment in participants’ daily lives.
*School District ratings included responses from both parents and residents without school-age children.
*The "worst-case" scenario assumed a minimum legally accessible environment in participants’ daily lives.
*School District ratings included responses from both parents and residents without school-age children.
Hierarchy Settings:
How should I prioritize and weight different attributes?
Hierarchy Settings:
How should I prioritize and weight different attributes?
Location
Score: 60
Price
60
School District
35
Size & Layout
40
Parking
23
Building Type
24
Built Year & Condition
27
View & Exterior
33
HOA
20
Location
Score: 60
Price
60
School District
35
Size & Layout
40
Parking
23
Building Type
24
Built Year & Condition
27
View & Exterior
33
HOA
20
Location
Score: 60
Price
60
School District
35
Size & Layout
40
Parking
23
Building Type
24
Built Year & Condition
27
View & Exterior
33
HOA
20
By converting extreme values into absolute numbers, I identified the attributes with the highest impact scores as the most influential in purchase decisions. These insights were incorporated into a treemap-style wireframe to visually represent their significance.
By converting extreme values into absolute numbers, I identified the attributes with the highest impact scores as the most influential in purchase decisions. These insights were incorporated into a treemap-style wireframe to visually represent their significance.
*School information was given a dedicated section due to detailed data requirements
*School information was given a dedicated section due to detailed data requirements
Visualization
Visualization
Composing the actual components.
Composing the actual components.
Badges
Photo
Exterior
View
Condition
Price
Monthly Cost
Layout: Beds
Layout: Baths
Size
Built Year
Location
Building Type
Status
Badges
Photo
Exterior
View
Condition
Price
Monthly Cost
Layout: Beds
Layout: Baths
Size
Built Year
Location
Building Type
Status
Badges
Photo
Exterior
View
Condition
Price
Monthly Cost
Layout: Beds
Layout: Baths
Size
Built Year
Location
Building Type
Status
Based on the survey results, I strategically placed key information in prominent areas of the cards. Additionally, I designed sections to highlight property conditions, provide supplementary details to assist with purchase decisions, and facilitate seamless connections with our services.
To ensure consistency across various pages and devices, I implemented a responsive design to prevent content truncation.
Based on the survey results, I strategically placed key information in prominent areas of the cards. Additionally, I designed sections to highlight property conditions, provide supplementary details to assist with purchase decisions, and facilitate seamless connections with our services.
To ensure consistency across various pages and devices, I implemented a responsive design to prevent content truncation.
Survey-Driven Values
Searching-Aid Values
Service-Connected Values
Here's my approach to structuring this data:
Open Sat 1pm - 4pm
1 day ago
Price changed: -$10,000
$2,749,000
12 Crescent Heights, Los Angeles, CA 90046
Single Family ResidenceㅣActive
7bd
10ba
5,428 sqft
2024 built
Est. $24,855/mo
Open Sat 1pm - 4pm
1 day ago
Price changed: -$10,000
$2,749,000
12 Crescent Heights, Los Angeles, CA 90046
Single Family ResidenceㅣActive
7bd
10ba
5,428 sqft
2024 built
Est. $24,855/mo
Open Sat 1pm - 4pm
1 day ago
Price changed: -$10,000
$2,749,000
12 Crescent Heights, Los Angeles, CA 90046
Single Family ResidenceㅣActive
7bd
10ba
5,428 sqft
2024 built
Est. $24,855/mo
This layout creates the following visual flow.
Size & Layout
Listing Price
Location
Building Type
Listing Status
Additional Information Badge
Property Images
Image Swiper
Additionally, key information from the research process was prominently displayed at the top of the property listing page.
Additionally, key information from the research process was prominently displayed at the top of the property listing page.
I can now display significantly more data on main screens with greater sophistication compared to prototype version.
I can now display significantly more data on main screens with greater sophistication compared to prototype version.
Takeaways
Takeaways
To enhance the prototype-level design, I restructured components based on survey data rather than intuition. This helped me understand key home-buying factors, adapt designs for various resolutions, and refine data definitions. By polishing micro-interactions, I created a more competitive and visually distinct UI.
Although limited user preference data made performance measurement difficult, curating and visualizing hundreds of MLS listings into useful information sets was invaluable. This experience not only improved my design skills but also strengthened my data-driven planning abilities.
To enhance the prototype-level design, I restructured components based on survey data rather than intuition. This helped me understand key home-buying factors, adapt designs for various resolutions, and refine data definitions. By polishing micro-interactions, I created a more competitive and visually distinct UI.
Although limited user preference data made performance measurement difficult, curating and visualizing hundreds of MLS listings into useful information sets was invaluable. This experience not only improved my design skills but also strengthened my data-driven planning abilities.
