Case study · Payments · White label

PROMETEUX
PROJECT RECEIPT
Demo store: Reino da imaginação
ROLE
COMPANY
TIMELINE
TEAM
PLATFORM
STATUS
ITEM
QTY
Flows
8
Screens
38
Ways to pay
3
Receipt channels
3
Paid with a phone. No card machine required.
THANK YOU
01 · The counter
In Brazil, most small stores take cards on a maquininha, a dedicated terminal. The Company’s product team, which I was part of, studied that market. The losses kept coming from the same four places.
A fixed monthly fee, whatever the store sells. One major acquirer’s smart terminal rents for R$ 99.90 a month.
Source: Cielo blog
small business owners had been victims of fraud, including swapped terminals that divert sales.
Source: CNDL/SPC Brasil, 2018
of small businesses would stop using a terminal because of its upkeep costs.
Source: Sebrae, 2021
asked for lower rates. 76% had picked their terminal by the cheapest fee.
Source: Sebrae, 2021
Market data from public sources, with their year. These figures describe the market, not this product.
The PCI MPoC standard set the security rules for taking card payments on ordinary phones with certified software. Tap to Pay brought it to iPhone in Brazil a year later. The hardware was no longer necessary.
Problem statement
Who feels it
The store pays rent, upkeep and fees, and carries the fraud risk.
The seller charges at the counter with a device that can fail or be swapped.
The white-label client wants to offer payments without building a terminal app.
Constraints
Certified security on an ordinary phone (PCI MPoC).
One core for every client, with brand, catalog and campaigns changing on top.
A buyer who has never seen the app must pay without help.
Success looks like
A seller takes card, Pix or link payments on their own phone.
A client launches under its brand without rebuilding the core.
A buyer finds the amount, places the card and gets a receipt alone.

02 · What we learned
By 2025 the big acquirers and fintechs already had their own Tap to Pay apps. That told us the technology worked and that sellers would use it. It also showed the gap. Each of those apps sold its own brand.
Conversations with people who charge at a counter every day: what slows a sale down and what makes them distrust a new device.
Studies released by other players in the payments market, used to size the costs of the terminal and to avoid repeating known work.
The products that were starting to add Tap to Pay: how they guide the buyer, confirm a read and handle receipts.
Work only with the acquirer that ships them. The store promotes someone else’s brand.
InfinitePay
Stone · Ton
Nubank
SumUp
Mercado Pago
Cielo
Any bank, acquirer, fintech or retailer ships its own terminal app, under its own brand. Few players were offering this.
Zoop
Our product
The Company builds payment security infrastructure for banks and fintechs. A white-label SoftPOS was added on top of that security, so other companies could put their own brand on it.
03 · Strategy
A white label sells the same product many times. So the strategy wasn’t about one perfect store. It was about deciding what every client can make their own, and what nobody can touch.
The brand layer belongs to each client and changes with every deployment. The core is identical everywhere, so it’s secured and tested once, and every improvement reaches all clients at the same time.
Changes per client
Visual identity
Catalog
Ad slot on the tap screen
Coupons
Games and prizes
The same for every client
Device check
Login + email code
Payment engine: Tap, Pix, Link
Digital receipts
“Reino da imaginação” is the proof: a fictional toy store with a loud personality, built to show prospective clients how far the brand layer goes without touching the core.
Each user meets the product in a different place, with a different amount of attention. That set the scope of this case.
Outside this case
Client and store managers
Brand and catalog
Campaigns, games and prizes
Seller accounts
This case
The seller, every sale
Find products and build the cart
Charge by card, Pix or link
Send the receipt
No app needed
The buyer, a few seconds
The tap screen, on the seller’s phone
An optional game
The receipt, by text, email or QR link
Every screen in the sale that follows answers to one of these.
A phone that takes cards has to prove it’s safe before anyone logs in, and the seller has to see it happen.
Item 01 · Open the till
Sellers start a sale in different ways, but each path joins the same screens. Fewer screens to secure, test and brand.
Items 02 and 08
Seller screens favor speed and repetition. The one screen the buyer sees favors clarity for a stranger.
Item 06 · The tap screen
Where the product earns attention, the client fills it: the ad slot, coupons and games carry the store’s brand, not the platform’s.
Items 04 and 06
Every strategic choice gave something up. These are the ones that shaped the product most.
Decision
What we gave up
What we gained
A fixed core
Custom flows per client
Security built and tested once; fixes reach every client
Seller-only mobile app
Managing the store from the phone
A faster counter; admin lives on the web
Digital receipts only
Paper for buyers who expect it
No printer, no hardware; QR needs no personal data
Opt-in games
Reach of every campaign
Buyer trust and room for local rules
04 · One sale, end to end
Item 01
09:36:10
A phone that takes cards has to be trusted before it sees one. Before anyone logs in, the app checks the device itself: suspicious apps, battery saver, audio and NFC. Then the seller logs in and confirms a 5-digit code sent by email.
Device
→
Login
→
Email code
→
Ready to sell
Why it matters: the client brand gets a terminal it doesn’t have to ship, and the store stops renting hardware that can be swapped.

