JO

Work

CV

JO

JO

Work

CV

AbbVie

Competitive Intelligence Search Tool

I led the design of a competitive analysis tool from 0-1. This tool has shipped to 8 R&D hubs around the world, impacting a total of 8,000+ users.

AbbVie

Competitive Intelligence Search Tool

I led the design of a competitive analysis tool from 0-1. This tool has shipped to 8 R&D hubs around the world, impacting a total of 8,000+ users.

Role

Lead Designer

Duration

Jan - Apr 2024

Team

Context

Competitive intelligence at AbbVie involves a systematic, multi-step process that enables the company to make smart decisions, particularly in the high-stakes area of developmental drugs. At a high level, it helps the company:

  • Stay ahead of competitors

  • Minimize risk and R&D waste

  • Have timely patient impact

Problem

Faced with a tedious and fragmented process, it takes AbbVie scientists hours to find up-to-date drug information. This inefficiency slows drug development and poses the risk of costing the company tens of millions of dollars annually.

Outcome

My team and I created a platform that helps 8000+ scientists find up-to-date drug information within seconds, which are used to make strategic business development choices that save the company tens of millions of dollars.

Scope

One of our largest clients, AbbVie, challenged my team to design a tool that would streamline the time-consuming process of competitive intelligence analysis. In early meetings, the client presented a clear vision of what they wanted, confident they understood both the user needs and the right solution.

I saw an opportunity to pause and challenge that assumption. By surfacing unanswered questions and highlighting the risks of assumption-based design, I advocated for and led a round of user interviews to ground the project in real-world needs. In those interviews, I uncovered critical insights that not only clarified real pain points in scientists’ workflows, but also revealed new opportunities we hadn’t initially considered.


These opportunities were shared back with the clients and added to our roadmap. However, given the upcoming launch at AbbVie's global innovation showcase, the Celebration of Science, we had to prioritize. While many of the new ideas were valuable, I made the strategic call to focus our MVP on the core experiences that would best demonstrate the platform’s reliability and potential. The goal wasn’t just to ship, it was to inspire confidence in scientists, decision makers, and people leaders and spark future adoption.


Balancing the tradeoff between perfection and velocity, I focused the first iteration on high-priority features aligned with client goals, while documenting future enhancements. I quickly ideated, wireframed, and prototyped key functionality in close collaboration with the engineers and client to ensure a fast, aligned build cycle. Ultimately, the MVP was presented with great success at the Celebration of Science, generating strong interest from scientists who could now depend on a tool to completely revolutionize the way they conduct competitive intelligence analyses.


I then packaged the final designs into a detailed developer-ready handoff file and stayed closely engaged through implementation. I conducted visual QA to ensure polish and usability as well. The launch exceeded expectations, earning enthusiastic support from stakeholders and laying the groundwork for future investments and iteration.

Show more

Scope

One of our largest clients, AbbVie, challenged my team to design a tool that would streamline the time-consuming process of competitive intelligence analysis. In early meetings, the client presented a clear vision of what they wanted, confident they understood both the user needs and the right solution.

I saw an opportunity to pause and challenge that assumption. By surfacing unanswered questions and highlighting the risks of assumption-based design, I advocated for and led a round of user interviews to ground the project in real-world needs. In those interviews, I uncovered critical insights that not only clarified real pain points in scientists’ workflows, but also revealed new opportunities we hadn’t initially considered.


These opportunities were shared back with the clients and added to our roadmap. However, given the upcoming launch at AbbVie's global innovation showcase, the Celebration of Science, we had to prioritize. While many of the new ideas were valuable, I made the strategic call to focus our MVP on the core experiences that would best demonstrate the platform’s reliability and potential. The goal wasn’t just to ship, it was to inspire confidence in scientists, decision makers, and people leaders and spark future adoption.


Balancing the tradeoff between perfection and velocity, I focused the first iteration on high-priority features aligned with client goals, while documenting future enhancements. I quickly ideated, wireframed, and prototyped key functionality in close collaboration with the engineers and client to ensure a fast, aligned build cycle. Ultimately, the MVP was presented with great success at the Celebration of Science, generating strong interest from scientists who could now depend on a tool to completely revolutionize the way they conduct competitive intelligence analyses.


