Solving for user trust

Redesigning the checkout experience for MakeO

Role

Design systems & checkout pod lead

Timeline

March 2023 – January 2024

Organization

MakeO (formerly Toothsi)

Team

3 designers

Markets

India and UAE

34%lower KLM score

42%drop-off reduction

-3 daysfrom PRD to design

MakeO offers two services: teeth alignment (toothsi) and laser hair removal and facial (skinnsi). Both were losing users in their main journey, which meant more marketing follow-ups and fewer conversions.

My role

I led the design systems team for this project.

  1. 01

    Built the system

    We created the design system elements.

  2. 02

    Delivered it with development

    We shipped it with developers and explained how tokens carry the system across the organization.

  3. 03

    Led accessibility

    I led accessibility for the project.

  4. 04

    Grew design-oriented QA

    I helped the QA team develop design-oriented testing skills.

8 in 10 users left before booking a scan or paying.

About half the users left each checkout journey before reaching payment. Across the whole journey, around 85% dropped off, so most people neither booked a scan nor paid.

Where it happened

Four steps lost the most users

Most of it happened at login, pincode, the map and slot selection.

  1. 01

    Login

    Onboarding
  2. 02

    Pincode

    Onboarding
  3. 03

    Map

    Booking
  4. 04

    Slot selection

    Booking

How we found out

Three sources, cross-checked in an audit

Microsoft Clarity

Session recordings showed where flows broke.

QA bugs

Bugs logged by QA flagged breakages and inconsistencies.

Marketing calls

User feedback from calls showed what confused people.

In the audit, we checked the marketing feedback against Clarity recordings to confirm where the flows broke.

Why it happened

No single cause, many small breakages

Screens had been built one at a time, with no shared rules for labels, feedback or layout, and the small breakages piled up on the busiest steps. So we decided to fix the system rather than patch screens.

Map, a major drop-off point

Heatmap and annotations of issues in the current location flow: smaller screen space, missing exact location and unclear usage of icons
  1. 01

    Smaller screen space

    The map is the key element, but familiar icons clutter it.

  2. 02

    Missing exact location

    The address line lacks a main keyword, so users can’t confirm the right place.

  3. 03

    Unclear usage of icons

    “Edit” isn’t self-descriptive and “Locate me” has no clear meaning.

Coupon code, also found in the audit

Annotated screens of issues in the current coupon flow: checkout disruption, missing delight and unclear label
  1. 04

    Checkout disruption

    Toothsi and skinnsi had no coupon codes, so a separate coupon page only interrupted checkout.

  2. 05

    Missing delight

    Validation gave no feedback, missing a moment of delight.

  3. 06

    Unclear label

    Users may not link “Promotions” with coupon codes.

We fixed the system behind the screens.

Instead of one checkout feature, we overhauled the design system and modernized the UX to lift conversion at scale. Three moves did it.

Klinikal design system file in Figma
01

A design system

Shared tokens and reusable components for consistent, faster design and production.

02

A revamped checkout flow

We rebuilt the screens both services share (pincode, map, slots and promotions) once, so each fix reached both.

03

Micro-interactions and loaders

Loaders, toasts and button feedback, so the flow no longer feels abrupt.

Initial findings

Rebuilding from the design system up fixed lots of small frictions at once, in validation, hierarchy and perceived speed, without breaking users’ mental model.

Trade-offs

Key decisions

Six calls that shaped the work, and why we made them.

A system over screen-by-screen fixes
With so many design files, keeping things consistent was hard. A system means everything new matches what is already there.
Both journeys at once
The team agreed early on the main checkpoints of toothsi and skinnsi, which sped things up.
Tokens over ad-hoc handover
Developers couldn’t always build components as designed, which added QA time and mismatches.
Small map updates over one rebuild
Toothsi and skinnsi maps ran on different code for the same component. We split the fix across journeys and shipped it step by step.
Bottom drawers over extra screens
Where we could, we moved steps into a bottom drawer so users go through fewer screens.
QA training over saved design time
It took design time, but QA now tests against our documentation, which saves design QA and development time later.

How do we solve the drop-out issue with speed?

Both journeys needed one shared approach to fix the inaccessible flows in the app.

A

Finding journey common pain points

Toothsi and skinnsi share components and screens, so fixing the common ones helps both flows fastest.

B

Introducing a design system

One source of truth for design, development and product, built on reusable components.

C

Building accessible user journeys

The design system and testing helped us build journeys that are accessible and easy to use.

Flows toothsi and skinnsi have in common

Toothsi and skinnsi share onboarding, then split into clinical treatment and e-commerce checkout. Orange marks the steps we worked on.

User journey flow: shared onboarding (log in, enter pincode, profile setup, app home), then clinical treatment (discover and book, consult and plan, pay and receive, monitor) and e-commerce checkout (browse and choose, build cart, checkout and pay, track order)

Shared onboarding

Log in and enter a pincode

Every user starts here, before picking a journey. We worked on log in and the pincode serviceability check.

Journey A · Clinical treatment

Discover and book

Users choose a category, book a consultation, pick a home visit or nearby centre, and schedule a time.

