A product that worked as designed. And a problem design couldn't solve.

TL;DR

• Reframed roommate matching from a listing problem to a trust problem
• Built a browse-first, progressive trust system based on research and behavioral data
• Reached ~4% conversion, roughly 50% of our estimated ceiling
• The ceiling itself was the real problem — and it wasn't a design problem

Overview

Roommate matching is not a listing problem. It's a trust problem.

But there's a deeper problem underneath that: it's a marketplace problem. Trust-based marketplaces only work when both sides show up. And getting both sides to show up requires either a lot of organic growth or a willingness to seed the market artificially. We had neither.

This case study is about the design work — which I'm proud of. It's also about the constraints that design couldn't fix, which I knew about the whole time.

My Role

Hybrid role: Senior Product Designer + Associate Product Manager, end-to-end ownership across a 12-month initiative. 12 stakeholders, 8 engineers, 2 QA, 1 PM, 1 junior designer, oversight from the General Manager.

The Problem

People looking for roommates aren't browsing inventory. They're making a high-stakes personal decision about who they're going to live with. The existing listing-based model treated this like an apartment search — which meant it optimized for the wrong thing.

The real challenge was building enough trust for users to take action, without requiring so much trust upfront that they never got started.