What the outcome can and cannot establish
Costco continued to expand the physical and operating system described before the December 1998 decision cutoff. It reported 345 warehouses and $1.447549 billion of property additions in fiscal 2001; the later record reaches 890 warehouses in fiscal 2024. That is an observed action and outcome path. It is not evidence that an actual December 1998 board considered or adopted Part A's hypothetical gated recommendation, and the public record does not disclose the relevant minutes, rejected alternatives, release gates, or site-by-site authorization process. [evidence.costco.cutoff.q1-expansion-footprint; evidence.costco.outcome.2001.footprint; evidence.costco.outcome.2001.cash-capex; evidence.costco.outcome.2024.footprint-weeks]
The endpoint is a large business success on the reported dimensions below, but it is not a one-variable experiment. The filings document merchandise sales, membership fees, merchandise gross margin, SG&A, operating income, cash flow, member and card counts, warehouses, and channel expansion as different objects. They do not identify what portion of the outcome was caused by membership fees, low prices, limited assortment, physical scale, vendor terms, inventory discipline, employees, ancillary businesses, geography, e-commerce, acquisitions, macro conditions, or any one management decision. [evidence.costco.cutoff.reported-results; evidence.costco.cutoff.gross-margin-1998; evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2024.results; evidence.costco.outcome.2024.digital]
Reported financial path
The first table preserves the $000 unit used in the early filings. The fiscal 2001 filing described the $660.016 million line as “membership fees and other revenue”; the fiscal 2005 ten-year table labeled that same historical amount “membership fees.” The label change is disclosed here rather than silently normalized. Gross margin is merchandise net sales less merchandise costs, while operating income is after the separately reported operating-expense structure. Membership fees therefore cannot be equated with gross margin, operating income, net income, or “profit.” [evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2001.operating-income; evidence.costco.outcome.2005.revenue; evidence.costco.outcome.2005.profit]
| Company-reported measure ($000) | FY2001 | FY2005 |
|---|---|---|
| Net sales | 34,137,021 | 51,862,072 |
| Membership fees / membership fees and other revenue | 660,016 | 1,073,156 |
| Gross margin | 3,538,881 | 5,515,111 |
| Operating income | 992,267 | 1,474,303 |
| Net income | 602,089 | 1,063,092 |
| Operating cash flow | 1,032,563 | 1,783,177 |
| Property additions | 1,447,549 | 995,431 |
Source lineage: [evidence.costco.outcome.2001.weeks-sales; evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2001.operating-income; evidence.costco.outcome.2001.cash-capex; evidence.costco.outcome.2005.revenue; evidence.costco.outcome.2005.renewal-margin; evidence.costco.outcome.2005.profit; evidence.costco.outcome.2005.net-income; evidence.costco.outcome.2005.cash-capex].
Fiscal 2001 had 52 weeks against 53 weeks in fiscal 2000. Costco attributed the fee-line increase partly to an approximately $5 increase beginning with October 2000 renewals and partly to membership sign-ups at 32 new warehouses. It separately said the Executive 2% reward reduced gross-margin percentage, offset in part by core merchandising and ancillary improvement. The filing thus supplies co-causes and an accounting separation, not proof that fee revenue flowed dollar-for-dollar to profit. [evidence.costco.outcome.2001.weeks-sales; evidence.costco.outcome.2001.fees-margin]
By fiscal 2005 Costco reported 433 warehouses outside the separately described Mexico joint-venture perimeter. It reported approximately 5.0 million primary Business memberships, 16.2 million Gold Star memberships, and about 4.2 million Executive upgrades, as well as 86% renewal without a geography in the captured span. Management attributed a nine-basis-point decline in gross-margin percentage to merchandise mix and increased Executive-program costs. Renewal, category counts, fee revenue, mix, and reward cost are distinct signals. [evidence.costco.outcome.2005.footprint; evidence.costco.outcome.2005.memberships; evidence.costco.outcome.2005.renewal-margin]
The later table preserves the filings' $ millions unit and as-reported period/perimeter. No inflation, constant-currency, acquisition, fiscal-week, or same-store adjustment has been inserted.
