CASE STUDY
Role: Product Designer
Tools: Figma, Figma Make, Claude Design, AI Plugins, LLMs
Duration: 5 months
Team: 1 Product Designer, 1 Head of Growth, 4 Developers, 2 Testers, 1 Paediatric Nurse
Valid Assessment
POST-RESULT DROP-OFF
WEEKLY ACTIVE USERS
3.4×
PAID CONVERSIONS
Click to zoom
Introduction
Great technology, broken funnel: parents signed up and went nowhere
Context
A regulated medical device app that lets parents use their phone camera to detect infant pain in seconds. Sold as a subscription, it brought clinically validated science into homes to give parents clarity when their baby was in distress.
The problem
Paid marketing brought people in, but the journey broke down immediately. An 81-day paid campaign drove 399 downloads, but only 20 parents completed a scan and just 7 converted to paying subscribers. High friction, invalid test results, and no guidance after the score caused parents to drop out when they needed the app most.
Core objective
Fix the core onboarding and assessment flow to eliminate drop-offs, provide clear next steps after each test, and transform a one-time score checker into an ongoing care companion.
Research
Discovery & Research Insights
10
In-home parent interviews
2
Independent clinical reviews with our partner paediatric nurse
4
Competitor app analyses
4
Measured business outcomes
Insights
Safety Checks Were Treated as a Formal Disclaimer: Because the pre-assessment check screen looked like legal small print, parents rushed past it or skipped it. This quietly invalidated their scans and caused 31% of users to leave the flow.
A Number Is Not an Answer: A raw clinical score gave parents data without direction. With no practical advice on how to comfort their baby or check for warning signs, 62% of users closed the app right after getting the result.
Nothing Existed Between Crises: Infant pain is episodic. Because tools for logging treatments, taking notes, and setting follow-up reminders lived outside the core flow, parents had no trigger to return, averaging just 0.4 logged treatments a month.
Solution: Rule out flow
Pre-Assessment Rule-Out Flow
Problem
Retention Features Sat Outside the Core Flow
The app requires parents to confirm physical baseline checks before scanning to produce a medically valid score. The original screen used dense, passive text that looked like a legal disclaimer. Stressed parents skipped it, which caused invalid scan results, hurt trust, and led 31% to quit mid-flow.
Click to zoom
Solution A
One check per screen with Yes/No
Guaranteed that every check was answered, but added too many steps and taps for an anxious parent holding a crying child at 2:00 AM.
Discarded
Click to zoom
Solution B
Single checklist screen with a progressive shortcut
Grouped all checks onto one screen using clear checkboxes and short supportive text. For repeat parents who already knew the checks, a "Select All" button appeared after 3 to 4 completed tests.
Shipped
Click to zoom
Soluition
Shipped a clean, single-screen interactive checklist that made safety checks quick and clear. Added a progressive "Select All" option for experienced parents to keep future tests fast while protecting accuracy.
THE RESULT
51% → 94%
increase in valid completed assessments
31% → 12%
reduction in mid-flow assessment abandonment
Solution: What to do next
Actionable Post-Assessment Guidance
Problem
Retention Features Sat Outside the Core Flow
After completing a scan, parents saw a raw pain number on a busy screen with zero guidance on what to do next. Left without practical support while managing a distressed infant, 62% of parents closed the app immediately after seeing the result.
Click to zoom
Solution
Horizontal swipeable guide cards
Collaborated with the partnered paediatric nurse to introduce three bite-sized, interactive quick guides directly below the result: How to comfort your child, Red flag assessment, and Head-to-toe assessment.
Shipped
Click to zoom
THE RESULT
62% → 29%
reduction in drop-offs immediately following results.
57%
of parents opened a clinical guide whenever pain was detected.
1.8 → 4.3
increase in monthly sessions per active user.
Solution: Log treatment
Disconnected Retention Features
Problem: Infant pain is episodic. Tools to log treatments, record notes, and track trends existed in isolated side menus where parents never found them. With just 0.4 treatments logged per user each month, the app was treated as a disposable calculator rather than an ongoing health companion.
Solution: Contextually embedded "Log Treatment", "Add Comment", and "Set Reminder" directly onto the results screen, and introduced a re-assessment reminder to prompt follow-up checks after soothing or administering medication.
Problem
Retention Features Sat Outside the Core Flow
After completing an assessment, parents received a score on a cluttered screen with no direction on what to do next. Left stranded with a distressed infant, 62% of parents exited the app immediately after seeing the result.
Click to zoom
Solution
Contextual entry points via existing screens
Surfaced direct action buttons on the post-assessment summary view that routed to existing native templates, keeping engineering effort low while surfacing the features in the moment.
Shipped
Click to zoom
THE RESULT
62%
of parents returning for a follow-up check.
11% → 35%
growth in weekly returning active users.
Design system
A design system to improve speed and quality
Click to zoom
Key takeaways
What stayed with me
Fix the Leak Before Pouring More Spend
Scaling ad spend into an unoptimized funnel simply magnifies lost spend. Solving core user activation must always come first.
Protect Safety Without Multiplying Friction
Multi-screen safety checks guarantee clinical compliance but trigger user abandonment under stress. A single-screen checklist preserved medical validity while keeping interaction costs low.
Design for Second and Tenth Visits
Progressive disclosure (such as exposing "Select All" only to repeat users) respects returning user proficiency without removing safety rails for first-time parents.