Rebuilding the main screen
for core business.
Kindoc had pivoted into an O2O health checkup booking platform, but the main screen still mirrored its earlier life as a hospital info app. I led the renewal — restructuring the visual hierarchy, repositioning the primary CTA, and introducing an entry point for the new Kindoc AI feature.
Health checkups became the core business. The main screen didn't know that.
By 2024, checkup bookings were Kindoc's primary revenue driver — but the main screen, untouched since launch, still treated them as one of many features.
Users couldn't find the booking flow.
Four years of short-term requests had left the screen drifting — unclear hierarchy, oversized components, primary and secondary actions competing for the same space. The checkup button, the single most important action, was visually indistinguishable from secondary entries.
Existing main screen
Four years of accumulated requests, no holistic redesign.
CTR drop
46% (Dec) → 32% (Jan). Beyond seasonal variance.
Five objectives. Four directions. One that earned the CTA.
Moving the CTA alone wouldn't fix it. The screen's hierarchy needed to reflect what the business had become. I set five objectives and explored four directions against them.
Lift CTR to the booking flow
Prior rates: 46% (Dec) → 32% (Jan). The single most important number to move.
Introduce Kindoc AI
Place the new generative AI assistant in a visible position without crowding the primary flow.
Clarify primary vs. secondary actions
Re-establish hierarchy through size, color, and placement.
De-emphasize underdeveloped features
The search bar's backend wasn't ready; it shouldn't read as a primary action.
Simplify the footer
Reduce navigation noise at the bottom of the screen.
Reference & Analysis.
I gathered mobile main screens across health, finance, mobility, and lifestyle services — looking for how each balanced a primary action against secondary entries.
Layout Patterns & CTA Placement
Recent box-style main screens often adopt masonry-style layouts and use color or motion to emphasize primary actions. CTA placement varied depending on layout context and the presence of top-area elements such as banners or ads.
- Primary and secondary actions were clearly differentiated through size and color.
- Visual flow was strongly influenced by elements occupying the top area.
- CTA placement was not fixed and required careful consideration of eye movement and layout consistency.
These observations shaped the four directions I explored next.
Direct list access. Not feasible under dev constraints.
Prominent CTA, but mid-screen ad competed with it.
Equal-weight squares flattened hierarchy.
Full-width banner + right-aligned CTA. Zigzag scan.
Three shifts that brought users back to the booking flow.
Eight detailed changes anchored the renewal. These three carried the conversion gain.
A primary CTA that earned its prominence.
Banner at the top-left, CTA at the bottom-right — aligning the Z-pattern scan flow with the thumb zone. The ad area shrank, same text-to-image ratio, less screen real estate.
A new entry for Kindoc AI, without crowding the flow.
The AI assistant got the secondary band — noticeable, scrollable into view immediately, but never competing with the booking CTA.
After-care features replacing dead weight.
Demoted the empty search bar. Replaced the drifted chat entry with a direct call button. Surfaced after-care features that returning users actually used — finding hospitals, checking expenses, self-assessments.
Documentation beyond Figma specs.
In addition to the Figma UI specs, I wrote a front-end guide (component states, edge cases, motion notes) and a brand designer guide (banner ratios, safe zones, tone). The renewal shipped without QA loops on UI fidelity.
For front-end developers
Component specs, states, motion notes, edge cases.
For marketing designers
Banner ratios, safe zones, tone alignment with the renewed UI.
+39% in one month, no campaign attached.
Launched May 3, 2024. No push notifications, no in-app promotions, no marketing spend.
Users didn't need new features. They needed the existing ones to lead somewhere.
The instinct in stuck-conversion situations is to add. The harder thing is to subtract — to ask which existing element is supposed to be doing the work, and why it isn't.