Credit Card Rewards Explained: How They Work and What They Are Worth

The short answer: cash back is money at a fixed rate; points and miles are a currency whose value depends on how you redeem it. On 20,000 USD of yearly spending, a no-fee 2% cash card pays 400 USD. A fee-paying points or miles card only beats that if your redemptions clear a break-even value per point — 1.23 US cents in the illustration below, and about 2 cents for a real 395 USD card earning 2x. If you carry a balance, the interest outweighs any reward; if you will not research transfers, take the cash.

Reward programmes are deliberately complex because complexity hides value — yours and theirs. Rewards are also close to universal: the CFPB notes that since 2019, more than 90 percent of general-purpose credit card spending in the United States has happened on rewards cards. Almost everyone is earning something. Very few people know what that something is actually worth. Here is the honest breakdown of the three currencies, starting with where the money comes from — because once you understand the funding, the rest of the design stops looking arbitrary.

Where Reward Money Actually Comes From

When you pay by card, the merchant does not receive the full ticket price. Their acquiring bank deducts a merchant service charge, and the largest single component of that charge is interchange — a fee handed from the merchant's bank to your card issuer on every transaction. Your issuer keeps part of it to cover fraud, funding and operations, and hands the rest back to you as rewards. Rewards are not a gift. They are a partial rebate of a fee the merchant already paid, and which merchants price into what everyone pays, cardholder or not.

That mechanism explains the single most confusing thing about global rewards: why the same brand pays wildly different rates in different countries. In the European Union, interchange is capped by law. Regulation (EU) 2015/751 states at Article 4 that providers "shall not offer or request a per transaction interchange fee of more than 0,3 % of the value of the transaction for any credit card transaction," with Article 3 setting a 0,2 % cap for debit. In the United States, by contrast, the Federal Reserve's Regulation II governs debit card interchange only — credit cards are not covered by it at all.

The size of that gap is visible in the Fed's own data. In its 2024 Regulation II reporting, debit transactions from capped ("covered") issuers averaged $0.23 per transaction, or 0.47% of value, while exempt issuers averaged $0.51, or 1.21% — roughly two and a half times more, on the same rails, for the same purchase. Caps compress the pool that funds rewards.

The practical consequence for you: in a capped market, no card can durably pay 2% back on general spending out of interchange alone. The money has to come from somewhere else — an annual fee, interest from people who revolve, or a co-brand partner's marketing budget. That is precisely why the generous cards outside the US are usually annual-fee products or airline co-brands, and why a flat 2% everyday cashback card is largely an American phenomenon. If someone in a capped market offers you an uncapped-looking rate, read the terms for the caps, the category exclusions, or the fee that is paying for it.

The other funding source is interest, and it deserves blunt treatment. If you carry a balance, the interest almost always exceeds the rewards by a wide margin. The UK's Financial Conduct Authority defines persistent debt as the state where customers are "paying more in interest, fees and charges than they are paying of their balance" — and estimated that firms addressing it properly could save customers up to £1.3bn a year in lower interest charges. Nobody in that position is winning on points. If that describes you, stop reading about reward currencies and go read about balance transfer cards instead.

Cashback

Cashback is the simplest and most transparent. A dollar back is a dollar, full stop. No valuation games, no blackout dates, no transfer charts. For anyone who does not want to optimise, cashback is almost always the right choice because you cannot misjudge its worth.

It has two further underrated advantages. First, it is currency-agnostic — it lands in the currency of your statement, so it never becomes stranded value in a programme you no longer use. Second, it is denominated, meaning the issuer cannot quietly reduce what you already earned. A points balance can be devalued overnight by changing the award chart. A cash balance cannot.

The trade-off is a ceiling. Cashback is normally paid at the rate the interchange pool supports, so it clusters in a narrow band. You will rarely see an outlier. That predictability is the product.

Points

Flexible bank points are worth more if you learn to use them — transferring to airline or hotel partners can yield outsized value, sometimes 1.5–2x a cash redemption. But that value only exists if you do the work of finding good transfer redemptions. Redeem points lazily, for statement credit or merchandise, and they are often worth less than plain cashback. Points reward effort.

