This side project started from a casual conversation with my roommate. She wanted to buy used furniture on a classified website. However, having no car and no friends with cars that could help her, she quickly dropped the idea.


In order to validate that selling and buying furniture was indeed a real problem for students, I interviewed:

  • 12 freshmen students
  • 3 students who were about to graduate
  • 4 recent alumni

With limited budgets, students, I interviewed, preferred having used furniture. However, quality and logistics were two main concerns from the buyer’s viewpoint.

this bar chart describes 4 people had experience of buying used products, 5 people would buy used products if they had a car, and 6 people were worried about the quality of used products
The bar chart shows 12 students' experience with used products.

On the other side, people had trouble with selling (or giving away) furniture. For example, the fear of allowing strangers to check items at home and the disappearance of buyers after accepting deals.


From interviews, I got many problems in used markets among students. What problems should I focus on?

In order to get feedback broadly and prioritize problems, I conducted an online survey and collected 100 responses from current students across the nation. The result showed that, for people who chose to buy (or not to buy) used products, their top three concerns were: quality, trust, and logistics.

Thus, I decided on focusing on these three problems.

from 100 responses of the survey, quality, trust and logistics are three top issues in used markets whether for 12% people who didn't consider used products and 88% people who considered it.
The online survey

Focus Group

In current used markets, there were two roles: Buyer and Seller. The logistics was something missed in between. Are people willing to help fill the gap?

Thus, I came up with a business mode of a three-sided marketplace: Buyer, Seller, and Delivery Partner.

In order to test the concept of adding Delivery Partners to solve the logistics problem, I recruited three students, playing one role for each, into a focus group activity. The person, who played the role of Delivery Partner, was willing to help deliver furniture with his car (both in this activity and the real life).

three participants attended the focus group activity
A photo of conducting a focus group with three students

While observing their interactions, I found that the difference between roles was the most critical insight. For example,

  1. The Buyer needed more item details to make decisions, but the Seller expressed it’s discouraging if it needed a lot of effort to describe an item.
  2. The Delivery Partner could help move a mattress for a $15 gas fee, but the Buyer didn’t agree with the price. (as seen in the sketch below)
a negotiation between a driver and a buyer about the gas fee
The difference in how much the gas fee should be charged between the Driver and Buyer


I observed the interactions by matching Buyers, Sellers, and Delivery Partners. Thus, I could not only see challenges and opportunities for innovations, but also validate if the business model of the three-sided used marketplace worked in the real world.

the relationship among buyers, sellers, drivers and me
My role as a people-person platform in the middle

From the $228.50 transactions I matched (as seen the transaction history below), it proved that the need of having a platform to make people connections happen. For example, people were willing to help deliver furniture for free or for a very small gas fee.

16 items observing from October 2018 to February 2019 value 208.50 dollars.
Transactions observed from October 2018 to February 2019

Design Decision

How can technology help students settle down within their budgets?

Using post-it notes to collect and arrange ideas by:

  1. Discussing with friends - a website or a mobile app was the best option because students already had these devices.
  2. Referring to a research report of Gfk: millennials used smartphones (73%) for shopping online more than desktops (59%) or tablets (35%) in the US.

Thus, I chose to design a mobile app.

An affinity diagram which has 7 categories: buyer, seller, driver, negotiate, people, quality, trivial
Categorizing ideas with post-it notes


I sketched out wireframes and their interactions on papers, which allowed me to discuss it with a senior student and to iterate on the design quickly.

the interaction design started from the key screens and connect them together
Interaction design: attached key screens on the wall to see interactions in the holistic view

Low-Fidelity Prototype

I moved the critical screens from papers into the digital format to make sure no major usability issues before going into building high-fidelity prototypes.

interactions among 5 tabs: buy, drive, sell, offer and me
Low-Fidelity prototypes


The black-and-white palette was used throughout the design to keep the focus on the content and to have better readability and accessibility (as seen in the color palette below). Moreover, the color contrast was at least 4.5:1 by following WCAG 2.1 AA.

