What happens when a number is wrong.

Two dates, not one

Pages carry two separate timestamps and they mean different things. Collected is when a figure was last read off the provider’s page. Checked is when a human editor last re-read that figure against the source and confirmed it. A page can be freshly collected and not yet checked; where that is true, it says so rather than implying a review that has not happened.

What we will not print

A figure with no source and no collection date does not render. Where a value is missing, the table prints an em dash. It is never a zero, never “TBD”, and never an estimate dressed as a fact. If a promotion has lapsed or cannot be confirmed, the number comes off the page until it can be.

What this site is not

We do not take into account your financial situation, needs or objectives, nor do we provide advice, recommendations or credit assistance.

Points Herald publishes comparisons, not recommendations. We cannot and do not recommend that any particular product is suitable for you. Whether a product fits your circumstances is a judgement only you, or a licensed adviser, can make.

How the tables are ordered

Ordering is computed, not sold. Rows sort by the metric named in the caption, then by annual fee low to high, then by issuer name alphabetically. No commercial arrangement influences this order, and no product is promoted, sponsored or placed for payment anywhere on this site.

How this site makes money

Points Herald currently earns nothing from credit card applications. Card links on this site are plain, untracked links to the issuer. If that changes, the link is marked as sponsored and this page says so before you click it.

Some travel links — award-search tools, eSIMs, delay-compensation services — are affiliate links today, and are marked as sponsored where they appear. A link’s presence never changes a figure in a table. The full position is on how this site makes money.

Before you chase a bonus

Chasing sign-up bonuses has costs that do not show up in a points table. Leaving an old card open can mean paying two annual fees; several applications in a short period can affect your credit file; and a high total credit limit can reduce how much a lender will advance you on a mortgage. Consider the timing and the wider effect on your finances before opening or closing an account.

Reporting an error

If a figure here disagrees with the provider, the provider is right and we want to know. Corrections are made against the source, and the collection date on that row moves when the figure does.

The log

Corrections are recorded here with the date the change was made and the pages it touched. A corrections page that lists no corrections is furniture.

7 September 2026 — three route pages asserted things no source states

Three business-cabin route pages published on 4 September carried claims that resolved to no provider page. Two of them — Melbourne to Haneda and Melbourne to Hong Kong — described a pattern in which front-cabin reward seats are released, saying they cluster when schedules load far ahead and return as scattered singles nearer departure. Neither programme publishes any such statement and this site had collected none. The third, Melbourne to Heathrow, asserted that no aircraft flies the city pair without a stop, again with nothing attached. The same page also called front-cabin reward inventory to Europe the scarcest in the programme, which is a ranking nobody publishes.

All three sentences have been removed and the three pages rewritten from stored figures only. No price changed. Separately, Melbourne to Hong Kong in economy described its sector as sitting comfortably inside its distance band when it in fact sits closer to a band ceiling than any other seat on this site; that wording has been corrected.

7 September 2026 — every published page swept for the claims the editorial gate cannot check

The sweep recorded above worked from a list of three pages rather than from a search of every published body, and it missed one. Melbourne to Singapore carried the same release-pattern sentence in hedged form, conceding in its second half that nobody outside the airline can see the inventory it had just finished describing. It has been removed. Nothing takes its place: the page owes the reader no statement on a subject it has collected nothing about.

The rewrite of Melbourne to Haneda then introduced an error of its own. It said the seat is least exposed to the larger of the two steps at its band edges, and that such a position is unusual on this site. Both halves fail against the rows: the nearer of the two boundaries is the band floor, and four of the six other city pairs published here sit further from a band ceiling than this one does. That sentence has been replaced by the fact which settles the question — a routing can only add distance to a great-circle figure, so the ceiling is the sole boundary this seat can ever cross. No price changed on either page.