The mechanism behind the upside is worth understanding, because it is not magic. An airline sells its miles to banks for cash, at a wholesale rate. When you transfer bank points into that airline, you are converting into a currency the airline can issue at will and price however it likes. When the airline prices a seat cheaply in miles — typically premium cabins on routes with unsold inventory — the arithmetic runs in your favour. When it prices dynamically against the cash fare, the arbitrage disappears and you are back to roughly cash-equivalent value. Transferable-points cards such as those earning Membership Rewards on the Amex Gold exist to give you optionality between those two worlds; the optionality is the asset, not the points themselves.

Two structural risks come with that optionality. Transfers are one-way and almost always irreversible, so a mistimed transfer converts flexible value into a specific airline's currency permanently. And bank points typically live in a programme account tied to your card. American Express states plainly that if you cancel and hold no other eligible Card or Checking account, the points you have not redeemed "will be forfeited immediately" — with one narrow exception on that same page, for New York Card Members, who as of 10 December 2023 "may be able to redeem for 90 days after notice of account closure". That wording was confirmed on American Express's own cancellation FAQ in July 2026. A large points balance is therefore a soft lock-in on the annual fee. Budget for that before you accumulate one.

Miles

Miles are airline-specific points. Powerful for committed flyers loyal to one alliance, frustrating for everyone else — award availability is limited, values swing, and they pressure you to keep flying one carrier. Only worth it if your travel already centres on that airline.

The core failure mode of miles is that you hold inventory in a currency whose issuer sets both the supply and the price, and has every incentive to adjust both. The CFPB has flagged exactly this, reporting that consumers describe issuers and merchant partners reducing the value of rewards already earned by increasing the number of points or miles needed for a redemption (CFPB), alongside balances vanishing on account closure and through expiration policies applied without prior notice. Treat a large mileage balance as a depreciating asset with no announced depreciation schedule. Earn towards a specific trip; do not hoard.

The Valuation Trap

Issuers advertise rewards at their best-case redemption value, which most people never hit. A "worth up to 2 cents per point" card is worth that only on optimal transfers — for typical use, value it lower and compare honestly.

The trap compounds with sign-up bonuses. The CFPB found that nearly 1-in-10 dollars earned by consumers in rewards are linked to sign-up bonuses — a large slice of total programme value that arrives once and never repeats. A card that looks brilliant in year one on a bonus-inclusive calculation can be mediocre in year two on earn rates alone. Always run the numbers twice: once with the bonus, once without. The second number is the one you will live with.

A Worked Example: One Year of Spending, Five Redemptions

The two cards below are illustrations, not real products — the rates are chosen to show how the arithmetic behaves, and no issuer is being described. Take 20,000 USD a year of card spending, split 4,000 dining, 5,000 groceries, 3,000 travel and 8,000 everything else. The method works in any currency, provided you keep the spending and the per-point valuations denominated in that same one.

A flat 2% cashback card with no fee pays 20,000 × 0.02 = 400 USD. Done, no decisions.

Now a points card charging a 250 USD annual fee, earning 4x on dining and groceries, 3x on travel and 1x elsewhere. That is (4,000 × 4) + (5,000 × 4) + (3,000 × 3) + (8,000 × 1) = 16,000 + 20,000 + 9,000 + 8,000 = 53,000 points. What those points are worth depends entirely on how you redeem them.

The break-even against the free 2% card is the number to memorise: you need 400 + 250 = 650 USD of value from 53,000 points, which is 650 ÷ 53,000 = 1.23 US cents per point. Redeem below that and the "premium" card has cost you money despite the flashier earn rate. Redeem at transfer values and it pays two to three times as much. The card did not change. Your behaviour did.

Note also what the table does not include: the effort. The business-award row assumes you found award space at a sensible price on dates you can actually travel, which is a real research task, not a formality. If you want to see what that work looks like end to end, the award-search walkthrough covers it. If you would not do it, your honest valuation is the statement-credit row.

