Dispute Intake

Role

• Discovered the opportunities
• Got buy in and helped get the work prioritized
• Collaborated with the design team as a thought partner
• Validated along the way with research
• Kept Product in the loop

Discovery

There were multiple pieces of research that alerted us to the issues:

• Analytics
  • The team had already identified that there was significant drop off on the landing page
• Usability Testing for fraud
  • Landing page information inapplicable for fraud (confusion and/or dropoff)
  • Unclear distinction between Triage and Form
• Usability Testing for unrelated changes
  • Ambiguous triaging
• Generative Research for Dispute Channels
  • More likely to call in for fraud
  • Easier to "get to the root of the issue" on the phone
  • This helped to underscore the importance

Solutions

The issues were a show stopper for fraud (significant portion of our case load), and a consistently large pain point/source of errors for non-fraud so it was imperative that we fixed these issues.

Three packages of work emerged that gave our team direction for the coming months but the two we will be looking at are as follows:

1. Sequencing/Landing Page
   • Moving the triage to the front would allow for more specific landing pages and create greater transparency on triage vs the actual form to be submitted
2. Triage Categorization
   • Minimize ambiguity and overlap, remove descriptions and flatten the tree
   • Strengthen the information scent
   • Reduce cognitive load, drop off and time taken

Sequencing/Landing Page

• Design Sprint for exploring concepts
• Usability Testing
  • Change in order and greater specificity of content increased understanding
  • 1 page > splitting over 2 pages
  • Eligibility content needed to be more explicit
• Iteration & Revalidation
• Implementation

Triage Categorization

• Baseline Tree Testing (qual & quant)
• Iteration based on problems identified
• Alternate Tree Testing (qual & quant)
  • Most of the changes were validated quantitatively with some areas to improve for the final version
  • 20% increase in overall task success
  • 18% increase in overall directness
  • 50% decrease in average time taken
  • Went from 7/15 tasks below benchmark to 1/14 below benchmark
• Workshop to arrive at final version with consensus
• Consultation with agents throughout
• Pushed final version out