Most diagrams of a card payment are wrong in the same way: they draw four boxes, connect them in a square, and call it done. The square is fine as a memory aid, but it hides where the money sits, who carries the risk, and why the network sits in the middle taking a cut from both sides. If you are building anything that touches card rails, getting this map exact is the difference between a model that predicts costs and one that surprises you at settlement.

This module draws the model properly. Later modules unpack the swipe, interchange, scheme fees, and routing. Here we just establish who the players are and what each one is actually on the hook for.

The five roles

The "four-party" name is a historical misnomer. There are five roles, and one of them (the network) is the reason the other four can transact at all.

Cardholder

The person or business that holds the card and initiates the purchase. The cardholder has a relationship with one party only: the issuer. They never contract with the merchant's bank, the network, or the acquirer. Everything downstream is invisible to them, which is the point.

Issuer

The cardholder's bank. The issuer extends the credit line or holds the deposit account, issues the physical or virtual card, sets the credit limit, and authorizes or declines each transaction in real time. Critically, the issuer carries the credit and fraud risk on the cardholder side. If the cardholder defaults or the card is used fraudulently, the issuer generally eats that loss. In exchange, the issuer earns interchange, the fee that flows toward it on every purchase.

Merchant

The business accepting the card. The merchant wants funds in its account and wants to offload as much fraud and reconciliation work as possible. It contracts with one party: the acquirer.

Acquirer

The merchant's bank, or more precisely the licensed entity that holds the merchant's account and connects it to the networks. The acquirer underwrites the merchant (a real risk decision, because the acquirer is liable for chargebacks if the merchant disappears with prepaid goods undelivered), routes transactions into the network, and pays the merchant the sale amount minus the merchant discount. Most "processors" and payment facilitators you deal with sit on top of an acquirer or are sponsored by one.

Network

Visa and Mastercard are the two large examples. The network owns the rails, the messaging standards, the rulebook, and the brand. It does not issue cards, hold cardholder funds, or pay merchants. It sets interchange schedules, charges its own scheme and network fees to issuers and acquirers, and clears and settles the net positions between them. It connects every issuer to every acquirer so that a card from one bank works at a terminal served by another.

The network's product is reach. Any card works at any accepting merchant because the network guarantees the connection and the settlement, and it charges both sides for that guarantee.

How money and risk actually move

The square diagram implies a loop. The reality is a sequence, and the fees attach at specific edges.

When a cardholder pays, the issuer authorizes, then debits the cardholder's account and settles the transaction amount to the acquirer, minus an interchange fee that the acquirer pays to the issuer. The acquirer then pays the merchant the sale amount minus a merchant discount, which has to cover interchange, the network's scheme fees, and the acquirer's own costs and margin.

So interchange flows from acquirer to issuer. Scheme and network fees flow from both the issuer and the acquirer to the network. The merchant never sees these as separate line items unless it is on interchange-plus pricing, but it pays for all of them inside the discount rate.

A worked example

Take a $100 purchase on a Visa or Mastercard consumer credit card.

The cardholder, meanwhile, paid exactly $100 and may have earned rewards funded out of that $1.80 of interchange the issuer collected. That is the engine: interchange pays for rewards, which drive card use, which generates more interchange.

Where three-party schemes differ

American Express and Discover were built on a different shape. In the classic three-party model, one company is the issuer, the acquirer, and the network at once. Amex issues the card to the cardholder, signs the merchant directly, and runs the rails in between. There is no separate interchange fee paid between two banks, because there are not two banks. There is one merchant fee, set and collected by Amex, which is why three-party merchant pricing has historically run higher than open-loop Visa and Mastercard rates.

The clean split between three-party and four-party has blurred. Through Global Network Services (GNS), Amex licenses third-party banks to issue Amex-branded cards and to acquire merchants on its network, which makes those flows look four-party in structure while Amex still controls authorization and card rules. Discover runs a comparable network-licensing arrangement. Regulation has pushed on this too: Australia and the EU brought GNS-style flows under interchange caps, which is why Amex withdrew from parts of that business in those regions.

The practical takeaway for builders: do not assume a card's economics from its brand. A card's role structure depends on how it was issued and acquired, not just the logo on the front.

The takeaway

Five roles, not four. The cardholder and merchant each contract with exactly one bank; the network connects those banks and charges both. Money settles in a sequence, with interchange flowing acquirer to issuer and scheme fees flowing to the network, all buried inside the merchant's discount rate. Three-party schemes collapse the issuer, acquirer, and network into one entity, then partially re-expand through licensing. Draw the model this way and every fee in the rest of this course has an obvious place to sit.

Next →
Anatomy of a Swipe: One Transaction, Step by Step