Healthcare
Medikou
0→1 Product Design
UX RESEARCH
USABILITY TESTING
Rapid Prototyping
Note: Designs shared below have been translated to english and some visual alignments may be off due to that.
Context
I joined a small Japanese startup to design a Minimum Viable Product for Medikou, a clinic discovery app aimed at helping young adults navigate Japan's overwhelming healthcare landscape of 180,000+ clinics. Working as the sole UX designer with a 12-month timeline and an external development team handling implementation, I needed to balance user needs with technical feasibility and handoff clarity. These constraints required me to make strategic scope decisions early and design for clarity over complexity.
My Role
I led the product from initial research to prototyping, working closely with a UI designer, and other cross-functional stakeholders
Timeline
12 month
Outcome
Delivered NPS 8 vs target of 7, 40% reduction in clinic search time
Built high-fidelity prototype to validate product-market fit
Completed final-design specs and documentation for external development team
Abundance Without Clarity
Japan's healthcare system lists over 180,000 clinics — and that abundance was quietly becoming a barrier. In 2021, users were spending an average of 12 minutes on clinic websites before churning, a signal that choice wasn't translating into confidence. Our task was to design Medikou, a product to help young Japanese citizens find the most suitable clinic for their needs. The measure of success was clear: an NPS of 7, reflecting genuine user satisfaction with the discovery experience.
80% of Users Were Failing Before We Started
Before fielding our own interviews, we grounded our approach in a published medical survey of 1,000 participants nationwide, using it to pressure-test our assumptions about where the real friction lived. That foundation shaped 20 user interviews focused specifically on mental models for search flow, information prioritization, and decision-making during clinic discovery.
Key Findings
The clearest strategic signal: young adults aged 22–35 emerged as the right user segment. Digitally native, willing to adopt new tools, and active healthcare consumers, they were the group most likely to change behavior if the product earned their trust. That framing shaped every subsequent design decision.
The Starting Point: Three Features Earned by Data
Research pointed directly to three features worth building: two credibility variations for surfacing clinic reputation, a preference-driven home screen with filtering by clinic type, location, and availability, and an appointment flow surfacing the essential details users needed to act. Each feature traced back to a specific finding, and each was treated as a hypothesis to validate rather than a decision already made. I designed a complete end-to-end booking solution that directly addressed observed user patterns:
Search & Discovery
Home screen with selectable preferences (clinic type, location, availability) - reflecting the 72.1% who shared similar prioritization patterns, allowing users to fast-track to their likely criteria
Standard search bar for symptom and location input - addressing the 55.8% already using Google-based search behaviour
Clinic Selection
Clinic cards highlighting reputation signals (ratings, review counts) - targeting the 75% who focused heavily on credibility and social proof
Essential information architecture (services, address, hours, contact)
Comparison capabilities for decision-making
Appointment Booking Flow
Complete in-app booking interface
Calendar integration for availability
Confirmation and reminder system
Discovering a Constraint
Four prototyping rounds and 12 A/B testing sessions gave us the feedback needed to find out what actually didn't hold up — and three findings directly changed the product's direction.
The Strategic Decision
Integrating with these fragmented systems would require:
Extensive backend development beyond our 6-month timeline
Ongoing clinic partnerships and onboarding
Complex API integrations with no standardization
Resources the startup couldn't support for MVP with external developers
I recommended a strategic pivot: Rather than delay indefinitely, shift from in-app booking to assisted connection. Using Google Maps API, we'd surface verified clinic contact information, enabling users to book directly via phone while maintaining our core validated value - helping them find the right clinic for their needs.
The Trade-off
What we sacrificed: Booking convenience, transaction data, retention through complete flow
What we gained: Speed to market, eliminated technical dependencies, maintained focus on validated discovery need, significantly reduced scope for external dev handoff
Trust depends on timing, not just relevance. We designed a "recommended clinics" feature to surface relevant options early in the search flow. 90% of users rejected the suggestions — not because the recommendations were poor, but because they arrived before users had oriented themselves. We deprioritized the feature based on that read, not despite the investment already made in it.
Credibility requires evidence, not just a score. We had prioritized rating scores as the primary credibility signal. Testing told a different story: 80% of users didn't find a rating score convincing on its own and looked for review counts to validate it. The design was updated to surface both.
A fragile booking flow would have broken user trust at the worst moment. There was a pivotal moment where feedback from 8 Tokyo-based clinics revealed that appointment management software varied significantly across facilities, making in-app booking operationally fragile for an MVP. Rather than ship a flow that would fail at the point of real-world contact, we restructured the experience to give users the information needed to book directly with clinics. It was a narrower outcome than originally scoped — and the right one given what we now knew.
Exceeding the Target by Solving the Right Problem
While the product didn't launch due to my contract ending at handoff, prototype testing with target users validated the revised approach:
NPS of 8 (exceeding target of 7)
40% reduction in clinic search time versus users' current Google-based process
Engagement with filter criteria reflected that users were finding the controls relevant to their actual decision-making process.
The NPS target wasn't arbitrary. Stakeholders set 7 as the goal, and our research gave us the context to understand what that number genuinely required. Users rated their existing clinic search experience via Google at a 4 — and given that Japanese respondents culturally tend to score conservatively, a 7 represented a meaningful threshold, roughly equivalent to a 9 in a Western context. Hitting 8 meant the product didn't just clear the bar stakeholders set; it demonstrated that the design had genuinely shifted how users felt about the experience relative to what they had before.
The number that matters most alongside NPS is the search time reduction. It's the most direct measure of whether the design solved the problem users described in research: the cognitive burden of manually navigating an overwhelming number of options to find a clinic that fit their needs. A 40% reduction means the product meaningfully changed that experience.
If I were to continue, the next steps would be to build re-engagement obsessively
The MVP's biggest structural gap is visibility into what happens after the handoff. Users contact clinics, but Medikou has no way to know if that contact turned into a booking, or what the experience was like. That blind spot limits both the product's ability to improve and its ability to bring users back.
I believe that post-visit prompt would have solved both problems at once, a touchpoint asking users to rate their clinic experience after their appointment would suggest 3 things:
It would create a return point
Generates review volume that 80% of users said they needed to trust ratings
Give product a signal on booking completion — which is the metric that actually matters for measuring end-to-end success.
What I would have focused on next:
Return visit rate within 90 days — healthcare isn't daily-use, so a longer window accounts for realistic visit frequency.
Post-visit prompt response rate — would suggest if the re-engagement touchpoint is landing at the right moment or creating friction.
Search-to-booking completion rate — metric that measures whether Medikou is actually getting users into clinics, not just helping them find them.
NPS on return visits — if remains the same or improves as the product adds complexity. If it worsens, we would be creating friction rather than adding value.