COmpany

International Business Machines (IBM): w3 Publisher

Domain

Technology

Scope

User research
Prototyping
Product design

Platform

Desktop

Context:

A young tool in need of iteration.

The image widget is one of many modular tools created for Publisher, IBM’s proprietary site builder. In addition to allowing user to embed images on site, the tool offered storage and management for uploaded images. What I noticed was beyond it’s basic functionality, the widget lacked many features found in comparable tools.

Problem:

Achieving a high-quality visual experience required technical know-how.

At the beginning of my tenure, Publisher’s image widget only offered sourcing from personal devices. As a result, our primary audience, non-technical users, had to either rely on third-party stock image sites or source their own images, risking a reduction in site quality. This created friction in everyday workflows, contributed to user dissatisfaction and was reflected in Publisher’s Net Promoter Score (NPS).

Solution:

Simplify asset handling and expand tooling to make utilizing the right image quick and intuitive.

Publisher's existing upload workflow could produce quality results, but only when users had the knowledge or guidance to select appropriate images. The opportunity was to introduce sourcing options and tools that reduced the technical knowledge required while maintaining quality. The goals realized from this were:

Short-term goal(s):

• Streamline existing workflows without reducing familiarity for long term users.

• Add image sourcing options that simplify accessing high-quality media.

Long-term goal(s):

• Expand configuration tools to reduce screen switching and cognitive load.

• Enhance sourcing options with AI-assisted workflows to better support non-technical users.

Impact

4 second

average time reduction when uploading and committing an image.

98.3%

task completion across during usability testing.

< 1%

detractors mentioned the image widget as a pain point during the first three months.

*Together, these results indicate improvements in both perceived product quality and the efficiency of core image-management workflows.

Challenges & trade-offs

Carbon consistency vs. familiar workflows

When Carbon lacked coverage for a workflow, I looked for existing components that could be adapted rather than introducing entirely custom patterns. Because these exceptions required Carbon team review and could affect our roadmap, they had to be intentional and sparingly used. The image card was a necessary exception because Carbon's tile component couldn't support the functionality required to maintain parity with Publisher's existing experience. On the other hand, other deviations, such as adapting the drawer into a split-panel experience, were justified by significant usability improvements.

Immediate impact vs. long-term investment

Time constraints required me to break larger improvements into the smallest deliverables that could independently address meaningful parts of the problem, then prioritize them by impact. Stock image sourcing, for example, provided an immediate improvement for non-technical users, while an AI-assisted workflow using Watsonx could eventually evaluate page context and recommend more relevant assets. Because the latter required significantly more investment, we prioritized the stock library as an interim solution while deferring AI-assisted curation.

Simplicity vs. discoverability

Removing friction sometimes meant deprioritizing lower-frequency controls or moving them to secondary locations, trading some discoverability for a simpler experience. After adding additional sourcing options, I found that exposing every action upfront created unnecessary complexity, so I designed a launchpad around the most common actions for each source while keeping niche tools accessible elsewhere. I applied the same principle to the editing tools and image cards.