As part of a consulting engagement with CommonSpirit Health—one of the largest healthcare systems in the U.S. with over 2,200 care sites across 24 states—I led efforts in visual design, UX, and UI for a major website relaunch. The primary focus was improving core functionality around physician and location search, a critical entry point for patient engagement.
Search is the foundation of CommonSpirit’s digital experience, acting as the primary bridge between patients and the care they need. However, previous design attempts had fallen short in accommodating the system’s scale and complexity. The physician and location search tools—vital to appointment scheduling and patient navigation—were underperforming due to fragmented patterns and inconsistent user flows.
Our work involved designing interoperable search tools that maintained consistency across the site, with particular attention to scalability and future-proofing the patterns. We collaborated closely with stakeholders to ensure that the tools aligned with CommonSpirit’s patient-centered mission.
We began by identifying the most common user journeys—most notably, patients starting their search with a specific provider in mind. This insight shaped our design direction. Our team focused on creating a seamless, intuitive search experience that could flexibly handle multiple use cases, from finding a specific doctor to locating nearby facilities.
At the time, CSH was one of the three largest health systems in the U.S., so there was a substantial amount of motivation to make sure the physician selection process supported the business. The agency had attempted to solve the physician and location search problems but had run into problems developing patterns that were robust enough to handle the daunting number of variables involved.
The journey to care often begins with a simple search. However, for many users engaging with CommonSpirit.org, that search resulted in confusion, frustration, or abandonment.Due to the vastly diverse range of potential users and motivations, user personas were not deemed an effective approach due to their narrow and fictionalized nature. Focused research was used instead, as it was more granular and could be linked to precise real-world interactions. These were then used to develop user archetypes to test against.
User data and qualitative feedback revealed key breakdowns in the site experience—beginning with the search functionality itself. In instances where users did not enable geolocation, the platform frequently returned null results, with no fallback suggestions, alternative prompts, or contextual help. This lack of support at a critical juncture increased user frustration and reduced task success rates.
The appointment-setting process also suffered from inconsistencies. There were exceptions in availability that contradicted what was initially shown to users. For example, appointment slots presented as available were occasionally revoked after selection, creating dissonance and eroding trust in the reliability of the system.
Another significant challenge was the handling of search facets. Users frequently encountered interfaces that either surfaced too many or too few filtering options, resulting in search errors. When overexposed, users felt overwhelmed; when underexposed, they were unable to meaningfully refine their results. This tension contributed to a broader pattern of user dissatisfaction during the search and discovery phase. Survey data reinforced these experiential observations.
Interestingly, users across age groups were generally open to AI-driven scheduling tools, so long as these tools demonstrated efficiency and offered error tolerance. The use of machine intelligence was not rejected outright, but expectations around precision, transparency, and fallback mechanisms were high.
Developing the journey maps for user search and appointment setting necessitated overlaying the quantitative and qualitative data, and distilling the key moments and inflection points. Four primary journeys were identified, all of which required some degree of search input/output.
Developing a component set – and kickstarting a nascent design system – began with the initial designs and grew organically over time. Deviations from primary traffic paths also needed to be addressed. These included the media section of the site which covers blogs, news, and events, as well as the treatments and conditions library.
Despite the broad scope of the improvements, issues remained. In particular, these were items that fell outside the initial scope. Noteworthy among them were remaining generic errors in the scheduling system when the inputs included a certain set of variables. Other areas for focus in the near future include the integration of limited patient records, and filling in areas of the physician and location templates that remained unpopulated.
Physician and location templates will be continuously optimized based on analytics data. Additional improvements are also planned to the post archives, with unique templates for the specific types of time-stamped content. There is also a plan in place to integrate third-party reviews in the physician templates, to replace the site-specific rating system that currently exists.
© 2023 Jon Gifford