I then packaged the final designs into a detailed developer-ready handoff file and stayed closely engaged through implementation. I conducted visual QA to ensure polish and usability as well. The launch exceeded expectations, earning enthusiastic support from stakeholders and laying the groundwork for future investments and iteration.

Show more

Scope

One of our largest clients, AbbVie, challenged my team to design a tool that would streamline the time-consuming process of competitive intelligence analysis. In early meetings, the client presented a clear vision of what they wanted, confident they understood both the user needs and the right solution.

I saw an opportunity to pause and challenge that assumption. By surfacing unanswered questions and highlighting the risks of assumption-based design, I advocated for and led a round of user interviews to ground the project in real-world needs. In those interviews, I uncovered critical insights that not only clarified real pain points in scientists’ workflows, but also revealed new opportunities we hadn’t initially considered.


These opportunities were shared back with the clients and added to our roadmap. However, given the upcoming launch at AbbVie's global innovation showcase, the Celebration of Science, we had to prioritize. While many of the new ideas were valuable, I made the strategic call to focus our MVP on the core experiences that would best demonstrate the platform’s reliability and potential. The goal wasn’t just to ship, it was to inspire confidence in scientists, decision makers, and people leaders and spark future adoption.


Balancing the tradeoff between perfection and velocity, I focused the first iteration on high-priority features aligned with client goals, while documenting future enhancements. I quickly ideated, wireframed, and prototyped key functionality in close collaboration with the engineers and client to ensure a fast, aligned build cycle. Ultimately, the MVP was presented with great success at the Celebration of Science, generating strong interest from scientists who could now depend on a tool to completely revolutionize the way they conduct competitive intelligence analyses.


I then packaged the final designs into a detailed developer-ready handoff file and stayed closely engaged through implementation. I conducted visual QA to ensure polish and usability as well. The launch exceeded expectations, earning enthusiastic support from stakeholders and laying the groundwork for future investments and iteration.

Show more

Role

Lead Designer

Duration

Jan - Apr 2024

Team

Research

The client stakeholders aimed to modernize the competitive intelligence workflow to accelerate and improve the drug development process. They mentioned several steps in the workflow as critical pain points that are time consuming and arduous. To validate this assumption and better understand the broader context, my team and I proposed conducting user interviews with scientists to learn how they currently approach competitive intelligence and identify key areas for improvement.


These interviews confirmed several pain points identified by the client stakeholders and surfaced additional insights we hadn't previously considered, including a need for visual representations of drug development data, the ability to annotate findings, and tools to save and share search history among others. To prioritize these ideas, I facilitated a workshop with both our team and client stakeholders, mapping the opportunities on a 2x2 matrix. This helped us align around the features that would deliver the greatest impact in the shortest time and guided the MVP direction.

Beyond the unmet needs revealed in user interviews, we also observed several behavioral challenges scientists faced during their workflow. These were issues that weren’t explicitly stated but became clear through watching how they worked. These behavioral friction points were slowing them down and contributing to inefficiencies in the competitive intelligence process. By identifying and addressing these subtle pain points, we saw an opportunity to meaningfully enhance the workflow and bring us closer to achieving a smoother, more intuitive experience for end users.

🔥

High-Friction When Googling

Scientists noted they felt pressure to work faster and expressed the concern of taking too much time to Google articles and press releases related to competitor drugs.

😖

Low Confidence

Scientists struggled to determine which results warranted deeper review, causing them to either miss important information or spend excessive time researching.

😵‍💫

Overwhelming Possibilities

Scientists found it daunting to sift through multiple pharmaceutical databases like the National Institute of Health and thousands of Google results at the start of their investigations.

Opportunity

Through observation, we noticed that while scientists could locate relevant content, the tools available to them weren't built for navigating complex medical data. This made the process inefficient and frustrating. Often, scientists relied on Google searches opened in multiple tabs and windows, followed by additional research to verify the credibility of the information they found. This constant switching between sources not only slowed them down but also eroded confidence in their findings.