Device check
Four checks, each one visible to the seller.

Login
No sign-up: the store creates each seller.

Email code
A second factor, with a resend timer.
Item 02
09:38:04
A customer at the counter might point at a shelf, name a toy or hand it over. So the seller can browse the catalog, search or scan the barcode. Every path ends on the same product page, which gets built and maintained once.
Browse
Search
Scan
→
One product page

Search
Recent and most searched items cut typing at the counter.

Results
Price and stock on every card.

Scan
The camera reads the tag already on the toy.

Product page
Stock is shown before the seller promises anything.

Added
The button confirms in place, with no extra screen.

Next step
The same button now leads to the cart.
Item 03
09:39:12
The home screen puts search, scan and categories within one thumb’s reach. The checkout lists every item with its quantity, then the subtotal, discount and total, in the order the buyer will read them. A discount can come from a coupon, a manual adjustment or a game.

Home
The client’s brand and catalog, the same structure for every client.

Checkout
The game icon in the header offers the buyer a chance at a discount.
Item 04
09:39:40
Before charging, the seller can offer the buyer a game from the checkout header. If the buyer wins, the prize goes straight into the cart. It gives the store a reason to be remembered, at the moment the buyer is already paying.
Optional
Only if the buyer accepts
Adults only
Offered to buyers 18+
Never money
Discounts or extra items
Set per client
Games, prizes, odds and limits in the web panel

1
Pick a game
The client decides which ones appear here.

2
Tap the wheel
The wheel is the button. Every prize is visible before the spin.

3
Claim the prize
“Aplicar desconto!” applies it and returns to the checkout.
The wheel is one of three games. All of them end the same way: one result, one button that names the prize, and a way back to the checkout. A discount says “Aplicar desconto!”. A gift says “Resgatar presente”.

Trinca Mágica
Gift prize

Carta Secreta
Gift prize
Item 05
09:40:05
For many small businesses in Brazil, Pix is the main way to get paid, so the app treats it as a first-class option, along with a payment link for buyers who aren’t at the counter. Installments show the price of each option, so nobody does math at the counter.

Payment method
Each option says how the buyer pays, not just its name.

Installments
Credit or debit, then the value of each installment.
Item 06
09:40:21
For a few seconds, the seller turns the phone around and the buyer becomes the user. I ordered the tap screen by the questions a buyer asks, from top to bottom.

1
2
3
4
5
6
7
1
Header
The seller’s way back, small on purpose.
2
Amount and payment type
How much, and how? The largest text on the screen.
3
Illustration
Where does the card go? Behind the phone, where the NFC antenna is.
4
Instruction
The same answer in words: “Aproxime o cartão atrás do celular.”
5
Signal bar
Is it working? It fills and darkens as the card gets closer, and a sound confirms the read.
6
Ad slot
Is there something for me? Each client runs its own campaigns here, like a coupon for the next visit.
7
Accepted brands
Will my card work? Answered before the buyer has to ask.
Why it matters: it’s the only moment the buyer looks at the store’s phone. The ad slot turns it into a marketing channel without adding a step to the payment.
Item 07
09:41:00
Approval leads straight to the receipt. The buyer chooses how to get it: a text to their phone, an email, or a QR code. With the QR code, the buyer scans the seller’s screen and opens the receipt on their own phone, without sharing a number or an email. It’s a transaction receipt, not a tax invoice.

Approved
One clear state, then two next steps.

Channel
Phone, QR code or email.

Phone number
A numeric keypad built for the counter.

Receipt
Order, items, discount and support in one view.
Item 08
Any time
Not every sale starts in the catalog: a service, a gift card, a toy without a tag. The Cobrar tab takes an amount directly, with shortcuts for common values. From there, it uses the same payment engine as the cart, so the tap screen, approval and receipt never had to be designed twice.
Cart
Typed amount
→
One payment engine

Empty
Continue stays disabled until there’s an amount.