The Same Test on Real Cards

The illustration above isolates the arithmetic. Here is the same test on four real US cards, using only what each issuer's own page stated when we read it on 16 October 2026. These are US products, open to US residents who pass the issuer's credit check; the method works in any market.

Three tests follow from that table.

A 395 USD miles card against a free 2% card. On 20,000 USD of everyday spending, Active Cash pays 20,000 × 0.02 = 400 USD. Venture X earns 20,000 × 2 = 40,000 miles and costs 395 USD. To draw level the miles must be worth 400 + 395 = 795 USD, which is 795 ÷ 40,000 = 1.99 US cents per mile. Capital One's page says the miles can cover travel purchases or transfer to "15+ travel loyalty programs", but it does not state a cents-per-mile value. So on everyday spending the card only wins if you redeem near 2 cents, or if you would have paid for its travel benefits anyway — and those benefits are worth what you would otherwise have spent, not their list value.

A 2% card abroad. Active Cash charges 3% on foreign transactions. Spend the equivalent of 1,000 USD abroad and you earn 20 USD and pay 30 USD: a loss of 10 USD on every thousand. Quicksilver earns less at home, 15 USD per thousand, but charges no foreign fee, so abroad it finishes 25 USD per thousand ahead. The best rewards card for a trip is often the one with the lower headline rate — see cards without foreign transaction fees for the alternatives outside the US.

A real calendar-year cap. The Gold card's supermarket rate is a live example of the calendar-year cap described further down: 4x on up to 25,000 USD of US supermarket purchases a year, then 1x. A household spending 30,000 USD a year at US supermarkets earns (25,000 × 4) + (5,000 × 1) = 105,000 points — a blended 3.5x, not the headline 4x. Restaurants run on their own 50,000 USD cap. Those points only become money once you decide what a point is worth to you, and the result has to clear an annual fee that the Amex page we read did not display, so check it on Amex's own site before you compare.

Which Currency Behaves How

How Rewards Are Earned, Posted and Lost

Valuing a reward is half of the subject. The other half is what happens between the moment you tap and the moment a reward is irreversibly yours. There are five stages, and a reward can quietly fail at every one of them.

The five stages between the tap and the money

1. Authorisation. The terminal asks your issuer to reserve the amount. This is a hold on your credit line, not a completed transaction. Nothing has moved and, for most programmes, nothing has been earned.

2. Settlement and posting. The merchant submits the transaction, it clears, and it appears on your account as posted. This is normally the point at which a reward can be calculated at all, because until now the final amount was not known. Hotels, car hire and restaurants with tips routinely settle for a different figure from the one authorised.

3. Accrual. The issuer applies the earn rate to the posted amount, using the merchant's category to decide which rate applies. This is where a purchase you thought was a bonus category turns out not to be.

4. Statement. The accrued rewards are added to your balance at the end of the cycle. Many programmes only make rewards redeemable after this point, which is why an app can show a number you cannot yet use.

5. Redemption. You convert the balance into a statement credit, a transfer, or a booking. Only here does the value become fixed. Everything before this is an entry in someone else's ledger.

The single most useful habit in this whole article follows directly from that list: a reward is provisional until it is redeemed. Every mechanism below is a way that a provisional reward becomes a smaller one.

Pending is not earned

The most common support question about rewards is why a purchase is not showing. The answer is usually stage one: the transaction is authorised, not settled.

That gap matters in three specific situations, and they are the expensive ones.

• Deposits and holds. A hotel or car hire firm authorises an estimate, then settles a different amount days later. The reward, when it comes, is calculated on the settled figure.
• Month-end and bonus deadlines. If a minimum-spend window closes on the 30th and the merchant settles on the 2nd, the purchase may land in the wrong window entirely. This is how people miss a bonus with money genuinely spent in time. Make the qualifying spend early in the window, never on the last day.
• Refunds in flight. A purchase that settles and is then refunded takes its reward with it.

