Editorial policy
Everything on this site comes out of one structured card database. There is no editor's thumb on the scale, because there is no editor — there is a data file and a ranking function, and both are checkable.
Last updated August 5, 2026.
Rankings are payout-blind
A card's position is computed from its terms: earn rate, spending cap, annual fee, category choice or activation requirements, and — for cards that earn points — the redemption value we have recorded for that currency. Compensation is not an input, and today there is none to be an input (how we're paid). If we ever take affiliate money, the ranking code still will not see it.
How we value points
A points card can only be compared with a cash-back card once somebody decides what a point is worth, and that decision moves every ranking on this site. Ours is deliberately conservative and written down: for each currency we record the published valuations we can find, and we use the lowest one. When The Points Guy says Amex Membership Rewards are worth 2.0¢ and Frequent Miler says 1.5¢, every number here uses 1.5¢. The build refuses to publish a value above the lowest source it cites, or below what the issuer will simply pay you in cash for the same point.
Each row that depends on a valuation prints the value, the sources it came from and their dates, so you can check our arithmetic against theirs. Where only one source exists, the row says single-source value. Where we could not find a defensible number at all — store-program points with no published redemption rate — the card is not ranked, and it is listed as unranked with the reason rather than quietly given a flattering 1.0¢. In the app, the value is a starting point you can overwrite per card: if you only ever cash out, say so and every number follows.
Every number carries a date
Card facts are researched from the issuer's own terms pages and stamped with the date we checked them. That date is printed next to the card wherever we show a rate. Issuers change terms constantly and without notice — a stamp that is months old is a warning, not a guarantee, and we would rather show you a stale date than hide one.
We never publish an unannounced rotating quarter, and we exclude a bonus we cannot verify rather than guess it. Cards where the research is thin are labelled unverified in the app and are kept out of the rankings entirely.
Corrections
Every card fact lives in a version-controlled data file, and each change is a separate commit. That history is our changelog — dated, public, and impossible to quietly revise. When you , we check it against the issuer's page; if you are right we change the data, and the change lands as a dated commit and a Card news entry. If we published something materially wrong, we say so there rather than editing in silence.
We will tell you not to get a card
The useful answer is often "you already own the right card" or "nothing here beats what you have." Advice funded by applications rarely says that; we would rather be the site that does. So:
- If your current wallet already wins a category, the recommendation is to use it — not to add plastic.
- If a card's annual fee outruns the credits you actually use, the verdict is cancel or product-change.
- If you carry a balance, no rewards rate beats your interest rate, and we will say that instead of selling you 5%.
What we are not
We are not a financial advisor, we are not paid to be neutral, and we are not comprehensive — the database covers a few hundred consumer cards, not every card in the country. Missing does not mean bad. See Terms.