Transforming Complex Legal Workflows into a Structured, High-Efficiency Case Management System.

Learning the domain before opening a design tool.

Immigration law is a specialized domain. Before I could design anything meaningful, I needed to understand how immigration cases actually work—the stages, the documents, the compliance requirements, and the legal stakes involved. I spent time studying the domain deeply before opening any design tool.

Replacing assumptions with evidence.

I mapped every screen of V1, documenting the existing user flows, navigation patterns, and interaction models. This created a factual baseline—not assumptions, but documented evidence of where the experience broke down and why.

Turning scattered feedback into a prioritized problem map.

The client was more of a tester for us in the first two months. The PM had collected client feedback as a rough problem list from direct client meetings and emails. I took this unstructured input and systematically mapped it to specific parts of the product.

Direct user observation

I conducted online observation sessions with users as they completed real tasks. I observed the Application Manager, PM, and development team members working through the product. I took detailed notes, documented exact pain points, and timed task completion. Screen recordings were taken for reference—no third-party tools, only direct observation.

Mapping every dependency and potential path.

I created detailed user flows for every feature and every user type—attorneys, paralegals, and admins. This revealed dependencies, gaps, and unnecessary steps that were not visible in the UI alone. It was the most critical step before ideation.

From a data dump to a working environment.

The analysis revealed dependencies, gaps, and unnecessary steps that were not visible in the UI alone. Two major issues were blocking confident, efficient case work.

Milestone-based progressive disclosure

Attorneys do not think in tasks—they think in stages. “What phase is this case in?” comes before “What do I do next?” Instead of showing all 34 tasks at once, I restructured the workspace around a milestone-based hierarchy. Nine milestones became visible checkpoints on the left panel: a persistent progress trail showing exactly where the case stood.

Inside each milestone, tasks were revealed progressively. Completed tasks were visually struck through, the current task was highlighted, and upcoming tasks remained visible but locked until dependencies were met. A “Next Task” CTA removed the decision entirely, while a progress counter gave instant case-level health at a glance.

The product thinking shifted from feature-based navigation to workflow-based navigation. The system was now organized around how work actually happens, not how the database was structured.

Motor memory: removing the visual hunt

Users faced action paralysis because buttons were scattered inconsistently across screens, with some steps showing up to seven competing CTAs. By anchoring primary actions to a persistent bottom bar, I replaced decision friction with a predictable motor-memory pattern.