How the numbers get here.

Collection

Every card figure — fee, earn rate, bonus, minimum spend, window, cap — is collected from the issuer’s own published page, and every reward-seat figure from the loyalty program’s published pricing. The collection date and source link render beside the data they support. Secondary sources are used only to find a figure, never to print one.

The two verification states

A table marked collected holds figures our pipeline took from the source on the stated date, awaiting a human editor’s re-check. A table marked verified has passed that re-check, and shows when. Neither state permits estimation: a figure we could not collect renders as an em dash with the fallback wording, full stop.

The arithmetic

Profile pages apply published earn rates to a stated monthly spending pattern, category by category, with uncategorised spending earning the card’s base rate. Twelve months of that earn plus any reachable sign-up bonus gives the first-year points position. A bonus counts only when the pattern’s spending clears the minimum-spend hurdle inside the published window — otherwise the tables say so and exclude it.

Earn thresholds

Most Australian rewards cards pay their headline rate only on the first slice of a period’s spending. NAB’s ladder runs to $5,000 a statement period and then to $20,000 until 1 October 2026, when the rate and both bands change; Westpac and St.George drop to a quarter of a point past $10,000 in a statement cycle; the Qantas Money Titanium drops past $12,500 of domestic spending while leaving international spending alone; the Qantas American Express Ultimate steps its everyday rate down once 100,000 points have been earned in a calendar year. Each threshold is quoted on the card’s fact sheet with the issuer page it came from, and every months figure on this site is computed through those bands rather than through the headline rate.

Two assumptions sit inside that arithmetic and are stated here rather than buried. An issuer applies a spending threshold in transaction order, which nobody can predict a month in advance, so spending inside each band is spread across the categories in proportion to their share of the month — what an even month looks like. And a calendar-year threshold is counted from the start of a calendar year, because a reader’s starting month is not something these pages know. A step-down lowers a rate and never raises one, so spending an issuer excludes outright keeps earning nothing above the threshold as well as below it.

Where an issuer has announced a change on a stated future date, the new figures are loaded ahead of time and dated: each rate and each threshold carries the day it starts applying and the day it stops. Tables compute whatever is in force on the day the page is built, so both structures sit in the database at once and only one of them renders. That is what keeps an October rate out of a September table.

A bonus the issuer publishes with no threshold stated on it sits outside the ladder and is applied per dollar rather than banded with the rest of the month, because it keeps paying after the spending ladder has stopped. NAB pays a bonus point per two dollars of international spending from 1 October 2026 on those terms, and the fact sheet shows the derivation from the issuer’s per-two-dollars wording.

Months-to-a-seat is published as two figures, kept in separate columns. The first counts the spending pattern alone at the card’s published earn rates, and holds for as long as that rate and the transfer ratio hold. The second adds whichever sign-up bonus the card currently advertises, counted from the month the pattern finishes putting the minimum spend through the card, since nobody holds a bonus before that spend has gone through. Tables rank on the first figure. A bonus running this quarter can be several times the price of a seat, which makes it capable of flattening every pattern and almost every route to the same one to three months — an answer about a promotion rather than about the reader’s spending. Both figures are true. Only the first one is still true after the offer closes.

Route mathematics converts card currencies into airline programs through published transfer ratios. Where no transfer bridge exists between a card and a program, no conversion is shown — we never assume parity. Value-per-point figures divide the fare a seat would otherwise cost, net of carrier charges, by the points the program asks for it.

What a card costs

Three of the eight issuers in these tables charge a rewards programme fee on top of the card fee, and it is a separate annual charge rather than part of the advertised annual fee. Every cost figure here is the sum of the two, with both printed beside it so the arithmetic is visible. Where an issuer publishes a programme fee of zero, that is a collected figure and adds nothing; where an issuer’s fee table has no such line, the card fee is the whole cost. Where nobody has checked which of those two it is, no all-in figure is given at all — an unchecked charge standing in as zero would understate what a card costs, and in the flattering direction.

A reduced first-year fee is recorded only where the issuer charges it to every new applicant. Several of these cards offer a lower first-year card fee to people who already bank with the issuer, which is a real offer and not one a reader arriving here can assume they qualify for, so it is printed as a note rather than counted as the price. A cashback advertised against the first year is not a reduced fee either: the fee charged is the standard one.

Fee changes carry dates. An issuer that has published a new fee has it loaded here against the date it takes effect, invisible until then, and set beside the figure in force today once the card page has a change to announce.

Transfer ratios and their dates

A ratio is a dated figure in the same way a rate and a fee are. Providers publish changes weeks ahead, so each ratio here is held against the day it starts applying and the day it stops, and the matrix shows whichever one is in force on the day you read it. A ratio a provider has announced for a future date sits in the same table and stays out of the matrix until that date, with the new figure and the day it lands printed in the footnote underneath.

The end date is exclusive: a ratio replaced on 30 September applies through the twenty-ninth and its successor applies from the thirtieth. Every months-to-a-seat answer on this site resolves its ratio the same way, so a page read after a change has landed computes on the new figure rather than carrying the old one until somebody notices.

Each ratio also carries whether a person has checked it. Two of the six here were read from an aggregator rather than from the provider, and one of them is contradicted by two of the issuer’s own change notices; both say so beside the figure rather than in a disclaimer elsewhere. Having a source is a weaker claim than having been verified, and the matrix keeps the two apart.

What fees are not

Fees are displayed beside points rather than subtracted from them. A point has no fixed dollar value, so netting a $450 fee against 100,000 points would smuggle a valuation into a table that claims not to make one. Readers can apply their own valuation; the route pages supply the redemption arithmetic to build it from.

Machines and editors

Collection and computation are automated, and so is publication. Every draft clears a written-quality and integrity gate — banned filler lexicon, sentence variation, length bounds, and a numeric check that rejects any figure absent from the row’s collected fields — before it can go live. What stays manual is verification: the reviewed-by field on a page is written by a person’s re-check of the figures against the source, and nothing automated may set it. A page that has had no such re-check says so rather than borrowing the credit for one.