Credit Card Rewards A to Z: 26 Rules and the Math Behind Them

Twenty-six rules for credit card rewards, one per letter. Each is a decision rule with the math attached rather than a definition, and each links to the CardSavvy page that works the problem in detail.

The letters exist to make the list memorable. The reason to read them is that rewards cards now carry 92% of general purpose credit card spending, and most of the ways people lose money on them are arithmetic errors that take about a minute to check.

Run your own numbers in the Wallet Optimizer →

The rule that comes before A

Do not optimize credit card rewards while carrying interest-bearing credit card debt.

In the CFPB's December 2025 report to Congress, the average APR on general purpose cards peaked at 25.2% in 2024. Consumers were assessed $160 billion in interest that year, up from $105 billion in 2022, against $47.5 billion earned in rewards. Those are market-wide totals rather than a comparison of identical people. The ratio is more than three dollars of interest assessed for every dollar of rewards earned.

A card earning 2% back is an expensive loan with a small rebate attached whenever its balance accrues interest at more than ten times that rate.

The rule: pay the full statement balance by the due date. If that is not reliably possible, skip the rest of this page and minimize borrowing costs first. Are you ready for rewards? walks the prerequisites.

The calculator below is the one that answers most of the letters at once. It compares what a wallet earns on paper against what it earns after routing mistakes, redemption shortfalls, and unused credits.

A is for Annual Fee

The fee is certain. The benefits are conditional.

An annual fee buys access to better earning rates, transfer partners, travel protections, lounge access, checked bags, hotel certificates, or statement credits. The common mistake is subtracting every advertised benefit from the fee at face value.

A $200 credit is worth $200 only when you would have spent that $200, at that merchant, on that schedule, without the card. A monthly credit that produces twelve orders you would not otherwise have placed is a coupon book that changed your behavior, and it should be valued at whatever those orders were worth to you.

Annual fees at large issuers reached $8.7 billion in 2024, the highest in the CFPB's data. The average general purpose annual fee rose 21% over two years, from $105 in 2022 to $127 in 2024, with premium products driving most of the increase.

The rule: at renewal, ask whether you would apply for this card today at its current price. The fee you already paid is gone either way. See the $95 fee tier compared and the Platinum break-even.

B is for Bonus

A welcome bonus is worth pursuing when the spending was already going to happen.

Welcome bonuses produce more first-year value than ordinary spending on most cards. In the CFPB's issuer data the average bonus actually earned was $311 in 2024, which is worth noting as a decline from $326 in 2022 rather than the growth the marketing implies.

The spending requirement is not an invitation to spend more. The right time to chase a bonus is when normal expenses already reach the threshold: insurance premiums, planned travel, tuition, daycare, or taxes when the processing fee math works.

Keep two numbers separate. First-year value is ongoing rewards plus the bonus plus benefits you use, minus fees. Ongoing value drops the bonus. A card can be an excellent one-year acquisition and a poor long-term keeper, which is the entire premise of the churning math.

The rule: never spend an extra dollar to earn a fraction of it back.

C is for Categories

The best bonus category is the one where you already spend.

A multiplier is a rate only after you convert points to dollars:

$$\text{reward rate} = \text{points per dollar} \times \text{value per point}$$

A card earning 4 points per dollar, on points you realistically redeem at 1.2 cents, returns 4.8% in that category. At 0.8 cents it returns 3.2%. Then apply that rate to your spending rather than a stranger's.

The rule: optimize around the spending you already have. Five categories carry most of the value for most households.

D is for Devaluation

A points balance can stay the same number while buying less.

Points are not a deposit. A program can raise award prices, drop transfer partners, remove redemption options, or change earning rates. The CFPB's May 2024 rewards issue spotlight, drawn from more than 1,200 complaints filed in 2023, named four recurring problems: unexpected promotional conditions, devaluation, redemption difficulty, and revocation.

There is also a regulatory gap worth knowing. Regulation Z requires 45 days' notice for significant changes to account terms, and the list it points to covers APRs, annual fees, transaction charges, the grace period, and late fees. Rewards terms are not on that list. The CFPB's Circular 2024-07 does say that devaluing rewards already earned, or revoking them on buried conditions, can be an unfair or deceptive act, but that is enforcement after the fact rather than advance notice.

Chase's June 2026 Sapphire Preferred refresh is a clean example: the headline was three new bonus categories, and the quieter change was a worse Hyatt transfer ratio.

The rule: earn points against a use you can name. A six-figure balance with no plan is exposure, not savings.

