CommonSpirit Health

We designed search tools that worked together and stayed consistent across the site, with an eye toward scaling and keeping the patterns useful over time.

We started by identifying the most common user journeys, most notably patients who start with a specific provider in mind. That insight shaped a search experience flexible enough to handle everything from finding a doctor to locating a nearby facility.

Stakeholder interviews surfaced the legacy pain points, and behavioral data confirmed them. A competitive audit showed where other health systems outperformed the legacy model, and overlaying that with decision points and component requirements gave us a roadmap everyone could work from. We wrote scenarios and edge cases to test whether future work would hold up in the real world.

Analytics also pointed to appointment setting as the next step in the journey, and where most users dropped out of the funnel.

What we learned became a roadmap for all future CSH digital properties: user, business, and technical requirements synthesized into a single visual form, ending in a generalized journey for any search aimed at finding a caregiver.

Result cards were one of the most critical pieces. They needed to show only what mattered, so a search stayed fast and matched what the user had asked for.

CommonSpirit was midstream on a new brand voice and visual identity, and that anchored our work. The client needed a shared place for new brand elements as they were created, so we built the smaller Brand Center site, a first step toward a style shared across all their properties.

Brand Center also gave us the beginnings of the CommonSpirit design system. We worked from mood boards to wireframes with reuse on the flagship site in mind, and mapped where the existing component set fell short of the competitive experiences from our research.

Care usually starts with a search, and for a lot of users that search ended in confusion or abandonment. Given the range of users and motivations, personas were too narrow and made-up to be useful. We used focused research instead, tied to specific real-world interactions.

Those interactions became 12 user archetypes to test against, reflecting key parts of the user base and validating the scenarios from our roadmap.

Rather than static wireframes, we built interactive prototypes that mimicked searching, filtering, and comparing results. Automated testing flagged friction at scale, while moderated sessions let us watch users in context, sharpening features like comparing multiple caregivers before deciding.

Most interactions still centered on finding a physician or facility, usually starting from the caregiver. When users didn't enable geolocation, the platform returned no results with no fallback or prompt, which is a bad place to lose someone.

Appointment setting was inconsistent: slots shown as available were sometimes revoked after selection, which hurt trust in the system.

Facets were the third challenge. Too many options overwhelmed users, too few left them unable to narrow results. Users across age groups welcomed AI-driven scheduling as long as it was fast and forgiving of mistakes.

Details like whether a provider spoke a given language mattered just as much. We designed overlapping filters that shifted based on the path a user was on: if a doctor wasn't taking new patients, we surfaced others at the same facility, and nearby practices that met the same criteria.

After several rounds of research and optimization, we moved on to the rest of the site: the diseases and conditions library, the blog, and utility pages, linking anchor content to user intent through a taxonomy built from the roadmap, and working with development on iterative visual testing.

The relaunch was a clear success, with metrics and feedback backing up the creative decisions. Low expectations for onsite healthcare search came from a lack of detail, which was our opportunity to add value with a more complete resource.

Six months in, the site was drawing positive feedback and generating additional revenue on a higher conversion rate, and the new component set gave the designs room to grow.

Physician and location templates will keep improving based on analytics data. More work is planned for the post archives, with unique templates for specific types of time-stamped content. There's also a plan to bring in third-party reviews on the physician templates, replacing the site-specific rating system that exists now.