← All work

Credit Card Fraud Detection & Risk Platform

A FICO-sponsored project helping fraud analysts catch real threats with less noise in credit-risk monitoring.

01

Project overview

This was an industry-sponsored project with FICO, the company behind the FICO credit score and a range of risk products for banks. FICO's UX VP and Product Manager defined the brief, provided the real dataset, and answered our questions in weekly sessions throughout, and the VP returned at the end to evaluate the final design. The brief was to design an analytics and operations dashboard for fraud managers working in credit card fraud call centers.

I owned the manager experience, designing dashboards that give managers real-time visibility into agent and queue performance so they can make fast, data-driven decisions. This worked in tandem with a parallel design for the agent side. While my partner designed the individual agent's workflow, I designed the manager's view of that same agent's performance, creating a unified product with two complementary perspectives.

A key requirement of the brief was integrating gamification elements to improve productivity, accuracy, and employee engagement in a high-pressure environment.

02

Problem & goal

Problem

Fraud managers lack real-time visibility into performance and fraud trends, leading to reactive decisions, inefficiencies, and employee burnout.

Goal

A centralized manager dashboard that combines real-time analytics, actionable controls, and gamification to improve decision-making, productivity, and employee experience.

03

Jobs-to-be-done

This project used a Jobs To Be Done (JTBD) framework because the problem space is highly operational and task-driven, with clearly defined roles and success metrics. Fraud managers’ effectiveness depends on completing time-sensitive tasks such as monitoring performance, reallocating resources, and responding to system signals, rather than on individual traits or preferences.

JTBD shifted the focus from who the users are to what they need to accomplish in specific situations. This made it easier to translate requirements into clear use cases and tasks, directly informing the dashboard’s information architecture, workflows, and action points.

Jobs-to-be-done map for a fraud manager balancing queues: job performer, target job, job steps, emotional and social jobs, needs and circumstances.
JTBD for Balance Queues (Manager). Click to open full size.

04

Conceptual models

Conceptual models were used to define how the system works before moving into screen design. They helped break down a complex product into clear objects, actions, data, and tasks, creating a shared structure for the system.

Using conceptual models reduced unnecessary complexity and made interactions more consistent and predictable. This foundation allowed the design to scale cleanly as more data, features, and workflows were added.

An Object × Action table defined which actions managers can perform on core system objects (investigators, queues, models, and cases), ensuring clear permissions and actionable UI controls.

Object × Action table: investigators can be transferred, tracked, shared and messaged; queues and cases can be transferred, tracked and shared; models can be tracked and shared.
Object × Action table

05

High-fidelity prototypes — iterations

Transfer agent

Initial prototype

The initial design feedback surfaced two key issues in the transfer-agent flow:

  • No visibility into how transferring an agent would affect queue load
  • Redundant agent selection, requiring multiple clicks across screens and overlays

Additional screen iterations

After receiving feedback on the initial design, I iterated to improve visual consistency, interaction clarity, and engagement. Feedback addressed:

  • Standardized screen dimensions across the flow
  • Established a consistent title hierarchy and labeling system
  • Expanded the transfer interaction to provide clearer, more flexible options (shown above)
  • Introduced gamification elements to increase engagement and system transparency
Dashboard after iteration: six KPI cards, a department leaderboard, a tabbed agent performance chart with forecast, and a status card per fraud model. Dashboard before iteration: four KPI cards, cases per queue, an agent performance table and a single model FPR chart.

Click the screen to see the before.

Expanded the KPI row from four to six metrics (adding agent performance, training completion, retention and ROI), added a department leaderboard, turned agent performance into a tabbed chart with a forecast, and gave each fraud model its own status card.

06

New screens

As part of the gamification requirement, I designed two new screens to boost agent performance, accuracy, and teamwork. The Game screen introduces progress tracking and achievement elements to increase engagement during work. The Training screen allows managers to assign courses and track agent progress, connecting skill development directly to performance outcomes.

Game screen: a department leaderboard ranked across accuracy, productivity, teamwork, training and retention, with performance and ROI charts and a quarterly budget bar.
Game. Department leaderboards across five performance categories, with performance, ROI and budget tracking.
Training screen: training modules with an Assign Training action, a training ROI chart, before and after training performance, and a training leaderboard.
Training. Managers assign modules, then track completion, ROI and how performance changes after training.

07

Learnings

  1. 01

    Data visualization

    Deepened my ability to convey complex operational data through charts, graphs, and dashboards by selecting the right visualization for each use case and user question.

  2. 02

    Micro-interactions

    Refined interaction patterns such as filtering, sorting, and drill-down to details, enabling users to explore data progressively without overwhelming the interface.

  3. 03

    Gamification

    Learned how to apply gamification thoughtfully to increase engagement, motivation, and system transparency without distracting from core tasks.

  4. 04

    Jobs To Be Done

    Strengthened my use of the JTBD framework to ground design decisions in user intent and outcomes, ensuring solutions addressed real operational needs rather than surface-level features.

More case studies