| Company-reported measure ($m) | FY2010 | FY2015 | FY2020 | FY2024 |
|---|---|---|---|---|
| Net sales | 76,255 | 113,666 | 163,220 | 249,625 |
| Membership fees | 1,691 | 2,533 | 3,541 | 4,828 |
| Gross margin | 8,260 | 12,601 | 18,281 | 27,267 |
| Operating income | 2,077 | 3,624 | 5,435 | 9,285 |
| Net income attributable to Costco | 1,303 | 2,377 | 4,002 | 7,367 |
| Operating cash flow | 2,780 | 4,285 | 8,861 | 11,339 |
| Property additions | 1,055 | 2,393 | 2,810 | 4,710 |
Source lineage: [evidence.costco.outcome.2010.results; evidence.costco.outcome.2010.cash-capex; evidence.costco.outcome.2015.results; evidence.costco.outcome.2015.gross-margin; evidence.costco.outcome.2015.cash-capex; evidence.costco.outcome.2020.results; evidence.costco.outcome.2020.gross-margin; evidence.costco.outcome.2020.cash-capex; evidence.costco.outcome.2020.capex; evidence.costco.outcome.2024.results; evidence.costco.outcome.2024.gross-margin; evidence.costco.outcome.2024.net-income; evidence.costco.outcome.2024.cash; evidence.costco.outcome.2024.capex].
These rows show that every selected nominal as-reported line was larger at the later observation dates. They do not supply site-level returns, capitalized lease commitments, a controlled fee elasticity, or a deterministic return-on-capital bridge. Property additions and operating cash flow are separate reported cash-flow measures; neither proves the marginal warehouse earned an acceptable return. The case therefore does not promote a target price, inventory-turn estimate, cash-conversion claim, or return metric. [evidence.costco.outcome.2010.cash-capex; evidence.costco.outcome.2015.cash-capex; evidence.costco.outcome.2020.cash-capex; evidence.costco.outcome.2024.cash; evidence.costco.outcome.2024.capex]
Footprint, member definitions, and renewal
The reported warehouse path was 345 in fiscal 2001, 433 in fiscal 2005, 540 in fiscal 2010, 686 in fiscal 2015, 795 in fiscal 2020, and 890 in fiscal 2024. The perimeter is not perfectly static: the 2010 disclosure put 32 Mexico locations in a separate 50%-owned joint venture, while later worldwide counts reflect a different consolidated perimeter. Geographic growth increases capacity and addressable demand, but aggregate warehouse counts do not reveal new-site cohorts, relocations, cannibalization, closures, or site-level economics. [evidence.costco.outcome.2001.footprint; evidence.costco.outcome.2005.footprint; evidence.costco.outcome.2010.footprint-channel; evidence.costco.outcome.2015.footprint; evidence.costco.outcome.2020.footprint; evidence.costco.outcome.2024.footprint-weeks]
The membership series changes vocabulary and scope:
| Fiscal observation | Exact reported population |
|---|---|
| 2001 | Approximately 4.4m Business memberships and 12.7m Gold Star memberships |
| 2005 | Approximately 5.0m primary Business memberships and 16.2m Gold Star memberships, including about 4.2m Executive upgrades |
| 2010 | 31.6m primary cardholders, 26.4m additional cardholders, and 58.0m total cardholders, excluding about 2.9m Mexico cardholders |
| 2015 | 44.6m total paid members, 36.7m household cards, and 81.3m total cardholders |
| 2020 | 58.1m total paid members, 47.4m household cards, and 105.5m total cardholders, after the methodology change described below |
| 2024 | 76.2m total paid members, 60.6m household cards, and 136.8m total cardholders |
Source lineage: [evidence.costco.outcome.2001.memberships; evidence.costco.outcome.2005.memberships; evidence.costco.outcome.2010.members; evidence.costco.outcome.2015.members; evidence.costco.outcome.2020.member-method; evidence.costco.outcome.2024.members].
