Cars by Overstock.com

Create a car shopping experience that allows people to buy car parts online, finance a new vehicle, and protect the one they own

UI/UX       Illustration        Branding

We needed data

When I began the project, very little research had been done and many ideas were going around the table to add new features. Convincing stakeholders to build based on user research was a real challenge; although a risk worth taking on.

 

After many conversations, we were able to present glimpses of data, which quickly opened their eyes. In doing so, it occurred to these leaders that our app needed to be created using a full design process; wireframes, journey maps, A/B testing, defining iterative successes, and so on.

Role: I was responsible for defining this design process, training the team on it, and integrating it company wide with in-house partners (dev, PM, etc).

The data we received:

• 25% buy a car within 90 days (12500)

• On these same leads, roadloans/carmax and carvana sell about 300 cars per month within the first 45 days (and 7500 in total do) , the 7200 who don’t buy from roadlaons/carmax/carvana generally go to another dealer.

• In the next 45 days after, about another 5000 people buy cars --> we’ll be marketing to these consumers

• In addition, we’ll work on extending our reach/drip campaign over time — we have about 1.3m leads historically

Building one experience

Credit IQ, Ally, and Overstock struggled to create a cohesive experience as a trifecta of partnerships.

Working together:

Rather than attempting to hook three API's together, we decided to work with their Product teams to create one single, cohesive experience that can fully be engaged with on Overstock.

I defined this flow, wireframed the process, designed the application, then gathered feedback.

Massive constraints

Overstock is limited with taxonomy. Car parts requires at least 15 levels deep and Overstock only supports 7.

 

So we took it to the whiteboard. We had to take a deep look at what the actual goal was; to purchase a car part. The reason car parts nowadays used 15-30 levels of taxonomy was because they required the user to drill-down without ever gathering information upfront.

Team & role

I coordinated and developed the user flows, UX architecture, surveys, user testing, and even brand creation for the extension of Cars. Furthermore, I conducted user research, user testing, and analytics.

Solutions

Accomplish the same goal using only 3-4 steps. Rather than drill downs, we ask the user to enter their current vehicle upfront, giving them only relevant parts for the car they're shopping for. Not only did this fix the problem, but it also massively reduces the amount of return orders placed on accidental car part purchases.

Next steps

Continue to validate success through user testing and user recordings. When a user spends more than 10 seconds finding a part, we have more work to do. Our goal is to get a user down to 5 secs without adding any new steps.

All rights reserved, ©2018 Copyright Drew Andersen