Year
June - Aug 2024
Role
Product Designer
Skills
User Research, Prototyping, Interaction Design, Product Strategy
Context
Persona’s Graph is a fraud detection product used by enterprises like Amazon, Walmart, Lyft, and DoorDash. It allows investigators to visualize relationships between accounts and quickly spot links that may indicate fraud, such as shared IP addresses, device fingerprints, or suspicious sign-up patterns.
The goal is simple: help investigators catch fraud with less effort. But the experience of drilling into nodes for detailed account information was creating friction, slowing down investigations, and frustrating customers.
The Challenge
When I joined, the product manager gave me a straightforward task:
“Customers want to make the node detail cards sticky so they don’t disappear when you move your cursor. Can you fix that?”
At first, this sounded like a small UI tweak. But after digging deeper, I realized the real issue wasn’t stickiness, it was comparison. Investigators weren’t just frustrated that cards disappeared. They were frustrated that they couldn’t easily compare multiple accounts side by side, which is the whole point of running an investigation.
This reframed the problem: Instead of a band-aid fix, we needed to redesign the entire comparison experience.
Research & Insights
Customer Conversations
I spoke to product managers, trust & safety leads, and risk ops teams at both customer companies and internally at Persona. Their frustrations echoed the same themes:
Hover cards disappeared instantly, making it impossible to scroll or keep context.
Only one card could be viewed at a time, limiting comparison.
Cards overlapped unpredictably on the canvas, making scanning difficult.
Market Research
To ground my explorations, I studied similar investigation and visualization tools (Plaid Beacon, Ravelin, Castle, TigerGraph, Salesforce, Sardine).
Across these tools, I saw common patterns emerge:
Allowing multiple details to be pinned or expanded at once.
Off-canvas panels (sidebars or drawers) for structured comparisons.
Lightweight hover states paired with persistent, user-controlled detail views.
These insights gave me a design vocabulary for tackling our own solution.
Explorations & Prototyping
This project was interaction-heavy, so prototyping in Figma became essential. I built high-fidelity, clickable prototypes for a range of approaches, including:
Sidebar comparison: pinned nodes displayed in a right-hand panel.
Bottom drawer: cards stacked horizontally for quick scanning.
Movable sticky cards: details could be grabbed and dragged around the canvas.
I returned to the same customers I’d interviewed and observed them working through real investigations using these prototypes. Watching how they naturally gravitated toward certain flows helped validate which concepts solved their pain points most effectively.
Final Solution – Milestone 1
We landed on a hybrid approach that:
Enables users to pin multiple cards for side-by-side comparison.
Keeps cards anchored consistently, avoiding overlapping chaos.
Provides a clear visual hierarchy for active vs background cards.
This design directly addressed the usability gaps customers had flagged, while also laying a scalable foundation for richer investigation tools.
Future Opportunities
Through this process, I uncovered that comparison was investigators’ most critical need. Looking ahead, I proposed:
Gallery View: batch view for browsing many accounts at once.
Focus Mode: zoom into a subset of nodes without distraction.
These ideas were logged as future explorations for deeper optimization.
Collaboration & Delivery
I partnered closely with engineering to pressure-test feasibility and edge cases throughout the design cycle. This tight alignment made handoff smooth, with engineers already bought into the interaction logic before specs were finalized.
The designs are scheduled to ship in Q4 2024.
Learnings
This project taught me the importance of challenging the initial ask. What started as a request for a “sticky card” became an opportunity to rethink the investigation workflow altogether.
By digging into why users wanted stickiness, I was able to:
Uncover the root problem (comparison).
Reframe scope with data-driven evidence.
Deliver a solution that solved not just the symptom, but the system.
