Product Entry Optimization
Student Loan Refinancing & In School Student Loans, SoFi

My role
I partnered with another designer (In School Student Loan product) when leading the Student Loan Refinance product at SoFi. Together, we collaborated with Product, Engineering, Data Science, Research and Operations to complete this project.

Timeline
8 days total, including card sorting, benchmarking current user experience of product entry points on multiple platforms and concept testing with members on Zoom.

Opportunity
Members are having trouble finding the right Student Loan products on our platforms (sofi.com, member home dashboard, mobile app, IVR/Member Support.) We are seeing the evidence of this happening from our funnel data, Glassbox recordings, and Operations call analysis.

Quantitative Research

• Users start two refinance applications within a week of each other:

• 6% (126/day) of daily start applicants started a second application within a week (same app_type)

• 4.5% (88/day) of daily start applicants started a second application within same day (same app_type)

• Conversion to funded loan of users who start two refinance applications within a week of each other (accounting for the duplicated apps):

• Applicants with multiple apps within a week has a higher conversion to fund. These users are high intent.

  • dupes (within 7day): 23.31% v.s. other: 15.5% start-> fund conversion - by borrowers

  • dupes  (within 7day): 13.82% v.s. other: 15.6% start-> fund conversion - by applications

• 4% of Pre-start ISL applications started later with an SLR application. 29% of these applicants ended up funding an SLR eventually (Typical Start to Fund conversation for SLR is only 14%.) 

• Through research via Voice Analytics, the team has found that 49% of ISL contacts are SLR. These contacts are being routed to ISL Skills via IVR and Chat.

• On a daily basis, 4 applicants starts a PSL application after starting a REFI application.
  
  
  
  

Qualitative Research

• ISL Member interviews

  • During SoFi Unfiltered (user research interviews) in Jan,  5 out of 6 users we recruited from our ISL Prestart drop-off list turned out to be SLR members. They also have no recollection of applying for an ISL. 

• Glassbox sessions review

Expected impact
Cleaner product entry should…

• Drive higher CTR into the funnel (Revenue)

• Increase funnel conversion (Revenue)

• Improve the member experience

Objective

Provide a simple, intuitive pathway for members to find the exact Lending product that they need, starting with the Loans page of member home for this phase of the initiative.

Process
Phase 1: Discovery

• Audit and gather existing data surrounding existing channels and traffic flows across web, SEM landing pages, MHF, mobile app, emails, and IVR 

• Audit competitors 

• Conduct card sorting exercises via UserZoom

• Conduct benchmarking user research with members to understand their current experience navigating on sofi.com and MHF

• Define user mental modal

Phase 2: Ideate

• Define hypothesis

• Ideate possible solutions

• Prioritize selected ideas and testing methods

Phase 3: Test, measure, learn

• Conduct usability studies on selected ideas

• Conduct A/B testings on selected ideas on sofi.com and MHF

• Iterate and retest

Phase 4: Implement

• Coordinate with relevant teams to prioritize and implement winning solutions

Principles

• Intuitive

• Clear

• Simple
  

Constraints
Outdated icons and value props within product cards at the time of designing

Discovery

Card sorting

• Two cardsorting exercises to understand user’s mental  model of product groupings.

• First exercise examined groupings within Lending Products

• Second exercise looked broadly across all SoFi Products as this solution would need to scale across Sofi-wide information architecture.
  

The results of both card sorting exercises helped us hone in on four mental models:

1. Life Stages

2. Roles

3. Intention

4. Purpose

Ideation

Next, we began work on the information architecture of the options we were organizing. To align with the mental models outlined above, we decided to simplify the “Private Student Loan” options into one, instead of four. This is something we plan to do in the future for the Refinance products but could not tackle in this effort due to engineering constraints.

Test, measure, learn

When the concepts were ready we conducted a benchmarking exercise and new concept testing with six SoFi members.

Implementation

Results of this informed our final concept refinement.

Results

The “purpose” mental model (Treatment D) had the highest clickthrough rate while increasing Student Loan Refinance application starts and decreasing In School Student Loans application starts.