Back to work
Before

Previous checkout screen

After

Redesigned checkout screen

Shipped

Flash · Checkout · 2025

Making room for
more ways to pay.

Flash was adding new payment methods, so the checkout screen had to be redesigned to fit them. I compared two designs in 126 unmoderated usability tests, and the parts they share went live.

Role
Product designer, from the PM's ticket to the analysis of the results
Company
Flash, consumer fintech
Year
2025
Tools
Figma, Maze
  • 126unmoderated usability tests on two checkout designs
  • 90%+task success on every task, in both designs
  • Heldcheckout conversion in the quarter after launch

Start here · before you read on

Try it yourself

Try the checkout yourself

Try a five-second screen test, then a checkout task, before you read the results. They're based on the Maze study I ran before launch. This is a prototype, so no money changes hands.

Take the testAbout 2 minutes

You vs the testers

Spotted the total in 5 seconds1 in 5?
Finished the checkout task9 in 10?
Time to pay44 to 60 s?

Testers’ numbers come from the 126 Maze tests. Your column fills in at the end.

01 · The brief

More payment methods, two constraints

Bills, recharges and payments all go through one checkout screen. Flash wanted more ways to pay on it: InstaPay, digital wallets and more cards. The live checkout listed payment methods in a dropdown, so every added method made the open list longer, and people would have to scroll to reach the one they wanted. The new screen had to hold them all, and two numbers could not get worse.

Before · the live checkout
The live checkout before the redesign: payment methods in a dropdown and every line of the summary expanded
All details openMethods in a dropdown

Constraint 1

Success rate must not drop

Checkout conversion is the revenue line. A design that looks cleaner but converts worse is a failure.

Constraint 2

Time to pay must not rise

Every added second at checkout is drop-off risk. New options couldn't slow down the people who pay today.

02 · Users and flows

Two paths through one checkout

People with a saved payment method and people with nothing saved land on the same checkout screen. Either of them can also apply an offer, add a donation or pay part of the total with Flash rewards before paying, and each one updates the total.

Existing userMethod saved

Keeps the selected method

  1. Checkout
  2. Tap Pay
  3. Payment success

Picks another saved method

  1. Checkout
  2. Methods bottom sheet
  3. Saved
  4. Tap Pay
  5. Payment success

Adds a new method

  1. Checkout
  2. Methods bottom sheet
  3. New
  4. Tap “Add card or wallet and pay”
  5. Add it, then pay
  6. Payment success

New userNothing saved

  1. Checkout
  2. Pick a method
  3. Tap “Add card or wallet and pay”
  4. Add it, then pay
  5. Payment success

Add-onsEither user, on the checkout, before paying

  1. Apply an offer
  2. Discount comes off, total updates
  1. Switch on the donation
  2. Donation added, total updates
  1. Switch on Pay with Flash
  2. Rewards pay part, the card pays the rest

Then

  1. Tap Pay, as above

03 · Directions

Choosing where to show payment methods

I first tried a separate screen for choosing a payment method. That meant leaving checkout for a single choice. A bottom sheet kept the checkout visible, so I took that direction forward and explored how to group saved methods and new options.

Payment method selection

Swipe for more
Own screen
Payment methods on a separate full screen, grouped into cards, wallets and other methods
Single list
A bottom sheet over the checkout with every method in one list
Grouped by type, A
A bottom sheet grouping cards, digital wallets and other methods into boxed sections
Grouped by type, B
A bottom sheet grouping cards, wallets and other methods as separate rows
Not chosenChosen

Payment summary

Swipe for more
Breakdown open under the headerBreakdown open
Checkout with the cost breakdown open under the header
Offers first, summary collapsedSummary collapsed
Checkout with offers first and the summary collapsed above Pay
Offers first, summary openSummary open
Checkout with offers first and the full summary open above Pay

More drafts

7 screens
Draft 1
Draft: dark summary at the top with the amounts paid by rewards and by card
Draft 2
Draft: summary card with the discount and Flash rewards listed
Draft 3
Draft: collapsed summary with offers and the donation below
Draft 4
Draft: method sheet with Flash rewards switched on at the top
Draft 5
Draft: method screen grouping cards and wallets together
Draft 6
Draft: summary open above offers
Draft 7
Draft: method sheet with InstaPay marked new

04 · The two designs

A bottom sheet to hold them all

Every payment method moves into a bottom sheet, so new options can join without crowding the screen. The summary collapses to a single total and expands on demand. The two designs share all of that. They differ only in what sits next to the Pay button, so both went through the test.

Bottom of design 1: the selected card, then a button that says Pay 200.00 EGP

Design 1

Payment method next to Pay

The selected method sits right above the button, and the button repeats the amount: Pay 200.00 EGP.

Bottom of design 2: the order summary with the total, then a button that says Pay now

Design 2

Summary next to Pay

The order summary sits right above the button. With the total already there, the amount on the button looked redundant, so I removed it and the button says Pay now.