These behaviors pointed us towards two key user needs. Streamlining the experience and building trust. After getting buy in from our engineers, I proposed consolidating relevant information into a single, unified interface to reduce the number of steps required to build a competitive landscape report, eliminating the need to jump between tools and databases.


To address the lack of trust we observed, we focused on surfacing reliable trust signals within the platform such as publication dates, source links, and confidence scores, so scientists could quickly evaluate the credibility and relevance of data at a glance. By reducing friction and increasing confidence, we aimed to make the competitive intelligence process not just faster, but more dependable as well.

Platform: Data Consolidation Tool

The way scientists begin the competitive intelligence workflow is critical as it shapes how efficiently scientists can produce a comprehensive analysis of competitive landscapes. To reduce friction and make the workflow more enjoyable for our users, I reimagined the way scientists discover and filter information related to developmental drugs.


Previously, scientists had to manually navigate multiple sources, such as the National Institute of Health and other databases, just to piece together basic insights. This not only slowed their progress but also left them overwhelmed by fragmented tools and inconsistent interfaces.


With the Competitive Intelligence Search Tool, we centralized this process. The new search experience pulls relevant data from trusted sources into one cohesive view, enabling scientists to explore, compare, and analyze information without switching between platforms. This shift significantly reduced time spent searching and made it easier to uncover meaningful insights.

🔧

Engineer's Corner

What seemed like a simple page on the surface required deep collaboration with engineers to get right. One of the biggest challenges was figuring out how to take unstructured inputs like PDFs, medical articles, and free-form text and transform them into structured, usable outputs that scientists could rely on.


Early in the process, I worked with engineers to understand how variables like data quality, transformation accuracy, and processing latency would affect performance and by extension user trust. We knew that if the information appeared incomplete or delayed without explanation, it could create confusion or doubt. In response, I designed clear fallback states to inform users when data was still processing or temporarily unavailable, helping to set expectations and avoid confusion.


Handling large datasets added another layer of complexity. Since much of the data transformation was happening dynamically or on the client side, we faced potential performance issues. To address this, I collaborated with engineering to implement lazy-loading components and caching strategies for high-traffic queries. I also introduced loading indicators, skeleton loading screens, and prioritized the rendering of high-value content first to keep the experience efficient.


We also addressed how data updates would be handled, both in terms of backend refresh logic and what users would see in the interface. To make the experience more transparent, we added clear timestamps showing when each piece of data was last updated. This gave users immediate context about the freshness of the information they were viewing. Behind the scenes, the data was set to refresh on a regular cadence, ensuring that the UI consistently reflected the most recent available data without requiring user action.


Finally, we had to account for incomplete data. Some fields would inevitably be missing or inconsistent. To address this, I helped design a modular layout that could handle gaps gracefully by labelling missing fields with clear, informative messages like "Not reported" or "Data pending," rather than leaving blank or confusing spaces.


The end result was a collaborative and rewarding work experience that produced a focused, reliable interface that delivered structured insights from messy inputs, while staying resilient to the limitations of real-world data.

Show more

🔧

Engineer's Corner

What seemed like a simple page on the surface required deep collaboration with engineers to get right. One of the biggest challenges was figuring out how to take unstructured inputs like PDFs, medical articles, and free-form text and transform them into structured, usable outputs that scientists could rely on.


Early in the process, I worked with engineers to understand how variables like data quality, transformation accuracy, and processing latency would affect performance and by extension user trust. We knew that if the information appeared incomplete or delayed without explanation, it could create confusion or doubt. In response, I designed clear fallback states to inform users when data was still processing or temporarily unavailable, helping to set expectations and avoid confusion.


Handling large datasets added another layer of complexity. Since much of the data transformation was happening dynamically or on the client side, we faced potential performance issues. To address this, I collaborated with engineering to implement lazy-loading components and caching strategies for high-traffic queries. I also introduced loading indicators, skeleton loading screens, and prioritized the rendering of high-value content first to keep the experience efficient.


