Customer-facing account servicing
Fraud and disputes investigation
I co-defined the experience strategy and co-designed the UX/UI for a digital servicing experience which has saved Capital One $5.1 million in business cost savings between June 2021 and November 2022. The value of this digital experience continues to grow as the feature scales.
Role
UI/UX Design, UX research, Content strategy
Company
Capital One
Timeline
2020-2021
Overview
Partners and stakeholders
iOS, Android, and web tech teams
Native and web product owners
Business intent owner
UX writer
UX researcher
Skills demonstrated
Understanding and designing for the underlying emotional causes of user behavior
Identifying and solving for specific customer problems in order to reach business objectives
Documenting and socializing designs to communicate value and garner buy-in from many stakeholders
Conducting usability testing to optimize the language used in multi-step investigation experiences
Quick links
Summary
My team and I modernized the experience of reporting a transaction problem for Capital One credit card customers. I combined a deep understanding of users’ needs with technical knowledge of credit card case management to improve the customer experience and save Capital One over $5 million in operational expenses and case dollar losses. The digital experience my team and I shipped:
Increased the rate of case wins by 8% over cases filed by trained agents
Increased customer NPS after case resolution by 15 points
Reduced operational expenses and case dollar losses by $5 million over the agent channel
Reduced operational expenses and case dollar losses by $11 million over the previous digital experience
The problem
The credit card industry defines fraud and disputes as different claim types, with different processes required to process the claims. However, consumers think of fraud and disputes as one-and-the-same, and often file their claims inaccurately as a result. Inaccurate claim processing results in incremental case-load and dollar losses for the business, and confusion for the customer.
Historically, digital fraud and disputes reporting experiences struggled to tease out what type of transaction problem the customer was having and whether a claim was needed. These experiences also struggled to determine which type of claim was right for the customer – in other words, which ones provided the highest chances of the customer getting their money back. As a result Capital One has relied heavily on agent conversations to classify and file cases.
Overall, the deliverables of this project bring value to users and the business by:
Reducing call volume, operational expenses, and deflecting unnecessary cases by alleviating customer confusion around unrecognized transactions
Providing accurate digital reporting experiences for optimal customer experience, and to reduce dollar losses for the business.
Giving customers confidence in filing and following up on claims digitally
Discovery
Defining the problem
The legacy experience of reporting fraud and disputes digitally relied on customers answering one question correctly, in order to get them to the right solution. Experience analytics showed that the screen in the flow with this question caused a large dropoff rate; moreover, customers that did proceed through the flow often selected responses inconsistent with the type of claim needed for their situation.
Ultimately, the legacy experience enabled customers to file claims without fully investigating the customer’s needs. Findings from analyzing the legacy experience called out two major areas of focus for this experience redesign:
Expand the investigation process to understand a customer’s needs and direct them to the right solution.
Emphasize content design to ask and frame questions in line with customers’ mental models around transaction problems. Make the investigation and claims process intuitive, with easy-to-answer questions.
Foundational research
Key questions:
What probing techniques do agents employ that enable them to deflect and triage claims effectively?
What pain points do customers have with our current claims reporting experiences?
How do customers describe, and what are their mental models around transaction problems?
Methodology:
Customer empathy research
Agent interviews
Agent side-by-side call observation
Knowledge mapping of existing research and institutional knowledge.
Research findings
1
Customer insight: The current experience is overwhelming.
The existing experience asks one question with many, highly-specific responses. The cognitive load of determining the difference between each response, and the pressure of selecting the right one create a difficult-to-navigate customer experience.
Takeaway: Break down complex questions into multiple choice questions with simple responses.
2
Customer insight: Customers don’t know the difference between fraud and disputes
When a customer has a problem with a transaction, they might say “ I want to dispute this fraudulent transaction.” In fact, the customer may not need to file a claim at all. They may just not recognize the transaction, and are worried about their account security.
Takeaway: Incorporate questions that enable Capital One to understand the customer’s problem without asking them to self-identify.
3
Agent insight: Probing powered by data clears confusion
The most effective probing questions are highly specific, to make the customer reconsider their “problem”. Agents use merchant transaction data to help the customer understand their problem, before filing any type of claim.
Takeaway: Build use-case specific probing questions which leverage transaction and merchant data into the experience.
Ideation and design
Experience framework
With these findings and principles in mind, we developed a new framework for investigation and claims intake. The new framework
Formalizes not filing a claim as a valid solution for the customer
Treats investigation and intake as one joined experience
Helps determine not only what type of transaction trouble the customer is experiencing, but also their disputes reason code (a part of the disputes claims process)
Is shaped by the customer’s specific situation
UX flow
I organized and designed a UX flow chart to share-out and validate the complex logic of transaction investigation in a low-fidelity format. The flow chart enabled the team to test our hypotheses about the content strategy and UX writing of the experience before investing in the UI.
The ideation and design process started with the insights from agent research: examples of questions they ask when helping a customer investigate transaction problems. As a team, we organized these questions into categories to determine their order within the experience and the data needed to power these questions. We ideated on clear indicators of a particular type of problem, and how the experience could use these indicators to pose questions which “shortcut” people to the right solution.
We validated these flows through several rounds of evaluative testing with our operational partners and fraud and disputes agents. We also stress-tested our logic using scenarios from real customer calls. Doing so forced us to focus on UX writing, because variation in how customers interpreted a question could significantly change the outcome of the flow.
High fidelity deliverables
Once I finalized the UX flows, I developed the UI to prepare the experience for further testing, and ultimately, development. Existing account servicing pattern libraries did not include patterns which showed transaction data to which customers were supposed to evaluate and respond. I pulled components from the existing account servicing design system and developed new patterns to facilitate the data-rich investigation experience.
The scale of the work and the many people involved required rigorous and consistent documentation. As a side-of-desk project, I developed a library of UI documentation components to standardize the notes and other information attached to the high-fidelity screens. This system enabled designers, engineers, product owners, and operational/process managers to enter into a file and understand a shared visual language, inclusive of arrows, color-coded comments, and flow-chart elements
Outcomes
Analysis of case wins and loss reasons confirm that cases reported in the new experience win more often than cases reported in the legacy experience and phone (agent) channels. The new experience is less prone to error and the cases are higher-quality.
Results, 1 year post-release:
Increased the rate of case wins by 8% over cases filed by trained agents
Increased customer NPS after case resolution by 15 points
Reduced operational expenses and case dollar losses by $5 million over the agent channel
Reduced operational expenses and case dollar losses by $11 million over the previous digital experience