Project

A furniture subscription for people who move often

Concept · Figma · 2025

Role
Research and Service Design Lead
Timeline
Sep to Dec 2025
An empty apartment beside a furnished living room, with a phone showing the delivery-tracking screen
The final service, with an empty apartment furnished in a single tracked delivery.

The Problem

Every late August, Boston has what locals call Allston Christmas. As leases turn over, students leave what they cannot carry at the curb, and the sidewalks of Allston, between BU, BC, and Harvard, fill with discarded furniture. The easy reading is a waste problem.

An Allston alley of overflowing dumpsters and discarded mattresses as students pass
A corduroy sofa left on the sidewalk beside overflowing trash and recycling bins
A dresser, mattress, cat tree, and office chair stacked on a residential curb
A cat tree, office chair, and beanbag abandoned at a crosswalk
Late August in Allston. When leases turn over on the same day, whatever cannot be carried is left at the curb.

But the waste is a symptom. A 2019 City of Boston report puts 39% of residents between 18 and 34, which is 18 points above the national average, so the city runs on people who move constantly. The people are not the problem. Ownership does not fit a transient life. Buying furniture asks people to commit to a place and a future they do not have yet, so they buy cheap, use it briefly, and leave it on the curb.

  1. A new lease

    A place and a future that are not settled yet.

  2. Cheap furniture

    Bought because the commitment cannot be justified.

  3. Brief use

    Until the next move.

  4. The curb

    Whatever cannot be carried is left behind.

Repeats with every lease turnover.

Research

I led five in-depth interviews with students across the city, three new arrivals and two who had moved several times within Boston. My two teammates recruited the participants and recorded the interviews and their notes. Two exercises did most of the work. In a Think-Out-Loud floor-plan exercise, participants sketched and narrated their spaces, and A Day in the Life traced how they actually used their furniture.

Floor-plan layout stimulus and furniture-symbol library used in the interviews
A Day in the Life timeline of a participant's daily furniture and space use
The two exercises turned vague preferences into decisions that could be observed. One is a floor-plan layout with a furniture-symbol library, the other a timeline of a participant's day across common area, private space, and time out of the house.

The conversations kept coming back to instability.

The research team around a table during Nancy's interview, a laptop and notes between them
“Financially stable and I am settling in one place, I will definitely buy better furnitures.”
Nancy
A researcher across the table reviewing notes on a laptop during Chayan's interview
“I don't know about next year because it really depends on my job situation.”
Chayan

People were avoiding a commitment they could not justify, and good furniture asked for one. A Day in the Life showed the same thing in their rooms. Residents ate meals in bed because a dining table was not worth the space or the money. That turned the project toward designing for impermanence instead of against it.

Cross-participant affinity map sorting interview notes from life-context to furniture-specific, resolved into three findings
Sorting every quote from life context to furniture specifics surfaced three findings. What people spend tracks how settled their life is, furniture carries identity, and small spaces have to flex.

Design

From the research I developed three directions, then pressure-tested each against feasibility and the waste it had to solve.

  • A subscription model, renting furniture on flexible terms and returning it or buying it out at the end, which removes the upfront cost and the commitment.
  • A personalized rental service that connects renters, apartments, and providers, so residents pick pieces that fit them and landlords keep tenants longer.
  • A convertible bed and dining table, a product answer to the finding about eating in bed that reclaims floor space in small apartments.

I carried the subscription model forward. It gave the provider real operational control, opened a market conventional furniture businesses had left untouched, and turned seasonal waste into inventory to repurpose rather than a cost to absorb.

To explore positioning before committing, I built three fictional furniture brands, each aimed at a different segment from secondary research on the Boston market. Those explorations mapped the touchpoints and service moments that shaped the final concept, which I then developed with the sponsor, ISHO Furniture (Bangladesh). The palette and moodboard came from their existing lines, so the service would read as part of their brand.

ISHO brand palette and moodboard
Aligning the service direction to ISHO's established aesthetic.

The Service

The final concept is a subscription that trades ownership for access. Customers get quality furniture without the commitment, and the provider gets a way to keep its furniture in circulation.

I mapped the full customer journey in a service blueprint, including the interactions and the logistics behind them, so the concept would work as an operation as well as an interface.

Service plan and blueprint
The service blueprint, built from the primary research and the business concepts, pans left to right as the page scrolls.

First-time movers told us budget was only part of the difficulty. They also did not know what they wanted or how to make the pieces in a room work together. So the service starts from rooms they like. A customer picks inspiration images, the system learns their aesthetic and generates scenes in that mood, and then it matches real pieces from ISHO's inventory to those scenes. The taste comes first and the products second, the reverse of the usual shopping flow.

The AI part of this is what the system shows of its own reading. Each generated scene carries the style it inferred as tags, such as Modern Scandinavian, Neutral Tones, and Wood Accent. The room it assumed, a studio living room for one person, sits behind an Edit button, so a customer can see what was inferred and correct it.

The usual shopping flow
  1. Browse products

    One item at a time, by category.

  2. Guess what fits

    The space and the other pieces.

  3. Buy, and hope it coheres
The service
  1. Pick inspiration images

    Rooms the customer likes.

  2. The taste is learned

    Mood-based scenes are generated.

  3. Pieces are matched

    Similar items from ISHO's inventory.

AI taste-matching from inspiration images
Matching inventory to generated scenes
On desktop, a customer picks inspiration images, then sees a generated scene labelled with the inferred style tags and an editable assumed room, before pieces are matched from ISHO's inventory.

Setup and activation happen on desktop. After that, the mobile app holds the ongoing relationship, with monthly feedback to keep the fit right, the option to buy pieces outright, and subscription and maintenance management. When a plan is about to expire, the app offers to sell the customer the pieces they have grown attached to.

  1. Set up on desktop

    Taste, plan, and activation.

  2. One tracked delivery

    An empty apartment furnished at once.

  3. Live with it on mobile

    Monthly feedback, maintenance, and plan management.

  4. The plan expires

    Return the pieces, renew, or buy the ones that stuck.

Final presentation of the service
Presenting ISHO CloudLiving.

Learnings

The brief looked like a waste problem, and the real lever turned out to be ownership. Once I stopped designing against transience and started designing for it, the rest of the service followed.

Acting the service out taught more than analyzing it. For the final presentation, we staged the service as a short skit. Acting it out exposed gaps between screens that had stayed invisible on the board, the kind that only show when a service is performed. The lesson I took from it is that trying the thing is the fastest way to find the real problem.

Updated January 2025@Northeastern University

More work

View all