TL;DR
A major prospect asked us to convert the Kanban-style triage list into a table view. Research with the prospect and existing clients showed the real issue was information display, not view type. The redesigned triage list matched how clinics actually work, and the prospect chose it.
Context & Existing Feature
Cardiologs is an AI-powered platform assisting cardiologists and electrophysiologists in analyzing long-duration ECGs. The Worklist (used tens to hundreds of times daily) managed patient recordings across hospitals, clinics, and IDTFs.
The original Worklist comprised four separate screens:
- A 3-column Kanban board (Processing, Pending Review, Reviewed)
- A separate Quality Control screen
- An Archive for recordings older than 14 days
- A Favorites section for starred ECGs
The Starting Point
The prospect asked us to convert the Worklist from Kanban to a list view. The request came from familiarity with their previous software, not from a clearly articulated need. But the existing Worklist did have real limitations: critical information scattered across screens, minimal card detail, and Quality Control functionality that was hard to find.
Before making changes, I proposed doing user research with both the prospect and existing clients.
Research
Interviews with 3-4 clinicians and IDTFs found:
- The triage process contained five stages, not three
- Cards lacked sufficient context: patient name, DOB, recording duration, urgency labels, and assigned reviewer were all missing
- The Favorites feature went unused
- Filtering and sorting were essential but hard to access
Two Directions, One Presentation
I designed and presented two solutions:
- A table view matching the prospect's initial request, enriched with the information gaps we'd identified
- An improved column view reflecting the real five-stage triage process, with richer cards and integrated Quality Control
Presenting both options showed that the Kanban structure matched actual clinic and IDTF workflows better than the requested table format.
What We Shipped
Phase 1 expanded the columns from three stages to five: To Do → Ongoing → Pending Quality Control → Ongoing Quality Control → Completed. We added a filter bar, search, and a clear upload entry point, removed Favorites, and turned Archive into a simple tab.
Phase 2 redesigned the cards to surface critical information: patient name, date of birth, file ID, recording duration, upload date, urgency and status labels, country, assigned reviewer, and last action timestamp.
Outcome
The prospect signed. The redesigned Worklist needed no post-launch iteration and drew no complaints, strong validation for a screen used tens to hundreds of times daily. One solution worked for every client, with no need to maintain separate views.
What I'd do differently
I would have implemented analytics tracking from day one (time-to-first-click, filter usage, column dwell time) to build a data foundation for future improvements.
