Healthcare Navigation Redesign

A redesigned information architecture cut secondary navigation items by 26%, lifted click-through to support resources by 4.5%, and cut bounce rate by 16.7%, all while improving the site's search performance. The client's navigation had grown to 7 primary and over 50 secondary items, burying the support content that mesothelioma patients and caregivers came to the site to find. I led a mixed-methods study, pairing moderated interviews and a card sort with an unmoderated card sort across 40 participants, to uncover users' mental models and then rebuilt the navigation around it.

Secondary nav items

-26%Secondary nav items

CTR to support resources

+4.5%CTR to support resources

Bounce rate

-16.7%Bounce rate

The client's site navigation had grown to 7 primary and over 50 secondary items. In interviews, both users and internal employees described the same struggle: they couldn't find what they were looking for, especially the support content for patients and caregivers navigating a mesothelioma diagnosis. That content existed on the site, but it was buried several levels deep in a structure organized around internal categories rather than how people actually thought about their situation, and it showed up as high exit and bounce rates on exactly the pages meant to help people in the moments they needed it most.

How might we restructure the navigation around how people affected by mesothelioma actually think about their journey, so the support they came for was never more than a click or two away, without disrupting the organic search traffic the site depended on?

The client was known internally for being skeptical of site changes and slow to approve them, so I treated rigor as the actual deliverable: every recommendation needed data strong enough to survive their scrutiny. Before designing the study, I ran alignment meetings with the client to understand their budget and tolerance for change, and to walk through what card sorting and interviews could and couldn't tell us, so nobody was surprised by the method or the findings later.

I designed a mixed-methods study: moderated interviews paired with a moderated card sort for 6 participants, plus an unmoderated card sort for 34 more, to balance the depth of qualitative conversation against a sample large enough to be quantitatively credible. Recruiting people with a direct connection to mesothelioma proved difficult, so the study widened to caregivers and patients across cancer types more broadly, on the assumption, validated in the interviews, that the information-seeking journey was similar enough to still be useful. I also coached a mid-level UX researcher I was working with on which method fit which question, pairing her on the moderated sessions while she ran a share of the unmoderated study independently.

“Doctor Google said, and I quote me because I'll never forget it, mesothelioma is an inevitably fatal disease with no known treatment.”

— Study participant, generative interview

Because the site's traffic depended heavily on organic search, I worked closely with the technical SEO and content teams throughout, checking every proposed structural change against which pages were actually driving traffic from Google. I also ran a competitive analysis, since the client tracked their top competitor closely and wanted to know how our recommendation would stack up, and I was able to use the research to help justify the new visual hierarchy and global header direction.

Card Sort Findings

• Respondents converged on the same six mental-model categories: Treatment, Support & Resources, About Cancer, Risk Factors, Legal Help, and About Us
• A meaningful share of participants independently proposed a dedicated Support or Resources category, even though nothing like it existed in the current navigation
• Caregivers and patients grouped content almost identically, so the new IA didn't need to fork by audience
• The number of primary categories people expected roughly matched what already existed; the real problem was one level down, where more than 50 secondary items had no clear hierarchy

To give the client confidence before committing to a full rollout, I built a pre/post analysis plan that benchmarked link clicks, bounce rate, and organic impressions on the pages most at risk, then proposed a phased rollout instead of a single cutover, so we could catch any negative signal early and roll it back before it became a bigger problem.

Research only mattered if it could withstand heavy client pushback. I grounded every recommendation in specific interview and card-sorting data, so decisions couldn't be dismissed as subjective. That shifted conversations from opinion to evidence and aligned both client stakeholders and our internal team.

Zoom reviews were slowing us down, with feedback filtered through layers of stakeholders and causing weeks of delay. I led an in-person workshop with executives, walking through the sitemap and updating the prototype live based on their input, which enabled real-time approvals instead of prolonged email chains. That single session unblocked more progress than months of remote reviews.

I also proposed a phased rollout with rollback plans for each stage, giving the client confidence to approve smaller, reversible steps rather than one large, high-risk launch.

The phased navigation rollout validated the approach with clear pre/post metrics: click-through to support resources rose 4.5%, and bounce rate on key pages fell 16.7%, showing users were finding what they needed more efficiently.

This improvement came without sacrificing organic search traffic. By integrating SEO from the start and throughout the phased rollout, we maintained rankings, meeting the client's core condition for approving any navigation changes.

• Reduced secondary navigation items by 26%
• Click-through rate to support resources up 4.5%
• Bounce rate down 16.7%, indicating stronger findability
• Updated visual design to improve content findability
• Rolled out in phases with SEO monitored throughout, protecting the organic traffic the client depended on