Amount entered
Cents in superscript, as on every price in the app.
05 · Total
The product went to development and was sold as a white label. I left The Company a few months after launch, so I don’t have post-launch data. What I can show is what was delivered and the framework I’d use to judge it.
PROMETEUX
DELIVERY RECEIPT
Reino da imaginação · SoftPOS white label · 2025
SCOPE
ARCHITECTURE
WAYS TO PAY
HANDOFF
DEMO STORE
THANK YOU
North Star metric
It only grows when all three users win: the seller keeps choosing the phone over the machine, the buyer gets through the tap screen, and the client’s volume rises. Revenue alone could grow from a few large stores while most sellers give up.
Each user of the product gets its own set, so a gain for one can’t hide a loss for another.
Activation
Share of new sellers with an approved payment within 7 days.
Time to charge
Median time from opening the checkout to approval.
Retention
Sellers still charging with the app 4 weeks after their first sale.
First-tap success
Approved reads on the first attempt, out of all payments.
Tap abandonment
Payments that reach the tap screen and never get approved.
Receipt delivery
Sales with a receipt sent, and the mix of phone, QR code and email.
Volume
Payment volume per store per month, by method.
Time to launch
Days from contract to a live branded app.
Campaign return
Sales influenced by games, coupons and the ad slot, against the cost of prizes.
Numbers that must not get worse while the KPIs improve.
Fraud and chargeback rate
Sales blocked by the device check
Support tickets per 1,000 payments
Discount given vs. sales
Brand settings that fail contrast
The payment funnel is where these KPIs meet. Each step gets an event, so a drop shows exactly which screen to work on.
Checkout
Carts started
Method
Method chosen
Tap screen
Time to first read
Read
First-tap success
Approved
Approval rate
Receipt
Receipts sent
30 days of events per client before changing anything. Each brand starts from its own numbers.
Find the biggest drop in the funnel, then watch it happen: counter observation and short interviews with sellers.
Test one change at a time, like the tap instruction or the illustration, against the KPI it should move and its guardrails.
What works goes into the core, so every client gets it at once. That’s the advantage of one core for every brand.
06 · Learnings
What happened
The tap screen is used by the buyer, not the seller. I ordered it by the questions a buyer asks, from the amount to the accepted brands.
Next time
Test it where it lives: the phone turned across a real counter, a physical card, a queue, glare. That’s where reads fail and instructions get skipped.
Applies to
Any screen shown to a second person: kiosks, check-in, signatures, handoffs.
What happened
Splitting a brand layer from a fixed core let one product serve many companies, and made the toy store possible.
Next time
Ship a configuration contract with the screens: minimum contrast for brand colors, logo safe area, copy length, image ratios. One client’s color can break accessibility for all of its sellers.
Applies to
Design systems, multi-brand and B2B2C products.
What happened
Games were optional, adults-only, and paid in discounts or items, never money.
Next time
Bring legal in from the first sketch and put the rules in the web panel: age confirmation, prize caps, end dates, a reminder of local authorization. Rules that live only in a contract get skipped.
Applies to
Promotions, rewards, anything with money or minors nearby.
What happened
I delivered the flows and left a few months after launch, before any data came in.
Next time
Hand off the framework in section 05 with the screens, every event specified, so the team keeps learning after the designer leaves.
Applies to
Every handoff.

07 · Next iteration
Each bet starts from a gap in this version and targets a KPI from section 05. I ranked them by impact on those KPIs and by effort, which is my estimate.
Impact on KPIs ↑
Do first
P1
P2
Plan
P3
Fill-ins
Later
L1
L2
Effort →
Now
P1
Recover a failed read
Now
P2
Show where every discount came from
Next
P3
Show the connection before charging
Later
L1
Feedback that doesn’t depend on sight
Later
L2
Self-serve brand setup with built-in checks
P1 · Protects the sale
Gap
The flows in this version follow a successful read. When the card moves too soon or isn’t contactless, the buyer needs a next step, not a dead end.
Change
A recovery state that names the likely cause, retries in place and offers Pix or a payment link in one tap.
Hypothesis
A specific cause plus a one-tap alternative will reduce payments lost at the counter.
KPIs
First-tap success
Tap abandonment
Guardrail: fraud rate
Concept sketch
Card read interrupted
Hold the card still behind the phone for 2 seconds.
Try again
Pay with Pix
Send a payment link
P2 · Protects the margin
Gap
Checkout and receipt show a single “Desconto” line, whether it came from a game, a coupon or the seller.
Change
Name the source in the checkout, on the receipt and in the panel’s reports.
Hypothesis
Naming the source will make prices easier to trust and campaigns easier to reconcile.
KPIs
Campaign return
Guardrail: discount vs. sales
Guardrail: support tickets
Before / after
Today
Desconto
- R$ 20,00
Next
Desconto · Roleta da sorte
- R$ 20,00
P3 · Protects trust
Gap
Connection was the top complaint about card machines (45% of small businesses, Sebrae 2021). A phone depends on it just as much.
Change
Connection status in the checkout header, a warning before the buyer taps, and receipts queued until the signal returns.
Hypothesis
Warning before the payment prevents failures during it.
KPIs
Tap abandonment
Receipt delivery
Retention
Concept sketch
Checkout
Weak signal
Check the connection first
Payments may fail. Move closer to the router or switch to mobile data.
Charge anyway
L1
Haptics and a spoken amount on the tap screen, so buyers with low vision can pay without asking the seller. It builds on the signal bar, which already uses length and sound, not just color.
L2
Clients upload their brand in the panel and see it live, with contrast and copy length checked before publishing. It turns the configuration contract into a tool and cuts the time to launch.

