Onsurity · 2023
Marketing wanted the banner. Support data wanted the claim button. I had four weeks and 5,000 daily users.
Rebuilding the homepage of a live healthcare app — replacing it piece by piece so nobody woke up to a product they no longer recognised.
Role
Senior Design Consultant
Timeline
4 weeks to handoff
Team
2 designers, PM, both platforms
Platform
Android & iOS
The product
Onsurity sells monthly healthcare memberships to small businesses. Employers buy it; employees live in the app. About 5,000 people open it every day — often standing at a hospital desk.
What had happened to it
The homepage had grown feature by feature into banners, swipe menus and nudges. Users came to do three things — raise a claim, find a cashless hospital, show their health card — and struggled to do any of them quickly.
Every new product launch made it worse, because there was nowhere for new products to go.
Three sources, one answer
Rather than trust assumptions, I triangulated. Where all three pointed at the same thing, that became the problem.
UX audit
Key tools tucked inside horizontal swipe menus nobody discovers. Rewards had said “Coming Soon” for three years.
Support tickets
“Where’s the claims section?” was the single most common navigation query in the queue.
App store reviews
Most negative reviews traced back to claims — delayed, unclear, or blocked by missing nominee details.
Top user intent was buried deepest in the navigation. All three sources said the same thing.
Sixteen opportunities, four weeks
Each mapped on user impact — frequency × severity, from support data — against build effort, sized with engineering. Phase 1 took the high-impact items needing no backend rework.
Phase 1
Shipped in Phase 1
Critical actions — claims, cashless hospitals, policy cards — upfront
Insurance and Wellness decoupled into separate sections
Active plans separated from upsell products
Profile as the single home for member details and claim history
Upcoming orders and appointments on the homepage
Support and FAQ restructured to deflect simple tickets
Positions are relative judgement, not a scored model.
Fig. 02 · The ten dots outside Phase 1 are the point — they show this was a decision, not a wishlist.
The argument
What owns the first fold?
Mid-fidelity wireframes let PM, CS, engineering and marketing argue about structure before pixels made it personal. Three options went on the wall.
Wireframe A
Promo banner first
Marketing’s preference
Maximises offer impressions. Also exactly why the homepage felt like an ad.
Wireframe B
Wellness feed first
The founder’s instinct
“Users come for claims but stay for wellness.” Good for engagement, wrong for urgency.
Wireframe C — chosen
Critical actions first
What support data said
Raise a claim, cashless hospitals, policy cards — what users actually open the app to do.
Fig. 03 · [Export A, B and C from FigJam at equal scale. This three-up is the most important image on the page.]
When someone opens this app at a hospital desk, seconds matter. Trust earned there is what makes wellness engagement possible later.
The compromise: announcements got a dedicated, bounded slot below the actions, and wellness got its own permanent section. The accepted trade-off — promo impressions likely dip. At this stage, trust beats conversion.
What shipped
Four structural changes, each answering something the research had surfaced.
Claims, cashless hospitals and policy cards sit above the fold — the three things support data said people open the app to do.
Upcoming orders and appointments surfaced
New Profile Section
Plans never mixed with things Onsurity wants to sell. Upsells live in clearly-labelled, swipeable cards — visible, but not pretending to be something you own.
The visual call
No brand guidelines existed — no rules for illustration, iconography or accent colour. A new illustration system was being built in parallel to fill that gap. It did not survive.
That left a choice: hold the redesign until there was a visual language, or ship the structural fix in the language that already existed. With 5,000 people opening the app daily, two simultaneous changes is how you lose the people who were coping with the old layout.
So the visual work went where it changed comprehension rather than where it would have been noticed — illustrated action cards instead of abstract icons, because iconography literacy varies across a blue-collar user base. Everything else held its existing form.
A full visual refresh belongs to a future revamp — one the instrumentation we added can actually inform.
Before and after
Before
After
Shipped in phased milestones — a Ship of Theseus rollout, replacing the app piece by piece so daily users never hit a wall of change.
What we can and cannot claim yet
The redesign shipped in phases and it is too early for conclusive numbers. Rather than dress anecdotes up as data, here is what the app was instrumented to track.
CS ticket volume for navigation queries
Whether “where is claims?” dies out
First-fold engagement
Tap-through on critical actions vs the old swipe menus
Order visibility
Whether surfacing orders reduces “where is my booking?”
App-store rating trend
Claims-related reviews specifically
The structural win is already banked: any new insurance or wellness product can now launch on the homepage without a redesign. That was the problem that triggered the project.
What I’d do differently
This redesign was supposed to ship with a new illustration system — the Neoclassicism × Madhubani work built in parallel. It was approved by both founders, then cancelled for business reasons, which left the homepage with no new visual language to draw on.
So the screens shipped close to the existing identity. I framed that as spending the innovation budget on structure rather than surface, and structurally it was the right call — navigation was the problem, not the palette. But it was also the only option left by then, and a case study that does not say so is telling half the story.
The lesson is about dependency rather than taste. I let one workstream’s visual layer rest on a parallel project whose fate I did not control. Scoping the illustration system as a gate for the homepage, instead of a track running alongside it, would have surfaced its cancellation as a decision to make rather than a fact to absorb.
The first-fold conflict was settled by evidence rather than opinion. Bringing ticket volumes to the meeting ended a debate that had been going in circles on taste.
Next project
Get in touch