Transforming a critical healthcare portal through evidence-based UX research, accessibility remediation, and iterative design — making claims management genuinely usable for all patients.
Client
CVS Health
My Role
UX Designer & Researcher
Domain
Healthcare · Insurance · Consumer Portal
Year
2019
Tools
Figma · UserTesting · Axe · NVDA
Overview
The CVS Health Claims Portal is the digital interface through which millions of patients manage their pharmacy insurance claims — submitting claims for reimbursement, checking claim status, understanding benefit coverage, and resolving disputes. Unlike most digital products, a poor experience here doesn't just cause frustration. It can mean missed reimbursements, delayed medications, and real financial harm to vulnerable users.
CVS engaged our team at Cognizant to conduct a comprehensive UX assessment and redesign of the claims portal with two imperatives: dramatically improve usability for the general population, and ensure full accessibility compliance for users with visual, motor, and cognitive disabilities — a population significantly over-represented among regular pharmacy customers.
As lead UX designer on the engagement, I owned research, interaction design, and accessibility strategy from discovery through design handoff.
The Challenge
The existing claims portal had been designed around internal CVS Health workflow requirements rather than around how patients actually think about and manage their healthcare finances. The result was a technically functional system that was genuinely difficult to use — particularly for the users who needed it most.
Opaque claim status language — Claim statuses used internal insurance terminology ("Adjudicated," "Pended," "EOB Issued") that patients couldn't interpret without calling support. The portal generated ~45,000 support calls monthly, most of which were patients asking "What does this status mean?"
Multi-step submission friction — Submitting a new claim required navigating 7 separate screens, with no progress indicators, no save-and-continue functionality, and automatic session timeouts that caused users to lose their work — particularly damaging for users on slow connections or assistive technology.
Critical accessibility failures — Initial audit revealed 140+ WCAG 2.1 violations — including color contrast failures on primary CTAs, form fields without programmatic labels, error messages not announced to screen readers, and interactive elements unreachable by keyboard navigation.
No mobile optimization — 58% of portal traffic was mobile, but the portal was designed exclusively for desktop. Mobile users encountered truncated tables, overlapping elements, and form fields too small to interact with on a touchscreen.
Research
The research program was designed to generate evidence across three distinct dimensions: usability failures (what's hard to use), accessibility violations (what's unusable for people with disabilities), and content comprehension failures (what's impossible to understand). Each dimension required different methods.
Expert review of the full portal against Nielsen's 10 heuristics and healthcare-specific UX principles — generating a prioritized list of 89 usability issues across severity levels.
Automated scanning with Axe combined with manual testing using NVDA screen reader and keyboard-only navigation — identifying 140+ WCAG 2.1 violations with remediation priority rankings.
20 moderated sessions — 14 general population users and 6 users with disabilities (visual impairment, motor disability, cognitive disability) performing 8 core portal tasks.
Analyzed 3 months of support call transcripts to identify the specific portal failures most frequently driving customers to phone support — quantifying the operational cost of UX debt.
Plain language assessment of all status messages, error states, and help content — testing comprehension with patients who had no insurance industry background.
Contextual research into when and where patients accessed the portal on mobile — understanding the environmental context (on the way to the pharmacy, in a waiting room) that mobile design needed to accommodate.
Critical Finding
Users with screen readers were completely unable to complete a claim submission — the process had 6 form fields with no programmatic labels and an error summary that was visually present but never announced to assistive technology. This wasn't a minor compliance gap; it was a complete exclusion of a significant user population.
Business Finding
Support call analysis revealed that 62% of monthly support volume was driven by just 3 comprehension failures — all of which were fixable through plain language rewrites and better status design. The business case for UX investment was straightforward: fix 3 content problems, reduce support costs by ~$2M annually.
Design Process
This project required a disciplined dual-track approach: an accessibility remediation track that systematically resolved WCAG violations, and a UX redesign track that rebuilt the core claim submission and status management flows from the ground up. Both tracks ran in parallel, with regular integration checkpoints.
Classified all 140+ accessibility violations by severity and user impact — creating a phased remediation roadmap aligned with development sprint capacity.
Rebuilt claim submission as a progressive, 3-step flow with persistent progress indicators, auto-save, and session extension warnings — eliminating the work-loss problem entirely.
Designed all screens mobile-first, with responsive breakpoints ensuring every interaction was fully operable on the smallest common screen sizes without horizontal scrolling.
Every design deliverable included an accessibility annotation layer — specifying ARIA labels, focus order, error announcement behavior, and keyboard interaction patterns for engineers.
Key Design Decisions
Plain Language Status System — Replaced all internal insurance terminology with patient-centered plain language. "Adjudicated" became "Processed — payment on the way." Every status was accompanied by a one-sentence explanation of what it meant for the patient and what action (if any) they needed to take.
3-Step Progressive Submission — Collapsed the 7-screen submission flow into a clearly signposted 3-step process with auto-save at each step, session extension prompts, and the ability to return and complete a saved draft — eliminating the most common frustration of lost work.
Full Accessibility Rebuild — Every form field received programmatic labels and descriptions. Error states were redesigned with both visual and programmatic notification. Focus management was implemented throughout, ensuring keyboard users could navigate every workflow without a mouse.
Responsive Claims Dashboard — Rebuilt the claims history from a fixed-width desktop table into a card-based responsive layout that presented the most critical claim information (status, amount, date) at a glance on any screen size.
Impact
140+
WCAG 2.1 violations resolved — achieving AA compliance across the full portal and enabling use by screen reader and keyboard-only users
↓ 58%
Reduction in portal-related support call volume within the first quarter — directly attributable to plain language status redesign
↑ 44%
Improvement in task completion rate for claim submission — measured in post-launch usability testing against the pre-redesign baseline
Key Learnings
Accessibility is a quality measure, not a compliance checkbox. Every accessibility fix we made also improved the experience for general users. Clearer labels, better error messages, and logical focus order made the portal better for everyone — not just users with disabilities.
The language IS the interface. In healthcare, the words on the screen carry more weight than the visual design. The plain language rewrite was the single highest-impact change we made — reducing confusion and support calls more than any structural redesign could have on its own.
Include users with disabilities in usability testing from day one. Testing only with the general population would have shipped a product that was still unusable for screen reader users. Including disabled users in research uncovered issues that automated auditing tools completely missed.