Overlake Medical Center & Clinics is a non-profit health system that enhances the health and wellness of those living in the greater Puget Sound region. The informative, actionable website supports organizational growth initiatives for new patient acquisition and retention.
In 2015, the health system entered into an engagement with an agency to overhaul the website property. The goal of this build revolved around a “Google-inspired hospital website.” The proposed solution bridged plain-language search intent with clinical taxonomy to produce highly personalized search results (see animation). Sadly, this solution cascaded into a waterfall of disappointment when building the custom taxonomy quickly became far more complicated and cumbersome than imagined. Ultimately, when the website went live in 2016, search never worked as expected, often returning results that were not in alignment with the query.
When I joined in the marketing team in 2018, we opted to work with a new partner agency and rebuilt the Drupal website from the ground up. As the internal expectation for a “Google-like search experience” remained, I soon realized that an investment in a hosted search solution would ultimately be necessary to meet or exceed the expectations that remained from years prior. The goal ultimately remained relatively the same: To simplify and accelerate the discovery of information that patients and visitors were seeking.
Following a full website overhaul in 2020, I began looking for a new tool that could solve for search intent, relevancy, type ahead and autocorrect. I was attending a continuing education webinar when I discovered another health system using Algolia for find a doctor/provider search. When budget was secured and a contract in place, they were selected as the hosted search partner for the website and we launched into building the framework for the solution architecture.
Throughout 2021, I worked toward gaining the executive buy-in to ensure ongoing funding would be available to support this solution. If there was anything the two years prior had taught me, the most complex—and by far most critical—section of the site to get right was the “Find a Doctor or Provider” index. As such, we built the proof of concept around this index to demonstrate how it solved for some of the most common, ongoing concerns.
Below are two screen recordings taken during the proof of concept to demonstrate how the product solves for two of the concerns. The first shows variations of how the “obstetrics and gynecology” nomenclature can be easily solved for with synonyms, and the second video demonstrates how the “out of the box solution” accounts for misspellings during search queries:
While the web development team at our partner agency began building and refining the data connection (from Drupal to Algolia), I worked alongside a graphic designer on our team to build the UX/UI for the platform. Together, we explored search and discovery patterns on websites across a myriad of industries, honing in on ways we could leverage these flows within the healthcare space.
As each index (services, doctors, locations, etc.) had a very distinct set of business needs, each relied heavily on section-specific facets and filters for search refinement. Each element was thoughtfully designed to ensure a delightful, pleasant experience for users while navigating them to the information they need as quickly as possible.
High-fidelity wireframes for the Find a Doctor or Provider search index on mobile devices is outlined below:
The website property had what seemed like an infinite amount of business rules that needed to be solved for within each index. These configurations also needed to build upon one another in order to produce a global search solution.
I began combing through historical search results captured in the organization’s Google Analytics account to help prioritize our focus. In addition to some of the more obvious terminology (e.g. “medical records,” “COVID test”), I wanted to account for terms highly utilized, but perhaps less clear. Some examples included: mapping zip codes to “City, State” locals, provider specialties to plain-language terminology, alternate forms of the same term (including abbreviations) including internal vs. external terminology.
The synonyms tool allowed for quick alignment of these terms to achieve the index-specific needs:
Furthermore, the organization’s online health library proved an invaluable resource for building synonym taxonomy between clinical keywords and lay terminology.
The final step brought everything together for the configuration of searchable assets, including connection of data to facets and filters. This required great thoughtfulness in how ranking and sorting should occur by index, while aligning the data to the intended UX/UI behavior: