Capital One: Digital Fraud Intake
At-a-glance
In late 2016, Capital One launched digital fraud filing to allow customers to report suspicious activity that our technology couldn’t catch. The result was less than positive: fraud report totals increased without a significant drop in customer service outreach. Analysis of the experience showed customers are filing cases incorrectly and are still confused about the state of their transaction even after filing.
Timeline
Four months
Lead product Designer
Collaboration
PM, tech lead
Deliverable
Mobile web
Problem
Customers need the full picture of information and the right tools to digitally report undetected suspicious transactions. They need to feel like the security of their account is protected in all circumstances.
Hypothesis
My role
Through customer research, content exploration, and working alongside the entire Fraud & Disputes team, we built a path of informative guidance that introduces situational details to help bring transaction clarity to the customer. We built an intentional flow so the customer can submit a claim for fraud or disputes with confidence.
empower customers to report fraudulent transactions while ensuring only legitimate fraud is filed?
HOW MIGHT WE
Using UserTesting.com, I conducted 10 interviews with a diverse pool of credit card users to get a sense of what content and information would surface the most recognition of a transaction. We explored what would potentially be filed as “false fraud”, our phrase for incorrectly filed fraud.
7/10
8/10
7/10
Key insights from the data
3 key insights emerged when looking at the data from the previous experience:
Disputes are accidentally filed as fraud because they aren’t aware of the option to dispute a transaction.
Customers don’t recognize a transaction due to not having enough information and will default to filing fraud.
Customers with a variety of filing outcomes select “Report a Problem” on a transaction.
Spent the most time reflecting on the question, “Do you recognize any details of this transaction?”
Used the location details to question the transaction
Believed “suspicious activity” meant true fraud and not a dispute.
Snippet from the low-fidelity screens used for testing
Data analysis & qualitative research
To get to the core of the issue, I examined the drop off points and associated data with the previous digital filing experience. In addition, I wanted to test potential content with customers to understand what words might lead them to the right outcome.
Content logic testing & discovery
Ideation
With my design lead and content design partner, I conducted several ideation sprints to explore the flow of the proper intake experience. We spent several sessions discussing the logic of what questions would help probe future questions and what the most efficient experience would look like given this was a content heavy workflow.
Low fidelity wireframes
To bring focus to the content given our design system is established for our customer facing site, we built low fidelity screens that only focused on content. When reviewing as a team, this helped to really dissect each step as its own opportunity for transaction clarity or fraud confirmation.
Final Design Solutions
System Checks
We incorporated system checks for transactions made by authorized users and recurring transactions to take the mental load off of the customer when beginning digital fraud intake.
False Fraud Probing Checklist
Our technology and merchant data is limited in what it can provide for our MVP state. With that in mind, we provided 3 points of reflection that our customer service agents use to check with customers when filing fraud. This is the final step before customers continue with fraud intake.
High Fidelity Prototype
Given that this solution was for mobile, we created the prototype for mobile only, specifically looking at all of the details included in the workflow that would ultimately help surface clarity for the customer.
Measuring Success
We hit our call reduction goal at 25%, 5% higher than expected in reductions! Our next step is to look at drop off rates and screens with the most amount of time spent on them to analyze the success of certain points in the flow.
The pilot is currently out to 20% of credit card customers as of October 2023.
Next Steps & Final Thoughts
In addition to monitoring, We are planning to move into the MVP+1 phase next year. We are simultaneously learning more about the machine learning opportunities to detect false fraud earlier on and reduce the number of steps that customers have to manually comb through.
Overall, I’m extremely proud of what we were able to uncover in this process. This project was high visibility given the past flow and subsequent losses. I’m excited to see what can be automated in the future to take the responsibility off of the customer and continue to build trust with them as a product.
