Returns Processing Operations Dashboard

Context

Associates complete high-volume, repetitive tasks; supervisors manage throughput and exceptions across lanes.

Users switched between 3 tools per item. Priority queues were unclear. Supervisors relied on end-of-day reports.

Problem

Processing delays and errors increased cost per return. Associates needed speed and clarity; supervisors needed situational awareness without micromanaging.

Leadership

Player-coach scope on the pod:

• Owned design direction with no direct reports — set standards for eng design reviews
• Advocated for role-based IA over separate tools; aligned ops leadership on shared data model
• Ran weekly working sessions with floor leads during pilot
• Defined scan-feedback patterns reused in subsequent floor-tool projects

Process

Floor-first discovery:

• Site visits during peak receipt at 2 processing centers
• Task timing — 40% of time lost to context switching
• Co-design sessions with associates on paper
• Pilot prototype on one lane before full build

Solution

Role-based views on a shared data model:

• Associate mode: scan → classify → disposition → next
• Supervisor mode: queue depth, error rate, active exceptions
• Large touch targets and immediate scan feedback
• Plain-language error recovery to reduce escalations

Impact

Pilot lane showed ~25% fewer mis-classifications. Rolled out to 800+ daily users. Scan-feedback patterns adopted as reference for other ops tools.

Reflection

Designing with associates on the floor surfaced constraints we wouldn't catch in Figma. Next time I'd run a longer pilot across shift changes.