Clinician Connection Platform

The platform connected clinicians with patients across telehealth, scheduling, and care management. Five years of organic growth had left the web and mobile experiences looking like they belonged to different companies. Basic tasks required different interaction patterns depending on the platform.

Patients were abandoning appointment flows at a 34% rate. Clinicians were burning time relearning the same features in each context. Trust in the product was eroding from both sides.

The interface audit came first, and the numbers made the case better than any argument: 14 button styles, 6 variants of the same appointment card, 3 different date pickers. Each reasonable in isolation, collectively incoherent.

Moderated sessions with 18 clinicians and 24 patients confirmed it. Users weren't failing because features didn't exist — they were failing because features looked and behaved differently across contexts. Knowledge didn't transfer from desktop to mobile. An accessibility audit (axe-core, plus manual testing with JAWS and VoiceOver) surfaced 200+ WCAG violations concentrated in forms and modals.

I proposed building the design system before touching product features — a hard sell when stakeholders want visible improvements. The argument that landed was cost: the team was spending roughly 30% of design time recreating components that already existed in slightly different forms, and every new feature added to the debt.

We started with the 12 core components that covered 80% of the interface. Each went through the same loop: inventory existing variants, identify the real interaction model, design an API that covered legitimate variation and killed the arbitrary kind, then build and test with the accessibility specialist embedded from day one.

Adoption was the other half of the job. Weekly pattern review gave feature designers a place to propose changes instead of forking, and office hours meant nobody was blocked long enough to justify a workaround.

The system — internally named Pulse — shipped with 48 components, each with usage guidelines, accessibility requirements, and responsive behavior documented. WCAG 2.1 AA was built into the foundations, not retrofitted.

With the system in place, we rebuilt the three highest-impact flows: appointment scheduling, secure messaging, and care note review. Same components on web and mobile, behavioral consistency for the first time. Scheduling went from 7 steps to 4, with inline validation and honest progress indication.

Design systems are political as much as technical. What converted the skeptics wasn't a slide deck — it was building the new appointment flow prototype in half the usual time because consistent components already existed.

The accessibility work was the most personally meaningful part. When testing showed a blind user completing the appointment flow unassisted for the first time, that was the moment I knew the system was doing its job.