Amanah Health

The Real Problem

Clinic assistants at Amanah's partner clinics were closing every patient visit in nine separate screens. Vitals in one, notes in another, billing codes in a third — and the same IC number typed into four of them. A routine visit took six minutes to document; a busy evening meant thirty visits and no dinner.

Worse, the system gave no sense of progress. Assistants would finish what looked like the last screen and discover two more hiding behind a dropdown. Nothing auto-saved. One misclick during a power flicker and fifteen minutes of careful typing was gone.

“I honestly cannot tell if I am done or not.”actual notes from the clinic floor

“I retype the same IC four times. Four!”

“After 8pm my only goal is: don't lose the form.”

The real issue wasn't one bad screen — it was that nobody had ever looked at the whole journey. Each screen had been added by a different vendor over five years and they had never been stitched into one flow.

Finding the Fix

I mapped all nine screens end-to-end — what data each one collected, where it duplicated, where people stalled. Support tickets and a survey of 40 assistants pointed at three things: no progress, no safety net, and screens organised by database table instead of by how a visit actually unfolds.

So the redesign focused on exactly three fixes: a visit progress rail that is always visible; auto-save with a persistent 'Saved just now' indicator; and collapsing nine screens into five steps that follow the real order of a consultation.

What Actually Happened

Prototypes went in front of clinic staff twice a week. My first grouping put billing next to prescriptions — logical to me, wrong to them: billing happens after the patient leaves. We reordered. My first auto-save toast vanished too fast to reassure anyone; a persistent timestamp fixed what a nicer animation couldn't.

We shipped to six pilot clinics first, tuned the copy on two confusing steps, then rolled out to all of them.

What Changed

Documentation time per visit dropped from six minutes to just over two. Lost-form tickets fell by 81% in the first month. In the follow-up survey, 9 out of 10 assistants rated the new flow easier — and the tenth asked for dark mode, which I choose to count as a win.

six minutes down to two 81% fewer lost forms 9 out of 10 found it easier

“I finish my last patient and I actually get to go home.”

What I Had to Work With

No direct access to patients (PDPA), a third-party component library we couldn't modify, and a hard 10-week window before the next clinic onboarding batch. The constraints picked the strategy: focused fixes over a grand rebuild.

What I'd Do Differently

Push for clinic shadowing earlier — one evening watching real check-ins taught me more than two weeks of tickets. And document the old flow properly before changing it; my before/after story would be twice as convincing with better evidence.

What I Learned

Progress you can see beats speed you can claim. Mental models beat database models. And the most valuable design work on this project — the saved-state indicator — is the part nobody screenshots.

This is sample content for portfolio purposes — the projects on this page are fictional placeholders.