We also addressed how data updates would be handled, both in terms of backend refresh logic and what users would see in the interface. To make the experience more transparent, we added clear timestamps showing when each piece of data was last updated. This gave users immediate context about the freshness of the information they were viewing. Behind the scenes, the data was set to refresh on a regular cadence, ensuring that the UI consistently reflected the most recent available data without requiring user action.


Finally, we had to account for incomplete data. Some fields would inevitably be missing or inconsistent. To address this, I helped design a modular layout that could handle gaps gracefully by labelling missing fields with clear, informative messages like "Not reported" or "Data pending," rather than leaving blank or confusing spaces.


The end result was a collaborative and rewarding work experience that produced a focused, reliable interface that delivered structured insights from messy inputs, while staying resilient to the limitations of real-world data.

Show more

🔧

Engineer's Corner

What seemed like a simple page on the surface required deep collaboration with engineers to get right. One of the biggest challenges was figuring out how to take unstructured inputs like PDFs, medical articles, and free-form text and transform them into structured, usable outputs that scientists could rely on.


Early in the process, I worked with engineers to understand how variables like data quality, transformation accuracy, and processing latency would affect performance and by extension user trust. We knew that if the information appeared incomplete or delayed without explanation, it could create confusion or doubt. In response, I designed clear fallback states to inform users when data was still processing or temporarily unavailable, helping to set expectations and avoid confusion.


Handling large datasets added another layer of complexity. Since much of the data transformation was happening dynamically or on the client side, we faced potential performance issues. To address this, I collaborated with engineering to implement lazy-loading components and caching strategies for high-traffic queries. I also introduced loading indicators, skeleton loading screens, and prioritized the rendering of high-value content first to keep the experience efficient.


We also addressed how data updates would be handled, both in terms of backend refresh logic and what users would see in the interface. To make the experience more transparent, we added clear timestamps showing when each piece of data was last updated. This gave users immediate context about the freshness of the information they were viewing. Behind the scenes, the data was set to refresh on a regular cadence, ensuring that the UI consistently reflected the most recent available data without requiring user action.


Finally, we had to account for incomplete data. Some fields would inevitably be missing or inconsistent. To address this, I helped design a modular layout that could handle gaps gracefully by labelling missing fields with clear, informative messages like "Not reported" or "Data pending," rather than leaving blank or confusing spaces.


The end result was a collaborative and rewarding work experience that produced a focused, reliable interface that delivered structured insights from messy inputs, while staying resilient to the limitations of real-world data.

Show more

Feature: Custom Entry Point

To surface the precise information scientists needed, I designed a powerful and flexible search component tailored specifically for developmental drug research. Rather than relying on a generic search bar, this component was built to reflect how scientists think by offering contextual filters that let users narrow or broaden their search based on their specific goals.


Scientists could input as many or as few parameters as they wanted, giving them control over the depth and scope of their queries. We hypothesized that this level of search customization would reduce the number of steps required to find relevant data and significantly improve the speed and accuracy of the experience.


Early and close collaboration with developers was essential to building this component, as it relied on complex but intuitive interaction design. I worked with the engineering team from the start to explore different interaction models, aligning on what was feasible and user-friendly. Throughout the development process, we continued meeting regularly to review progress and ensure pixel perfect implementation. This tight feedback look allowed us to catch and simplify overly complex interactions early on, which not only improved the user experience but also made implementation smooth and more efficient.

🔧

Engineer's Corner

The search component of the competitive intelligence tool was one of the most complex parts of the product, both in terms of interaction design and underlying functionality. It had to support intricate queries that allowed scientists to generate the specific comparison landscapes they needed. This meant enabling Boolean intersections across multiple entity types like diseases, genes, and compounds within a streamlined, easy-to-use UI. To maintain focus and reduce initial complexity, we deferred support for operators like “OR” and “NOT” to a later phase of the roadmap.


Because of the complexity of the search logic, I collaborated closely with engineers to ensure that the data was indexed appropriately to support the interactions I designed. We held multiple alignment sessions to define how the system should handle different data types, query patterns, and edge cases, with the goal of balancing user experience with performance constraints.


