Social Care
B2B/SaaS
End-to-end Design
Shipped
Unite Us is a social healthtech company that develops platforms used by community organizations, hospital systems and government agencies to connect people to food, shelter and other essential services.
I led the redesign of their referral system in order to address legacy pain points and to lay the foundation for future releases.
Disclaimer: The following case study contains no PHI or sensitive information.
Role
Research, userflow mapping, wireframes,
high-fidelity mockups, interactive prototypes
Team
• Sr UX Designer (me)
• UX Designer x1
• Researcher x1
• Product Manager x4
• Engineer Teams x4
Platforms
Desktop web,
EHR integration
Impact
• 66% faster referral completion speed
• 37% less clicks required
• 10 million lives reached in 2024
DESIGN PREVIEW
BACKGROUND
The referral platform hadn't been updated in ages.
The referral workflow is what users must complete in order to connect their clients to a service provider (food, housing, transportation, etc.). However, it hadn't been updated in years and was not up to the task of meeting current or future business needs.
I partnered with our Lead Researcher and PM to conduct multiple user studies to identify the needs and painpoints of both Unite Us and NowPow users. Through in-person observations, zoom interviews, surveys and shadowing sessions, we uncovered the following themes:
INSIGHT A
Poor information disclosure leading to client rejections
Roughly one-third of referrals submitted through the Unite Us platform were getting rejected due to the client being ineligible for the program. Looking at the referral form, it didn't provide adequate information about eligibility requirements such as citizenship status, age, gender or location.
The user was not being presented with the information they needed to make a successful referral.
INSIGHT B
Confusing business logic leading to drop-offs
Pendo indicated that 40% of dropoffs occurred when the user tried to group referrals together. If a user created a combination of programs that violated any one of 8 rules, they would get an error. Interviewed users did not hesitate to voice their frustration.
"It punishes me for selecting the wrong program in a group and then I kind of am just done with it."
INSIGHT C
Incomplete referral submissions leading to rejections
In addition to sending a referral, users sometimes needed to submit additional forms, such as consent or payment requests, which were located in entirely separate areas of the platform.
This was resulting in referrals being rejected because they were missing crucial forms.
PROBLEM
The outdated referral platform was not empowering users to send successful referrals, leading to clients not getting the help they need.
PLANNING
Making sure everything fits into place
It was important that the new features be seamlessly integrated into the existing referral system. This included making sure that there were solid connections from each entry point, whether it be from the platform itself or via the EHR integration. This was also an opportunity to integrate other essential workflows from across the platform directly into the referral flow, such as payments authorization and patient consent.
Fitting in with the design system
For this new platform, I utilized our recently implemented UU Design System which was being rolled out across the main customer platform. This took care of roughly 80% of components needed for the new features, with the remaining 20% needing to be designed and approved through the design system governance process.
SOLUTIONS
Boosting relevant resources with revamped search page
Addresses:
Poor information disclosure
To replace the single-page referral form, I designed a more intuitive, search-driven experience modeled after familiar shopping interfaces. Users could now browse a dynamic list of results shaped by our search algorithm, apply filters, and view critical details upfront, such as eligibility based on demographics, location, or citizenship status.
Reducing drop-offs by automatically grouping programs
Addresses:
Drop-offs
To remedy drop-offs due to dead-end errors, I designed a checkout flow that automatically groups programs according to business rules. This allows users to focus on selecting the best programs for their clients, and prevents them from making any invalid groupings.
Combining workflows to ensure complete submissions
Addresses:
Incomplete submissions
With great help from engineering teams across several domains, the checkout process was expanded to support additional steps, enabling other teams to insert their own layer into the referral process. This reduced rejections due to incomplete submissions by 60 percent.
OUTCOMES
Impact
The new search page and checkout flow led to several positive user outcomes:
• 32% increase in successful referrals
• 60% decrease in error-related drop-offs
• 66% faster referral completion speed
• 10 million lives reached in 2024
These features were formally released to all users in May 2025 as part of the "new experience", which you can read more about here.
TAKEAWAYS
Design leads the way
A significant takeaway for me was the importance of using design as a tool to cut through ambiguity. Although there were several concepts that I worked hard on that didn't make the final cut, each one gave the team a better idea as to what direction we wanted to head in, and what things we should leave on the table. Rejections pave the way to the goal.
Guardrails are good
The positive performance of the referral preparer illustrated how adding guardrails to the user experience can provide better clarity. Instead of letting the user do whatever they want (and suffer the consequences), guiding them down a path while still allowing them a degree of freedom within bounds is an effective way to balance flexibility with positive outcomes.