Refunds, returns and the reward that goes backwards

If a purchase is refunded, disputed successfully, or otherwise reversed, the reward earned on it is generally reversed too. This is not a penalty; the transaction the reward was paid on no longer exists.

It becomes a real problem in one pattern in particular: buying something expensive to hit a minimum spend and then returning it. The spend is undone, the reward on it is undone, and if the return drops you below the bonus threshold the bonus can be undone as well. The rule we give in hitting a sign-up bonus without burning cash is the one that survives this — only ever count money you were going to spend anyway, on things you are going to keep.

It is worth knowing the direction the protection runs, too. Card networks and consumer-credit law give you real leverage against a merchant who will not resolve a dispute; they give you nothing at all against a rewards programme reversing an accrual on a transaction you cancelled yourself. The dispute rights are covered in chargebacks, Section 75 and Regulation Z; the reward reversal is simply the arithmetic correcting itself.

The merchant category code decides what counts

Here is the mechanism behind the second most common complaint: a purchase that obviously belongs in a bonus category did not earn the bonus rate.

Issuers define bonus categories by the merchant's category code — a classification attached to the business, not to your basket. It describes what the merchant is, and it is not chosen by you, by the till, or by what you bought. The practical consequences are consistent everywhere:

• A supermarket concession inside a department store can code as the department store.
• A warehouse club, a delivery platform and a petrol station with a shop attached each code as themselves, not as the groceries you bought there.
• A travel agent, an airline and an online travel platform can code differently from one another for what is functionally the same booking.
• Paying through an intermediary — a wallet, a marketplace, a bill-pay service — can change the code your issuer sees entirely.

Your issuer's own terms are the only document that states which codes qualify, and issuers vary. Two things follow. First, ask for that list in writing before you build a strategy on a category. Second, test it cheaply: make one small purchase at the merchant, wait for it to post, and read what actually accrued. One small transaction settles the question permanently; a year of assuming does not.

How a cap is actually measured

A capped bonus rate is not one design, it is several, and they behave very differently on identical spending. The measurement basis is usually in the terms and almost never in the headline.

That last row is not an accident of design. A category that requires quarterly activation is priced on the assumption that a meaningful share of holders will not activate, and the ones who forget are subsidising the ones who do not.

A worked example: the same spending, two cap designs

Take 12,000 a year of grocery spending, unevenly distributed — 1,500 a month for eight months, and 0 for four. Two cards, both advertising the same headline.

Card A: 4% on groceries, capped at 500 of spending per statement cycle, 1% thereafter. In each of the eight active months, 500 earns 4% (20) and 1,000 earns 1% (10), so 30 a month, 240 a year. The four empty months earn nothing and the unused cap does not carry forward. Blended rate on 12,000: 2.0%.

Card B: 4% on groceries, capped at 6,000 of spending per calendar year, 1% thereafter. The first 6,000 earns 4% (240); the remaining 6,000 earns 1% (60). Total 300. Blended rate: 2.5%.

Same headline rate. Same spending. A 60 difference, entirely from how the cap is measured — and the second card wins precisely because this spender's pattern is lumpy. Reverse the pattern to a flat 1,000 a month and the gap narrows sharply.

The instruction that comes out of this is short: never compare headline rates. Compute the blended rate against your own twelve months of spending. Our rewards calculator does the arithmetic; three months of statements is enough input to get the shape right.

Expiry, forfeiture and what happens when the account closes

Three separate clauses decide whether a balance survives, and they are usually in three different places in the terms.

Expiry on a timer. Some balances expire a fixed period after they are earned. Where this exists it is generally stated plainly, and it is the easiest of the three to plan around.

Expiry on inactivity. More common and much easier to trip. The clock is reset by activity on the account rather than by earning, so a card left in a drawer can lose a balance that was never going anywhere.

Forfeiture on closure. The one that costs the most, because it interacts with a decision people make for unrelated reasons. Closing an account — or having it closed for you — commonly forfeits an unredeemed balance outright. If you are considering dropping a card because the annual fee no longer earns its keep, redeem the balance first and read why downgrading beats closing before you call.

