Mary Kate Larson
CleanMind
CASE STUDY · IOS APP · UX/UI DESIGN · 2025
ROLE
TIMELINE
TOOLS
PLATFORM
Solo UX/UI Designer
3 months · Sep–Dec 2025
Figma · FigJam
Interactive prototype: all screens clickable with Smart Animate transitions
iOS · Mobile-first
— OVERVIEW
— RESEARCH & DISCOVERY
A companion for the moment cravings hit, not just the moments between meetings.
The streak mechanic is the problem.
There are 21 million Americans in recovery. Most formal support therapy, AA, sponsor calls, runs on a schedule. Cravings don't.
CleanMind fills the gap with a system built around one core insight: people in recovery don't need more self-awareness. They need to know what to do next. Daily check-ins track mood, sleep, cravings, exercise, and meetings attended, feeding into a real-time risk score. When that score is elevated, CleanMind doesn't just surface a warning. It generates a Next Best Action plan: call your sponsor, attend a meeting, use the in-app breathing tool. Specific steps, right when you need them
— PROBLEM STATEMENT
People in sobriety need in-the-moment support, but existing apps only log what happened, but they never tell you what to do next. When a craving hits at 11pm on a Tuesday, a streak counter doesn't help. There's a gap for something that gives people a real plan exactly when they need one.
I audited five apps: I Am Sober, Sober Grid, Nomo, Bearable, and Headspace, alongside published research on digital health interventions for substance use disorder. The pattern was consistent: streak mechanics that reset to zero on relapse. Users reported deleting the app entirely rather than facing a zeroed counter. The app's primary motivational tool was also its primary churn driver.
WHAT EXISTING APPS GOT WRONG
DESIGN OPPORTUNITIES I IDENTIFIED
Streak reset on relapse → immediate abandonment
Cold, clinical UI language created emotional distance
No actionable guidance, apps log feelings but never tell you what to do next
Check-ins designed as data collection, not genuine support
No guidance for the first 72 hours of a new attempt
Milestone model that accumulates, never resets
Warm, conversational UI language
Risk score that turns daily check-ins into a real-time vulnerability indicator
Check-ins that feel like a conversation, not a form
Progress framing that survives setbacks
— DESIGN PROCESS
From research to a system that holds together.
01
02
03
04
Jobs-to-be-done mapping
High-fidelity prototype with Smart Animate
Lo-fi wireframes — 12 screens
Design system built before hi-fi
My first dashboard put three competing actions on one screen. Peer feedback was immediate: "I don't know where to start." I rebuilt around one primary action per screen from that point on.
Before wireframing, I mapped four core user jobs: track sobriety, manage a craving right now, log how I'm feeling, and connect with community. Each job became its own flow.
Color tokens, type scale, spacing system, icon set, all before high-fidelity screens. Every screen is a system decision, not a one-off.
Full interactive prototype: onboarding → dashboard → check-in → craving response → milestones. Transitions use Smart Animate to feel like a real iOS app.
— KEY DESIGN DECISIONS
Three choices that defined the product, starting with the system underneath the screens.
Decision 1 : Milestones replace streaks. CleanMind tracks total milestones across the entire journey. A relapse doesn't erase anything, it becomes a data point. The number always goes up. This came directly from the research and is the single most important UX decision in the project.
Decision 2 : The risk score system. Every daily check-in, mood, sleep, cravings, exercise, whether you attended a meeting, feeds into a real-time risk score. This wasn't just a feature decision. It was an architectural one: turning passive logging into active intelligence. A score of 68 doesn't just mean 'elevated risk.' It triggers a specific Next Best Action plan so the user always knows what to do next, not just how they're feeling.
Decision 3 : Warm visual language as a therapeutic choice. Sage greens, parchment off-whites, rounded typography are not just aesthetic choices. Research on therapeutic digital environments shows warmer visuals reduce psychological defensiveness. The color palette has a rationale, not just a vibe. This also informed the risk score UI, a gauge showing 68 'Elevated Risk' needed to feel serious without feeling alarming. Warm amber tones rather than harsh red.
— OUTCOMES
What this project proves.
12
5
4
Screens designed end-to-end
Apps competitively audited
Distinct user flows prototyped
The most important design decision wasn't which shade of green to use. It was asking: when someone's risk score hits 68 at 11pm, what do they actually do next? Answering that question designed the whole product.
"
— Mary Kate Larson
The biggest gap is the absence of moderated usability testing with people actually in recovery. Every decision is research-informed, but research about others, not feedback from real users of this specific app. The craving response flow is the highest-stakes screen in the product and deserves real validation, not just peer review.
I'd also push deeper into accessibility. Someone using this app in acute distress isn't reading carefully, that means larger touch targets, simpler microcopy, and WCAG AA compliance at minimum.
— REFLECTION
What I would do differently
Mary Kate Larson
© 2025 Mary Kate Larson · Boston, MA