Scaled a brand-agnostic components system and conversational interface to expand Merck’s global reach to healthcare providers across 72 countries.

My role:

  • Design lead for the entire workstream (4 additional Merck product projects)

  • Defined UX strategy from sitemap consolidation to component-level authoring flexibility

  • Partnered with product and engineering to align on roadmap and localization needs

  • Led collaboration with marketers, researchers, and copywriters to ensure scalability and compliance

  • Designed and prototyped the conversational interface, tone, and testing approach

what markets and users needed

Regulatory complexity & localization needs: Merck’s 50+ HCP sites were shaped by local MLR constraints, with minimal global standards and wide structural variation.

Need for product-specific branding: Drug sites needed to reflect individual product identities while maintaining a clear connection to the Merck brand.

Desire for marketer autonomy: Local teams lacked tools to build or update content without developers, creating bottlenecks, especially in content-heavy franchises like fertility.

These insights shaped my proposal: to maintain the integrity of the AEM backend while giving drug teams control over the front end, offering an authorable UI layer and CSS flexibility that allowed each market to reflect product branding, without compromising global consistency or compliance.

defined UX architecture for a fragmented global ecosystem

  • Audited Merck’s sites across regions and consolidated them into a unified, flexible sitemap structure.

  • Identified the ideal entry points from Merck’s consumer .com into the HCP gated experience to streamline way-finding and prevent drop-off.

  • Designed with localization in mind—factored in variables like RTL language support (Arabic), extended French text, and strict U.S. legal overlays.

  • Created custom components for country-specific needs:

    • U.S.: Persistent safety info occupying 25% of the screen.

    • France: Removed character limits to allow for longer prescription copy, with guidance via authoring tooltips.

    • Korea: Addressed CSS bugs that broke spacing between Hangul characters.

built a brand configurable UI toolkit

Designed a library of 180 modular AEM components to support marketer-authored sites—customizable by drug brand while retaining Merck’s global header and footer.
Provided documentation and visual rules for how and when brand elements could flex, enabling consistency without rigidity.

Scaling the platform with a conversational agent that helps providers quickly access prescribing information in high-pressure clinical moments.

why scaling the tech made sense

Despite having robust content on Merck’s HCP platform, customer support continued receiving frequent questions from providers, especially around dosage, contraindications, and administration.

This revealed a gap in how easily providers could access critical information in the moment. To address it, we built a proof of concept for a conversational assistant that could deliver fast, personalized answers, making existing content more accessible in clinical settings.

insights from prescribing doctors

I partnered with a UXR lead to interview six prescribing HCPs (neurologists, oncologists, and generalists) and Merck’s customer support agents.
 The key insights were:

  • Speed is critical: Routine info like dosage or side effects is often looked up between patient visits.

  • Escalation is expected: Off-label use, trial data, or financial help should route to a rep.

  • Skepticism runs high: Most users had low trust in bots from past experiences.

“I’ve got the next patient waiting for me… I don’t want to spend 10 minutes hunting down some abstract.”

designing intent driven conversations

I designed the assistant’s conversation architecture, defining intents, follow-ups, and escalation triggers based on real prescribing scenarios. I translated complex drug content into structured, navigable flows and collaborated with a copywriter to script responses and chip suggestions in a tone that balanced clarity and clinical credibility.

I also worked closely with engineering to align on platform constraints—leveraging Dialogflow for NLP, Firebase for session management, and SendGrid for email delivery, all within GDPR guidelines.

guerilla testing

To validate the experience, I designed and co-led Wizard of Oz testing in Slack, simulating real-time interactions before development began.

  • Simulated real-time answers using scripted replies and visual cards

  • Tested tone, flow logic, and escalation clarity

global impact, long term partnership

  • The conversational interface prototype validated HCP demand for self-serve prescribing info and laid the foundation for a broader AI-driven service roadmap.

  • Adoption of toolkit by 72 countries

  • The account generated ~$12M in business revenue.

  • Delivered a flexible, brand-agnostic component system that enabled drug teams to quickly publish compliant, locally branded HCP experiences.

what i learned

  • Working with emerging technology tested our team’s comfort with ambiguity. With varying levels of confidence across disciplines, it became crucial to over-communicate, align often, and test early.

  • Documentation and onboarding for local marketers should have been prioritized earlier to streamline requests from regional marketers.

  • Local regulations vary dramatically, making upfront alignment on edge cases essential.

case studies: