Home

Helping doctors actually start

I improved doctor onboarding for a medical-imaging collaboration tool. The goal was to cut drop-off so more doctors reach the main product, and to stop them from making duplicate workspaces.

Visit Medicai ↗
Role
Product Designer
Paired w/ 1 designer
@Atta Systems
Timeline
Nov 2023
~1 month
Status
Not shipped
Designed intent + measurement plan
Skills
UX research
Interaction & visual design
Cross-platform
Impact

I redesigned the onboarding flow for a medical-imaging collaboration tool. I removed friction so more doctors reach the main product, and guided them away from making duplicate workspaces.

−33%
onboarding clicks
12 → 8
~50%
fewer clicks to
invite members
2 solved
drop-off +
duplicate workspaces
Problem

Doctors stalled before they began

Many quit partway through setup and never reached the main product. Those who got in often made duplicate workspaces instead of joining the existing one, which flooded support with deletion requests.

Drop-out
Duplicate workspaces
x%
reached the core
[ Duplicate-workspace symptom ]
Research

Borrowing the users I couldn't reach

I couldn't talk to doctors directly, so I compared the two data sources to find the real causes. The invite looked like spam, and the first page asked for 9 fields at once.

Hotjar
Hotjar
Where doctors dropped off, in numbers.
Sales interviews
Daily contact with doctors, cross-checked against the data.
Solution

Earn trust, then make every step lighter

01

An invite doctors actually trust

Doctors mistook the invite for spam. The new version starts with names they recognize and clear info about the platform, before it asks for anything.

[ As-is / To-be. Invitation email ]
02

Nine fields, made manageable

The first page asked for 9 fields at once, and this is where most people quit. I split it into two simple steps with quick-fill suggestions.

03

Inviting doctors, without the chaos

A 5 Whys showed the duplicates came from admins who couldn't invite people easily at scale. The fix wasn't one screen. It was a whole access system.

01 One-click invite
02 Bulk invite via CSV
03 Link options
04 Domain-based routing

Designing every path, end to end

Whether a doctor creates, joins, or requests a workspace depends on three things: their invite, their email domain, and the admin's permission setting. I mapped every path and designed each screen, down to the pending and declined states, so nothing was left undefined.

Part 1 · before account setup
Part 2 · after account setup

Hover over a diagram to zoom in. The lens follows your cursor.

Key screens from both onboarding paths. The main screen is enlarged; states and domains are shown as detail crops.

Craft

Consistent on every screen

I designed for desktop and mobile web, one coherent experience.

Next steps

What I'd measure once it ships

The redesign didn't ship. To show it really worked, I'd track these four signals, each tied to one of my starting assumptions.

M1
Email open rate
Does the invite read as trustworthy?
M2
Onboarding completion
Do fewer fields lift finish rate?
M3
Perceived security
Is the spam doubt actually gone?
M4
Duplicate-workspace rate
Does routing stop the duplicates?
Learnings

What I'd carry forward

A proxy is filtered truth
Sales gave me the user's voice when I couldn't reach doctors. But it's only useful when you check it against real data.
Flag your assumptions
When you design on an unproven guess, make it visible and add a way to check it.