Senior UX Designer & Researcher
(SaaS/B2B and E-Commerce)
3 min read
New App:
Vroom Transaction Manager (VTM)
(Leveraging AI)
"The products and features Beth designed have significantly contributed to company initiatives, increased website conversion, and improved customer experience as a result."
—Thomas McNair, VP of Product
The Problem
The “Documents Team" at Vroom would review and validate all of the documents that customers submitted for buying or selling a car. Salesforce was the main tool that they used. Salesforce was expensive and inefficient in that the information the team needed was spread throughout the application. This current process was difficult and slow. This new application, Vroom Transaction Manager (VTM), would leverage artificial intelligence to automatically validate the bulk of the data (ie: Legal names, addresses, birthdates etc) and help the documents team work more efficiently.
Business Goals
• Get users out of Salesforce
• Decrease time to move a deal to the next step
• Automation (automatically validate customer data)
• Reduction in force (RIF)
• Apply this AI technology to the customer-facing app (separate project)
My Role
I worked closely with a Product Manager on this project. We immersed in all of the early research and then brought in other teams for feedback and to shape each iteration. I led interviews, ideation, user reviews, and stakeholder reviews.
Methodologies
-
Shadow sessions
-
Interviews
-
Ideation
-
Stakeholder feedback, iterations
-
User feedback, iterations
-
Follow up shadow sessions and working sessions with users
The Process
• User Story
I, as a member of the documents team, need to be able to look up a deal, know the status of individual documents, and review/validate specific data points so that I can quickly process customer documents.
• Intake
At the beginning of every large project, I use an intake form. This is a fast way to orient all of us who are on the team to make sure we are aligned right from the start. Basically, we gather all of the information we already know, and then I use this intake form as a reference throughout the project. The intake process answers basic questions such as: “What is the problem” (as we understand it at this point), “what are our goals,” “what are the known pain points,” “is there existing research,” “who are our stakeholders.” We also talk about the schedule, personas, expected assets, how we will measure success, business value and how this project will feed our company strategic initiatives.
• Shadow Sessions
We scheduled about 6 shadow sessions to begin to understand the different roles on the documents team. As we observed, the goal was to better understand what they do and to look for specifics as well. I documented which tools they use, their pain points, definitions pertaining to their work, the abilities they currently had and the abilities they would need in the new VTM application.
I used "Dovetail" to take notes during our shadow sessions and interviews. This tool allows for quick tagging which makes it easy to surface insights
• Interviews
After the shadow sessions the team gathered. Did we still have any questions? With quick interviews, we were able to get specific answers to the few blindspots we still had.
• Iterations and Feedback from Stakeholders & Users
I quickly put together the first round of mockups and brought them back for discussion with the PM and developer. We did more ideation together, and after a few working sessions, we were more sure of the direction.
From that point, we met with internal stakeholders to talk through each screen. We were on a cadence of meeting 3 times per week with stakeholders. With users, we were on a similar cadence. We would put iterations in front of them for feedback and refining as needed. We did not do a formal usability study on this project. The design patterns were common and, with a tight deadline, we chose to leverage follow-up shadow sessions to iron out any kinks.
The MVP included specific data points in a drawer at the top, an indicator to show where the errors were and how many, and the ability to drill into individual documents
Users could now drill in, verify, and correct the outstanding errors.
Outcomes
Slow success: We initially released the MVP to a few people on the documents team. I was especially excited to see if we were able to decrease the time to complete each deal. For the first few weeks, we did not see a time improvement, it stayed the same. The difficulty was that our MVP gave users the basic needs, but there were still features that were only in Salesforce. At this point, it was easier for them to just keep using Salesforce or they would go back and forth between the two applications. Over the following weeks, we had regular feedback sessions & shadow sessions, and we continued adding the necessary features. To encourage adoption, we also held short training sessions.
Success: After a few weeks, this project became a huge success. The VTM solution met the goals in that we were able to pull those users away from Salesforce, and give them what they needed in order to do their jobs more efficiently. Leveraging AI, we were able to automatically validate most datapoints and then allow users to quickly validate or correct additional datapoints. We beautifully reduced the overall time to process these deals, and yes, soon after this, there was a reduction in force (RIF) in that department. Over the next year, this application was built out to meet the needs of other internal teams.
Customer-facing application: With the success of this application and learnings, we applied this AI technology to the customer-facing application. AI was used to quickly show the customer if there were errors. For instance, it would look to confirm that it was the correct document and would check specific data points such as expiration dates. If errors were found, the customers would be notified in real time and could make corrections.