top of page

Designing a privacy first healthy eating app

Key Question: How can we help Medicaid members use their ‘food as medicine’ benefits to eat healthier, by using their health data to guide and encourage better choices?   

CLIENT

Major Retailer in North America (confidential)

MY ROLE

Lead UX Researcher 

Impact

The app post launch had 10,000 signups and has currently been scaled to a large scale nutrition program

The problem

Jacks (pseudonym), a major grocery retailer in North America, wanted to increase participation in the 'Food as Medicine' program, through which Medicaid members receive a $50/month stipend to spend on healthy foods.

 

Jacks partnered with us to design a health app that converts members' health history into personalized meal plans while prioritizing privacy and data sharing best practices 

MIT_Food-Diabetes-01_0.jpg
Epic 1: Consent and data sharing

How might we design a consent experience that helps Medicaid members feel informed, in control, and confident enough to share their data in exchange for personalized benefits?

My role and methodology: I led this sprint end-to-end, using a combination of concept testing and task-based usability testing across moderated 1:1 sessions.

  • Three variants of the consent and data sharing flow were designed and tested to evaluate which structure best supported comprehension, surfaced concerns, and felt trustworthy

  • Participants were also asked to navigate the app and locate their data sharing settings, testing whether the information architecture actually supported user autonomy after the initial consent moment.

Key Insights: 

 

  • Transparency builds willingness, not fear. Users weren't opposed to sharing data. They needed to understand what they were agreeing to and what the value of providing their data was 

  • Control is the price of trust. Users expected data sharing toggles to be off by default, and wanted granular control over categories like third-party sharing. Equally important was knowing they could change their mind later, and have a clear path to do so

  • Clean design signals credibility. Across all three variants, participants responded more positively to layouts that were visually uncluttered and used plain language, icons, and imagery. Users exclaimed that unlike typical privacy policy notices that made them sceptical, the clean design actually made them trust the app more

  • Insurance coverage was as important as consent. Users wanted confirmation that their insurance would cover the benefits they redeemed through the program, and they wanted that assurance in the same moment they were being asked to share their data, not buried elsewhere in the experience.

Screenshot 2026-05-04 at 7.50.16 PM.png
Screenshot 2026-05-04 at 7.52.28 PM.png

The privacy checkup screen we implemented post interviews was always accessible through account settings 

Final privacy policy screens we implemented

Epic 2: Meal planning and eating behavior 

The goal of this phase was to clearly define what process should an AI engine follow, how to ensure that the engine was not just producing noise and is actually aligned with business priorities and is taking into account what was feasible 

My role:  I designed a mixed-methods research plan combining exploratory research with concept testing. My team and I recruited medicaid patients across the US and and conducted interviews with 18 users with clinically diagnosed obesity across North America.

Key Insights: 

 

  • Users wanted help eating healthier, but rejected rigid meal plans that felt controlling. They needed to feel like they were still making choices, even if those choices were being guided.

  • Users wouldn't follow recommendations they didn't understand. They needed to know why a meal was being suggested and how it connected to their health goals

  • Many of the interviewees had tried using apps like myfitnesspal and noom but found it too difficult to stick with them without additional support

Strategic Recommendations: 

 

  • Changed meal plans to meal ideas offering 5-7 options per meal time vs a rigid plan to follow 

  • Each meal idea was tagged (e.g., "low sodium," "high fiber") mirroring the advice provided by the user's plan approved dietician, linking advice and action

  • We made the decision to go beyond meal ideas to include recipes. In interviews when we asked users how they would decide which meal to cook they said they would think about cooking times, portion sizes, effort they want to expend. These later became filters and categories in the app

Outcome 
10,000 signups on the app within the first 3 months 
40-45% monthly active users 
The client has now taken the solution internally and has scaled it to multiple states 

Have something interesting we can work on together or just want to chat about a sticky problem?
Reach out to me!
 

WhatsApp Image 2026-05-04 at 3.32.45 PM.jpeg
bottom of page