Memberships, primary cardholders, paid members, household cards, affiliates, additional cardholders, and total cardholders are not interchangeable. In 2020 Costco standardized its global counting method to the U.S./Canada method, adding approximately 2.0 million total cardholders, including 1.3 million paid members, to the 2020 count. It did not apply that change to 2019 or 2018 and said fee income and renewal calculations were unaffected. A trend that silently splices the old and new definitions would manufacture precision. [evidence.costco.outcome.2020.member-method]
Renewal also needs its reported definition. Costco reported 88% U.S./Canada renewal in 2010. In 2015 it reported approximately 91% in the U.S./Canada and 88% worldwide, using a trailing calculation covering renewals seven to eighteen months before the reporting date; in 2020 it again reported 91% and 88% on that trailing basis. In 2024 the figures were 92.9% and 90.5%, with Business affiliates excluded and membership counts including active memberships plus memberships not renewed within the prior twelve months. These percentages are useful issuer-reported retention signals, but they are not identical to annual churn or a cohort survival model. [evidence.costco.outcome.2010.renewal; evidence.costco.outcome.2015.members; evidence.costco.outcome.2020.renewal-executive; evidence.costco.outcome.2024.members]
What evolved inside the operating system
The strongest interpretation is an integrated member-value system, not a standalone fee engine. At the cutoff Costco described low prices, limited branded and private-label selection, high volume, efficient distribution, reduced handling, rapid-turn aspirations, vendor financing, and no-frills warehouses. It reported roughly 3,600-4,400 active SKUs and screened products for meaningful member savings. Those are management's mechanism claims. The frozen record does not contain a governed “treasure-hunt” cadence or economics, so the familiar label is not used as a causal metric. [evidence.costco.cutoff.operating-model; evidence.costco.cutoff.vendor-working-capital; evidence.costco.cutoff.sku-value-policy]
Vendor terms and inventory discipline are likewise mechanisms to test, not ratios to improvise. The cutoff filing said mature warehouses could sell a substantial portion of inventory before paying vendors while taking early-payment discounts, but the episode does not contain consistent average inventory, payable aging, receipts, shrink, and stable-perimeter merchandise-cost data across the horizon. No inventory-turn or payable-days value is derived in prose. [evidence.costco.cutoff.vendor-working-capital; evidence.costco.cutoff.balance-1998; evidence.costco.cutoff.q1-balance]
Physical scale and geography remained material. The warehouse sequence and expanding country footprint show repeated investment in capacity, while the 2010 filing still identified Walmart, Target, Kohl's, and Amazon as significant general-merchandise competitors. Scale could improve purchasing, distribution, brand awareness, and convenience; it could also conceal weak marginal cohorts or cannibalization. Competition and generic retail execution remain co-causes and falsifiers of a fees-only story. [evidence.costco.outcome.2010.footprint-channel; evidence.costco.outcome.2010.competition; evidence.costco.outcome.2020.footprint; evidence.costco.outcome.2024.footprint-weeks]
Digital developed from an option into a material complement. In 2010 Costco described U.S. and Canadian e-commerce businesses offering additional products and services. Online sales were approximately 3% of net sales in fiscal 2015. Fiscal 2020 e-commerce comparable sales increased 50%, including 80% in the second half, during the pandemic. By fiscal 2024 e-commerce was approximately 7% of net sales and the broader “digitally originated” definition—including e-commerce, business delivery, travel, and same-day grocery—was approximately 9%; comparable e-commerce sales increased 16% on comparable retail weeks. Revenue share, digitally originated share, and comparable growth are different measures and reveal neither incremental demand nor fulfillment margin by themselves. [evidence.costco.outcome.2010.ecommerce; evidence.costco.outcome.2015.channel-footprint; evidence.costco.outcome.2020.ecommerce; evidence.costco.outcome.2024.digital; evidence.costco.outcome.2024.weeks-comps]
Employees were also part of the expanding operating perimeter. Costco reported approximately 63,000 employees in 1998. By 2024 it reported 333,000 worldwide, approximately 95% in warehouses and distribution and approximately 5% represented by unions; management described compensation and benefits as its largest expense after merchandise cost. These facts establish scale and management emphasis. They do not establish that pay, benefits, tenure, turnover, training, promotion, safety, scheduling, or union relations independently caused the financial result; that requires comparable operating and employee panels. [evidence.costco.cutoff.employees-1998; evidence.costco.outcome.2024.workforce; evidence.costco.outcome.2024.compensation-claim]