Typography following Apple Design Principle; using 21 icons and black-and-white color palette in this application.
Typography, iconography and color design

First-Round Iteration

This round focused on the quality verification of used products and basic features: selling and buying. Students, who I talked to, sold furniture when they were about to graduate, and the amount of items was usually a lot. Dumping furniture happened due to the time limits.

two couches and one mattress were in trash cans
Furniture was dumped into the trash.

In order to encourage students to sell (or give away) their furniture with minimal clicks, I designed a tagging feature for multiple-items posts. The existing solutions were a little difficult to know, for example, which products were sold and the price. (as seen the photo on the right below: a post on my school's classified website)

with the tagging feature, users were able to distinguish what were selling in the photo
My design versus an existing design

In my design, for Buyers, it was clear to match items in the picture and description (as seen the video on the left below); For Sellers, it saved the time from posting an item at a time (as seen the video on the right below).

How did it save time by posting multiple items in a post? It reduced the amount of photo-taking and data entry.

for posting one item at a time, it needed 10 pages to post 5 items; for posting multiple items in a post, it needed 7 pages to post 5 items.
Single-Item post versus multiple-items post

In the first round of usability testing, four participants, who had experience of shopping online, tried the buying and selling features.

From participants' opinions: the multiple-items post was better when Sellers wanted to save time. However, the single-item post was promising when Sellers wanted to pay more attention to higher-price items by having one product in a post.

Moreover, there was a usability issue in the testing: participants kept switching hands to hold the phone because certain buttons were aligned to one side. (as seen in an example below) Thus, I made all buttons have full width in the next iterations.

a screenshot of one button at the buttom-right of the screen.
One button at the bottom-right made users switch hands.

At the end of testing with each participant, I asked them to prioritize what they cared about the most while shopping online. By averaging the results, the item description showed in the next iterations in the following order:

  • Price
  • Condition
  • Size
  • Material
  • Availability
  • Delivery
  • Distance
  • Payment
  • Seller identification check
first hi-fi testing with 4 participants
Card sorting: 4 participants were arranging what they cared about the most while shopping online.

For testing live-streaming sales, I used the video streaming feature of WhatsApp. It worked well and I made real transactions. (The case below was I sold my pillow…)

In the first-round iteration, the takeaway for the quality assessment of furniture was: depending on the nature of needs and environments, people (both Buyers and Sellers) chose different media (photo, video or live-streaming) and different ways (single-item or multiple-items posts) to build connections.

Second-Round Iteration

The second round focused on features of building trust and providing logistics to fill the gap between Buyers and Sellers. I recruited 10 international students into the testing. I asked participants to make a transaction by requesting a Delivery Partner to pick up furniture for them and observed what Buyers looked at while viewing used products.

For trust, the testing result showed: in addition to 5-star ratings, the relationship (as seen in the relationship description below) helped a lot when deciding if purchasing an item or believing a review.

at the item page and shopping confirmation page, it showed the relationship between the user and the seller, reviewers and drivers
Showing relationships helped Buyers know people online.

Communication was a critical factor while shopping online. For example, Buyers needed more pictures from Sellers. Through the survey earlier, people had different opinions on how long it should take for Sellers to respond to their questions. Thus, the Seller's response time helped Buyers move forward on transactions.

showing the seller's response time under the seller's picture
Showing Sellers' response time helped Buyers make decisions.

I was curious to know if the length of reviews impacted Buyers' perception of reviews, so I conducted an A/B test on two review styles (as seen in two versions below) by randomly selecting 5 participants to view each version.

using longer version reviews and shorter ones to see if it impacted users' behaviors
Longer-version reviews versus shorter-version ones

8 (out of 10) participants read reviews as how they usually did, no matter how short or long reviews were. However, one interesting insight I found in the testing was a contradiction that people relied on reviews a lot, but they never provided one.

The takeaways in the second round were 1) cooperating with people to deliver furniture was helpful for budget-concerned students when the total price (item + delivery fee) was lower than having a new one and 2) knowing relationships made people confident about who they were doing transactions with - a way to build trust.


During these 5 months, I talked to many people. I got to know this is an underground need in my student community. It was satisfying to help students solve their furniture problems.

I won't stop here but will continue to listen and learn. Some day, I will implement this design into a real mobile app to help more students.

Thanks for reading :)