Pre-authorization & referral submission

Providers were guessing their way through a tangled list of care types. Plans spent half their day fixing bad submissions. We segmented the catalog, guided the flow, and cut submission errors by about 90%.

Role
Senior Product Designer (research → ship)
Team
Ayin platform · Provider & plan workflows
When
Enterprise redesign with external partners
Impact
~90% fewer submission errors
90%
Fewer errors needing plan-side fix-ups
50% → less
Plan time spent repairing submissions (was ~half the day)
Card sort
Care types co-designed with plans & offices
Stepper
Only the fields that care type needs
90% fewer errorsResearchProvider + plan
Healthcare professionals collaborating
Partner research — plan staff and provider offices in the same problem

The problem

Pre-authorization and referral submission lived in the membership view. A provider office picks a patient, then tries to submit an auth or referral to the plan. In one-on-ones, I watched office staff fight the form—too many options, unclear labels, everything mixed together.

I took that suspicion to our health plan partners. We reviewed their reports and sat with their workflow. The pain wasn’t theoretical.

“We spend about 50% of our time fixing authorizations rather than approving or denying them. I spend a lot of time on the phone with providers trying to help them submit correctly.” — Health plan partner

Research

I also brought in people who know pre-auth workflows but don’t live in our tool every day. They ran the current product and said the quiet part out loud:

“This list is ridiculous—which one is the right one? I’ll just pick this and write what I want in the comments.” — Usability session participant

Authorizations and referrals were jumbled. Staff couldn’t tell what the plan actually needed. Comments became a dumping ground for missing structure.

Hypothesis

Segment pre-authorizations and referrals into clear care types (and sub-types) so the form only asks for data that type needs—and so the plan receives something they can act on.

Goal: fewer bad submissions, less plan-side rework and coaching.

What we built

Card sorting with plan partners and provider office staff: primary care types and sub-types. That segmentation powered a wizard-style stepper—each step controlled by the care type, not a single overloaded form.

Stepper flow for care type submission
Guided stepper — fields depend on the care type selected

We validated with prototypes (Sketch/InVision at the time), then refined through development with continuous usability checks.

Results

Plan effort spent fixing bad submissions
Before~50%Of staff time repairing & coaching
After~5%After segmentation + guided flow

≈90% reduction in error-driven rework

From partner workflow reports after rollout; templates later cut provider time further.

Error drivers before redesign (ranked)
  • Wrong care type / mixed catalog38%
  • Missing required clinical data27%
  • Ambiguous free-text in comments21%
  • ICD / coding mistakes14%

Relative share of plan-side correction reasons before the redesign.

Feedback from plans and providers improved across the workflow. We later added templates to speed repeat submissions, and an ICD-10 lookup so offices can check billable codes without leaving the tool.

Care types configuration and submission UI
Shipped experience — structured care types instead of a flat guessing list

Why this still matters

Multi-step clinical and plan workflows fail the same way: too many choices, not enough structure, humans papering over the product in free text. Segmentation + guided steps is still how I approach dense healthcare forms—claims, auths, referrals, assessments.

← All case studies