E is for Effective Rate

"Up to 5%" describes one category on one card. The number that matters is what the whole wallet returns.

$$\text{effective rate} = \frac{\text{rewards and benefits realized} - \text{fees and costs}}{\text{spending you would have done anyway}}$$

The denominator is where this goes wrong. It should hold the spending that existed before the rewards program, not purchases created to earn rewards.

For scale: across general purpose rewards cards, rewards earned came to roughly 1.6% of purchase volume in the CFPB's most recent data, up from 1.4%. That aggregate covers many cards and many people, and it is a useful check against the impression that 5% is the normal outcome.

The gap between the advertised rate and the realized one has seven measurable sources, worked through in detail there.

The rule: judge the wallet by its realized value. The largest multiplier printed on one card describes one category.

See what your wallet returns after fees →

F is for Foreign Transaction Fee

A travel reward can vanish at the border. A card charging 3% on foreign transactions and earning 2% returns a negative rate abroad.

The fee can also apply to an online purchase processed overseas, depending on the terms. Bilt cardholders found this the hard way when charges posted 0.2% high in 2026.

Separately, decline dynamic currency conversion when a foreign merchant offers to bill you in dollars. That markup is the merchant's, and it is on top of whatever your card does.

The rule: carry at least one widely accepted card with no foreign transaction fee when you travel internationally.

G is for Grace Period

The grace period is what lets a credit card work as a payment tool instead of a loan.

It is the window between the close of a billing cycle and the payment due date. Most cards offer one on purchases, though issuers are not required to. When the card has one and you are not revolving, paying the statement balance in full by the due date avoids purchase interest.

Two things break it. Cash advances usually accrue interest immediately with no grace period at all. And once you carry a balance, new purchases can start accruing interest from the transaction date until the grace period is restored, which is also why adding everyday spending to a card holding a promotional balance transfer gets expensive quietly.

The rule: protect the grace period. No cash advances on a rewards card.

H is for Hard Inquiry

Applying usually means a hard inquiry, a new account, and a lower average account age.

Credit card applications are not grouped the way rate shopping is. FICO's own guidance limits that grouping to mortgages, auto loans, and student loans, so several card applications count separately.

One inquiry rarely matters much on an established file. Several before a mortgage, auto loan, or apartment application can. The real cost is not the score points, it is the chance of complicating a much larger transaction.

The rule: pause applications ahead of an important credit event. Chase applicants have a second constraint on top of this one, explained in the 5/24 rule.

I is for Interest

Rewards marketing describes what a card gives. The cardholder agreement describes what borrowing costs.

The two are not close. In 2024 the average general purpose APR peaked at 25.2% while rewards earned averaged about 1.6% of purchase volume. By the end of that year 49% of active general purpose accounts were revolving a balance, up from 45% in late 2021.

Interest is typically calculated on an average daily balance, so it accrues day by day rather than waiting for the next statement. A single month of revolving can erase a year of category optimization on ordinary spending.

This is also why proposals to cap card interest cut both ways for rewards optimizers, which a 10% cap would demonstrate quickly.

The rule: the phrase "free money" requires confirming first that no interest is being paid.

J is for Justifying the Fee

Once the fee is paid it is gone, and spending more does not recover it. Value each credit at the lowest of its face amount, what you would otherwise have paid, and the value of the alternative you gave up. CardSavvy publishes what premium credits are realistically worth against their marketed face value, and scores which ones are real value versus forced spending.

K is for Keep It Simple

A six-card setup beats a two-card setup in a spreadsheet and often loses in practice, to forgotten activations, wrong-card purchases, missed credits, and cards left at home. Two cards beat most trifectas, and the second card is worth roughly thirteen times the third.

L is for Loyalty

An airline or hotel card can clear its fee on one benefit. After every major US carrier moved the first checked bag to $45 each direction in April 2026, a couple checking one bag each on a single round trip avoids $180, which covers a $99 co-brand fee on the first trip of the year.

The cost is flexibility. An award seat can quietly justify a worse schedule, a further airport, or a hotel you would not have chosen, and none of that shows up as a loss anywhere. Flexible points reduce the dependence at the price of a harder valuation problem.

The rule: be loyal while the program keeps serving the travel you were going to do anyway. Hotel card break-evens run the numbers per program, and bag fee math shows which co-brands pay for themselves.

M is for Merchant Category Code

The card rewards the merchant's classification, not your description of the purchase.