The issuer's history also contains structural breaks that should not be coded as organic performance. The 1993 Price/Costco merger was accounted for as a pooling of interests that retrospectively combined the companies and conformed policies and interim periods. Shareholders later approved the Price/Costco, Inc. to Costco Companies, Inc. name change; that legal-name and ticker transition is not itself an economic gain. In fiscal 2020 Costco acquired a 35% interest in Navitus and acquired Innovel's final-mile capabilities. Merger accounting, legal-name presentation, acquisitions, consolidation perimeter, and organic operations must stay separate. [evidence.costco.cutoff.merger-accounting; evidence.costco.cutoff.name-change-1997; evidence.costco.outcome.2020.acquisitions]
Fiscal 2020 is not a steady-state test
Fiscal 2020 combines an operating result with a global membership-count methodology change, pandemic demand and channel shifts, $564 million of incremental wage and sanitation cost, and the Navitus and Innovel transactions. Its 50% e-commerce comparable-sales increase occurred in that setting. Those facts do not negate the long-run system hypothesis, but they prevent fiscal 2020 from serving as a clean controlled observation of membership economics or warehouse expansion. [evidence.costco.outcome.2020.member-method; evidence.costco.outcome.2020.ecommerce; evidence.costco.outcome.2020.covid-cost; evidence.costco.outcome.2020.acquisitions]
The 2024 fee event
On July 10, 2024, Costco announced that U.S./Canada Gold Star, Business, and Business add-on fees would rise by $5 to $65 effective September 1. It said the Executive total fee would rise from $120 to $130 and the maximum annual 2% reward from $1,000 to $1,250, affecting around 52 million memberships, a little over half Executive. Reuters contemporaneously described the action as the first increase in seven years. [evidence.costco.outcome.2024.fee-release; evidence.costco.outcome.2024.reuters-headline]
The announcement arrived with strong five-week operating context: Costco reported $24.48 billion of June net sales, up 7.4%, and e-commerce comparable sales of 18.4%, or 19.1% excluding gasoline-price and foreign-exchange effects. These are issuer-reported five-week observations, not annual economics or evidence that the fee action caused sales. [evidence.costco.outcome.2024.june-sales]
At fiscal 2024 year-end Costco reported 35.4 million Executive members, and Executive-member sales penetration of approximately 73.3% of worldwide net sales. The fee action became effective September 1, 2024—the fiscal year-end date. This episode therefore contains the decision and exposed population but no post-effective-date realized retention, spending, reward-cost, fee-revenue, or profit outcome. Fiscal 2024 also had 52 weeks versus 53 in fiscal 2023, and the filing used comparable retail weeks for its 2024 comparable-sales measure. [evidence.costco.outcome.2024.executive-fee; evidence.costco.outcome.2024.executive-sales; evidence.costco.outcome.2024.footprint-weeks; evidence.costco.outcome.2024.weeks-comps]
Competing causal explanations
Primary hypothesis — integrated member value. Low prices and limited assortment create member surplus and volume; scale purchasing, distribution, vendor terms, inventory discipline, and employee execution support the price promise; fees and Executive rewards monetize and reinforce engagement; warehouses, geography, ancillary services, and digital access extend the system. The repeated coexistence of these mechanisms and the later reported scale is consistent with contribution. It does not isolate causality. [evidence.costco.cutoff.operating-model; evidence.costco.cutoff.vendor-working-capital; evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2015.executive; evidence.costco.outcome.2024.digital; evidence.costco.outcome.2024.executive-sales]
Rival hypothesis — footprint, geography, and retail execution. Much of the result may instead be explained by openings, geographic white space, locations, merchandise and ancillary mix, gasoline, private label, real estate, and generic purchasing and distribution scale. The warehouse path and management's own margin explanations make this both a rival and a complement to membership economics. It would gain force if matched market and site evidence showed that these variables dominate after controlling for member characteristics. [evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2005.renewal-margin; evidence.costco.outcome.2010.footprint-channel; evidence.costco.outcome.2024.footprint-weeks]
Rival hypothesis — later channel, shocks, acquisitions, and execution. E-commerce, logistics, acquisitions, later management and technology, currency, gasoline, inflation, labor markets, and the pandemic accumulated after 1998. Fiscal 2020 gives direct evidence of several such breaks. This hypothesis would gain force if a controlled decomposition found that the later variables explain most of the endpoint after holding the legacy member system and footprint constant. [evidence.costco.outcome.2020.ecommerce; evidence.costco.outcome.2020.covid-cost; evidence.costco.outcome.2020.acquisitions; evidence.costco.outcome.2024.digital]
The evidence cannot choose cleanly among these explanations. A credible causal test would need stable-definition member cohorts linked to baskets and renewal, matched price gaps, SKU and vendor panels, site cash flows and capital, employee outcomes, channel economics, geography, acquisitions, and macro controls. The conclusion is contribution by an integrated system with material co-causes, not attribution to membership fees or to the hypothetical 1998 recommendation.
