By the time interchange splits off (module 3) and the scheme takes its cut (module 4), most people assume the issuer is swimming in transaction fees. It is not. A credit card program is a lending business wearing a payments costume. If you build, price, or model card products, the single most useful correction we can offer is this: the transaction side roughly breaks even, and the lending side pays for everything else, including the rewards.
This matters because product and growth teams routinely optimize the wrong line. They chase spend volume to grow interchange, fund richer rewards to chase that spend, and watch unit economics quietly invert. Understanding which lever actually moves profit is the difference between a program that scales and one that subsidizes its own decline.
The four revenue lines, ranked by what they contribute
A credit card issuer earns from four places: interest on carried balances, interchange, cardholder fees, and scheme incentives. They are not close to equal.
Federal Reserve research on credit card profitability put the structure in plain terms. Around 80 percent of issuer profit comes from interest income. Roughly 15 percent comes from usage fees such as late fees and foreign exchange fees. The remainder comes from balance transfers and miscellaneous items. Net interchange revenue, after rewards and processing costs, is slightly negative.
Read that last point twice. After you net rewards and the costs of running the transaction, the swipe itself is a small drag on profit, not the engine. The Fed estimated the transaction function at roughly negative 4 percent of aggregate credit card profitability.
Why interchange does not carry the program
Interchange is real money, more than $30 billion a year across US issuers, but it does not survive contact with rewards. On general-purpose cards, issuer interchange income averages close to 1.82 percent of purchase volume while rewards cost around 1.57 percent of that same volume. The gap left over has to cover fraud, processing, servicing, and the cost of funding the cardholder's float.
For a customer who pays in full every month, a transactor, that gap usually does not cover those costs. The transactor is a net expense unless an annual fee or some eventual revolving balance closes it. This is the structural fact behind every premium card with a high annual fee: the fee exists because the spend alone does not pay.
The lending engine
The profit lives in net interest income. Issuers borrow at one rate and lend to cardholders at a far higher one, and the spread is where a program either works or does not.
The CFPB has documented that the APR margin, the difference between the average card APR and the prime rate, reached an all-time high, roughly 14 percentage points on revolving accounts in its 2024 reporting. Average APRs hit about 25.2 percent on general-purpose cards and 31.3 percent on private-label store cards in 2024, the highest since at least 2015. Consumers were assessed about $160 billion in interest in 2024, up from roughly $105 billion in 2022.
That spread is not free profit. Against it the issuer carries the cost of funds and credit losses. The charge-off rate on credit card loans at commercial banks rose from about 1.72 percent at the start of 2022 to roughly 4.24 percent by the end of 2024. Net interest income only becomes profit after losses and funding are subtracted, which is why issuer profitability swings hard with the credit cycle even when spend looks healthy.
Revolvers versus transactors
This splits the customer base into two economies. Revolvers carry a balance and generate interest. Transactors pay in full and generate interchange minus rewards. The lending math means up to 80 percent of program profit can trace to interest on revolving balances, so issuers structure rewards, minimum payments, and promotional APRs to keep a healthy share of spend converting into carried balances.
A worked example
Take a cardholder who spends $20,000 a year on a 1.5 percent flat-cashback card.
Interchange to the issuer at about 1.82 percent of volume is roughly $364. Rewards paid out at 1.5 percent are $300. That leaves $64 before the issuer pays for fraud, processing, servicing, and funding the float on purchases. On a pure transactor, that $64 is gone and then some. The account is a small loss, which is exactly the Fed's negative-transaction-margin finding at the level of a single account.
Now assume the same customer revolves an average balance of $4,000 at a 24 percent APR. That is about $960 a year in interest. Even after subtracting an assumed 4 percent annualized loss rate on the balance, around $160, and a few points of funding cost, the lending side contributes several hundred dollars of margin. One revolver is worth more than a dozen clean transactors. The whole reward structure, signup bonuses, accelerators, and 0 percent intro offers, is engineered to acquire and retain that revolving balance.
How rewards are actually funded
Rewards are not paid out of interchange, though the industry often frames it that way. Interchange covers part of the rewards on heavy spenders, and the rest, plus the program's fixed costs and its profit, comes from interest income and fees.
That has two consequences worth holding onto. First, richer rewards only pencil out if they pull in spend that eventually revolves or pay an annual fee large enough to cover the gap. Second, anything that compresses interest income, a rate cap, a balance-transfer war, or a shift toward disciplined transactors, puts direct pressure on how generous rewards can be. We will see this tension again in module 10, where the pressure on the model bites hardest on exactly these lines.
Takeaway
A card program is priced on the lending desk, not the payments rail. Interchange roughly funds rewards and not much more, fees fill a narrow band, and the spread on revolving balances pays for the program and its profit. If you are modeling a card business, start with revolve rate, APR margin, and loss rate. Spend volume tells you how big the program is. The credit book tells you whether it makes money.