People wouldn’t share health data before knowing where the trial was
A Fortune 100 clinical research organization averaged 65% pre-screening abandonment across four previous trial sites. Support mail kept asking one question: “Where is this trial?” Locations stayed hidden until after screening, and the client wanted another site on the same template.
Constraints: six weeks to winter enrollment, WCAG 2.1 AA and HIPAA-safe, and no turning dosing facilities into a support desk. I was the sole UX/UI designer.
Three decisions that flipped the funnel
1. Cities before the screener, facility contacts after
Eight interviews with prospective participants (ages 24–61) said the same thing. One put it plainly: “I see no point in divulging my medical history if this trial is in California and I’m in Ohio.” Stakeholders hid locations to protect understaffed facilities, which was a fair worry with the wrong fix. I put a searchable directory of test-site cities up front and held ZIP-based facility matching until after qualification. Scroll data later showed about 85% of mobile users heading straight to the directory.
2. Make the screener earn the health data
Think-aloud retests surfaced privacy anxiety, unclear progress, and mis-taps on 32px targets. The fix was a three-step flow (city directory → branching screener → consent and facility match), a plain-language privacy banner (“We don’t store your information until you consent”), and conversational progress labels instead of 1/3–3/3 bars.
3. Design for the commute, not the desk
About 67% of traffic was already mobile, because the ads reached people on lunch breaks and commutes. I went mobile-first, with readable body text and 48px minimum touch targets.
Results: abandonment fell from 65% to 37%
The site launched on the six-week clock with a React eligibility screener integrated with Miracle Software, and it holds steady at 10,000+ monthly users. Compared with the organization’s four prior campaign sites (first ~90 days):
| Measure | Before | After | Source |
|---|---|---|---|
| Form abandonment | 65% | 37% | Funnel / screener completion |
| Bounce rate | 62% | 44% | Analytics vs prior baselines |
| Screener time | 4:20 | 2:30 | Moderated sessions (n=8) |
| Verified participant matches | — | +23% | Enrollment reporting, first 90 days |
| Completions on mobile | — | 67% | Analytics |
Test sites reported no increase in premature contact. Two regional sites hit enrollment quotas 11 days early, and the client reused this system on two more studies.
What I’d do again (and differently)
- The bottleneck was structural. Executing the industry template well wouldn’t have fixed it.
- When ops fear exposure, design the minimum viable transparency. Cities, not contacts.
- Next time, A/B the location placement. Qualitative evidence won the argument under a six-week crunch; a controlled test would have made it airtight.