Update, August 3, 2026: the section below that admitted our own Costco dates were wrong no longer describes the present. Every Costco period end this site stores has since been restamped onto the filed fiscal calendar — twenty-eight of forty-seven quarters moved, from the filed 10-K/10-Q report dates where the filing index reaches and from the 52/53-week calendar those filings confirm where it does not. No stored value moved; this was a date defect, not a data one. Two numbers below were computed from the old stamps and are corrected: Costco's spread between its longest and shortest quarter now reads 28 days in the sweep, not 35, and its shortest quarter reads 84 days, not 77 — the figures that section already gave as the filed truth. The section is kept, in the past tense, because the sweep it qualifies was run on the old stamps.
This is the first article this site has published about Costco. It has appeared here once before, in a single line of our February Walmart preview, named as the only comparable trading at Walmart's multiple.
Its most recently completed fiscal year ended with revenue going from $63,205M to $86,156M — a 36.3% sequential jump.
That is not even the biggest one Costco has produced. Its four largest sequential revenue increases are +47.1%, +46.6%, +43.2% and +38.4%, and every one of them lands on the same quarter of the fiscal year — the fourth, the long one, which ends in the last days of August or, in a 53-week year, the first days of September. The two largest, in fiscal 2023 and fiscal 2017, are exactly those 53-week years, where that quarter runs seventeen weeks rather than sixteen — fiscal 2023 confirmed below from its filed dates, fiscal 2017 from the same 52/53-week calendar projected back, since the filing index in this repo does not reach 2017.
Costco did not sell 36% more. That quarter is four weeks longer than the one before it.
The calendar
Costco is a 52/53-week filer. Three of its quarters run twelve weeks; the fourth runs sixteen. From the filings — SEC XBRL, CIK 0000909832, quarterly contexts as filed:
| FY2025 period | window | length | revenue |
|---|---|---|---|
| Q1 | 2024-09-02 → 2024-11-24 | 84d — 12 weeks | $62,151M |
| Q2 | 2024-11-25 → 2025-02-16 | 84d — 12 weeks | $63,723M |
| Q3 | 2025-02-17 → 2025-05-11 | 84d — 12 weeks | $63,205M |
| Q4 | 2025-05-12 → 2025-08-31 | 112d — 16 weeks | $86,156M |
| full year | 2024-09-02 → 2025-08-31 | 364d — 52 weeks | $275,235M |
Day counts are inclusive of both endpoints, which is why they land on exact multiples of seven: 84 days is twelve weeks, and 112 is sixteen.
The fourth-quarter figure is derived — a fiscal Q4 has no quarterly XBRL context of its own, so it is the annual less the three filed quarters: $275,235M − $189,079M = $86,156M. The reason to trust that derivation is that it reproduces the value already stored in our Costco data for the quarter, to the dollar.
Per week of trading
| revenue | weeks | per week | |
|---|---|---|---|
| FY2025 Q3 | $63,205M | 12 | $5,267M |
| FY2025 Q4 | $86,156M | 16 | $5,385M |
+2.2%.
Thirty-four of the thirty-six points are the calendar. Nothing in the series, the chart, or any sequential comparison drawn from them says that the two bars measure different lengths of time.
This is not the argument that the Rule of 40 does not suit retailers. Our own explainer already says the metric was built for high-gross-margin software, and our Walmart coverage has made the point. This one survives even if you accept the metric for retail, and it applies to any sequential comparison at all: revenue growth, margin trends, a cash-flow line, anything where two consecutive Costco quarters are set beside each other. One of them is a third longer.
It is also not the question of which window to measure over, which we looked at when Netflix's quarterly score moved 32 points over a window in which its trailing-twelve-month score moved three. That is a choice between two well-defined windows. This is a window that is not a fixed size.
It also has a layer underneath it. FY2023 was a 53-week year — 2022-08-29 to 2023-09-03, 371 days — so that year's fourth quarter was seventeen weeks, not sixteen. Any Costco Q4-against-Q4 comparison that crosses it is comparing sixteen weeks with seventeen, which is the same trap one level down.
