Overview
Role
Principal UX & Product Design Lead
Scope
UI Design
UX Design
Design System
Research
Information Architecture
Primary Users
Payment Integrity Specialists, Special Investigation Unit, Auditors, Case Managers, Admins, Executives
Healthcare payers lose billions annually to fraudulent claims, improper billing, and upcoding schemes. Optum’s Payment Accuracy Control System (PACS) was designed to help Special Investigation Units (SIU) and Data Mining auditors detect these anomalies.
The Problem
Investigators were doing real financial detective work by reviewing fraud alerts, comparing billed versus paid amounts, and deciding which providers to escalate for recovery. However, the interface worked against them at nearly every step: alert volume with no triage, ambiguous financial figures, truncated trend data, and the single most vital piece of evidence buried below the fold where every participant missed it.
The Solution
Through field research and usability benchmarking, we overhauled the system: accelerating investigation speed and accuracy across high-exposure claims by eliminating reliance on external spreadsheets, resolving a 1-in-25 false-positive noise baseline, and driving discovery of organised fraud rings from 0% to 100% through an exposure-first triage model, sticky claim workspaces, and interactive Social Network Analysis (SNA).
Designing the Experience
Exposure-First Triage and Alert Redesign
Investigators prioritise their day around recoverable dollars, but the original cards only showed a generic value figure that hid whether money had actually been paid out. We redesigned the triage cards to lead with true financial exposure by labelling Amount Paid clearly, surfacing provider specialities alongside unique NPI identifiers, and making filter states persistent so supervisors could triage queues without constantly losing their place.
A High-Density Claims Workspace
Analysts were routinely abandoning the product for Excel because the core line-item data was buried beneath slow, unhelpful charts. We pinned a sticky header with vital alert metadata to anchor the page, then elevated the raw claim lines directly underneath it with one-click copy interactions to keep users inside the tool. We also expanded the historical view from three months to three years, giving investigators the control limits and peer baselines they needed to validate spikes without second-guessing the data.
Interactive Fraud Ring Mapping
Collusion networks were previously invisible because the critical risk table sat below the fold, resulting in zero discoverability during initial testing. We rebuilt the Social Network Analysis layout to scale nodes by offender significance and edge thickness by financial flow, pulling the community tables directly into the main viewport alongside the visual graph. Pairing this with an interactive geospatial timeline allowed investigators to scrub across dates and watch suspicious billing clusters expand across physical locations.
Dedicated Workflows
The original interface treated all users as single-provider fraud investigators, which clashed with the mental model of data mining auditors who manage systemic overpayment scenarios. We split the core architecture into two purposeful experiences: single-entity forensic deep dives for investigation teams, and scenario-based inventory management for data miners. This gave auditors the multi-claim filtering, bulk status updates, and batch refund workflows they needed to process thousands of claims across multiple providers at once.
Outcome
100% Fraud Ring Discovery
Redesigning the SNA layout and hierarchy ensured all tested investigators immediately located and explored high-risk provider networks.
Triage Clarity
Defaulting work queues to Amount Paid and providing 3-year peer baselines enabled teams to bypass the 95% false-positive noise baseline.
Zero Spreadsheet Drop-Off
Restructuring claim line tables with one-click copy affordances kept analysts working inside the secure, auditable platform.