Even seemingly minor features, like the paste function, required thoughtful design. Users often pasted multiple terms into the search field, but for the platform to parse them correctly, each item needed to be comma-separated. To reduce user friction, we designed the input field to automatically insert commas where possible and included clear visual cues to communicate the expected input format. While we saw opportunities to improve this interaction further, we prioritized core functionality over perfecting secondary flows at this stage.

Show more

🔧

Engineer's Corner

The search component of the competitive intelligence tool was one of the most complex parts of the product, both in terms of interaction design and underlying functionality. It had to support intricate queries that allowed scientists to generate the specific comparison landscapes they needed. This meant enabling Boolean intersections across multiple entity types like diseases, genes, and compounds within a streamlined, easy-to-use UI. To maintain focus and reduce initial complexity, we deferred support for operators like “OR” and “NOT” to a later phase of the roadmap.


Because of the complexity of the search logic, I collaborated closely with engineers to ensure that the data was indexed appropriately to support the interactions I designed. We held multiple alignment sessions to define how the system should handle different data types, query patterns, and edge cases, with the goal of balancing user experience with performance constraints.


Even seemingly minor features, like the paste function, required thoughtful design. Users often pasted multiple terms into the search field, but for the platform to parse them correctly, each item needed to be comma-separated. To reduce user friction, we designed the input field to automatically insert commas where possible and included clear visual cues to communicate the expected input format. While we saw opportunities to improve this interaction further, we prioritized core functionality over perfecting secondary flows at this stage.

Show more

🔧

Engineer's Corner

The search component of the competitive intelligence tool was one of the most complex parts of the product, both in terms of interaction design and underlying functionality. It had to support intricate queries that allowed scientists to generate the specific comparison landscapes they needed. This meant enabling Boolean intersections across multiple entity types like diseases, genes, and compounds within a streamlined, easy-to-use UI. To maintain focus and reduce initial complexity, we deferred support for operators like “OR” and “NOT” to a later phase of the roadmap.


Because of the complexity of the search logic, I collaborated closely with engineers to ensure that the data was indexed appropriately to support the interactions I designed. We held multiple alignment sessions to define how the system should handle different data types, query patterns, and edge cases, with the goal of balancing user experience with performance constraints.


Even seemingly minor features, like the paste function, required thoughtful design. Users often pasted multiple terms into the search field, but for the platform to parse them correctly, each item needed to be comma-separated. To reduce user friction, we designed the input field to automatically insert commas where possible and included clear visual cues to communicate the expected input format. While we saw opportunities to improve this interaction further, we prioritized core functionality over perfecting secondary flows at this stage.

Show more

Feature: Tools for Deeper Investigation

To help scientists verify and trust the information they were reviewing, I designed a detailed drug view page optimized for deeper investigation. The goal was to give scientists the context they need at a glance, while still giving them the flexibility to easily pursue additional trust signals if needed. Our hypothesis was that showing these trust signals in one place would increase user confidence in the data and make it easier for scientists to validate or act on the information efficiently.


The information architecture of the page was shaped by early user interviews, combined with design intuition grounded in scientists’ workflows. One of the key insights was that sharing and exporting findings played an essential role in cross-functional collaboration. To support this need, I placed the “Share” and “Export” actions in a prominent position at the top right of the screen, making them easy to access without disrupting the user’s flow.


Below the top actions bar, I presented high-level metadata about the selected drug, disease, or gene to immediately orient users and reinforce the context of their investigation. The rest of the page was divided into tabbed sections, each supporting a deeper layer of research—such as comparing the current asset to in-house products or reviewing related publications.


The “Overview” tab was selected as the default landing tab, as it contains the most frequently referenced content, including the drug’s development timeline and description. This structure allowed scientists to quickly find key information while providing flexible paths for deeper exploration as needed.

🔧

Engineer's Corner

To effectively surface trust signals like data sources, publication dates, and last-updated timestamps, I first aligned with engineering on where the data was coming from, how often it was refreshed, and how it would be processed. Once we established a shared understanding of the data logic and update cadence, I designed a clear and unobtrusive way to present this information. The goal was to give scientists confidence in the data they were viewing and help establish the platform’s credibility.


