The RealReal
Role
Lead Product Designer, Growth Team
Key Contributions
  • 190bps+ uplift to conversion at top of Seller Funnel
  • 120bps+ uplift in number of Buyer's Saved Searches
  • 900bps+ uplift to new users with sign-up funnel overhaul with focus on SSO (surpassed 1MM users)
  • 30bps+ uplift to qualified Seller leads due to Pricing widget
  • $1MM in annual savings of warehouse labor costs designing tool that helped humans train dataset for AI Machine Vision Model
Platform
The RealReal is an online marketplace app for users to buy/sell authenticated luxury items, with over 10,000 items added daily, 1 million registered users, and annual Gross Merchandise Value sold exceeding $1B. For users selling luxury items The RealReal provides a white glove experience—handling the entire process of shipping, storage, authentication, photography, and pricing. For users shopping for luxury items the app has powerful personalized search tools and filters that help customers find the items they're most interested in amongst the millions of items for sale.
Seller Funnel
KPIs
Seller Funnel Conversion/Bounce Rates
Session Length
User Problem & Insights
Within the Seller Funnel, we noticed an increase in bounce rates and a decrease in conversion. Through analytics, we determined that the step where a user had to create a shipping list had the highest bounce rate, and session lengths were extending past a minute. To further understand the issue at hand, we ran a usability study using UserTesting software between us and our competitors, and discovered that our packing list required far more time and clicks than other apps.
Hypothesis
If we enable the user to more expediently "add items" in the seller funnel, then we will see a measurable increase in seller funnel conversion and qualified leads.
Design Solutions
On the growth team it was important to partner with our engineers early to determine the scope of design that could be launched within 1-2 sprints (e.g. 4 weeks at most.) I created wireframe prototypes that addressed the user problem by letting them search for the brand easily, instead of cumbersome drop downs. I worked with growth team engineers to determine the dev effort involved with the prototype, we aligned on a scope that was achievable in the sprints and would strike a balance that resulted in actionable insights when launching as A/B tests. Once aligned, high fidelity designs with production ready assets and layouts are presented to stakeholders and a direction is chosen.
Validate, Iterate, Launch, Results
Using prototypes of the proposed design I validated against existing user flow and saw users complete the task up to twice as fast. I then translated the design to css parameters for engineers to launch to Optimizely. The test was launched to 10% of mobile web users and saw promising results, with 190bps+ increase for users converting on first step of the funnel.
Buyer Saved Searches
KPIs
DAU (Daily Active Users)
MAU (Monthly Active Users)
User Problem & Insights
Sign ups were nominal, but daily active users were flat YoY. This was particularly unexpected as our platform also had a steady increase of items being listed YoY. Our most active users were on iOS and had more than five Saved Searches and visited their Saved Search PLP daily. 10,000+ items are listed on the site daily, so a user with a healthy amount of saved searches will be notified that new “item posts” via app push notification and emails. The frequent users also happened to be our highest converting users, and they converted mainly off of their Saved Search and Favorited PLPs (product listing pages.)
Identify Opportunity/Hypothesis
If we enhance the UI and language of saved search button, then we will see a measurable change in saved search KPI (and by proxy should provide uplift to DAU.)
Testable Designs
Informed by our user interviews with our most active users, the hierarchy and usability of the button was poor due to the location and lack of icon. I designed a test that used a new location for the button as well as a heart icon (to leverage semiotics and not just words.)

Another dynamically triggered module was designed that used our quantitative data of when searches were saved, and would insert itself further down the page if a certain number of rows were scrolled or it was near the end of the items returned for a particular search.
Validate, Iterate, Launch, Results
The variant along with its alert module banner led to 120+ bps increase in saved searches. YoY DAU started to climb but needed several months of data to reach statistical significance as the statistical power was a smaller signal.