Payment networks assign each merchant a four-digit merchant category code. Visa's merchant data standards manual defines it as a code describing the merchant's primary business, used for activity tracking, reporting, and risk management. Issuers then build bonus categories on top of that classification, which is where the surprises come from.

A cafe inside a hotel may not code as a restaurant. Stadium food may code as the stadium. An online marketplace often codes differently from the product being sold. The most-cited example on this site is that the Sapphire Reserve does not earn 3x on Airbnb, while the Preferred does.

The rule: treat category coding as a processing rule. Check the code before a large purchase.

N is for Net Value

Net value is the number that decides whether a card belongs in the wallet.

$$\text{net annual value} = \text{rewards realized} + \text{credits used} - \text{annual fee} - \text{costs} - \text{spending the card caused}$$

That last term is missing from nearly every rewards calculator. A card that induces $300 of extra spending to generate $30 of extra rewards has not produced $30 of value.

For a card you are considering rather than holding, the comparison is against your current wallet rather than against zero:

$$\text{incremental net value} = \text{value with the new card} - \text{value with the wallet you have}$$

The rule: gross rewards describe what the card produced. Net value describes whether it improved your finances. The optimizer glossary defines how CardSavvy computes each term.

Compare a new card against the wallet you already have →

O is for Opportunity Cost

Every swipe has an alternative, and the alternative is usually another card you already own.

Suppose a new dining card returns 4.5% after converting points to dollars, and the card in your wallet returns 3%. The improvement is 1.5 percentage points. On $5,000 of annual dining that is $75 a year, which does not cover a $95 fee on dining alone.

The same logic applies to redemptions. Booking through an issuer portal for extra points can cost you a lower direct price, elite night credit, easier changes, and better service recovery when something breaks.

The rule: compare against the best realistic alternative, never against earning nothing. Three of the assumptions that move this answer most are documented in where our own optimizer goes wrong.

P is for Points

Points look like currency and behave like a claim.

The same point can be worth several different amounts depending on whether it is redeemed for cash back, travel, merchandise, gift cards, an issuer portal, or a transfer partner. Program terms generally allow changes to earning rates, redemption options, award pricing, and partner lists.

Expiration is its own trap. Amex says Membership Rewards points have no expiration date, and its terms separately describe the conditions under which you can lose them anyway, including closing the last enrolled product.

The rule: convert points to dollars before comparing cards. How to value points without lying to yourself is the long version.

Q is for Quarterly Categories

Rotating category cards need activation each quarter and cap the elevated rate. Chase's Q1 2026 Freedom and Freedom Flex categories were dining, American Heart Association donations, and Norwegian Cruise Line, at 5% on up to $1,500 in combined purchases. If your natural spending in a quarter's categories is $300, the cap is irrelevant and buying gift cards to fill it is a way to lose money slowly. Several rotating category cards charge no annual fee.

R is for Redemption

Rewards are not value until they are used.

A large balance feels like wealth while remaining exposed to devaluation, program changes, account closure, and simple forgetting. The average rewards account in the CFPB's issuer data held about $192 in unused rewards at the end of 2024.

A good redemption clears three bars: it beats your cash-value floor, it buys something you wanted, and it does not require extra spending or unreasonable effort to arrange. That is a lower standard than the best redemption anyone has ever posted about, and it is the right one.

The rule: do not let the search for a perfect redemption prevent a good one. Bilt cash redemptions ranked by hurdle rate shows the method.

S is for Statement Balance

The minimum payment keeps the account current. It does not avoid interest.

The current balance can include purchases made after the statement closed, and paying it is normally unnecessary to preserve the grace period. The number that matters is the full statement balance by the due date.

Automate that specific amount, keep enough in the payment account to cover it, and add a second reminder. Autopay set to the minimum is the single most expensive default setting in personal finance.

The rule: autopay the statement balance.

T is for Transfer Partners

Transferable points lose their flexibility the moment they transfer.

Moving bank points into an airline or hotel program reaches the best redemptions available. It also hands control to the receiving program. Amex advises checking availability and the points required before transferring because transfers are final. Capital One states plainly that transferred miles become subject to the receiving program and cannot be transferred back.

A transfer bonus is not a reason to move points speculatively. A 20% bonus on points that then sit unused, in a program that later reprices its awards, is a loss.

The rule: find the award, confirm it, then transfer. Ratios and sweet spots are mapped for Chase and Amex.

U is for Utilization

Utilization is the reported balance against the limit, and it can affect a score even when you pay in full every month. Paying down before the statement reports lowers it when an application is imminent. What it does not require is interest: FICO calls the carry-a-balance advice a myth in those words. Approval also depends on more than a score, as a declined brand spokesperson demonstrated.