The defence against all three is the same, and it is not loyalty. It is liquidity. Redeem regularly. A balance you can spend this year at today's rate is worth more than a larger balance under next year's rules.

The exchange rate you never see

A cashback balance is denominated in money and cannot be repriced. A points balance is denominated in a unit the issuer or its partner programme controls, and the number of points a redemption costs is set by them — after you earned it.

The regulator has looked directly at this. The CFPB's rewards issue spotlight documents consumers reporting rewards devalued, revoked, or made harder to redeem than the marketing implied, and its Circular 2024-07 of 18 December 2024 warns operators that materially reducing the value of rewards consumers have already earned, or revoking them on vague conditions, can be unfair or deceptive.

Read that as a description of the asset class rather than as an accusation. A points balance is an unindexed, revocable claim on a company with a commercial incentive to reprice it. That is not a reason to refuse points — transferable points can be worth considerably more than cash to somebody who redeems them well, and turning everyday points into premium travel is how that is actually done. It is a reason not to treat a balance as savings.

What to check, and what to ask

Six answers, all obtainable from your issuer, that decide more of your return than the earn rate does.

1. When does a reward move from pending to posted, and does the earn date or the posting date determine which promotional window it falls in?
2. Which merchant category codes qualify for each bonus category? Ask for the list, in writing.
3. How is the cap measured — statement cycle, calendar quarter, calendar year, or anniversary year?
4. Does anything require activation, and does it repeat?
5. What expires, and on what trigger — a timer, inactivity, or account closure?
6. What happens to the balance if I close, downgrade, or am closed?

Who This Is Wrong For

Rewards optimisation is oversold. It is the wrong project for you if any of the following is true.

• You carry a balance. Interest on revolving debt dwarfs any realistic reward rate. Clear the balance first; the maths is not close.
• Your spending is low or lumpy. On modest annual spend, the difference between a good and a mediocre earn rate is small in absolute terms, and an annual fee will usually swallow it. Run the break-even calculation above with your real numbers before assuming otherwise.
• You will not redeem strategically. If you know you will take the statement credit, buy the card that pays the statement credit best. There is no shame in it and it is frequently the higher-net answer.
• You travel rarely or unpredictably. Award availability rewards flexibility. If you must travel on fixed dates in peak periods, miles and hotel points will consistently underperform their advertised value.
• You are building or repairing credit. Approval odds, reporting quality and cost matter far more than earn rate at this stage. Start with building credit from scratch.

Questions People Actually Ask at This Point

Are points better than cashback? Only if you redeem them above roughly 1.2–1.3 cents each on a fee-paying card, and above 1 cent on a no-fee card. Otherwise cashback wins. Work out your own break-even rather than accepting a headline valuation. The cashback versus travel points comparison goes further into the choice.

Should I hold points or transfer them? Hold, until you have identified a specific award you can book. Transfers are one-way, and holding keeps your value flexible against devaluation in any single partner.

Do points expire? It depends on the programme, and the more important risk is not the clock but the account. Bank points are typically forfeited if the underlying card account closes, as Amex's own terms confirm above. Closing a card is a decision about your points balance as much as your annual fee.

Is an annual fee ever worth it? Yes, when the fee buys either a materially better earn rate on categories you genuinely spend in, or credits you would have spent the money on anyway. The test is whether the fee survives contact with your actual statement, not the marketing sheet — see when annual fees are worth it for how to run that test properly.

The Simple Rule

Want zero effort and guaranteed value: cashback. Willing to learn transfers for trips: flexible points. Loyal to one airline and fly a lot: miles. Match the currency to how much effort you will actually put in — not the marketing.

If you are somewhere in the middle — happy to learn transfers but unwilling to restructure your travel around one carrier — a transferable-points card such as the Amex Gold is the usual compromise, because it lets you decide later whether a given balance becomes a flight or a statement credit. That deferred decision is the whole value proposition. Just remember to check the annual fee against the break-even arithmetic above, with your own spending, before you commit.

Not financial advice — reward values and terms change; confirm before relying on them.