Journey B · E-commerce checkout

Browse, build a cart, pay

From the Shop tab to checkout: select products, apply promo codes, choose an address and pay by UPI, card, wallet or net banking.

Design components

We listed the essential components first to keep the UI consistent across products. The full story is in the Klinikal design system study.

Design system component library: toasts, accordions, inputs, radio buttons, tabs and calendar

Design decisions

Pastel palette, accessibility-checked

Contrast check passing WCAG AA and AAA

Figma + Strapi, consistent icons

Figma Strapi

For rapid icon shipment

Mobile-first, phase 1

Mobile dental aligners offer screen

Token-based components

Component tokens defined as CSS variables

Company-wide, not designer-only

Design system documentation for mobile toasts

Instance swap over one-size-fits-all

Five flows, rebuilt on the system

What changed in pin code, location, coupon, slots and motion, and how each one looks now.

Pin code flow

The location selector now adapts to users in India and the UAE, using their IP address to skip steps.

Before

Heatmap of the previous pin code screen

User problems

  1. Users had to select a country when changing a pin code, adding steps.
  2. Inconsistent bottom-drawer heights shifted the button’s hit area.

Updating the pincode flow to become user-relevant…

After
India pin-code-based flowA pin code field helps users find their location fast.
After
UAE city-based flowThe UAE defaults to a city-based detector to show relevant clinics.

Location selection

Users enter their home address for the partner’s visit. The old screen reloaded fully on every map refresh. The new one adds a skeleton loader, search, a better “locate me” and cleaner controls, fixing friction points 1 to 3.

Before Previous map location screen reloading as the map is refreshed
Full-screen reloadThe whole screen reloads on every map refresh
After Updated map location screen with a loader and search
Loader and searchA loader appears when the location changes, and search adds structure

Coupon code

Coupons moved into a bottom sheet over the checkout summary, which now shows whether any offers are available, fixing friction points 4 to 6.

Before Previous coupon flow: a separate Apply Promotions screen
A separate promotions screenUnclear labels, little feedback
After
Without coupons“No offers available” keeps focus on payment
After
Coupons availableShows how many offers are available; codes are applied in a bottom sheet

Macro interactions

Slot selection left users guessing. Date tabs now show availability, a button jumps to the next open date, and the toast no longer covers the main button. The reminder was reworded so its button doesn’t imply the booking still needs confirming.

Slot selection

Before Previous slot selection: full dates, a toast covering the continue button and an empty screen, with numbered pain points
Dead ends and clutterFull dates led nowhere, and the toast covered the continue button.
After New slot selection with date tabs showing availability and a list of times
Availability upfrontDate tabs show which days have slots, and times are listed right away.
After Fully booked date offering a button to go to the next open date, 18 May
A way forwardA fully booked date offers a button that jumps to the next open date.

Friendly reminder

Before Previous friendly reminder sheet with a confirm booking button
“Confirm booking” after bookingThe old reminder’s button appeared after booking and left users unsure whether they had booked.
After New friendly reminder sheet with an Ok, got it button
Clear deadline, no confirmationClearer wording and the full deadline, without asking users to confirm again.

Add to cart button

BeforeAfter
From “Add to Cart” to “Checkout”One clip shows the old, abrupt jump and then the smooth animation that replaced it.

Micro interactions

Meaningful interactions and illustrations across the process

Skeleton loaders

Skeleton loaders across MakeO screens, shown while pages load
Pages keep their shape while loadingSkeleton loaders show each page’s layout as content arrives, instead of several loaders at once.

Illustration language

Address screen with no saved addresses: an illustration of a clinic between trees, and an Add new address button
No saved addressAn illustration and one clear action: add a new address.
Checkout summary with an empty cart illustration and suggested services
Empty cartSuggested services turn a dead end into a next step.
Notifications screen with a smiling bell illustration: there are no new notifications yet
Nothing newA friendly empty state, with a way to keep exploring.
Notifications screen with a sleepy bell illustration: notifications are off, with an Enable notifications button
Notifications offExplains what’s being missed and offers to turn them on.

14+ components entered the design system for product and tech teams, with 100+ enhancements across checkout and onboarding.

Reduced KLM score

34%

An AI score predictor showed flow time falling from 30.43 to 20.82 seconds.

Drop-off reduction

42%

Microsoft Clarity tracked the drop-off reduction in the revamped flow.

PRD to design

-3 days

The design system cut cognitive load for design and production, speeding the path from PRD to design and build.

What the development team said

“Standardizing our icon pipeline with Strapi and the bottom drawer made developers self-sufficient. Icons became consistent, load times improved, and designers could adapt it easily.”

Ami Gami Director of Technology

“The Design Revamp made us realize how fast our system actually was. Design became the means to a better customer experience through a faster checkout process and user perception.”

Japan Doshi CPTO

Key learnings

A design culture

Design is also a culture of attention to detail. Refinement raised quality, cut QA time and streamlined work between development, design and product.

Systemic implementations

Structuring large projects around a design system, as with journey refinement, aligns teams on process and unifies implementation.