Counterfactual review
Gated membership-first scale. Part A recommended funding already-committed openings and relocations while releasing follow-on warehouse and digital capital against same-definition cohort renewal, member-value, unit-economics, inventory/vendor, employee, and downside-liquidity gates. The observed path is compatible with continued investment, but public filings do not reveal whether those gates existed or whether staging would have improved or harmed results. [evidence.costco.cutoff.expansion-1998; evidence.costco.cutoff.q1-expansion-footprint; evidence.costco.outcome.2001.footprint]
Faster fee or margin harvesting. A company could raise fees and merchandise margins faster, broaden assortment, or compress operating and employee cost. The 2001 and 2005 filings show why this cannot be scored from fee revenue alone: fee prices and new sign-ups affected the fee line, while Executive rewards and merchandise mix separately affected gross margin. A valid test would require controlled member surplus, renewal, visits, spending, reward cost, and lifetime value. [evidence.costco.outcome.2001.fees-margin; evidence.costco.outcome.2005.renewal-margin]
Slower, asset-light, or digital-led growth. A company could slow owned openings while testing relocations, joint ventures, ancillary partnerships, and digital fulfillment. The record demonstrates a Mexico joint-venture perimeter in 2010 and later e-commerce scale, but it provides no matched return comparison between owned warehouses, joint ventures, partnerships, and digital delivery. Slower density might preserve capital or might weaken convenience, purchasing scale, and vendor economics. [evidence.costco.outcome.2010.footprint-channel; evidence.costco.outcome.2010.ecommerce; evidence.costco.outcome.2024.digital]
Transferable rules, false positives, and boundary conditions
- Preserve member surplus before fee or margin harvest. Use the rule only where customers voluntarily pay for measurable price, quality, convenience, or service value. High fee revenue, high renewal, or a long delay between increases can be false positives if definitions, grace periods, switching friction, overlapping memberships, or underinvestment obscure deterioration. The counterexample boundary is a format facing technological obsolescence: generous legacy economics cannot by itself repair a structurally obsolete channel.
- Gate repeated physical growth on stable-scope unit economics. Aggregate sales and warehouse growth can hide weak marginal cohorts, cannibalization, preopening cost, working-capital strain, or employee-capacity failures. Use staged releases only where capital is divisible; scarce sites, network density, and binding commitments can make delay costly. The counterexample is scale pursued without sufficiently proven unit economics, not an inference that every large footprint is bad.
- Preserve definitions before trends. Store the exact membership/card label, renewal method, fiscal weeks, geography, consolidation perimeter, acquisition status, currency and gasoline treatment, channel definition, and public date. Branch or abstain when there is no deterministic bridge. This measurement rule prevents false precision; it does not itself explain performance. [evidence.costco.cutoff.merger-membership-overlap; evidence.costco.outcome.2001.weeks-sales; evidence.costco.outcome.2010.members; evidence.costco.outcome.2020.member-method; evidence.costco.outcome.2024.members; evidence.costco.outcome.2024.footprint-weeks]
Final assessment
Costco's later reported financial, warehouse, member, employee, and digital scale is consistent with a durable and adaptive business system. The most defensible learning is architectural: recurring access revenue can reinforce a low-price operating model when the company continually funds the merchandise, vendor, inventory, distribution, employee, physical, and digital capabilities that create member value. Membership fees are neither the merchandise gross margin nor a standalone measure of profit.
That lesson remains conditional. The record does not grade the actual 1998 decision process, prove every warehouse was optimal, bridge every member definition, normalize every fiscal period and acquisition, or isolate causality. The 2024 fee action ends at effectiveness, before any realized test. A financial-analysis agent should retain the exact source definitions and two clocks, keep company attributions separate from analyst judgment, require deterministic calculations for derived metrics, and abstain from valuation or causal precision when capitalization, unit-economics, conflict, or evidence inputs are incomplete.