TTarek

Spoiler: this was a bad idea. The 5-second test shows why.

05 · Prototypes

Prototypes that behave like the real app

I first tried a regular click-through prototype with people in the office. Dead taps, a donation toggle with no feedback and a total that never updated kept stalling them with questions.

~60%

longer to finish checkout on the regular click-through prototype

Time to pay was one of the two numbers that could not get worse, so fake friction would have skewed the timing results.

So I built both test prototypes with Figma variables and conditional logic. They respond like the app: methods change, offers apply and the total updates live. Doing that the regular way, with one screen for every state, would have taken more than 100 screens for a single test flow. The conditional version covers it with 29.

Disclaimer: the regular prototype is only here for show. I never actually built 123 screens.Tarek

Regular prototype · 123 screensCanvas of the regular prototype: a wall of 123 near-identical checkout screens, one for each state, buried under prototype arrowsConditional prototype · 29 screensCanvas of the conditional prototype: 29 screens covering the same flow, with a few short arrows per row and one line down the side94 fewer screens

Disclaimer: the regular prototype is only here for show. I never actually built 123 screens.Tarek

What the prototype keeps track of

4 variables
  • payment methodInstaPaychanges when you pick another one
  • offer appliedtruethe discount comes off the total
  • donationoffthe toggle answers when you tap it
  • total180.00 EGPrecalculated after every change

06 · The test

126 unmoderated usability tests, run in Maze

Tasks mirrored real checkout, and every task ran on both designs. Existing users changed a payment method and applied an offer. New users added a payment method and applied an offer. Every participant confirmed upfront that it was a test and no real money would move.

  • 2 designstested against each other and the live checkout
  • 2 audiencesexisting power users invited by SMS in batches of 200, plus new users
  • 3 signalstask success, time to complete, and which details people remember
  • + 5-second testcan people spot the total amount at a glance?

07 · Results

What the head-to-head found

Both designs stayed at or above 90% task success. Design 2 was a few points higher on every task, and Design 1 was faster on every task.

  • Design 1: the selected card above a button that says Pay 200.00 EGP
    Design 1Payment method next to PayThe button says Pay 200.00 EGP
  • Design 2: the order summary above a button that says Pay now
    Design 2Summary next to PayThe button says Pay now

Task success rate

Change payment method · existing users

Design 1: 90%
Design 2: 94%

Apply an offer · existing users

Design 1: 92%
Design 2: 97%

Apply an offer · new users

Design 1: 93%
Design 2: 96%

Time to complete

Change payment method · existing users

Design 1: 35 s
Design 2: 38 s

Apply an offer · existing users

Design 1: 29 s
Design 2: 34 s

Apply an offer · new users

Design 1: 44 s
Design 2: 60 s

Experience rating

Average answer, out of 5, to “How would you rate this checkout experience?”

Design 14.7 out of 5
Design 24.3 out of 5

33 to 45 participants per task for existing users, 14 to 23 for new users.

What people noticed, and what they chose

5-second test

People who remembered the total after five seconds

Design 112 of 4030%
Design 23 of 368%

One dot per tester. Filled: remembered the total.

Most remembered the Vodafone logo. Some mistook the 100 EGP rewards message for the amount due.

Redesign or the live checkout

Which one people chose when shown both

33% chose Design 167% chose the live checkout

61% chose Design 239% chose the live checkout

TTarek

21 of 37 new users picked InstaPay as their payment method.

08 · What went live

What went live, and why

Every test, on both designs, ran through the same two parts: payment methods in a bottom sheet and a summary collapsed to one total. Both designs cleared the success and time constraints.

On what sits next to Pay, the result was split. Design 2 had higher task success and won preference, and Design 1 was faster on every task. So we shipped the two shared parts and parked the Pay button decision. In the quarter after launch, checkout conversion held.

Shipped
Shipped checkout: the payment summary under one total amount
Shipped checkout: saved and other payment methods in a bottom sheet
Summary in one totalOne totalMethods in a bottom sheetBottom sheet

09 · Learnings

What the study taught me

Three things I would do differently, or do next.

01

We should have tested in Arabic

We built and tested in English only. Nobody caught it, me included, even though most Flash users speak Arabic. Arabic first is now my default.

02

The total belonged in the Pay button

Design 1's button said Pay 200.00 EGP, and 12 of 40 people recalled the total. Design 2's said Pay now, because the summary sat right above it and the amount felt redundant. Recall fell to 3 of 36. Looking back, repeating the amount was helping people.

03

What I'd run next

The split result says neither design was the whole answer. I'd design a third option that aims for Design 2's success rate at Design 1's speed, starting with the amount back in the Pay button, and test it in Arabic.

Next case study
Contact

Get in touch

TTarek

Have a role or project in mind? Send me an email.

Email me