Both of those were found by comparing every published body against the database rather than against the record of what had been fixed, so the same comparison was then run across all fourteen. It returned three more claims the gate has no way to test. Melbourne to Denpasar in economy described the quantities in which reward inventory is released and ranked a few days of date flexibility above every other technique for finding a seat; no programme publishes either, and the paragraph now stops at the arithmetic and says why. Melbourne to Singapore in economy opened by calling a tie the least common outcome on this site, when a tie and a Qantas-cheaper result each occur once among the six seats a second programme prices; it now states the fact instead, which is that exactly one seat here is priced at the same number by two programmes. And Melbourne to Hong Kong in business said the scoring sends most rows to the cheaper Velocity figure, while the table immediately beneath it sends every row to the Qantas price, because no published bridge in these tables carries a Qantas balance into Velocity. Five pages changed today and no price moved on any of them.

8 September 2026 — the ranking counted bonus points that arrive a year late

Card rows store a sign-up bonus as the advertised total across every instalment, with the minimum spend that unlocks only the first one stored beside it. The engine added the whole total to twelve months of spending earn, and the strategies pages rank on that figure. Four cards are paid in two instalments, so the first-year position printed for each of them was overstated: the Westpac Altitude Rewards Black by 100,000 points, the St.George Amplify Signature by 60,000, the ANZ Frequent Flyer Black by 40,000 and the NAB Qantas Rewards Signature by 30,000. The ordering of the ranking moved with them.

The same total was also being spent inside the months-to-a-seat arithmetic at the moment the minimum spend cleared, which returned some seats sooner than the points to book them exist. Both now count only the instalment the stated spend actually pays, and the deferred instalment is named beside the answer with the date it lands rather than folded into it. The advertised total is still printed, in its own column, because it is what the issuer advertises.

Four further cards — the American Express Platinum and Qantas Ultimate, the Qantas Premier Titanium and the Westpac Altitude Qantas Black — have collected bonus figures, but whether each is paid in one instalment or two has not been read off the issuer’s page. No first-year position is computed for them until it has. Assuming a single instalment would have been the same defect in the opposite direction, and it would have been the flattering one. All four were read off the issuers the same morning; the entry below records what each turned out to be.

8 September 2026 — the four unread splits, and a bonus the issuer does not publish

The four cards held out of the first-year ranking that morning were read off the issuers’ own pages the same day. The Qantas Premier Titanium pays in a single instalment: footnote 1 awards the whole 150,000 on $5,000 of eligible purchases within 90 days of approval, and the issuer states the points are ordinarily credited within two months of the spend criteria being met. That footnote sits behind a disclosure the 7 September attempt could not open, and no reading was taken from the headline’s silence while it stayed shut.

The Westpac Altitude Qantas Black pays 90,000 of its advertised 150,000 in year one. The remaining 60,000 is a second hurdle rather than a delayed payment: it requires a further $6,000 of eligible purchases within 90 days of the year-two card anniversary. The Qantas American Express Ultimate was stored at 70,000, which turned out to be the first instalment sitting in the field that holds the advertised total. American Express advertises 120,000 across two instalments, of which 70,000 is paid on the stated spend and 50,000 requires a further $5,000 in the first three months of the second year. The advertised total on that row was corrected upward and the figure the ranking uses was left where the issuer puts it.

The American Express Platinum Card turned out to have no published sign-up offer at all. Its product page and its acquisition page both state annual fee, card type and cash flow period and stop there, while the same template on sibling American Express cards leads with the bonus. The 300,000-point figure this site carried traced only to an aggregator, and the row also held an offer end date of 3 November 2026, which is the published end date of the American Express Explorer offer rather than this card’s. Both have been withdrawn to an unknown rather than restated, and the card prints no sign-up figure and counts none. Seven of the eight cards now carry a first-year position, and every card row on the site is sourced to the issuer that publishes it.

A gate rule was added the same morning for the class of claim that produced the three corrections above: a superlative whose sentence names no set for it to range over. Every numeral in those three sentences was correct, which is why the numeric check passed them and a person reading the rendered page found them. The rule was proved by putting one of the retracted claims back into a published template, watching the gate fail on it, and removing it again.

POINTS HERALD · COMPILED IN MELBOURNE, AUSTRALIA