It is the only ticker on this site with the problem
Costco is not an example of a class. Sweeping every ticker that carries at least nine stored revenue period ends — 63 of them — and measuring the spread between its longest and shortest consecutive period over its last nine period ends:
| ticker | spread | shortest | longest |
|---|---|---|---|
| COST | 28d | 84d | 112d |
| KO | 9d | 87d | 96d |
| HD | 7d | 91d | 98d |
| QCOM | 7d | 91d | 98d |
Exactly one ticker exceeds 20 days, and it does so by a factor of three over the runner-up. Every other tracked company reports in windows that vary by nine days or fewer — close enough that a sequential comparison means what it appears to mean.
Costco's longest here is 112 days, not the 119 of a seventeen-week fourth quarter: fiscal 2017 and fiscal 2023, the 53-week years named above, both predate this nine-period window.
This table read 35d/77d for COST when this piece was published, on the stamps the next section describes. It now reads the filed 28d/84d. No other row moved.
Our own dates were wrong here, and it mattered for this piece specifically
The sweep above was run on the period ends this site stores, and when this was published several of Costco's were wrong. They have since been repaired; what follows is the record of what they were, because it is what the sweep ran on.
Its first fiscal quarter was stamped at calendar month-end rather than at the filed period end: 2024-11-30 against a filed 2024-11-24, and 2025-11-30 against a filed 2025-11-23. The same month-end convention recurred across the ticker's stored history — the fiscal-year end stamped 08-31 regardless of the filed date, and several Novembers stamped 11-30 — so those two were the recent face of a longer-running pattern rather than isolated slips. Twenty-eight of the forty-seven stored quarters were affected; the widest miss, 2017-05-31 against a period ending 2017-05-07 — projected from the fiscal calendar, since the filing index does not reach 2017 — was out by twenty-four days. The revenue values were correct throughout — a date defect, not a data one, which is why nothing above except the two lengths named here was touched by the repair.
It was worth naming here rather than in a footnote, because it contaminated the exact quantity this article is about. Computed from the old stamps, Costco's November quarter read 91 days against a filed window of 84. Its spread between longest and shortest quarter read 35 days rather than the filed 28 (84 to 112) — the 77-day minimum was itself an artefact of a stamp seven days late. The two late Novembers differed by a day and it is worth being exact: the November 2024 stamp was six days late and produced a 78-day gap, the November 2025 stamp seven days late and produced the 77.
One thing the defect never did was print a wrong date at a reader. The "Period ended" line on our Costco earnings pages reads from the press-release record, not from the series, and it said November 23 throughout. The stored stamp was a join key — which is why it survived: it sat inside every tolerance downstream of it, including the seven-day window the earnings pages use to match a release to a series point, where the November 2025 stamp was sitting exactly on the boundary.
Every length quoted in this piece came from the filings for that reason. An article whose argument is that nobody notices these periods are different sizes could not take its sizes from the field that had them wrong. It now agrees with that field.
What to watch
Costco's next fourth quarter will be sixteen weeks again, and it will again produce a sequential jump of roughly a third that means roughly nothing. Our schedule marks the print as estimated for late September, so treat the date as provisional.
The useful number when it lands is revenue per week against $5,385M. That comparison is stable across the seam; the headline one is not.
The four largest sequential revenue increases, and every day-count computed from stored stamps, are from our Costco data; the fiscal-2017 week count is projected from the 52/53-week calendar rather than read off a filing, because our filing index does not reach 2017. Costco period windows, lengths and revenue are from SEC XBRL, CIK 0000909832, RevenueFromContractWithCustomerExcludingAssessedTax, using the periods as filed; day counts are inclusive of both endpoints. The fourth-quarter figure is derived as the fiscal year less the three filed quarters and reproduces the value stored in our Costco data exactly. The corpus sweep covers the 63 companies we track carrying at least nine revenue period ends, measured over the last nine, and runs on stored period ends; it read a 35-day spread for Costco when published, against a filed 28, and reads 28 since the repair described above. No other ticker's row moved. No price, valuation or share-move figure appears above, and the September print date is the estimate carried in our schedule rather than a company confirmation.