A working model · two seconds, at real speed
Followthe money
You tap, you swipe, or you insert the card.This is everything that happens next, live.
Three people live inside every card payment: the one who taps, the shop that accepts, and the people who build the rails between them. This page works from any of those seats. Everything below is playable, so tap what invites tapping, and carry one question through all six acts: where is the money right now, and who is on the hook if this step fails?
First, choose this tap's fate
Then tap the card
02 · The economics
The tap is free. The system is not.
So who pays for the two seconds? Not you, at least not directly. The merchant receives slightly less than you paid, and the difference is divided among everyone who carried the question: the issuer's share is called interchange, the network takes its fee, the acquirer keeps a margin. Drag the amount and watch the split — this small arithmetic is the business model of every card on earth.
- Interchange · to the issuer
- $1.70
- Network fee · to the scheme
- $0.15
- Acquirer margin
- $0.45
- The merchant keeps
- $97.70
- If you carry the card
- Your rewards are not a gift. Points and cash-back are paid out of interchange, which means the shop just funded your miles. It is also why the corner store sets a card minimum: on a small ticket, the fixed part of the fee eats the margin.
- If you run the shop
- This split is what card acceptance actually costs you, and most of it is set by the networks and not up for discussion. The acquirer margin is the one line you can shop around on. That is what you are really comparing when you compare processors.
- If you build the rails
- Interchange is the gravity of the whole system. It funds card programs, decides which products issuers push, and explains why every fintech eventually wants to issue a card. Follow it and most strategy in this industry starts to make sense.
03 · That night, and the next day
No money moved when you paid.
If the approval took two seconds, when does the shop actually get the money? Here is the part almost nobody knows: at the moment of approval, not one naira or cent has moved. The approval was a promise, the issuer saying I am good for this, hold my word. The money keeps that promise overnight — press play and watch eighteen hours in ten seconds.
5:03 PM
04 · When it goes wrong
The argument after the money moved.
Disputes are the part of payments people feel the most and understand the least, and they are not one thing. A stolen card, an order that never came, a subscription you cancelled twice, a charge that only looks doubled: same rail, four different arguments. So this act is a case file rather than a diagram. Pick one, and it plays out a filing at a time. You will make the calls.
05 · Before the tap
Every card is a promise someone else keeps.
Four acts started at the tap and moved outward. This one goes backward. Before any tap there had to be a card, and a card is a promise four parties make together behind one logo. A company that is not a bank can hand you one, and the question that carried you here turns with it: whose money is behind this card, and who promised the network it would be there? Sit in the program manager's seat and build one.
First, choose who you are
You pay drivers the moment a ride ends, onto a card you issue, instead of a bank transfer on Friday.
The cast you must recruit
Four seats, and you fill only the last. Each one asks something of you before your first card can tap.
- 01
The network
Lends the card its reach, and the rules that come with it.
Asks of you · A certified program on a member bank's card range, and a fee on every transaction it carries.
- 02
The sponsor bank
The issuer of record. Its name is in the small print on your card.
Asks of you · Due diligence on you, reserves against what your cardholders might spend, and the final say on compliance.
- 03
The issuer processor
Answers in the two seconds, keeps the ledger, and gives you an API instead of a mainframe.
Asks of you · Your program's rules in advance, and an answer in milliseconds when it asks you one.
- 04
You, the program
The product, the customers, the funding, and the support.
Asks of you · Of yourself: a reason the card should exist, and the money to stand behind it.
Then, decide where the money sits
This is the choice that decides who is on the hook. Act II showed that no money moves at the tap; here is where it was waiting.
- The night before
- Your money, already at the sponsor bank, in an account set aside for the program.
- At the tap
- The processor checks the balance it holds and answers alone. Nothing is asked of you.
On the hook
You, and you already paid. The money was there before the tap.
For a gig platform · You load each driver's card after the ride; the float is yours until they spend it.
Then, replay the first tap
Act I gave the issuer nine hundred milliseconds and drew it as one stop. Here it is opened up, for the card you just built: the machine that answers, the seat that is sometimes asked, and the book that remembers.
The issuer's window: 900 ms
In from the network →
Finally, who is on the hook
The question this page carries, answered for the card you built. Three nights it could fail, and who pays on each.
- The tap, and your service is down
Nobody notices
The processor answers from the balance it already holds. Your outage is your problem and not the cardholder's; the card keeps working on money that was posted yesterday.
- Settlement, and you cannot pay
The sponsor bank, from your account
There is nothing to fail. The bank pays the network from the program account you funded, and the only way it comes up short is if you loaded less than your cardholders spent, which the processor would not have let happen.
- A dispute, a month later
The cardholder's balance, then the merchant
A chargeback credits the cardholder at the sponsor bank and pulls the money back from the acquirer. Your float is untouched; your support desk is not.
Behind this card: your money, posted in advance. The sponsor bank is on the hook to the network, you are on the hook to the sponsor bank, and the cardholder is never asked. The safest card to run, and the one that ties up the most cash.
- If you carry the card
- The name on the front is not the bank. Turn the card over and the small print names the bank that actually holds your money, and that is the name worth looking up. A decline at the till is usually a rule the program wrote, not a judgement the shop made.
- If you run the shop
- A card from a company founded last year taps like one from a bank founded last century, because to your terminal it is one: the range belongs to a member bank, and the rules arrived with it. You are paid the same way, on the same night, whoever is on the front.
- If you build the rails
- Where the money sits is the whole design. Prefunded ties up cash and sleeps well; just-in-time frees the cash and makes your uptime the card's uptime; credit pays the most and bills you a year later. Pick the funding first, and the rest of the architecture follows from it.
06 · What it costs to run
A card program is a business, not a feature.
The card is built and the cast is paid. So does it pay? Act two showed where a hundred dollars goes at one tap; this act adds the taps up for a month, against what the network, the processor, the fraud and your own team cost, and asks the page's question one last time, now as a consequence of your numbers. Three stories to start from, then every number is yours to move.
- Gig payouts$102,833 a month · pays
Prepaid, 25,000 drivers, paid the moment the ride ends and spent by the end of the shift. The card from act five.
Largest line: processor, per account · breaks even at 2,613 cards.
- Neobank debit-$48,556 a month · underwater
Debit, 50,000 customers, no monthly fee, balances at the sponsor bank and a team to pay. Underwater until something moves; try the fee.
Largest line: fixed costs · breaks even at 62,052 cards.
- Credit builder$74,750 a month · pays
Credit, 15,000 cards, the highest interchange, and a loss line that grows with the balances rather than the cards.
Largest line: credit losses · breaks even at 3,760 cards.
The sheet needs more room than this column has, so it sits in a room of its own: build a card program.
I have spent ten years building the systems in this story, on the issuing side and the acquiring side, in Lagos, Toronto and Vancouver. The two seconds have been my working life.
If you want the longer version, the work is here.
Further reading: The Anatomy of the Swipe, Ahmed Siddiqui — the book that maps this territory in full.
Open to new roles
Open to founding and senior product roles, Head of Product to Director, in Vancouver or remote across Canada.
If you are a founder with a first product still to ship, I design it end to end and build the first version with your engineers. If you run product at a scale-up or a bank, I take the regulated, operationally heavy workflows and make them feel simple.
Twenty minutes on a call is the quickest way to find out whether the fit is real. The calendar link books straight into my week.