While I typically design with scalability in mind, our user research showed that most scientists conducted their work on desktop devices. With that in mind, the engineers and I prioritized optimizing the experience for desktop viewports, while still creating flexible components that could adapt to different screen sizes and accommodate dynamic content in the future.


Testing and QA were a collaborative effort as well. With engineering, test cases were defined together and worked on iteratively to ensure the implementation was accurate, polished, and faithful to the design intent ahead of launch.

Show more

🔧

Engineer's Corner

To effectively surface trust signals like data sources, publication dates, and last-updated timestamps, I first aligned with engineering on where the data was coming from, how often it was refreshed, and how it would be processed. Once we established a shared understanding of the data logic and update cadence, I designed a clear and unobtrusive way to present this information. The goal was to give scientists confidence in the data they were viewing and help establish the platform’s credibility.


While I typically design with scalability in mind, our user research showed that most scientists conducted their work on desktop devices. With that in mind, the engineers and I prioritized optimizing the experience for desktop viewports, while still creating flexible components that could adapt to different screen sizes and accommodate dynamic content in the future.


Testing and QA were a collaborative effort as well. With engineering, test cases were defined together and worked on iteratively to ensure the implementation was accurate, polished, and faithful to the design intent ahead of launch.

Show more

🔧

Engineer's Corner

To effectively surface trust signals like data sources, publication dates, and last-updated timestamps, I first aligned with engineering on where the data was coming from, how often it was refreshed, and how it would be processed. Once we established a shared understanding of the data logic and update cadence, I designed a clear and unobtrusive way to present this information. The goal was to give scientists confidence in the data they were viewing and help establish the platform’s credibility.


While I typically design with scalability in mind, our user research showed that most scientists conducted their work on desktop devices. With that in mind, the engineers and I prioritized optimizing the experience for desktop viewports, while still creating flexible components that could adapt to different screen sizes and accommodate dynamic content in the future.


Testing and QA were a collaborative effort as well. With engineering, test cases were defined together and worked on iteratively to ensure the implementation was accurate, polished, and faithful to the design intent ahead of launch.

Show more

Summary

By the end of the project, we successfully delivered on our north star by creating a platform that reduces the effort required to generate competitive landscape reports and removes key friction from the competitive intelligence workflow.


The platform also increases user confidence by surfacing trust signals like source links, publication dates, and validation statuses—helping scientists make decisions with greater speed and assurance.


Scientists can now search for developmental drug information using flexible inputs, enabling quick comparisons and deeper exploration of individual compounds. What was once a fragmented, manual process is now centralized, faster, and more intuitive.


We exceeded industry standards with a strong Net Promoter Score of 58 that reflected high user satisfaction and energized the team to pursue additional features scientists had initially requested, including:


  • Improved performance through deeper engineering collaboration

  • Real-time alerts to track drug development activity

  • Drug history comparison tools

  • Integration with existing AbbVie platforms


Not only do these opportunities inspire more confidence in our users and reduce friction, they offer a clear path to expand the platform’s impact and further support scientists in their decision-making.

Takeaways

This project was a lesson in balancing stakeholder expectations with user needs under tight timelines. While AbbVie came to us with a clear solution in mind and a fast-approaching showcase deadline, early user research revealed gaps between their vision and scientists' actual workflows. Rather than push back entirely, I took a hybrid approach. I moved forward with the core of their concept to meet the immediate goal, while layering in insights from user interviews to shape the design in ways that would be more useable and scalable long-term.


This experience taught me that user-centered design isn’t always about pushing back. Sometimes, it’s about strategic alignment. Meeting stakeholders where they are, building momentum with early wins, and embedding user value incrementally. By doing this, I was able to build trust, gain influence, and pave the way for deeper user validation in future iterations.

Prism

© 2025

© 2025

© 2025

Jung Oh.

Jung Oh.

Jung Oh.

Made in Toronto, Ontario.

Made in Toronto, Ontario.

Made in Toronto, Ontario.

in

in

in

@

@

@