V is for Valuation

A point is worth the cash it saves you.

$$\text{cents per point} = \frac{\text{cash price you would have paid} - \text{taxes and fees on the award}}{\text{points redeemed}} \times 100$$

"Would have paid" is the load-bearing phrase. A $10,000 first class ticket is not a $10,000 redemption if the alternative you would have bought was a $1,200 economy seat, or no trip at all.

Valuation should also absorb the things that do not show up in the headline number: award availability, cancellation rules, positioning flights, portal markups, and the hours spent finding the seat.

For comparing cards, use either your own average redemption value from the last year or the program's cash-equivalent floor. Both are defensible. A blog's aspirational valuation applied to your wallet is not.

The rule: put your own numbers into the calculator, not someone else's.

W is for Who Pays

Rewards are funded by the payment system, and the funding is not neutral.

Merchants pay interchange fees to accept cards. When they charge everyone the same price regardless of payment method, part of that cost spreads across cash, debit, and card customers alike. A 2026 NBER working paper estimates interchange transfers roughly $30 billion a year from cash and debit users to credit card users, and finds that consumer sorting and merchant fee differences shrink that transfer by about 25% without eliminating it. Roughly $9.2 billion a year flows from households under $150,000 to higher-income ones.

A Federal Reserve working paper finds a similar pattern inside the card market. Cardholders who come out ahead gain about $15.1 billion a year. Cardholders who come out behind pay roughly $4.1 billion a month, and the difference between those two flows is bank income. The authors are explicit that the simple rich-versus-poor framing is incomplete: the dividing line is financial sophistication, and it runs at every income level.

None of this makes earning rewards responsibly unethical. It does mean the system is a transfer with identifiable payers, and that a strategy built on paying interest puts you on the wrong side of it. That is the argument behind the CardSavvy philosophy.

The rule: understand who funds the rewards, and avoid becoming one of them.

X is for Excluded Transactions

Card agreements commonly exclude balance transfers, cash advances, cash-like transactions, interest, fees, and refunded purchases from earning. Some person-to-person payments, account funding transactions, and gambling transactions are excluded or treated separately. A transaction can also earn the base rate while missing the category multiplier entirely because of how it codes.

The rule: before a large or unusual payment, confirm it is eligible, find the processing fee, and compute the net return.

Y is for Your Behavior

The largest uncertainty in a rewards model is the person operating it.

A 2026 study in the Journal of Banking & Finance used account-level data from a US financial institution and found that cardholders who joined a 1% cash back program increased card spending by 32% and card debt by 8%, with larger responses among more liquidity-constrained and less financially literate consumers. That is one promotion and one population, so it should not be applied mechanically to everyone.

Other evidence is genuinely mixed. A 2021 replication in the Journal of Retailing and Consumer Services failed to reproduce the classic credit card spending effect across all four of its own studies, though its meta-analysis still found an overall effect that had weakened over time and varied by where the data was collected. A 2021 fMRI study in Scientific Reports by Banker, Dunfield, Huang and Prelec found that card payment cues activated reward-related brain networks.

Behavior belongs inside the equation as a term, with its own sign. Ask whether you would buy this with cash, whether you would choose this merchant without the credit, and whether you spend differently while chasing a bonus.

The rule: a 1% increase in spending consumes most of the value of an ordinary rewards strategy.

Z is for Zero Interest

Zero is the highest-value rate on this page.

A working strategy targets zero interest, zero missed payments, zero purchases made to earn rewards, zero credits counted at face value unless they replace real expenses, and zero speculative transfers.

The goal is the most realized net value from spending that was already part of your life. For some households that is a carefully optimized multi-card wallet. For many it is two cards. For some it is one no-annual-fee cash back card, and when a card makes spending harder to control, the right number is zero for a while.

Rewards are also a small lever compared to the ones next to them, which is the subject of what to do after you finish optimizing.

Where to start

If you read one line from this page: pay in full, value everything honestly, compare against the alternative, and never let the reward decide the purchase.

The Wallet Optimizer handles the arithmetic. It computes which cards fit your spending, what each is worth net of its annual fee, and how much a new card would add over the wallet you already have. The hard part is feeding it assumptions that match how you spend and redeem.

Start with your real spending →

CardSavvy publishes educational information and math tools. This is not individualized financial, tax, or legal advice. Card terms, benefits, and reward programs change; check the current issuer agreement before applying or making a large purchase.