AI-Assisted Medicare Advantage Enrollment
Overview
Client: UnitedHealthcare | Status: Live
Problem: Medicare shopping is confusing. Only 2% enroll online. Most users call instead, enduring 42-minute calls while agents juggle up to 25 tools per enrollment.
Goal: Simplify Medicare shopping for seniors and support agents with fast, personalized plan insights and streamlined tools. Guide Medicare shoppers with real-time, AI-driven answers to reduce call volume and boost digital enrollments.
Role and team: I lead UX design for this ongoing initiative, partnering with product, engineering, and compliance teams to deliver a conversational assistant that’s accessible, accurate, and Medicare-compliant for older adults and their caregivers.
Design Process & Iteration Highlights
Early Exploration: We began with a floating icon entry point, but testing showed it was too easy to miss. We moved to an inline module on plan pages, which boosted engagement.
User Journey Mapping: We mapped common user pathways from search → plan comparison → enrollment to identify key decision moments where the chatbot could help.
Testing Takeaway: Seniors responded better to simplified prompts (“What plan is cheaper?” vs “Compare plan premiums”) — this informed our conversational tone.
Early concept: an AI-powered product recommendation tool. Due to compliance and MVP timeline constraints, it pivoted to a simpler chatbot—expansion planned.
User flows for the early product recommendation concept.
Phased rollout:
Phase 1 – Q4 2024: Launched MVP on Plan Details pages. Conversational AI provides real-time answers to plan-specific questions.
Phase 2 – Q1 2025: Expanded to Plan Compare and Summary pages. Usability testing shaped entry points and interactions.
Phase 3 – Q2 2025+: Scaling across the Medicare shopping journey to provide personalized, real-time guidance.
Results:
Higher digital adoption, reduced call handling time, and faster, smoother enrollments.
Image 1: Initial chatbot prompt. Image 2: Plan-specific question. Image 3: AI response. Image 4: AI-generated answer with source docs.